From 6f9f6c63e7c637f87fd37bcd34aec29b2b1b3ddb Mon Sep 17 00:00:00 2001 From: bailian-bot Date: Wed, 15 Jul 2026 14:04:30 +0800 Subject: [PATCH 01/13] chore: update bailian-docs-llm-wiki (2026-07-15) --- skills/bailian-docs-llm-wiki/SKILL.md | 28 - skills/bailian-docs-llm-wiki/llms.txt | 963 +++--- .../models/families.jsonl | 160 +- .../models/groups/Kimi-K2.json | 498 --- .../models/groups/MiniMax-M2.1.json | 178 - .../groups/MiniMax-speech-market-place.json | 222 -- .../models/groups/aitryon-parsing-v1.json | 65 - .../models/groups/aitryon-plus.json | 54 - .../models/groups/aitryon-refiner.json | 52 - .../models/groups/aitryon.json | 54 - .../groups/animate-anyone-detect-gen2.json | 63 - .../models/groups/animate-anyone-gen2.json | 65 - .../groups/animate-anyone-template-gen2.json | 65 - .../models/groups/cosyvoice.json | 432 --- .../models/groups/deepseek.json | 554 +--- .../models/groups/embedding.json | 246 -- .../models/groups/emo-detect-v1.json | 63 - .../models/groups/emo-v1.json | 65 - .../models/groups/emoji-detect-v1.json | 63 - .../models/groups/emoji-v1.json | 65 - .../models/groups/facechain-facedetect.json | 56 - .../models/groups/facechain-generation.json | 58 - .../models/groups/farui-plus.json | 99 - .../models/groups/fun-asr-flash.json | 78 - .../models/groups/fun-asr-realtime.json | 144 - .../models/groups/fun-asr.json | 96 - .../models/groups/fun-music.json | 151 - .../models/groups/glm-4.5.json | 836 ----- .../models/groups/glm-fast.json | 76 - .../models/groups/gui-plus.json | 98 - .../models/groups/gummy-chat-v1.json | 60 - .../models/groups/gummy-realtime-v1.json | 60 - .../models/groups/happyhorse-i2v.json | 82 +- .../models/groups/happyhorse-r2v.json | 178 - .../models/groups/happyhorse-t2v.json | 102 +- .../models/groups/happyhorse-video-edit.json | 88 - .../models/groups/image-erase-completion.json | 65 - .../groups/image-instance-segmentation.json | 65 - .../models/groups/image-out-painting.json | 65 - .../groups/kimi-models-market-place.json | 352 -- .../groups/kling-models-market-place.json | 297 -- .../models/groups/liveportrait-detect.json | 63 - .../models/groups/liveportrait.json | 65 - .../groups/minimax-models-market-place.json | 357 -- .../models/groups/paraformer-8k-v1.json | 54 - .../models/groups/paraformer-8k-v2.json | 66 - .../models/groups/paraformer-mtl-v1.json | 54 - .../groups/paraformer-realtime-8k-v1.json | 52 - .../groups/paraformer-realtime-8k-v2.json | 42 - .../models/groups/paraformer-realtime-v1.json | 54 - .../models/groups/paraformer-realtime-v2.json | 48 - .../models/groups/paraformer-v1.json | 54 - .../models/groups/paraformer-v2.json | 48 - .../groups/pixverse-c1-market-place.json | 335 -- .../models/groups/pixverse-market-place.json | 317 -- .../groups/pixverse-v6-market-place.json | 345 -- .../models/groups/qvq-max.json | 108 - .../models/groups/qvq-plus.json | 110 - .../groups/qwen-audio-realtime-flash.json | 116 - .../groups/qwen-audio-realtime-plus.json | 116 - .../models/groups/qwen-audio-tts.json | 164 - .../models/groups/qwen-coder-plus.json | 109 - .../models/groups/qwen-coder-turbo.json | 109 - .../models/groups/qwen-deep-research.json | 97 - .../models/groups/qwen-doc-turbo.json | 99 - .../models/groups/qwen-embedding.json | 416 --- .../models/groups/qwen-flash-character.json | 101 - .../models/groups/qwen-flash.json | 139 - .../models/groups/qwen-image-2.0-pro.json | 83 - .../models/groups/qwen-image-2.0.json | 83 - .../models/groups/qwen-image-edit-max.json | 62 - .../models/groups/qwen-image-edit.json | 117 - .../models/groups/qwen-image-max.json | 78 - .../models/groups/qwen-image-plus.json | 156 - .../models/groups/qwen-long.json | 192 -- .../models/groups/qwen-math-plus.json | 380 --- .../models/groups/qwen-math-turbo.json | 109 - .../models/groups/qwen-max.json | 117 - .../models/groups/qwen-mt-flash.json | 79 - .../models/groups/qwen-mt-image.json | 81 - .../models/groups/qwen-mt-lite.json | 78 - .../models/groups/qwen-mt-plus.json | 78 - .../models/groups/qwen-mt-turbo.json | 78 - .../groups/qwen-omni-turbo-realtime.json | 178 - .../models/groups/qwen-omni-turbo.json | 185 -- .../models/groups/qwen-plus-character.json | 102 - .../models/groups/qwen-plus.json | 459 --- .../models/groups/qwen-rerank.json | 246 -- .../models/groups/qwen-tts-realtime.json | 180 -- .../models/groups/qwen-tts.json | 128 - .../models/groups/qwen-turbo.json | 143 - .../models/groups/qwen-vl-embedding.json | 144 - .../models/groups/qwen-vl-max.json | 117 - .../models/groups/qwen-vl-ocr.json | 286 -- .../models/groups/qwen-vl-plus.json | 124 - .../models/groups/qwen-voice-design.json | 67 - .../models/groups/qwen-voice-enrollment.json | 60 - .../models/groups/qwen2.5.json | 95 - .../groups/qwen3-asr-flash-filetrans.json | 80 - .../groups/qwen3-asr-flash-realtime.json | 78 - .../models/groups/qwen3-asr-flash.json | 54 - .../groups/qwen3-coder-30b-a3b-instruct.json | 106 - .../qwen3-coder-480b-a35b-instruct.json | 106 - .../models/groups/qwen3-coder-flash.json | 108 - .../models/groups/qwen3-coder-plus.json | 114 - .../qwen3-livetranslate-flash-realtime.json | 90 - .../groups/qwen3-livetranslate-flash.json | 81 - .../models/groups/qwen3-max.json | 290 -- .../groups/qwen3-omni-30b-a3b-captioner.json | 80 - .../groups/qwen3-omni-flash-realtime.json | 102 - .../models/groups/qwen3-omni-flash.json | 107 - .../groups/qwen3-tts-flash-realtime.json | 61 - .../models/groups/qwen3-tts-flash.json | 64 - .../qwen3-tts-instruct-flash-realtime.json | 80 - .../groups/qwen3-tts-instruct-flash.json | 87 - .../models/groups/qwen3-tts-vc-realtime.json | 61 - .../models/groups/qwen3-tts-vc.json | 60 - .../models/groups/qwen3-tts-vd-realtime.json | 81 - .../models/groups/qwen3-tts-vd.json | 61 - .../models/groups/qwen3-vl-flash.json | 145 - .../models/groups/qwen3-vl-plus.json | 140 - .../models/groups/qwen3.5-flash.json | 282 +- .../qwen3.5-livetranslate-flash-realtime.json | 92 - .../models/groups/qwen3.5-ocr.json | 99 - .../groups/qwen3.5-omni-flash-realtime.json | 101 - .../models/groups/qwen3.5-omni-flash.json | 103 - .../groups/qwen3.5-omni-plus-realtime.json | 101 - .../models/groups/qwen3.5-omni-plus.json | 103 - .../models/groups/qwen3.5-plus.json | 261 +- .../models/groups/qwen3.5.json | 582 ---- .../models/groups/qwen3.6-flash.json | 200 +- .../models/groups/qwen3.6-max.json | 82 +- .../models/groups/qwen3.6-plus.json | 222 +- .../models/groups/qwen3.6.json | 132 +- .../models/groups/qwen3.7-max.json | 248 +- .../models/groups/qwen3.7-plus.json | 232 +- .../models/groups/qwen3.json | 2388 -------------- .../models/groups/qwq-plus.json | 108 - .../models/groups/sambert.json | 2586 --------------- .../models/groups/shoemodel-v1.json | 65 - .../models/groups/siliconflow-models.json | 376 --- .../models/groups/speech-biasing.json | 70 - .../groups/stepfun-models-market-place.json | 103 - .../groups/tongyi-intent-detect-v3.json | 97 - .../groups/tongyi-xiaomi-analysis-flash.json | 95 - .../groups/tongyi-xiaomi-analysis-pro.json | 95 - .../groups/tripo-models-market-place.json | 172 - .../groups/vanchin-models-market-place.json | 660 ---- .../models/groups/video-style-transform.json | 65 - .../models/groups/videoretalk.json | 66 - .../vidu-image-models-market-place.json | 378 --- .../groups/vidu-models-market-place.json | 1917 ----------- .../models/groups/virtualmodel-v2.json | 65 - .../models/groups/voice-enrollment.json | 44 - .../models/groups/wan-image-edit.json | 396 --- .../models/groups/wan-image-to-video.json | 1108 ------- .../models/groups/wan-reference-to-video.json | 306 -- .../models/groups/wan-text-to-image.json | 702 ---- .../models/groups/wan-text-to-video.json | 586 ---- .../models/groups/wan-video-edit.json | 96 - .../groups/wanx-background-generation-v2.json | 65 - .../groups/wanx-poster-generation-v1.json | 65 - .../groups/wanx-sketch-to-image-lite.json | 67 - .../models/groups/wanx-style-repaint-v1.json | 65 - .../models/groups/wanx-virtualmodel.json | 65 - .../models/groups/wanx-x-painting.json | 67 - .../models/groups/wanx2.1-vace-plus.json | 85 - .../models/groups/wordart-semantic.json | 65 - .../models/groups/wordart-texture.json | 66 - .../groups/xiaomi-models-market-place.json | 89 - .../models/groups/z-image-turbo.json | 76 - .../groups/zhipu-models-market-place.json | 264 -- .../bailian-docs-llm-wiki/models/index.json | 2863 +---------------- skills/bailian-docs-llm-wiki/models/index.md | 382 +-- .../bailian-docs-llm-wiki/models/models.jsonl | 376 +-- ...pi-contactcenterai-2024-06-03-changeset.md | 0 ...centerai-2024-06-03-analyzeconversation.md | 0 ...contactcenterai-2024-06-03-analyzeimage.md | 0 ...i-contactcenterai-2024-06-03-createtask.md | 0 ...-contactcenterai-2024-06-03-createvocab.md | 0 ...-contactcenterai-2024-06-03-deletevocab.md | 0 ...api-contactcenterai-2024-06-03-getvocab.md | 0 ...pi-contactcenterai-2024-06-03-listvocab.md | 0 ...-contactcenterai-2024-06-03-updatevocab.md | 0 ...actcenterai-2024-06-03-analyzeaudiosync.md | 0 ...centerai-2024-06-03-generalanalyzeimage.md | 0 ...ontactcenterai-2024-06-03-gettaskresult.md | 0 ...ontactcenterai-2024-06-03-runcompletion.md | 0 ...enterai-2024-06-03-runcompletionmessage.md | 0 ...api-contactcenterai-2024-06-03-endpoint.md | 0 ...api-contactcenterai-2024-06-03-overview.md | 20 +- ...-style-request-body-signature-mechanism.md | 0 .../technology-integration-scheme.md | 0 .../tongyi-xiaomi-ccai-update-announcement.md | 0 ...pi-bailianvoicebot-2025-01-01-changeset.md | 0 ...ailianvoicebot-2025-01-01-bridgewebcall.md | 0 ...anvoicebot-2025-01-01-createapplication.md | 0 ...bot-2025-01-01-createapplicationversion.md | 0 ...anvoicebot-2025-01-01-deleteapplication.md | 0 ...ilianvoicebot-2025-01-01-getapplication.md | 0 ...ianvoicebot-2025-01-01-listapplications.md | 0 ...oicebot-2025-01-01-listbackgroundmusics.md | 0 ...ailianvoicebot-2025-01-01-listnlumodels.md | 0 ...i-bailianvoicebot-2025-01-01-listvoices.md | 0 ...bailianvoicebot-2025-01-01-previewvoice.md | 0 ...ot-2025-01-01-publishapplicationversion.md | 0 ...anvoicebot-2025-01-01-updateapplication.md | 0 ...bot-2025-01-01-updateapplicationversion.md | 0 ...ianvoicebot-2025-01-01-createclonevoice.md | 0 ...ianvoicebot-2025-01-01-deleteclonevoice.md | 0 ...ilianvoicebot-2025-01-01-listclonevoice.md | 0 ...oicebot-2025-01-01-listclonevoicemodels.md | 0 ...ianvoicebot-2025-01-01-updateclonevoice.md | 0 ...ianvoicebot-2025-01-01-createvocabulary.md | 0 ...ianvoicebot-2025-01-01-deletevocabulary.md | 0 ...ianvoicebot-2025-01-01-exportvocabulary.md | 0 ...ailianvoicebot-2025-01-01-getvocabulary.md | 0 ...ianvoicebot-2025-01-01-importvocabulary.md | 0 ...ilianvoicebot-2025-01-01-listvocabulary.md | 0 ...ianvoicebot-2025-01-01-updatevocabulary.md | 0 ...voicebot-2025-01-01-disablesubscription.md | 0 ...lianvoicebot-2025-01-01-getsubscription.md | 0 ...nvoicebot-2025-01-01-updatesubscription.md | 0 ...bot-2025-01-01-createvoiceaccessprofile.md | 0 ...bot-2025-01-01-deletevoiceaccessprofile.md | 0 ...cebot-2025-01-01-listvoiceaccessprofile.md | 0 ...ianvoicebot-2025-01-01-listvoiceengines.md | 0 ...bot-2025-01-01-updatevoiceaccessprofile.md | 0 ...ilianvoicebot-2025-01-01-createvariable.md | 0 ...ilianvoicebot-2025-01-01-deletevariable.md | 0 ...bailianvoicebot-2025-01-01-listvariable.md | 0 ...ilianvoicebot-2025-01-01-updatevariable.md | 0 ...bot-2025-01-01-generatefileuploadparams.md | 0 ...bot-2025-01-01-getdatachannelcredential.md | 0 ...api-bailianvoicebot-2025-01-01-overview.md | 0 .../api-bailianvoicebot-2025-01-01-ram.md | 0 .../product-0verview.md | 0 .../api-anytrans-2025-07-07-changeset.md | 6 +- .../api-anytrans-2025-07-07-ram.md | 2 +- .../guidelines-for-use.md | 179 -- ...api-contactcenterai-2024-06-03-overview.md | 117 - ...ces-for-automatic-work-order-generation.md | 141 - ...rvice-quality-inspection-best-practices.md | 138 - .../best-practices/summary-best-practices.md | 118 - ...lysis-aio-application-through-task-type.md | 563 ---- ...-ccai-dialogue-analysis-aio-application.md | 225 -- ...analysis-by-uploading-offline-task-data.md | 197 -- ...e-prompt-to-call-tongyi-xiaomi-ccai-aio.md | 385 --- ...plate-id-to-call-tongyi-xiaomi-ccai-aio.md | 404 --- .../product-overview-1.md | 67 - .../application-management.md | 38 - ...an-application-based-on-a-custom-method.md | 60 - ...cation-based-on-dialogue-analysis-agent.md | 140 - .../hot-phrase-management.md | 48 - .../using-the-knowledge-base.md | 74 - .../operation-guide.md | 85 - .../api-bailianchatbot-2024-11-05-ssechat.md | 0 .../api-bailianchatbot-2024-11-05-overview.md | 0 .../api-bailianchatbot-2024-11-05-ram.md | 0 ...orkflow-integration-video-understanding.md | 4 +- .../api-aimiaobi-2023-08-01-createdataset.md | 0 .../api-aimiaobi-2023-08-01-deletedataset.md | 0 .../api-aimiaobi-2023-08-01-getdataset.md | 0 .../api-aimiaobi-2023-08-01-listdatasets.md | 0 .../api-aimiaobi-2023-08-01-updatedataset.md | 0 .../knowledge-base/rag-optimization.md | 42 +- .../voice-clone-design-http-api.md | 16 +- .../sound-reengraving/voice-clone-java-sdk.md | 16 +- .../voice-clone-python-sdk.md | 16 +- .../text-embedding-synchronous-api.md | 95 +- .../vidu-reference-to-video-api-reference.md | 4 + .../get-started-with-models/models.md | 8 +- .../get-started-with-models/rate-limit.md | 22 +- .../tripo-3d-generation-guide.md | 14 +- .../model-experience/tts-model.md | 10 +- .../tpm-reservation.md | 11 +- .../release-notes/model-release-notes.md | 24 +- .../model-user-guide/test-1/model-pricing.md | 8 + skills/bailian-docs-llm-wiki/raw/test/test.md | 1 + .../wiki/api/3d-generation.md | 136 +- .../wiki/api/application-call.md | 191 +- .../application-component-api-reference.md | 234 +- .../wiki/api/file-management-api.md | 54 +- .../wiki/api/frameworks.md | 168 +- .../wiki/api/image-generation.md | 73 +- .../wiki/api/knowledge.md | 66 +- .../wiki/api/long-term-memory-new.md | 256 +- .../wiki/api/managed-agents-api.md | 182 +- .../wiki/api/model-production.md | 50 +- .../wiki/api/more-about-models.md | 143 +- .../wiki/api/more-models.md | 182 +- skills/bailian-docs-llm-wiki/wiki/api/more.md | 149 +- .../wiki/api/omni-realtime-api.md | 221 +- .../wiki/api/preparations.md | 130 +- .../wiki/api/qwen-api-reference.md | 78 +- .../wiki/api/toolkits-and-frameworks.md | 174 +- .../wiki/api/vector-and-sort.md | 178 +- .../wiki/api/video-generation-api.md | 115 +- .../comparisons/api-invocation-comparison.md | 75 - .../comparisons/app-eval-vs-model-eval.md | 70 - .../app-monitoring-vs-model-monitoring.md | 70 - .../comparisons/app-vs-model-evaluation.md | 64 - .../comparisons/app-vs-model-monitoring.md | 66 - .../application-call-vs-bailian-calling.md | 45 - .../application-calling-comparison.md | 57 - .../application-evaluation-vs-monitoring.md | 45 - .../application-observation-vs-evaluation.md | 59 - .../comparisons/application-orchestration.md | 81 + .../application-vs-managed-agents.md | 56 - .../data-sources-for-rag-comparison.md | 61 - .../evaluation-app-vs-model-comparison.md | 52 - .../wiki/comparisons/evaluation-compare.md | 50 - .../wiki/comparisons/evaluation-comparison.md | 71 - .../extension-framework-comparison.md | 38 - ...ine-tuning-vs-compression-vs-deployment.md | 73 - .../comparisons/fine-tuning-vs-compression.md | 55 - .../fine-tuning-vs-model-compression.md | 55 - .../generation-modalities-comparison.md | 54 - .../comparisons/getting-started-comparison.md | 60 - .../image-video-3d-generation-comparison.md | 78 - .../comparisons/image-video-3d-generation.md | 70 + .../image-vs-video-vs-3d-generation.md | 41 - .../knowledge-base-vs-memory-library.md | 57 - .../knowledge-data-management-comparison.md | 78 - .../comparisons/knowledge-memory-compare.md | 69 - .../knowledge-memory-comparison.md | 59 - .../wiki/comparisons/knowledge-vs-memory.md | 63 - .../managed-agents-vs-llm-application.md | 58 - .../comparisons/media-generation-compare.md | 48 - .../media-generation-comparison.md | 72 - .../memory-and-knowledge-solutions.md | 69 - .../memory-approaches-comparison.md | 59 - .../comparisons/memory-feature-comparison.md | 43 - .../wiki/comparisons/memory-solutions.md | 61 + .../model-call-vs-application-call.md | 43 - .../model-customization-methods-comparison.md | 64 - .../model-customization-options.md | 43 - .../comparisons/model-deployment-options.md | 51 + ...odel-deployment-vs-high-speed-inference.md | 45 - .../model-deployment-vs-inference.md | 51 - .../model-evaluation-monitoring.md | 57 + ...el-evaluation-vs-application-evaluation.md | 62 - .../model-evaluation-vs-monitoring.md | 55 - ...-experience-vs-production-vs-deployment.md | 55 - .../comparisons/model-lifecycle-comparison.md | 69 - ...el-monitoring-vs-application-monitoring.md | 54 - .../comparisons/model-optimization-compare.md | 42 - .../model-optimization-comparison.md | 85 - .../model-optimization-methods-comparison.md | 69 - .../monitoring-app-vs-model-comparison.md | 51 - .../wiki/comparisons/monitoring-compare.md | 60 - .../wiki/comparisons/monitoring-comparison.md | 57 - .../multimodal-generation-api-comparison.md | 76 - .../multimodal-generation-comparison.md | 71 - .../qwen-api-vs-app-call-vs-agents-api.md | 74 - ...-api-vs-omni-realtime-vs-managed-agents.md | 57 - .../wiki/concepts/agent-application.md | 81 - .../wiki/concepts/agent-orchestration.md | 90 - .../wiki/concepts/api-key.md | 93 - .../wiki/concepts/async-invocation.md | 64 - .../wiki/concepts/billing.md | 85 - .../wiki/concepts/context-window.md | 32 - .../wiki/concepts/cross-session-memory.md | 112 - .../wiki/concepts/dashscope-sdk.md | 99 - .../wiki/concepts/embedding.md | 54 - .../wiki/concepts/evaluation.md | 77 - .../wiki/concepts/fine-tuning.md | 89 - .../wiki/concepts/function-calling.md | 62 +- .../wiki/concepts/knowledge-base.md | 87 - .../wiki/concepts/long-context.md | 45 - .../wiki/concepts/long-term-memory.md | 48 + .../wiki/concepts/model-deployment.md | 92 - .../wiki/concepts/model-selection.md | 83 - .../wiki/concepts/multi-modal.md | 80 + .../wiki/concepts/multimodal.md | 45 - .../wiki/concepts/openai-compatible-api.md | 82 + .../concepts/openai-compatible-interface.md | 79 - .../wiki/concepts/plugin.md | 56 + .../wiki/concepts/prompt-caching.md | 59 - .../wiki/concepts/prompt-engineering.md | 74 - .../wiki/concepts/rag.md | 78 +- .../wiki/concepts/rate-limiting.md | 71 - .../wiki/concepts/region.md | 80 - .../wiki/concepts/rerank.md | 61 - .../wiki/concepts/sdk.md | 54 - .../wiki/concepts/streaming-output.md | 85 +- .../wiki/concepts/streaming.md | 47 - .../wiki/concepts/token-billing.md | 69 - .../wiki/concepts/token.md | 84 +- .../wiki/concepts/vector-embedding.md | 57 - .../wiki/concepts/vpc-private-access.md | 55 - .../wiki/concepts/workflow.md | 75 - .../wiki/concepts/workspace.md | 63 - .../wiki/guides/application-evaluation.md | 178 +- .../wiki/guides/application-monitoring.md | 161 +- .../application-permission-management.md | 156 +- .../application-publishing-and-sharing.md | 92 +- .../wiki/guides/application-support.md | 98 +- .../wiki/guides/application-use-cases.md | 114 +- .../guides/bailian-application-calling.md | 236 +- .../wiki/guides/data-connection-overview.md | 91 +- .../wiki/guides/fine-tuning.md | 91 +- .../wiki/guides/get-started-with-models.md | 99 +- .../wiki/guides/knowledge-base.md | 79 +- .../wiki/guides/llm-application.md | 191 +- .../wiki/guides/managed-agents.md | 87 +- .../wiki/guides/memory-library-overview.md | 187 +- .../wiki/guides/model-compression.md | 87 +- .../wiki/guides/model-context-protocol.md | 119 +- .../wiki/guides/model-data-overview.md | 178 +- .../wiki/guides/model-deployment-1.md | 132 +- .../guides/model-evaluation-introduction.md | 123 +- .../wiki/guides/model-experience.md | 79 +- .../wiki/guides/model-high-speed-inference.md | 112 +- .../wiki/guides/model-monitoring.md | 111 +- .../wiki/guides/plug-in.md | 170 +- .../wiki/guides/prompt.md | 124 +- .../wiki/guides/release-notes.md | 61 +- .../wiki/guides/security-and-compliance.md | 253 +- .../wiki/guides/skill.md | 137 +- .../wiki/guides/start-using.md | 86 +- .../wiki/guides/support.md | 120 +- .../wiki/guides/test-1.md | 73 +- .../wiki/guides/token-plan-guide.md | 84 +- .../wiki/guides/use-cases.md | 91 +- .../use-chat-client-or-development-tool.md | 81 +- skills/bailian-docs-llm-wiki/wiki/index.md | 29 +- 427 files changed, 5418 insertions(+), 48154 deletions(-) delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/embedding.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/emo-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/farui-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-music.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/glm-fast.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/gui-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/liveportrait.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qvq-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-long.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/sambert.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/videoretalk.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-changeset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/technology-integration-scheme.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/tongyi-xiaomi-ccai-update-announcement.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/product-0verview.md (100%) delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-overview.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-ram.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-search-data-source => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-createdataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-search-data-source => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-deletedataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-search-data-source => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-getdataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-search-data-source => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-listdatasets.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-search-data-source => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-updatedataset.md (100%) create mode 100644 skills/bailian-docs-llm-wiki/raw/test/test.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/billing.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/region.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md diff --git a/skills/bailian-docs-llm-wiki/SKILL.md b/skills/bailian-docs-llm-wiki/SKILL.md index e5ff0782..9d0c87dc 100644 --- a/skills/bailian-docs-llm-wiki/SKILL.md +++ b/skills/bailian-docs-llm-wiki/SKILL.md @@ -92,19 +92,7 @@ description: >- | `TG` | 文本生成 | | `Reasoning` | 推理 | | `VU` | 视觉理解 | -| `IG` | 图像生成 | | `VG` | 视频生成 | -| `TTS` | 语音合成 | -| `ASR` | 语音识别 | -| `Realtime-ASR` | 实时语音识别 | -| `Realtime-Text-to-Speech` | 实时语音合成 | -| `Realtime-Audio-Translate` | 实时音频翻译 | -| `Realtime-Omni` | 实时全模态 | -| `Multimodal-Omni` | 全模态 | -| `ME` | 多模态嵌入 | -| `TR` | 翻译 | -| `3D-generation` | 3D 生成 | -| `Realtime-Chatting` | Realtime-Chatting | 一个模型常常带多个 capability，`index.md` 中按 `capabilities[0]`（主能力）归类， 查找时按中文标签即可定位章节。 @@ -148,22 +136,6 @@ description: >- | **按家族筛选**：按 primaryCapability / providers / itemCount / maxContextWindow 找家族 | `models/families.jsonl`（一行一家族，含 items[] 摘要） | | 模型家族总览 / 按能力分桶浏览 | `models/index.md` | | 主题页 / API 文档（按功能领域查找） | `wiki/index.md`（完整索引入口） | -| 函数调用（Function Calling） | `wiki/concepts/function-calling.md` | -| 检索增强生成（RAG） | `wiki/concepts/rag.md` | -| Token 与计量计费 | `wiki/concepts/token.md` | -| 流式输出 | `wiki/concepts/streaming.md` | -| OpenAI 兼容接口 | `wiki/concepts/openai-compatible-interface.md` | -| API Key 鉴权 | `wiki/concepts/api-key.md` | -| 异步调用与任务轮询 | `wiki/concepts/async-invocation.md` | -| 业务空间（Workspace） | `wiki/concepts/workspace.md` | -| 模型微调与生产链路 | `wiki/concepts/fine-tuning.md` | -| 评测体系 | `wiki/concepts/evaluation.md` | -| 多模态能力 | `wiki/concepts/multimodal.md` | -| 图像、视频与3D生成对比 | `wiki/comparisons/media-generation-compare.md` | -| 模型微调、压缩与高速推理对比 | `wiki/comparisons/model-optimization-compare.md` | -| 知识库与长期记忆对比 | `wiki/comparisons/knowledge-memory-compare.md` | -| 应用监控与模型监控对比 | `wiki/comparisons/monitoring-compare.md` | -| 应用评测与模型评测对比 | `wiki/comparisons/evaluation-compare.md` | > 实际文件名以 `wiki/index.md` 为准；上表若有出入应回到索引页查找。 diff --git a/skills/bailian-docs-llm-wiki/llms.txt b/skills/bailian-docs-llm-wiki/llms.txt index 16975914..95cdb18f 100644 --- a/skills/bailian-docs-llm-wiki/llms.txt +++ b/skills/bailian-docs-llm-wiki/llms.txt @@ -4,24 +4,19 @@ ## 模型使用指南 -- **接入客户端/开发工具** - - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) - - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) - - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) - - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) - - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) - - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) +- **开始使用** + - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) + - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) + - [选择模型](raw/model-user-guide/get-started-with-models/models.md) + - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) + - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) + - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) +- **产品计费** + - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) + - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) + - [节省计划与资源包](raw/model-user-guide/test-1/savings-plan-and-resource-package.md) + - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) + - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) - **Token Plan（团队版）** - **最佳实践** - [工具调用](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) @@ -31,40 +26,43 @@ - [联网搜索](raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) - - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [Token Plan（团队版）概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team.md) + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-faq.md) -- **产品计费** - - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) - - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) - - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) - - [节省计划与资源包](raw/model-user-guide/test-1/savings-plan-and-resource-package.md) - - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) -- **开始使用** - - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - - [选择模型](raw/model-user-guide/get-started-with-models/models.md) - - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) - - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) - **模型体验** - - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) + - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - [语音合成](raw/model-user-guide/model-experience/tts-model.md) - - [语音识别](raw/model-user-guide/model-experience/asr-model.md) + - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) + - [语音识别](raw/model-user-guide/model-experience/asr-model.md) - [全模态](raw/model-user-guide/model-experience/omni.md) -- **模型部署** - - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) - - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) - - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) + - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) +- **接入客户端/开发工具** + - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) + - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) + - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) + - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) + - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) + - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) + - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) + - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) + - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) + - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) + - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) + - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) + - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) + - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) + - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) + - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) + - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) +- **模型推理** + - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) + - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) - **模型调优** - **千问模型调优** - [模型调优简介](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) @@ -75,21 +73,26 @@ - [CosyVoice模型调优](raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) -- **模型推理** - - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) +- **模型部署** + - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) + - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) + - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) + - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型评测** - - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - [评测维度](raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) + - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - **模型压缩** - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) - **用量统计与性能监控** - - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) + - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) +- **模型数据** + - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) + - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) - **安全合规** - **传输安全** - - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) + - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - **安全存储** - [配置终端节点并发起连接](raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) @@ -101,104 +104,99 @@ - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - [合规资质与隐私说明](raw/model-user-guide/security-and-compliance/privacy-notice.md) -- **模型数据** - - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) +- **服务支持** + - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) + - [相关协议](raw/model-user-guide/support/related-agreements.md) - **实践教程** - **三方模型调用教程** - - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek-阿里云](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) + - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [Kimi](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - [GLM](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) - - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) + - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) + - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Stepfun-阶跃星辰](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - [高效搭建 AI 智能体与工作流应用](raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) + - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - [文生文Prompt指南](raw/model-user-guide/use-cases/prompt-engineering-guide.md) - - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) + - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - [文生视频/图生视频Prompt指南](raw/model-user-guide/use-cases/text-to-video-prompt.md) + - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](raw/model-user-guide/use-cases/model-training-best-practices.md) - - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - [借助大模型将文档转换为视频](raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) -- **服务支持** - - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - - [相关协议](raw/model-user-guide/support/related-agreements.md) + - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - **产品动态** - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) ## 应用使用指南 -- **记忆库** - - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) - - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) +- **应用开发** + - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) + - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) + - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) + - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) + - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) + - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) - **开始使用** - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) -- **Prompt** - - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) - - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) - - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) - - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - **Managed Agents** - - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) + - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [构建 Agent](raw/application-user-guide/managed-agents/managed-agents-agent.md) - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) - - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) -- **应用开发** - - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) - - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) - - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) - - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) - - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) - - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) -- **Skill** - - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) -- **数据连接** - - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) + - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) +- **Prompt** + - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) + - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) + - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) + - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) + - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) +- **记忆库** + - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) + - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) + - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - **知识库（RAG）** - - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) + - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) - [知识库日志与监控](raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) + - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) + - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) -- **插件** - - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) - - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) - **MCP** - - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) - - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) + - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) + - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) +- **Skill** + - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) +- **数据连接** + - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) +- **插件** + - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) + - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) + - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) - **应用发布与分享** - - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) + - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [UI设计器](raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) -- **应用观测** - - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **应用调用** - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) + - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - **应用评测** - **新版应用评测** - [新版评测集](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) @@ -209,6 +207,15 @@ - [手动评测](raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - **应用广场** + - **官方应用-通义音频播客生成** + - **API参考** + - **API目录** + - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) + - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) + - [通义音频播客生成产品介绍](raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md) - **官方应用-通义拍照解题辅导** - **API参考** - **API目录** @@ -218,54 +225,45 @@ - [服务接入点](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-ram.md) - [通义拍照解题辅导产品介绍](raw/application-user-guide/application-gallery/edu-tutor/brief-introduction-of-edu-tutor.md) - - **官方应用-通义音频播客生成** - - **API参考** - - **API目录** - - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) - - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) - - [通义音频播客生成产品介绍](raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md) - **官方应用-多模态交互开发套件** - **使用指南** - [应用创建](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-creation.md) - [应用配置](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-configuration.md) - [应用体验与发布](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-experience-and-publishing.md) - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [指令列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/instruction-list.md) + - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) - [三方Agent接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-integration-a2a.md) + - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) + - **API参考** + - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) + - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) + - [调用官方Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/official-agent.md) + - [调用三方语音模型](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/third-party-voice-integration.md) + - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) + - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) + - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) + - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) + - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) - **SDK安装** - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) + - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) + - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) - - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) - - **API参考** - - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) - - [调用官方Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/official-agent.md) - - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) - - [调用三方语音模型](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/third-party-voice-integration.md) - - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) - - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) - - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) - **最佳实践** + - **接入拍照问答Agent** + - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) + - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - **接入百炼及三方Agent** - [百炼及三方Agent直连调用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/agent-direct-call.md) - [接入百炼智能体应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-app.md) - [接入百炼工作流应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-workflow.md) - - **接入拍照问答Agent** - - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - **接入图像生成Agent** - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) @@ -274,11 +272,11 @@ - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) - [接入多模态备忘录Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/multimodal-memo-agent.md) - - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) + - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [动作情绪控制实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/action-emotion-control-practice.md) - - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) + - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) - [基于RTOS SDK (License模式) 实现聊天能力](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/chat-capability-based-on-rtos-sdk.md) @@ -286,204 +284,203 @@ - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) - [多模态交互开发套件常见问题](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-faq.md) - **官方应用-全妙轻应用系列** - - **计费说明（全妙轻应用）** - - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) - - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) - - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) - - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) - - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) - - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) - **使用指南** - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) - - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) + - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) + - **计费说明（全妙轻应用）** + - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) + - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) + - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) + - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) + - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) + - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) + - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) + - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) + - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) + - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) - **开发文档** - - **最佳实践** - - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) - - [挖掘VOC信息和数据分析的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-mining-voc-information-and-data-analysis.md) - - [阿里云百炼工作流集成视频理解最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md) - **API参考** - **数据结构** - [ModelUsage](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-struct-dir/api-quanmiaolightapp-2024-08-01-struct-modelusage.md) - **API目录** + - **传媒/零售文章风格与格式学习** + - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **电商零售推广文案写作** - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) - - **传媒/零售文章风格与格式学习** - - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **影视互娱剧本创作** + - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) - - **影视传媒智能拆条** - - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) - - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) - - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) - - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) - - **泛企业VOC挖掘** - - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - **影视传媒视频理解** - - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) + - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) - [UpdateVideoAnalysisTasks - 视频理解-批量取消任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistasks.md) + - **影视传媒智能拆条** + - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) + - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) + - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) + - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) + - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) + - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) - **车机网络热点信息互动问答** - - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) - [RunHotTopicChat - 播报单（热榜）问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicchat.md) + - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) - **泛企业线索挖掘** - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) - - **网络内容安全审核** - - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) - - **作文批改** - - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) - - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) - - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) - - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) - **其他** - [GenerateBroadcastNews - 播报单（热榜）热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-generatebroadcastnews.md) - - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) + - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) + - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) - - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC挖掘异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submitenterprisevocanalysistask.md) - - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) + - **泛企业VOC挖掘** + - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) + - **作文批改** + - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) + - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) + - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) + - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) + - **网络内容安全审核** + - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-ram.md) + - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-changeset.md) + - **最佳实践** + - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) + - [挖掘VOC信息和数据分析的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-mining-voc-information-and-data-analysis.md) + - [阿里云百炼工作流集成视频理解最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md) - [全妙轻应用更新公告](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-update-announcement.md) - [常见问题](raw/application-user-guide/application-gallery/quanmiao-light-application-series/quanmiao-lightapp-faq.md) - - **官方应用-伶鹊CCAI-语音对话机器人** - - **API参考** - - **API目录** - - **MQ消息订阅配置** - - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) - - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) - - **变量管理** - - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) - - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) - - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) - - **热词管理** - - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) - - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) - - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) - - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) - - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) - - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) - - **三方语音配置** - - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) - - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) - - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) - - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) - - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) - - **克隆音管理** - - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) - - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) - - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) - - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) - - **应用管理** - - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) - - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) - - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) - - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) - - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) - - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) - - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) - - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) - - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) - - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) - - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) - - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) - - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) - - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md) - - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) - **官方应用-伶鹊CCAI-对话分析AIO** - **使用指南** - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) + - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) - **API参考** - **API目录** - **热词管理** - - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) - - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) - - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) - - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) - - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) - - **其他** - - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) + - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) + - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) + - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) + - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) + - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) - **不推荐或白名单开放** - - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) - - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) - - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) - - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) - - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) - - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) + - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) + - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) + - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) + - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) + - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) + - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) + - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) + - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) - **接口调用示例** - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) + - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md) - **最佳实践** - - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) + - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) - - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) + - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) - - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md) + - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) + - **官方应用-伶鹊CCAI-语音对话机器人** + - **API参考** + - **API目录** + - **MQ消息订阅配置** + - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) + - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) + - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) + - **变量管理** + - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) + - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) + - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) + - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) + - **三方语音配置** + - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) + - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) + - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) + - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) + - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) + - **热词管理** + - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) + - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) + - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) + - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) + - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) + - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) + - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - **克隆音管理** + - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) + - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) + - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) + - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) + - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) + - **应用管理** + - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) + - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) + - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) + - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) + - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) + - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) + - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) + - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) + - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) + - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) + - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) + - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) + - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) + - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) + - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) + - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md) + - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) + - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) - **官方应用-伶鹊CCAI-客服对话Agent** - **API参考** - **API目录** - - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) + - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) - **通义点金** - **API参考** @@ -491,54 +488,54 @@ - **平台能力-文档库** - [UpdateDocumentChunk - 更新文档块内容](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocumentchunk.md) - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) + - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) - [CreateLibrary - 创建文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createlibrary.md) - [GetLibrary - 获取文档库详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrary.md) - - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) - - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [GetDocumentUrl - 获取文档的下载链接](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumenturl.md) + - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) + - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) - - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) + - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) + - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) - [RecallDocument - 文档召回](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-recalldocument.md) - [GetParseResult - 获取文档解析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getparseresult.md) - - [UpdateLibrary - 更新文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatelibrary.md) - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) + - [UpdateLibrary - 更新文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatelibrary.md) - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) - [RunLibraryChatGeneration - 文档库会话生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-runlibrarychatgeneration.md) - - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - [GetHistoryListByBizType - 根据业务类型获取对话历史记录](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-gethistorylistbybiztype.md) + - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - **平台能力-应用** - [EndToEndRealTimeDialog - 语音实时对话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-endtoendrealtimedialog.md) - [RunDialogAnalysis - 会话分析结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rundialoganalysis.md) - - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) - - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) + - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) + - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) + - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) + - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [GetDialogAnalysisResult - 获取会话分析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoganalysisresult.md) - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) + - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) + - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) + - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) + - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) - [UpdateQaLibrary - 更新QA问答库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-updateqalibrary.md) - - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - [SubmitChatQuestion - 提交问题列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-submitchatquestion.md) - - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) - [RunChatResultGeneration - 对话结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runchatresultgeneration.md) + - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - **其他** - [DashscopeAsyncTaskFinishEvent - Dashscope异步任务完成回调事件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-other/api-dianjin-2024-06-28-dashscopeasynctaskfinishevent.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) @@ -552,56 +549,56 @@ - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) - - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) + - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) + - **官方应用-通义深度搜索** + - **API参考** + - **API目录** + - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) + - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) + - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) + - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) + - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) + - [服务接入节点](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-service-access-point.md) + - [错误码-通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-error-code.md) + - [通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/tongyi-deep-search-introduction.md) + - [操作指南](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-guide.md) - **官方应用-通义多模态翻译** - **API参考** - **API目录** - **文本翻译** - - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) - [BatchTranslate - 批量文本翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-batchtranslate.md) + - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) - [SubmitLongTextTranslateTask - 提交长文本翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submitlongtexttranslatetask.md) - - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) + - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) + - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) - - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - **图片翻译** - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) - **文档翻译** - - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) + - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-overview.md) - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md) - [通义多模态翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/official-application-tongyi-translate-overview.md) - [网页翻译JSSDK](raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md) - - **官方应用-通义深度搜索** - - **API参考** - - **API目录** - - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) - - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) - - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) - - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) - - [服务接入节点](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-service-access-point.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) - - [错误码-通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-error-code.md) - - [通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/tongyi-deep-search-introduction.md) - - [操作指南](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-guide.md) - **官方应用-千问联网检索Agent** - **API参考** - - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/service-access-point.md) - - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) + - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) - [生成对话](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat.md) + - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) - [千问联网检索Agent产品简介](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-guide.md) - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) - **通义 UI Agent** @@ -611,8 +608,8 @@ - **使用指南** - **AI妙笔** - **功能界面** - - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) + - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) @@ -621,29 +618,29 @@ - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [素材库](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/material-library.md) - [文章风格和格式学习](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/style-imitation.md) - - [AI妙策](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/ai-miaoce.md) - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) + - [AI妙策](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/ai-miaoce.md) - [深度写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/deep-writing.md) - **文本写作指导** - **传媒类文体写作指导** - [快速写一篇传媒稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/quick-media-writing-prompt.md) - - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) - [没有思路，要谋篇布局](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/use-amb-to-help-writing.md) + - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) - **政务公文写作指导** - - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) + - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) - [常见FAQ](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/faq-for-using-quanmiao-series-products.md) - **更新公告** - **功能更新** - - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2025年2月26日更新-妙笔](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/february-26-2025-update-miaobi.md) - [2024年3月11更新-AI全妙系列 V2.2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-11-ai-quanmiao-v2-2.md) + - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) + - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) + - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) - [计费说明（PPT生成）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/ppt-generation-billing.md) - [计费说明（妙策-自定义数据源）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-document-miaoce-custom-data-source.md) - [计费说明（视频混剪）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-description-video-mixing.md) @@ -652,112 +649,118 @@ - [妙搜](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaodou-and-miaodu-guidelines-for-use/ai-miaosou.md) - [计费说明（妙搜和妙读）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaosou-miaodu-api-billing.md) - **开发文档** - - **更多** - - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) - - [妙搜数据集管理通过API引入数据源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaosou-introduce-data-source-through-api.md) - - [全妙iframe嵌入方案](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/iframe-embedding-scheme.md) - - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) - - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) - - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) + - **最佳实践** + - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) + - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) + - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) + - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) + - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) + - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) + - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - **API参考** - **数据结构** - - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) + - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) - - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) + - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) - [WritingStyleTemplateField](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatefield.md) - **API目录** - **通用接口** - [CreateToken - 获取授权token](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-createtoken.md) - [ListDialogues - 生成历史列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listdialogues.md) - - [GetProperties - 获取配置信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-getproperties.md) - [ListVersions - 获取版本信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listversions.md) + - [GetProperties - 获取配置信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-getproperties.md) - **通用接口-文件上传下载** - - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - - **通用接口-异步任务管理** - - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) - - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) - - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) + - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) - **通用接口-通用配置** - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) + - **通用接口-异步任务管理** + - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) + - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) + - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) + - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - **妙笔-创作文章** - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) - - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) + - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) + - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) + - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - [RunWriteToneGeneration - 文风改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritetonegeneration.md) + - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) - [RunSummaryGenerate - 摘要生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runsummarygenerate.md) - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) + - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) - [ListBuildConfigs - 获取系统自定义预设](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-listbuildconfigs.md) - - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - [GenerateImageTask - 生成智能配图任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-generateimagetask.md) - - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - [FeedbackDialogue - 反馈对话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-feedbackdialogue.md) + - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - **妙笔-文体仿写** - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) - - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) + - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - [DeleteStyleLearningResult - 删除自定义文体](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-deletestylelearningresult.md) - - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) + - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) + - **妙笔-视频审校** + - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) + - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文章审校-规则库管理** - [SubmitAuditNote - 提交自定义规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-submitauditnote.md) - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) - - [DownloadAuditNote - 下载规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-downloadauditnote.md) - [DeleteAuditNote - 删除规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-deleteauditnote.md) - [GetAuditNotePostProcessingStatus - 获取规则库后处理进度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnotepostprocessingstatus.md) - - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) + - [DownloadAuditNote - 下载规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-downloadauditnote.md) - [GetAvailableAuditNotes - 查询可用规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getavailableauditnotes.md) - - **妙笔-视频审校** - - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) - - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) + - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) - **妙笔-文章审校-词库管理** - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) - - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) + - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) + - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) - [FetchExportTermsTask - 获取导出词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchexporttermstask.md) - - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - **妙笔-文章审校-事实性审核** - [SubmitFactAuditUrl - 提交事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-submitfactauditurl.md) - - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - [GetFactAuditUrl - 获取事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-getfactauditurl.md) + - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - **妙笔-文章审校** - - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md) - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md) + - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md) - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md) - - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - [GetSmartAuditResult - 查询智能审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-getsmartauditresult.md) - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) + - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - [ExportAuditContentResult - 导出智能审校报告](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-exportauditcontentresult.md) - **妙笔-文档管理** - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) - - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) + - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) - - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) + - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) + - **妙笔-素材库** + - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) + - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) + - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) + - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) + - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - **妙笔-素材库-自定义文本** - [GetCustomText - 获取自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-getcustomtext.md) - [UpdateCustomText - 更新自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-updatecustomtext.md) @@ -765,14 +768,6 @@ - [SaveCustomText - 保存自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-savecustomtext.md) - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - - **妙笔-素材库** - - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) - - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) - - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - - **公文库检索** - - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-自定义数据源** - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) @@ -782,34 +777,36 @@ - [AsyncCreateClipsTimeLine - 创建剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstimeline.md) - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - [AsyncUploadVideo - 异步上传视频剪辑素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncuploadvideo.md) + - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) - - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) - **妙策-选题热点** - - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) + - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) - - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) + - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) + - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) - [ListWebReviewPoints - 获取网友视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listwebreviewpoints.md) - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) + - **公文库检索** + - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-自定义话题** - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) + - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) - **妙策-openapi** - [SubmitDocClusterTask - 提交内容聚合任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitdocclustertask.md) - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) + - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) - [SubmitCustomTopicSelectionPerspectiveAnalysisTask - 提交自定义热点选题视角分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitcustomtopicselectionperspectiveanalysistask.md) - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - **妙策-新闻播报** @@ -818,159 +815,161 @@ - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - **妙策-企业VOC挖掘** - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) - - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) + - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) - **妙搜-数据源** - - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) - [GetDatasetDocument - 数据源-获取文档详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdatasetdocument.md) + - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) - [UpdateDatasetDocument - 数据源-修改文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedatasetdocument.md) - [ListDatasetDocuments - 数据源-文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasetdocuments.md) - - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) + - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - **妙搜-智能搜索** - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) - - **系统配置-信源管理** - - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) - - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - **系统配置-干预配置** - [ListInterveneCnt - 获得所有干预项的数量](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenecnt.md) - [ListIntervenes - 列出干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenes.md) - - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [ImportInterveneFile - 同步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefile.md) - [ImportInterveneFileAsync - 异步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefileasync.md) + - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) + - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - [DeleteInterveneRule - 删除干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-deleteintervenerule.md) - [ExportIntervenes - 导出干预项内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-exportintervenes.md) - [GetInterveneImportTaskInfo - 获得干预项目导入任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneimporttaskinfo.md) + - **系统配置-信源管理** + - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) + - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - **妙读-基础操作类** - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) - [UploadBook - 书籍上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploadbook.md) - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) - - [DeleteDocs - 批量删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-deletedocs.md) - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) + - [DeleteDocs - 批量删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-deletedocs.md) - **妙读-生成类** - [RunMultiDocIntroduction - 多文档聚合摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runmultidocintroduction.md) - - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) + - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) + - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) - [RunCommentGeneration - 客户之声预测](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runcommentgeneration.md) - **妙读-抽取类** - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - - **妙读-其他** - - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) - - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) - **妙读-问答类** - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) + - **妙读-其他** + - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) + - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) + - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) - **深度写作** - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) - - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) + - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) - - **标书生成** - - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) - - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) - - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) - - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) - **PPT生成** - - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [ListEnterprisePptTemplates - 查询企业专属PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listenterpriseppttemplates.md) - - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) + - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) + - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) - [ExportPptArtifact - 导出PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-exportpptartifact.md) - [GetPptArtifact - 查询PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifact.md) + - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ListPptArtifacts - 查询PPT作品列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listpptartifacts.md) - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) - [InitiatePptCreation - 初始化用来创建PPT的会话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreation.md) - [GetPptConfig - 获取PPT组件配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptconfig.md) - - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) + - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) + - **标书生成** + - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) + - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) + - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) + - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) + - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) + - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) + - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - **其他** - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) - - [SaveOrUpdateOssConfig - 配置-云存储-参数配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-saveorupdateossconfig.md) - [SubmitSmartClipTask - 提交智能一键成片任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitsmartcliptask.md) - - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) + - [SaveOrUpdateOssConfig - 配置-云存储-参数配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-saveorupdateossconfig.md) - [CreateDataPermissions - 权限-批量添加](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdatapermissions.md) - - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) - - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md) - - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) + - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) + - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) + - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md) + - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md) - [FetchParseDocumentLayoutTask - 获取排版任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-fetchparsedocumentlayouttask.md) - - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md) - - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md) - - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) + - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md) + - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md) + - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-ram.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) - - **最佳实践** - - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) - - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) - - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) - - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) - - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) - - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) + - **更多** + - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) + - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) + - [妙搜数据集管理通过API引入数据源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaosou-introduce-data-source-through-api.md) + - [全妙iframe嵌入方案](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/iframe-embedding-scheme.md) + - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) + - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) + - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) - [官方应用-通义听悟Agent](raw/application-user-guide/application-gallery/official-application-tingwu-agent.md) - [官方应用-析言GBI](raw/application-user-guide/application-gallery/xiyan-gbi.md) - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) +- **应用观测** + - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **权限管理** - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) -- **服务支持** - - [常见问题](raw/application-user-guide/application-support/application-faq.md) - - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) - **实践教程** - [在网站上增加一个AI助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) + - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [基于本地知识库构建RAG应用](raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) +- **服务支持** + - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) + - [常见问题](raw/application-user-guide/application-support/application-faq.md) ## 模型 API 参考 - **使用 API** - - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) + - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) - [错误码](raw/model-api-reference/preparations/error-code.md) - **图像生成** - **千问** + - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - **万相** + - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-文生图V2版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) + - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) - [万相-涂鸦作画API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-图像局部重绘API参考](raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - **Z-Image** - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) @@ -981,13 +980,13 @@ - **创意工具** - [人像风格重绘API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - [图像画面扩展API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - [虚拟模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) + - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) + - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [图像背景生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) - - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) + - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [创意文字WordArt锦书](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) - [常见问题](raw/model-api-reference/image-generation/image-faq.md) @@ -1002,95 +1001,95 @@ - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - **更多模型** - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) + - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - [Qwen-OCR API参考](raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) +- **模型生产** + - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) + - [模型部署](raw/model-api-reference/model-production/deployments-api.md) - **工具包/框架** - [OpenAI Chat接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) + - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI兼容-Batch Chat](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI Embedding接口兼容](raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) -- **模型生产** - - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - - [模型部署](raw/model-api-reference/model-production/deployments-api.md) - **更多** - [生成临时API Key](raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) + - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [子业务空间的模型调用](raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - [上传本地文件获取临时URL](raw/model-api-reference/more-about-models/get-temporary-file-url.md) + - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - **视频生成** - **HappyHorse** - [HappyHorse-文生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) + - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - - **人像驱动** - - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - - [图生播报视频-灵动人像LivePortrait](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) - **万相** - **万相-早期视频模型（2.1-2.6）** - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) + - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-视频编辑API参考（2.1）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) + - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [万相-视频换人API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) + - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) + - **人像驱动** + - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) + - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) + - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) + - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) + - [图生播报视频-灵动人像LivePortrait](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) + - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) - **爱诗** - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) + - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - **可灵** - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **Vidu** - [Vidu-文生视频API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) + - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - **音频** - **语音识别** + - **实时语音识别（Qwen-ASR-Realtime）** + - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) + - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) + - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) + - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) + - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - **实时语音识别（Fun-ASR）** - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) - - **实时语音识别（Qwen-ASR-Realtime）** - - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) - - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) - - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) - - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) - **实时语音识别（Paraformer）** - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) - - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) + - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) - - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) + - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) - **录音文件识别（Fun-ASR）** - [Fun-ASR录音文件识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - [Fun-ASR录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) @@ -1098,11 +1097,11 @@ - [Fun-ASR录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - [Fun-ASR录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) - **录音文件识别（Paraformer）** + - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - [Paraformer录音文件识别RESTful API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) - [Paraformer录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) - - [Paraformer录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) - - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - [Paraformer录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) + - [Paraformer录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) - **定制热词** - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) @@ -1110,20 +1109,20 @@ - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) - [录音文件识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - **语音合成** + - **实时语音合成（Qwen-TTS-Realtime）** + - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) + - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) + - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) + - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) - - **实时语音合成（Qwen-TTS-Realtime）** - - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) - - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) - - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) @@ -1133,126 +1132,126 @@ - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) + - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) - - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) + - **非实时语音合成（MiniMax）** + - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - **声音复刻** - - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) + - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - [声音复刻Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md) - - **非实时语音合成（MiniMax）** - - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - [非实时语音合成（Qwen-TTS）API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md) - [声音设计API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/voice-design-api-references.md) + - **音乐生成** + - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音翻译** - **实时音视频翻译（Qwen-Livetranslate-Realtime）** - [客户端事件](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/live-translator-client-events.md) - [服务端事件](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/live-translator-server-events.md) - - [实时音视频翻译（Qwen-LiveTranslate）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-java-sdk.md) - [实时音视频翻译（Qwen-LiveTranslate）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-python-sdk.md) + - [实时音视频翻译（Qwen-LiveTranslate）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-java-sdk.md) - [音视频翻译-通义千问 API 参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/qwen3-livetranslate-flash-api.md) - - **音乐生成** - - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音对话** - **实时语音对话** - [Qwen-Audio 实时语音对话客户端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md) - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话服务端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md) - **向量与排序** + - **通用文本向量** + - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) + - [同步接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) - **多模态向量** - [Multimodal-Embedding API详情](raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) - **排序模型（Rerank）** - [文本排序](raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - - **通用文本向量** - - [同步接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) - - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - [文本生成模型API参考](raw/model-api-reference/qwen-api-reference.md) - [文件管理](raw/model-api-reference/file-management-api.md) ## 应用 API 参考 - **Managed Agents** + - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [API 总览与认证](raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) - - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) - [File](raw/application-api-reference/managed-agents-api/files-api.md) - [Skill](raw/application-api-reference/managed-agents-api/skills-api.md) -- **长期记忆** - - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **应用调用** - - **DashScope API** - - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - **Responses API** - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) + - **DashScope API** + - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) + - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) - **应用组件** - **API目录** - - **知识库** - - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) - - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) - - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) - - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) - - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) - **数据连接（原应用数据）** - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - - [ListFile - 文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) + - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) + - [ListFile - 文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) + - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) + - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) + - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) + - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [ChangeParseSetting - 修改类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) + - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [AddConnector - 新增连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) + - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) + - **知识库** + - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) + - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) + - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) + - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) + - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) + - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) + - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) + - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) + - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) + - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) + - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) + - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) + - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) + - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) - **Prompt工程** - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) + - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - **其他** - **长期记忆（旧）** - - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [CreateMemory - 创建长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) + - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) + - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - [DeleteMemoryNode - 删除记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) + - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) + - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) + - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [版本说明](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) +- **长期记忆** + - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **框架** - **Spring AI Alibaba** - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) @@ -1260,7 +1259,7 @@ - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) - **更多** - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [知识库SearchFilters](raw/application-api-reference/more/how-to-use-search-filters.md) + - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [知识检索与问答](raw/application-api-reference/knowledge.md) diff --git a/skills/bailian-docs-llm-wiki/models/families.jsonl b/skills/bailian-docs-llm-wiki/models/families.jsonl index f40a1a06..a3275cc6 100644 --- a/skills/bailian-docs-llm-wiki/models/families.jsonl +++ b/skills/bailian-docs-llm-wiki/models/families.jsonl @@ -1,169 +1,11 @@ -{"slug":"Kimi-K2","name":"Kimi","description":"Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["moonshot-ai"],"itemCount":5,"items":[{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","contextWindow":262144,"capabilities":["TG","Reasoning"]},{"model":"kimi-k2.5","name":"Kimi-K2.5","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"kimi-k2.6","name":"Kimi-K2.6","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/Kimi-K2.json","maxContextWindow":262144} -{"slug":"MiniMax-M2.1","name":"MiniMax","description":"MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["mini-max"],"itemCount":2,"items":[{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","contextWindow":204800,"capabilities":["Reasoning","TG"]},{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","contextWindow":204800,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/MiniMax-M2.1.json","maxContextWindow":204800} -{"slug":"MiniMax-speech-market-place","name":"MiniMax-Speech系列语音模型","description":"由MiniMax提供的MiniMax-Speech系列语音模型API服务。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["mini-max"],"itemCount":4,"items":[{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","capabilities":["TTS"]},{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","capabilities":["TTS"]},{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","capabilities":["TTS"]},{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","capabilities":["TTS"]}],"detailPath":"groups/MiniMax-speech-market-place.json"} -{"slug":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","description":"图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","capabilities":["IG"]}],"detailPath":"groups/aitryon-parsing-v1.json"} -{"slug":"aitryon-plus","name":"AI试衣-Plus版","description":"aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-plus","name":"AI试衣-Plus版","capabilities":["IG"]}],"detailPath":"groups/aitryon-plus.json"} -{"slug":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","description":"图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","capabilities":["IG"]}],"detailPath":"groups/aitryon-refiner.json"} -{"slug":"aitryon","name":"AI试衣-基础版","description":"aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon","name":"AI试衣-基础版","capabilities":["IG"]}],"detailPath":"groups/aitryon.json"} -{"slug":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","description":"AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-detect-gen2.json"} -{"slug":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","description":"AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-gen2.json"} -{"slug":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","description":"AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-template-gen2.json"} -{"slug":"cosyvoice","name":"CosyVoice大模型","description":"基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen","qwen-domain-model"],"itemCount":7,"items":[{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","capabilities":["TTS"]},{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","capabilities":["TTS"]},{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","capabilities":["TTS"]}],"detailPath":"groups/cosyvoice.json"} {"slug":"deepseek","name":"DeepSeek","description":"DeepSeek是由深度求索提供的开源模型，包含 V3.1、V3、R1以及基于Qwen2.5系列蒸馏的大语言模型。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["deepseek"],"itemCount":12,"items":[{"model":"deepseek-r1","name":"DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning"]},{"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","contextWindow":131072,"capabilities":["Reasoning"]},{"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-14b","name":"DeepSeek-R1-Distill-Qwen-14B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-32b","name":"DeepSeek-R1-Distill-Qwen-32B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-7b","name":"DeepSeek-R1-Distill-Qwen-7B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-v3","name":"DeepSeek-V3","contextWindow":65536,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v3.1","name":"DeepSeek-V3.1","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","contextWindow":1000000,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/deepseek.json","maxContextWindow":1000000} -{"slug":"embedding","name":"通义多模态向量","description":"基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。","primaryCapability":"ME","capabilities":["ME"],"providers":["qwen-domain-model"],"itemCount":3,"items":[{"model":"multimodal-embedding-v1","name":"通用多模态向量","contextWindow":0,"capabilities":["ME"]},{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","capabilities":["ME"]},{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","capabilities":["ME"]}],"detailPath":"groups/embedding.json"} -{"slug":"emo-detect-v1","name":"悦动人像EMO-detect","description":"EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emo-detect-v1","name":"悦动人像EMO-detect","capabilities":["VG"]}],"detailPath":"groups/emo-detect-v1.json"} -{"slug":"emo-v1","name":"悦动人像EMO","description":"EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emo-v1","name":"悦动人像EMO","capabilities":["VG"]}],"detailPath":"groups/emo-v1.json"} -{"slug":"emoji-detect-v1","name":"表情包Emoji-detect","description":"表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emoji-detect-v1","name":"表情包Emoji-detect","capabilities":["VG"]}],"detailPath":"groups/emoji-detect-v1.json"} -{"slug":"emoji-v1","name":"表情包Emoji","description":"表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emoji-v1","name":"表情包Emoji","capabilities":["VG"]}],"detailPath":"groups/emoji-v1.json"} -{"slug":"facechain-facedetect","name":"FaceChain人物图像检测","description":"对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"facechain-facedetect","name":"FaceChain人物图像检测","capabilities":["IG"]}],"detailPath":"groups/facechain-facedetect.json"} -{"slug":"facechain-generation","name":"FaceChain人物写真生成","description":"基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"facechain-generation","name":"FaceChain人物写真生成","capabilities":["IG"]}],"detailPath":"groups/facechain-generation.json"} -{"slug":"farui-plus","name":"通义法睿-Plus-32K","description":"通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"farui-plus","name":"通义法睿-Plus-32K","contextWindow":12000,"capabilities":["TG"]}],"detailPath":"groups/farui-plus.json","maxContextWindow":12000} -{"slug":"fun-asr-flash","name":"Fun-ASR-Flash","description":"百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","capabilities":["ASR"]}],"detailPath":"groups/fun-asr-flash.json"} -{"slug":"fun-asr-realtime","name":"Fun-ASR实时语音识别","description":"通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","capabilities":["Realtime-ASR"]},{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","capabilities":["Realtime-ASR"]}],"detailPath":"groups/fun-asr-realtime.json"} -{"slug":"fun-asr","name":"Fun-ASR语音识别","description":"通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"fun-asr","name":"Fun-ASR语音识别","capabilities":["ASR"]},{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","capabilities":["ASR"]}],"detailPath":"groups/fun-asr.json"} -{"slug":"fun-music","name":"音乐生成","description":"百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":2,"items":[{"model":"fun-music-preview","name":"音乐生成 Preview","capabilities":["TTS"]},{"model":"fun-music-v1","name":"音乐生成","capabilities":["TTS"]}],"detailPath":"groups/fun-music.json"} -{"slug":"glm-4.5","name":"GLM","description":"GLM是由智谱提供的开源模型。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["zhipu-ai"],"itemCount":7,"items":[{"model":"glm-4.5","name":"GLM-4.5","contextWindow":131072,"capabilities":["TG"]},{"model":"glm-4.5-air","name":"GLM-4.5-Air","contextWindow":131072,"capabilities":["TG"]},{"model":"glm-4.6","name":"GLM-4.6","contextWindow":202752,"capabilities":["Reasoning","TG"]},{"model":"glm-4.7","name":"GLM-4.7","contextWindow":202752,"capabilities":["TG","Reasoning"]},{"model":"glm-5","name":"GLM-5","contextWindow":202752,"capabilities":["TG","Reasoning"]},{"model":"glm-5.1","name":"GLM-5.1","contextWindow":202745,"capabilities":["TG","Reasoning"]},{"model":"glm-5.2","name":"GLM-5.2","contextWindow":1048576,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/glm-4.5.json","maxContextWindow":1048576} -{"slug":"glm-fast","name":"GLM-5.2-Fast","description":"GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。","primaryCapability":"TG","capabilities":["TG"],"providers":["zhipu-ai"],"itemCount":1,"items":[{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/glm-fast.json","maxContextWindow":1048576} -{"slug":"gui-plus","name":"GUI-Plus","description":"GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"gui-plus","name":"GUI-Plus","contextWindow":256000,"capabilities":["VU"]}],"detailPath":"groups/gui-plus.json","maxContextWindow":256000} -{"slug":"gummy-chat-v1","name":"一句话识别及翻译V1.0","description":"多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","capabilities":["ASR"]}],"detailPath":"groups/gummy-chat-v1.json"} -{"slug":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","description":"多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen"],"itemCount":1,"items":[{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/gummy-realtime-v1.json"} {"slug":"happyhorse-i2v","name":"HappyHorse-I2V","description":"HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","capabilities":["VG"]},{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-i2v.json"} -{"slug":"happyhorse-r2v","name":"HappyHorse-R2V","description":"HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","capabilities":["VG"]},{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-r2v.json"} {"slug":"happyhorse-t2v","name":"HappyHorse-T2V","description":"HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","capabilities":["VG"]},{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-t2v.json"} -{"slug":"happyhorse-video-edit","name":"HappyHorse-Video-Edit","description":"HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":1,"items":[{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","capabilities":["VG"]}],"detailPath":"groups/happyhorse-video-edit.json"} -{"slug":"image-erase-completion","name":"图像擦除补全","description":"图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-erase-completion","name":"图像擦除补全","capabilities":["IG"]}],"detailPath":"groups/image-erase-completion.json"} -{"slug":"image-instance-segmentation","name":"人物实例分割","description":"人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-instance-segmentation","name":"人物实例分割","capabilities":["IG"]}],"detailPath":"groups/image-instance-segmentation.json"} -{"slug":"image-out-painting","name":"图像画面扩展","description":"图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-out-painting","name":"图像画面扩展","capabilities":["IG"]}],"detailPath":"groups/image-out-painting.json"} -{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["moonshot-ai"],"itemCount":4,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":262144} -{"slug":"kling-models-market-place","name":"可灵AI","description":"由可灵AI提供的高质量视频与图像生成及编辑模型。","primaryCapability":"VG","capabilities":["VG","IG"],"providers":["kling"],"itemCount":4,"items":[{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","capabilities":["IG"]},{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","capabilities":["IG"]},{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","capabilities":["VG"]},{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","capabilities":["VG"]}],"detailPath":"groups/kling-models-market-place.json"} -{"slug":"liveportrait-detect","name":"灵动人像LivePortrait-detect","description":"LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","capabilities":["VG"]}],"detailPath":"groups/liveportrait-detect.json"} -{"slug":"liveportrait","name":"灵动人像LivePortrait","description":"LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait","name":"灵动人像LivePortrait","capabilities":["VG"]}],"detailPath":"groups/liveportrait.json"} -{"slug":"minimax-models-market-place","name":"MiniMax文本模型","description":"由MiniMax提供的MiniMax-M系列文本模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["mini-max"],"itemCount":4,"items":[{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","contextWindow":1048576,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/minimax-models-market-place.json","maxContextWindow":1048576} -{"slug":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","description":"Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-8k-v1.json"} -{"slug":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","description":"Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","capabilities":["ASR"]}],"detailPath":"groups/paraformer-8k-v2.json"} -{"slug":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","description":"Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-mtl-v1.json"} -{"slug":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","description":"Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-8k-v1.json"} -{"slug":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","description":"推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-8k-v2.json"} -{"slug":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","description":"Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-v1.json"} -{"slug":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","description":"推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-v2.json"} -{"slug":"paraformer-v1","name":"Paraformer语音识别-v1","description":"Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-v1","name":"Paraformer语音识别-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-v1.json"} -{"slug":"paraformer-v2","name":"Paraformer语音识别-v2","description":"推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-v2","name":"Paraformer语音识别-v2","capabilities":["ASR"]}],"detailPath":"groups/paraformer-v2.json"} -{"slug":"pixverse-c1-market-place","name":"PixVerse C1","description":"由爱诗科技提供的PixVerse C系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-c1-market-place.json"} -{"slug":"pixverse-market-place","name":"PixVerse V5.6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-market-place.json"} -{"slug":"pixverse-v6-market-place","name":"PixVerse V6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-v6-market-place.json"} -{"slug":"qvq-max","name":"QVQ-Max","description":"千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-max","name":"QVQ-Max","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-max.json","maxContextWindow":131072} -{"slug":"qvq-plus","name":"Qwen-QVQ-Plus","description":"千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-plus","name":"QVQ-Plus","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-plus.json","maxContextWindow":131072} -{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":8192} -{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":8192} -{"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"} -{"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072} -{"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072} -{"slug":"qwen-deep-research","name":"qwen-deep-research","description":"千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-deep-research","name":"qwen-deep-research","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-deep-research.json","maxContextWindow":1000000} -{"slug":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","description":"快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen-doc-turbo.json","maxContextWindow":262144} -{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen-domain-model"],"itemCount":6,"items":[{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json"} -{"slug":"qwen-flash-character","name":"Qwen-Flash-Character","description":"千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash-character","name":"Qwen-Flash-Character","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/qwen-flash-character.json","maxContextWindow":8192} -{"slug":"qwen-flash","name":"Qwen-Flash","description":"Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash","name":"Qwen-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-flash.json","maxContextWindow":1000000} -{"slug":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","description":"Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0-pro.json"} -{"slug":"qwen-image-2.0","name":"Qwen-Image-2.0","description":"Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0.json"} -{"slug":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","description":"千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit-max.json"} -{"slug":"qwen-image-edit","name":"Qwen-Image-Edit-Plus","description":"千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image-edit","name":"Qwen-Image-Edit","capabilities":["IG"]},{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit.json"} -{"slug":"qwen-image-max","name":"Qwen-Image-Max","description":"千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-max","name":"Qwen-Image-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-max.json"} -{"slug":"qwen-image-plus","name":"Qwen-Image-Plus","description":"千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image","name":"Qwen-Image","capabilities":["IG"]},{"model":"qwen-image-plus","name":"Qwen-Image-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-plus.json"} -{"slug":"qwen-long","name":"Qwen-Long","description":"Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"qwen-long","name":"Qwen-Long","contextWindow":10000000,"capabilities":["TG"]},{"model":"qwen-long-latest","name":"Qwen-Long-Latest","contextWindow":10000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-long.json","maxContextWindow":10000000} -{"slug":"qwen-math-plus","name":"Qwen-Math-Plus","description":"Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":4,"items":[{"model":"qwen-math-plus","name":"Qwen-Math-Plus","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","contextWindow":4096,"capabilities":["TG"]}],"detailPath":"groups/qwen-math-plus.json","maxContextWindow":4096} -{"slug":"qwen-math-turbo","name":"Qwen-Math-Turbo","description":"Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","contextWindow":4096,"capabilities":["TG"]}],"detailPath":"groups/qwen-math-turbo.json","maxContextWindow":4096} -{"slug":"qwen-max","name":"Qwen-Max","description":"千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-max","name":"Qwen-Max","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/qwen-max.json","maxContextWindow":32768} -{"slug":"qwen-mt-flash","name":"Qwen-MT-Flash","description":"基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-flash.json","maxContextWindow":16384} -{"slug":"qwen-mt-image","name":"Qwen-MT-Image","description":"专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-image","name":"Qwen-MT-Image","capabilities":["IG"]}],"detailPath":"groups/qwen-mt-image.json"} -{"slug":"qwen-mt-lite","name":"Qwen-MT-Lite","description":"基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-lite.json","maxContextWindow":16384} -{"slug":"qwen-mt-plus","name":"Qwen-MT-Plus","description":"基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-plus.json","maxContextWindow":16384} -{"slug":"qwen-mt-turbo","name":"Qwen-MT-Turbo","description":"基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-turbo.json","maxContextWindow":16384} -{"slug":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","description":"千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","contextWindow":32768,"capabilities":["Realtime-Omni"]},{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","contextWindow":32768,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen-omni-turbo-realtime.json","maxContextWindow":32768} -{"slug":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","description":"千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","contextWindow":32768,"capabilities":["Multimodal-Omni"]},{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","contextWindow":32768,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen-omni-turbo.json","maxContextWindow":32768} -{"slug":"qwen-plus-character","name":"Qwen-Plus-Character","description":"千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-plus-character","name":"Qwen-Plus-Character","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/qwen-plus-character.json","maxContextWindow":32768} -{"slug":"qwen-plus","name":"Qwen-Plus","description":"千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":4,"items":[{"model":"qwen-plus","name":"Qwen-Plus","contextWindow":1000000,"capabilities":["Reasoning","TG"]},{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-plus.json","maxContextWindow":1000000} -{"slug":"qwen-rerank","name":"Qwen-Rerank","description":"基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen","qwen-domain-model"],"itemCount":3,"items":[{"model":"gte-rerank-v2","name":"深度文本重排序","contextWindow":30000,"capabilities":["TR"]},{"model":"qwen3-rerank","name":"千问3-Rerank","contextWindow":30000,"capabilities":["TR"]},{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","capabilities":["TR"]}],"detailPath":"groups/qwen-rerank.json","maxContextWindow":30000} -{"slug":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","description":"Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","contextWindow":8192,"capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","contextWindow":8192,"capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-tts-realtime.json","maxContextWindow":8192} -{"slug":"qwen-tts","name":"Qwen-TTS","description":"千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-tts","name":"Qwen-TTS","contextWindow":8192,"capabilities":["TTS"]},{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","contextWindow":8192,"capabilities":["TTS"]}],"detailPath":"groups/qwen-tts.json","maxContextWindow":8192} -{"slug":"qwen-turbo","name":"Qwen-Turbo","description":"千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-turbo","name":"Qwen-Turbo","contextWindow":131072,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-turbo.json","maxContextWindow":131072} -{"slug":"qwen-vl-embedding","name":"Qwen-VL-Embedding","description":"基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。","primaryCapability":"ME","capabilities":["ME"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","capabilities":["ME"]},{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","capabilities":["ME"]}],"detailPath":"groups/qwen-vl-embedding.json"} -{"slug":"qwen-vl-max","name":"Qwen-VL-Max","description":"Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-vl-max","name":"Qwen-VL-Max","contextWindow":131072,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-max.json","maxContextWindow":131072} -{"slug":"qwen-vl-ocr","name":"Qwen-VL-OCR","description":"Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen-domain-model"],"itemCount":3,"items":[{"model":"qwen-vl-ocr","name":"QwenVL-OCR","contextWindow":38192,"capabilities":["VU"]},{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","contextWindow":34096,"capabilities":["VU"]},{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","contextWindow":38192,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-ocr.json","maxContextWindow":38192} -{"slug":"qwen-vl-plus","name":"Qwen-VL-Plus","description":"Qwen-VL-Plus，即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-vl-plus","name":"QwenVL-Plus","contextWindow":131072,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-plus.json","maxContextWindow":131072} -{"slug":"qwen-voice-design","name":"Qwen-声音设计","description":"Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-design","name":"Qwen-Voice-Design","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-design.json"} -{"slug":"qwen-voice-enrollment","name":"Qwen-声音复刻","description":"千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-enrollment.json"} -{"slug":"qwen2.5","name":"Qwen2.5-开源模型","description":"Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","contextWindow":32768,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen2.5.json","maxContextWindow":32768} -{"slug":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","description":"Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","capabilities":["ASR"]}],"detailPath":"groups/qwen3-asr-flash-filetrans.json"} -{"slug":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","description":"Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","capabilities":["Realtime-ASR"]}],"detailPath":"groups/qwen3-asr-flash-realtime.json"} -{"slug":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","description":"Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","capabilities":["ASR"]}],"detailPath":"groups/qwen3-asr-flash.json"} -{"slug":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","description":"基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-30b-a3b-instruct.json","maxContextWindow":262144} -{"slug":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","description":"基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-480b-a35b-instruct.json","maxContextWindow":262144} -{"slug":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","description":"基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-flash.json","maxContextWindow":1000000} -{"slug":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","description":"基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-plus.json","maxContextWindow":1000000} -{"slug":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","description":"Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3-livetranslate-flash-realtime.json","maxContextWindow":53248} -{"slug":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","description":"Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","contextWindow":53248,"capabilities":["Realtime-ASR"]}],"detailPath":"groups/qwen3-livetranslate-flash.json","maxContextWindow":53248} -{"slug":"qwen3-max","name":"Qwen3-Max","description":"千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3-max","name":"Qwen3-Max","contextWindow":262144,"capabilities":["TG","Reasoning"]},{"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","contextWindow":262144,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3-max.json","maxContextWindow":262144} -{"slug":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","description":"千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","contextWindow":65536,"capabilities":["ASR"]}],"detailPath":"groups/qwen3-omni-30b-a3b-captioner.json","maxContextWindow":65536} -{"slug":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","description":"Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","contextWindow":65536,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3-omni-flash-realtime.json","maxContextWindow":65536} -{"slug":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","description":"Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","contextWindow":65536,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3-omni-flash.json","maxContextWindow":65536} -{"slug":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","description":"Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash-realtime.json"} -{"slug":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash.json"} -{"slug":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","description":"通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} -{"slug":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-instruct-flash.json"} -{"slug":"qwen3-tts-vc-realtime","name":"Qwen3-TTS-VC-Realtime","description":"Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vc-realtime.json"} -{"slug":"qwen3-tts-vc","name":"Qwen3-TTS-VC","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vc.json"} -{"slug":"qwen3-tts-vd-realtime","name":"Qwen3-TTS-VD-Realtime","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd-realtime.json"} -{"slug":"qwen3-tts-vd","name":"Qwen3-TTS-VD","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd.json"} -{"slug":"qwen3-vl-flash","name":"Qwen3-VL-Flash","description":"Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-flash.json","maxContextWindow":262144} -{"slug":"qwen3-vl-plus","name":"Qwen3-VL-Plus","description":"Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-plus.json","maxContextWindow":262144} -{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} -{"slug":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","description":"Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json","maxContextWindow":53248} -{"slug":"qwen3.5-ocr","name":"Qwen3.5-OCR","description":"Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","contextWindow":65536,"capabilities":["VU"]}],"detailPath":"groups/qwen3.5-ocr.json","maxContextWindow":65536} -{"slug":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash-realtime.json","maxContextWindow":262144} -{"slug":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash.json","maxContextWindow":262144} -{"slug":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus-realtime.json","maxContextWindow":262144} -{"slug":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus.json","maxContextWindow":262144} +{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG","VU"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} {"slug":"qwen3.5-plus","name":"Qwen3.5-Plus","description":"Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-plus","name":"Qwen3.5-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.5-plus.json","maxContextWindow":1000000} -{"slug":"qwen3.5","name":"Qwen3.5开源模型","description":"Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":4,"items":[{"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-27b","name":"Qwen3.5-27B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-35b-a3b","name":"Qwen3.5-35B-A3B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-397b-a17b","name":"Qwen3.5-397B-A17B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.5.json","maxContextWindow":262144} {"slug":"qwen3.6-flash","name":"Qwen3.6-Flash","description":"Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-flash","name":"Qwen3.6-Flash","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.6-flash.json","maxContextWindow":1000000} {"slug":"qwen3.6-max","name":"Qwen3.6-Max","description":"Qwen3.6原生Max模型，相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提升；长尾知识能力进一步升级。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","contextWindow":262144,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen3.6-max.json","maxContextWindow":262144} {"slug":"qwen3.6-plus","name":"Qwen3.6-Plus","description":"Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-plus","name":"Qwen3.6-Plus","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.6-plus.json","maxContextWindow":1000000} {"slug":"qwen3.6","name":"Qwen3.6开源模型","description":"Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。","primaryCapability":"VU","capabilities":["VU","TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.6-27b","name":"Qwen3.6-27B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","contextWindow":262144,"capabilities":["VU","TG","Reasoning"]}],"detailPath":"groups/qwen3.6.json","maxContextWindow":262144} {"slug":"qwen3.7-max","name":"Qwen3.7-Max","description":"Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.7-max","name":"Qwen3.7-Max","contextWindow":1000000,"capabilities":["Reasoning","TG"]},{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","contextWindow":1000000,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3.7-max.json","maxContextWindow":1000000} {"slug":"qwen3.7-plus","name":"Qwen3.7-Plus","description":"Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.7-plus","name":"Qwen3.7-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.7-plus.json","maxContextWindow":1000000} -{"slug":"qwen3","name":"Qwen3开源模型","description":"Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。","primaryCapability":"VU","capabilities":["VU","Reasoning","TG"],"providers":["qwen"],"itemCount":20,"items":[{"model":"qwen3-14b","name":"Qwen3-14B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","contextWindow":81920,"capabilities":["Reasoning","TG"]},{"model":"qwen3-32b","name":"Qwen3-32B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-8b","name":"Qwen3-8B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","contextWindow":262144,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3.json","maxContextWindow":262144} -{"slug":"qwq-plus","name":"Qwen-QwQ-Plus","description":"千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。","primaryCapability":"Reasoning","capabilities":["Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwq-plus","name":"QwQ-Plus","contextWindow":131072,"capabilities":["Reasoning"]}],"detailPath":"groups/qwq-plus.json","maxContextWindow":131072} -{"slug":"sambert","name":"Sambert语音合成","description":"提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":43,"items":[{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","capabilities":["TTS"]},{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","capabilities":["TTS"]},{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","capabilities":["TTS"]},{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","capabilities":["TTS"]},{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","capabilities":["TTS"]},{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","capabilities":["TTS"]},{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","capabilities":["TTS"]},{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","capabilities":["TTS"]},{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","capabilities":["TTS"]},{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","capabilities":["TTS"]},{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","capabilities":["TTS"]},{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","capabilities":["TTS"]},{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","capabilities":["TTS"]},{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","capabilities":["TTS"]},{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","capabilities":["TTS"]},{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","capabilities":["TTS"]},{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","capabilities":["TTS"]},{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","capabilities":["TTS"]},{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","capabilities":["TTS"]},{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","capabilities":["TTS"]},{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","capabilities":["TTS"]},{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","capabilities":["TTS"]},{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","capabilities":["TTS"]},{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","capabilities":["TTS"]},{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","capabilities":["TTS"]},{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","capabilities":["TTS"]},{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","capabilities":["TTS"]},{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","capabilities":["TTS"]},{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","capabilities":["TTS"]},{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","capabilities":["TTS"]},{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","capabilities":["TTS"]},{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","capabilities":["TTS"]},{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","capabilities":["TTS"]},{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","capabilities":["TTS"]},{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","capabilities":["TTS"]},{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","capabilities":["TTS"]},{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","capabilities":["TTS"]},{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","capabilities":["TTS"]},{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","capabilities":["TTS"]},{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","capabilities":["TTS"]},{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","capabilities":["TTS"]},{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","capabilities":["TTS"]},{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","capabilities":["TTS"]}],"detailPath":"groups/sambert.json"} -{"slug":"shoemodel-v1","name":"鞋靴模特","description":"鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"shoemodel-v1","name":"鞋靴模特","capabilities":["IG"]}],"detailPath":"groups/shoemodel-v1.json"} -{"slug":"siliconflow-models","name":"SiliconFlow DeepSeek","description":"由硅基流动提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["deepseek"],"itemCount":4,"items":[{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","contextWindow":163840,"capabilities":["TG"]},{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","contextWindow":163840,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/siliconflow-models.json","maxContextWindow":163840} -{"slug":"speech-biasing","name":"语音识别热词","description":"热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"speech-biasing","name":"语音识别热词","capabilities":["ASR"]}],"detailPath":"groups/speech-biasing.json"} -{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} -{"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192} -{"slug":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","description":"通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768} -{"slug":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","description":"通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768} -{"slug":"tripo-models-market-place","name":"Tripo","description":"AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。","primaryCapability":"3D-generation","capabilities":["3D-generation"],"providers":["tripo"],"itemCount":2,"items":[{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","capabilities":["3D-generation"]},{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","capabilities":["3D-generation"]}],"detailPath":"groups/tripo-models-market-place.json"} -{"slug":"vanchin-models-market-place","name":"Vanchin DeepSeek","description":"由快手万擎提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["deepseek"],"itemCount":6,"items":[{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","contextWindow":8192,"capabilities":["VU","TG"]},{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","contextWindow":131072,"capabilities":["TG"]},{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/vanchin-models-market-place.json","maxContextWindow":1048576} -{"slug":"video-style-transform","name":"视频风格重绘","description":"视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"video-style-transform","name":"视频风格重绘","capabilities":["VG"]}],"detailPath":"groups/video-style-transform.json"} -{"slug":"videoretalk","name":"声动人像VideoRetalk","description":"VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"videoretalk","name":"声动人像VideoRetalk","capabilities":["VG"]}],"detailPath":"groups/videoretalk.json"} -{"slug":"vidu-image-models-market-place","name":"Vidu AI生图","description":"由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。","primaryCapability":"IG","capabilities":["IG"],"providers":["vidu"],"itemCount":4,"items":[{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","capabilities":["IG"]},{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","capabilities":["IG"]}],"detailPath":"groups/vidu-image-models-market-place.json"} -{"slug":"vidu-models-market-place","name":"Vidu AI生视频","description":"由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。","primaryCapability":"VG","capabilities":["VG"],"providers":["vidu"],"itemCount":20,"items":[{"model":"vidu/viduq2_reference2video","name":"ViduQ2_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2_text2video","name":"ViduQ2_text2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_img2video","name":"ViduQ2-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_reference2video","name":"ViduQ2-Pro_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_start-end2video","name":"ViduQ2-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro-fast_img2video","name":"ViduQ2-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_img2video","name":"ViduQ2-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_start-end2video","name":"ViduQ2-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3_reference2video","name":"ViduQ3_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-ad_reference2video","name":"ViduQ3-Ad_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-drama_reference2video","name":"ViduQ3-Drama_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-mix_reference2video","name":"ViduQ3-mix_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_img2video","name":"ViduQ3-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_start-end2video","name":"ViduQ3-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_text2video","name":"ViduQ3-Pro_text2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro-fast_img2video","name":"ViduQ3-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_img2video","name":"ViduQ3-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_reference2video","name":"ViduQ3-Turbo_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_start-end2video","name":"ViduQ3-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","capabilities":["VG"]}],"detailPath":"groups/vidu-models-market-place.json"} -{"slug":"virtualmodel-v2","name":"虚拟模特V2","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"virtualmodel-v2","name":"虚拟模特V2","capabilities":["IG"]}],"detailPath":"groups/virtualmodel-v2.json"} -{"slug":"voice-enrollment","name":"大模型声音复刻及声音设计","description":"大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","capabilities":["TTS"]}],"detailPath":"groups/voice-enrollment.json"} -{"slug":"wan-image-edit","name":"Wan-Image","description":"指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":5,"items":[{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","capabilities":["IG"]},{"model":"wan2.6-image","name":"Wan2.6-Image","capabilities":["IG"]},{"model":"wan2.7-image","name":"Wan2.7-Image","capabilities":["IG"]},{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","capabilities":["IG"]},{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","capabilities":["IG"]}],"detailPath":"groups/wan-image-edit.json"} -{"slug":"wan-image-to-video","name":"Wan-I2V","description":"图片生成视频内容，稳定保持图像主体、风格和文字等细节信息","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":14,"items":[{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","capabilities":["VG"]},{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","capabilities":["VG"]},{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","capabilities":["VG"]},{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","capabilities":["VG"]},{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","capabilities":["VG"]},{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","capabilities":["VG"]},{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","capabilities":["VG"]},{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","capabilities":["VG"]},{"model":"wan2.6-i2v","name":"Wan2.6-I2V","capabilities":["VG"]},{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","capabilities":["VG"]},{"model":"wan2.7-i2v","name":"Wan2.7-I2V","capabilities":["VG"]},{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","capabilities":["VG"]},{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","capabilities":["VG"]},{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","capabilities":["VG"]}],"detailPath":"groups/wan-image-to-video.json"} -{"slug":"wan-reference-to-video","name":"Wan-R2V","description":"参考视频中的人或物，精准保持形象和声音，支持多参考合拍","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":3,"items":[{"model":"wan2.6-r2v","name":"Wan2.6-R2V","capabilities":["VG"]},{"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","capabilities":["VG"]},{"model":"wan2.7-r2v","name":"Wan2.7-R2V","capabilities":["VG"]}],"detailPath":"groups/wan-reference-to-video.json"} -{"slug":"wan-text-to-image","name":"Wan-T2I","description":"文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":9,"items":[{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","capabilities":["IG"]},{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","capabilities":["IG"]},{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","capabilities":["IG"]},{"model":"wan2.6-t2i","name":"Wan2.6-T2I","capabilities":["IG"]},{"model":"wanx-v1","name":"wanx-t2i","capabilities":["IG"]},{"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","capabilities":["IG"]},{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","capabilities":["IG"]},{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","capabilities":["IG"]},{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","capabilities":["IG"]}],"detailPath":"groups/wan-text-to-image.json"} -{"slug":"wan-text-to-video","name":"Wan-T2V","description":"文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":6,"items":[{"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","capabilities":["VG"]},{"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","capabilities":["VG"]},{"model":"wan2.6-t2v","name":"Wan2.6-T2V","capabilities":["VG"]},{"model":"wan2.7-t2v","name":"Wan2.7-T2V","capabilities":["VG"]},{"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","capabilities":["VG"]},{"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","capabilities":["VG"]}],"detailPath":"groups/wan-text-to-video.json"} -{"slug":"wan-video-edit","name":"Wan-VideoEdit","description":"通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","capabilities":["VG"]}],"detailPath":"groups/wan-video-edit.json"} -{"slug":"wanx-background-generation-v2","name":"图像背景生成","description":"图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-background-generation-v2","name":"图像背景生成","capabilities":["IG"]}],"detailPath":"groups/wanx-background-generation-v2.json"} -{"slug":"wanx-poster-generation-v1","name":"创意海报生成","description":"创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-poster-generation-v1","name":"创意海报生成","capabilities":["IG"]}],"detailPath":"groups/wanx-poster-generation-v1.json"} -{"slug":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","description":"万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","capabilities":["IG"]}],"detailPath":"groups/wanx-sketch-to-image-lite.json"} -{"slug":"wanx-style-repaint-v1","name":"人像风格重绘","description":"人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-style-repaint-v1","name":"人像风格重绘","capabilities":["IG"]}],"detailPath":"groups/wanx-style-repaint-v1.json"} -{"slug":"wanx-virtualmodel","name":"虚拟模特","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-virtualmodel","name":"虚拟模特","capabilities":["IG"]}],"detailPath":"groups/wanx-virtualmodel.json"} -{"slug":"wanx-x-painting","name":"万相-图像局部重绘","description":"万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-x-painting","name":"万相-图像局部重绘","capabilities":["IG"]}],"detailPath":"groups/wanx-x-painting.json"} -{"slug":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","description":"万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","capabilities":["VG"]}],"detailPath":"groups/wanx2.1-vace-plus.json"} -{"slug":"wordart-semantic","name":"WordArt锦书-文字变形","description":"WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"wordart-semantic","name":"WordArt锦书-文字变形","capabilities":["IG"]}],"detailPath":"groups/wordart-semantic.json"} -{"slug":"wordart-texture","name":"WordArt锦书-文字纹理生成","description":"WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","capabilities":["IG"]}],"detailPath":"groups/wordart-texture.json"} -{"slug":"xiaomi-models-market-place","name":"MiMo文本模型","description":"由小米MiMo提供的MiMo文本模型API服务","primaryCapability":"TG","capabilities":["TG"],"providers":["xiaomi"],"itemCount":1,"items":[{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/xiaomi-models-market-place.json","maxContextWindow":1048576} -{"slug":"z-image-turbo","name":"Z-Image-Turbo","description":"Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"z-image-turbo","name":"Z-Image-Turbo","capabilities":["IG"]}],"detailPath":"groups/z-image-turbo.json"} -{"slug":"zhipu-models-market-place","name":"智谱GLM系列文本模型","description":"由智谱提供的GLM系列文本模型API服务","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["zhipu-ai"],"itemCount":3,"items":[{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","contextWindow":204800,"capabilities":["TG"]},{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","contextWindow":1048576,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/zhipu-models-market-place.json","maxContextWindow":1048576} diff --git a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json deleted file mode 100644 index 15fdc4d5..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json +++ /dev/null @@ -1,498 +0,0 @@ -{ - "name": "Kimi", - "description": "Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。", - "features": [ - "cache", - "function-calling", - "model-experience", - "structured-outputs", - "web-search", - "prefix-completion" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi-k2.7-code", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "VU", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-06-14T16:59:14.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 229376, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "kimi-k2.7-code", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -s -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Image", - "Text", - "Video" - ] - }, - "description": "kimi-k2.6是Kimi最新最智能的模型，具备更强更稳的长程代码编写能力，指令遵循和自我纠错能力显著提升，同时支持文本、图片与视频输入，思考与非思考模式，对话与Agent任务。", - "features": [ - "cache", - "function-calling", - "model-experience" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi-k2.6", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-04-21T09:55:34.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 229376, - "inferenceProvider": "bailian", - "name": "Kimi-K2.6", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi-k2.5", - "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-01-30T01:49:03.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 229376, - "inferenceProvider": "bailian", - "name": "Kimi-K2.5", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "kimi-k2-thinking模型是月之暗面提供的具有通用 Agentic能力和推理能力的思考模型，它擅长深度推理，并可通过多步工具调用，帮助解决各类难题。", - "features": [ - "model-experience", - "cache", - "function-calling" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi-k2-thinking", - "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-11-10T07:24:34.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 229376, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Kimi-K2-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Kimi-K2是月之暗面提供的国内首个开源万亿参数MoE模型，激活参数达 320 亿，具有卓越的编码和工具调用能力。", - "features": [ - "model-experience", - "cache", - "function-calling" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "Moonshot-Kimi-K2-Instruct", - "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-07-16T14:56:31.000+00:00", - "contextWindow": 131072, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Moonshot-Kimi-K2-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json deleted file mode 100644 index bd2b5e97..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json +++ /dev/null @@ -1,178 +0,0 @@ -{ - "name": "MiniMax", - "description": "MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.1", - "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 172032, - "inferenceProvider": "bailian", - "name": "MiniMax-M2.1", - "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.5", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-02-24T15:07:02.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 196608, - "inferenceProvider": "bailian", - "name": "MiniMax-M2.5", - "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json deleted file mode 100644 index abeefc7d..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json +++ /dev/null @@ -1,222 +0,0 @@ -{ - "name": "MiniMax-Speech系列语音模型", - "description": "由MiniMax提供的MiniMax-Speech系列语音模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", - "features": [], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/speech-2.8-turbo", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T08:44:53.000+00:00", - "maxInputTokens": 10000, - "inferenceProvider": "mini-max", - "name": "speech-2.8-turbo", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-2.8-turbo\"\n }'", - "docUrl": "https://help.aliyun.com/document_detail/3021951.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", - "features": [], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/speech-2.8-hd", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T08:46:11.000+00:00", - "maxInputTokens": 10000, - "inferenceProvider": "mini-max", - "name": "speech-2.8-hd", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-2.8-hd\"\n }'", - "docUrl": "https://help.aliyun.com/document_detail/3021951.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", - "features": [], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/speech-02-turbo", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T08:44:37.000+00:00", - "maxInputTokens": 10000, - "inferenceProvider": "mini-max", - "name": "speech-02-turbo", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-02-turbo\"\n }'", - "docUrl": "https://help.aliyun.com/document_detail/3021951.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", - "features": [], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/speech-02-hd", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000, - "usage_limit_field": "characters", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T08:45:04.000+00:00", - "maxInputTokens": 10000, - "inferenceProvider": "mini-max", - "name": "speech-02-hd", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-02-hd\"\n }'", - "docUrl": "https://help.aliyun.com/document_detail/3021951.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json deleted file mode 100644 index 261b3171..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "AI试衣OutfitAnyone-图片分割", - "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "aitryon-parsing-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-15T13:13:06.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣OutfitAnyone-图片分割", - "docUrl": "https://help.aliyun.com/document_detail/2865249.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json deleted file mode 100644 index 2516bb8d..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "AI试衣-Plus版", - "description": "aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "aitryon-plus", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-04-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣-Plus版", - "docUrl": "https://help.aliyun.com/document_detail/2881846.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json deleted file mode 100644 index 5fa687a4..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json +++ /dev/null @@ -1,52 +0,0 @@ -{ - "name": "AI试衣OutfitAnyone-图片精修", - "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "aitryon-refiner", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣OutfitAnyone-图片精修", - "docUrl": "https://help.aliyun.com/document_detail/2796663.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json deleted file mode 100644 index a1747bfc..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "AI试衣-基础版", - "description": "aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "aitryon", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-05-24T10:29:05.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣-基础版", - "docUrl": "https://help.aliyun.com/document_detail/2796626.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json deleted file mode 100644 index 5f83317b..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json +++ /dev/null @@ -1,63 +0,0 @@ -{ - "name": "舞动人像AnimateAnyone-detect", - "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "animate-anyone-detect-gen2", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-10T06:28:33.000+00:00", - "inferenceProvider": "bailian", - "name": "舞动人像AnimateAnyone-detect", - "docUrl": "https://help.aliyun.com/document_detail/2786465.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json deleted file mode 100644 index 49465b13..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "舞动人像AnimateAnyone", - "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "animate-anyone-gen2", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-10T06:25:42.000+00:00", - "inferenceProvider": "bailian", - "name": "舞动人像AnimateAnyone", - "docUrl": "https://help.aliyun.com/document_detail/2786464.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json deleted file mode 100644 index a6232ced..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "舞动人像AnimateAnyone-template", - "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Video" - ] - }, - "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "animate-anyone-template-gen2", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-10T06:27:17.000+00:00", - "inferenceProvider": "bailian", - "name": "舞动人像AnimateAnyone-template", - "docUrl": "https://help.aliyun.com/document_detail/2807955.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json deleted file mode 100644 index 62f0c0f7..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json +++ /dev/null @@ -1,432 +0,0 @@ -{ - "name": "CosyVoice大模型", - "description": "基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "CosyVoice-v3.5-Flash是通义实验室CosyVoice系列的高性能语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", - "collectionTag": "", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v3.5-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-02-27T09:13:23.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成CosyVoice-v3.5-flash大模型", - "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v3.5-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-02-27T09:13:30.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成CosyVoice-v3.5-plus大模型", - "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v3-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-11-17T06:35:36.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成CosyVoice-v3-flash大模型", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v3-flash\"\nvoice = \"longanyang\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v3-flash\";\n private static String voice = \"longanyang\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v3-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-03T01:45:34.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成CosyVoice-v3-plus大模型", - "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "情感", - "key": "emotion", - "default": "neutral", - "tip": "合成音频说话的情感" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v3-plus\"\nvoice = \"longanyang\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v3-plus\";\n private static String voice = \"longanyang\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-clone-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "声音复刻CosyVoice大模型", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", - "inferenceProvider": "bailian", - "name": "语音合成CosyVoice大模型", - "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v2", - "capabilities": [ - "TTS" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成cosyvoice-v2大模型", - "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "字级别时间戳", - "key": "enableWordTimestamp", - "default": false - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json index 00cd5780..da8f4c8d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json +++ b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json @@ -19,6 +19,9 @@ "web-search" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v4-pro", "prices": [ { @@ -71,73 +74,20 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V4-Pro", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-pro\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-pro\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-pro\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-pro\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -160,6 +110,9 @@ "web-search" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v4-flash", "prices": [ { @@ -213,73 +166,20 @@ "name": "DeepSeek-V4-Flash", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -303,6 +203,9 @@ "batch" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v3.2", "prices": [ { @@ -398,63 +301,20 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -476,6 +336,9 @@ "web-search" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v3.2-exp", "prices": [ { @@ -527,63 +390,20 @@ "inferenceProvider": "aliyun-bailian", "name": "Deepseek-V3.2-Exp", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2-exp\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2-exp',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2-exp\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2-exp',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2-exp\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2-exp\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2-exp\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2-exp\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -606,6 +426,9 @@ "cache" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v3.1", "prices": [ { @@ -663,63 +486,20 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -742,6 +522,9 @@ "cache" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-v3", "prices": [ { @@ -811,57 +594,20 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"deepseek-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"deepseek-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"deepseek-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"deepseek-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -883,6 +629,9 @@ "web-search" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1-0528", "prices": [ { @@ -933,36 +682,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-0528\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-0528',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-0528\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-0528',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-0528\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-0528\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-0528\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-0528\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -984,6 +716,9 @@ "cache" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1", "prices": [ { @@ -1052,36 +787,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1100,6 +818,9 @@ "model-experience" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1-distill-qwen-7b", "prices": [ { @@ -1150,36 +871,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-7B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-7b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-7b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-7b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-7b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1198,6 +902,9 @@ "model-experience" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1-distill-qwen-32b", "prices": [ { @@ -1248,36 +955,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-32B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-32b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-32b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-32b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-32b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1296,6 +986,9 @@ "model-experience" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1-distill-qwen-14b", "prices": [ { @@ -1346,36 +1039,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-14B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-14b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-14b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-14b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-14b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1394,6 +1070,9 @@ "model-experience" ], "provider": "deepseek", + "limit": { + "message": "model not exist" + }, "model": "deepseek-r1-distill-qwen-1.5b", "qpmInfo": { "model-default-actual": { @@ -1424,36 +1103,19 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-1.5b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-1.5b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-1.5b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-1.5b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/embedding.json b/skills/bailian-docs-llm-wiki/models/groups/embedding.json deleted file mode 100644 index 387fe2bf..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/embedding.json +++ /dev/null @@ -1,246 +0,0 @@ -{ - "name": "通义多模态向量", - "description": "基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "tongyi-embedding-vision-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "ME" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-23T09:09:49.000+00:00", - "inferenceProvider": "bailian", - "name": "视觉向量-flash", - "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-flash\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-flash\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "tongyi-embedding-vision-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "ME" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-23T09:09:31.000+00:00", - "inferenceProvider": "bailian", - "name": "视觉向量-plus", - "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-plus\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-plus\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "multimodal-embedding-v1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "ME" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 0, - "latestOnlineAt": "2024-12-23T11:55:31.000+00:00", - "contextWindow": 0, - "maxInputTokens": 512, - "inferenceProvider": "bailian", - "name": "通用多模态向量", - "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"multimodal-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"multimodal-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json deleted file mode 100644 index 114ba204..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json +++ /dev/null @@ -1,63 +0,0 @@ -{ - "name": "悦动人像EMO-detect", - "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "emo-detect-v1", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-11-07T14:16:51.000+00:00", - "inferenceProvider": "bailian", - "name": "悦动人像EMO-detect", - "docUrl": "https://help.aliyun.com/document_detail/2786463.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json deleted file mode 100644 index d2b6cb99..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "悦动人像EMO", - "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "emo-v1", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-11-07T14:16:42.000+00:00", - "inferenceProvider": "bailian", - "name": "悦动人像EMO", - "docUrl": "https://help.aliyun.com/document_detail/2786461.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json deleted file mode 100644 index ae9f1a1a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json +++ /dev/null @@ -1,63 +0,0 @@ -{ - "name": "表情包Emoji-detect", - "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "emoji-detect-v1", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-16T09:55:26.000+00:00", - "inferenceProvider": "bailian", - "name": "表情包Emoji-detect", - "docUrl": "https://help.aliyun.com/document_detail/2865371.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json deleted file mode 100644 index 5c8c9829..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "表情包Emoji", - "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "emoji-v1", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-16T09:46:00.000+00:00", - "inferenceProvider": "bailian", - "name": "表情包Emoji", - "docUrl": "https://help.aliyun.com/document_detail/2865374.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json deleted file mode 100644 index b5948c96..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json +++ /dev/null @@ -1,56 +0,0 @@ -{ - "name": "FaceChain人物图像检测", - "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "facechain-facedetect", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T08:10:47.000+00:00", - "inferenceProvider": "bailian", - "name": "FaceChain人物图像检测", - "docUrl": "https://help.aliyun.com/document_detail/2712507.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json deleted file mode 100644 index d32c1391..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json +++ /dev/null @@ -1,58 +0,0 @@ -{ - "name": "FaceChain人物写真生成", - "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "facechain-generation", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T08:12:23.000+00:00", - "inferenceProvider": "bailian", - "name": "FaceChain人物写真生成", - "docUrl": "https://help.aliyun.com/document_detail/2712501.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json deleted file mode 100644 index ec9bd5b0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json +++ /dev/null @@ -1,99 +0,0 @@ -{ - "name": "通义法睿-Plus-32K", - "description": "通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "farui-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "output_tokens", - "count_limit": 4, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "output_tokens", - "count_limit": 4, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 2000, - "latestOnlineAt": "2024-05-14T13:32:40.000+00:00", - "contextWindow": 12000, - "maxInputTokens": 12000, - "inferenceProvider": "bailian", - "name": "通义法睿-Plus-32K", - "docUrl": "https://help.aliyun.com/document_detail/2778998.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json deleted file mode 100644 index 14412c98..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Fun-ASR-Flash", - "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "fun-asr-flash-2026-06-15", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-06-17T08:56:52.000+00:00", - "inferenceProvider": "bailian", - "name": "Fun-ASR-Flash-2026-06-15", - "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/2869541.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json deleted file mode 100644 index 79fb2a3d..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json +++ /dev/null @@ -1,144 +0,0 @@ -{ - "name": "Fun-ASR实时语音识别", - "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "fun-asr-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-23T11:05:02.000+00:00", - "inferenceProvider": "bailian", - "name": "Fun-ASR实时语音识别", - "docUrl": "https://help.aliyun.com/document_detail/2842554.html", - "predictConfig": [ - { - "name": "开启语义断句", - "key": "semantic_punctuation_enabled", - "default": false, - "tip": "开启语义断句后则将关闭VAD(语音活动检测)断句，具体见说明文档" - }, - { - "name": "VAD静音阈值", - "key": "max_sentence_silence", - "default": 1300, - "tip": "VAD(语音活动检测)断句的静音时长阈值(单位为ms)", - "range": [ - 200, - 6000 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='fun-asr-realtime',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"fun-asr-realtime\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "fun-asr-flash-8k-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-ASR" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "equivalentSnapshot": "fun-asr-flash-8k-realtime-2026-01-28", - "latestOnlineAt": "2026-02-12T11:34:14.000+00:00", - "inferenceProvider": "bailian", - "name": "Fun-ASR-Flash-8k实时语音识别", - "docUrl": "https://help.aliyun.com/document_detail/2842554.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='fun-asr-flash-8k-realtime',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"fun-asr-flash-8k-realtime\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json deleted file mode 100644 index eac84c5d..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json +++ /dev/null @@ -1,96 +0,0 @@ -{ - "name": "Fun-ASR语音识别", - "description": "通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "fun-asr", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-11-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Fun-ASR语音识别", - "docUrl": "https://help.aliyun.com/document_detail/2880903.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport dashscope\nimport os\nimport json\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ntask_response = Transcription.async_call(\n model='fun-asr',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "fun-asr-mtl", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-25T06:03:29.000+00:00", - "inferenceProvider": "bailian", - "name": "Fun-ASR-MTL", - "docUrl": "https://help.aliyun.com/document_detail/2978300.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport dashscope\nimport os\nimport json\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ntask_response = Transcription.async_call(\n model='fun-asr-mtl',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json deleted file mode 100644 index 3751a275..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json +++ /dev/null @@ -1,151 +0,0 @@ -{ - "name": "音乐生成", - "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "fun-music-v1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-05-06T12:15:28.000+00:00", - "inferenceProvider": "bailian", - "name": "音乐生成", - "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030448.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "fun-music-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-06-01T07:20:31.000+00:00", - "inferenceProvider": "bailian", - "name": "音乐生成 Preview", - "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030448.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json deleted file mode 100644 index ad33f246..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json +++ /dev/null @@ -1,836 +0,0 @@ -{ - "name": "GLM", - "description": "GLM是由智谱提供的开源模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。", - "features": [ - "cache", - "function-calling", - "model-experience", - "structured-outputs" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-5.2", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 2000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 2000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-06-16T08:16:52.000+00:00", - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "GLM-5.2", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。", - "collectionTag": "", - "features": [ - "cache", - "function-calling", - "model-experience", - "structured-outputs" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-5.1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-04-14T11:34:59.000+00:00", - "contextWindow": 202745, - "maxInputTokens": 202745, - "inferenceProvider": "bailian", - "name": "GLM-5.1", - "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.7", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", - "contextWindow": 202752, - "maxInputTokens": 169984, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "GLM-4.7", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-5", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", - "contextWindow": 202752, - "maxInputTokens": 169984, - "inferenceProvider": "bailian", - "name": "GLM-5", - "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM新一代旗舰模型，核心能力较4.5全面提升。总参数量为3550 亿，激活参数320亿，上下文窗口扩展至200K。", - "features": [ - "model-experience", - "cache" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.6", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-10-21T06:26:06.000+00:00", - "contextWindow": 202752, - "maxInputTokens": 169984, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "GLM-4.6", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-4.5采用混合专家（MoE）架构，总参数量为3550 亿，激活参数320亿，在复杂推理、代码生成及智能体交互等通用能力上实现了能力融合与技术突破。", - "features": [ - "model-experience" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.5", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-08-06T11:38:01.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "inferenceProvider": "bailian", - "name": "GLM-4.5", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-4.5-Air采用混合专家（MoE）架构，总参数量为1060亿，激活参数120亿，相较GLM-4.5更紧凑、轻量，适用于对模型规模和资源消耗有一定限制的场景。", - "features": [ - "model-experience" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.5-air", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-08-06T12:02:34.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "inferenceProvider": "bailian", - "name": "GLM-4.5-Air", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json deleted file mode 100644 index 69f63105..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json +++ /dev/null @@ -1,76 +0,0 @@ -{ - "name": "GLM-5.2-Fast", - "description": "GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。", - "features": [ - "cache", - "function-calling", - "prefix-completion", - "structured-outputs", - "web-search" - ], - "provider": "zhipu-ai", - "model": "glm-5.2-fast-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-07-09T02:50:34.000+00:00", - "inferenceSpeeds": [ - "fast" - ], - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "inferenceProvider": "bailian", - "name": "GLM-5.2-Fast-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "category": "Third-party", - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2-fast-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2-fast-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json deleted file mode 100644 index 3f5f3c4c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json +++ /dev/null @@ -1,98 +0,0 @@ -{ - "name": "GUI-Plus", - "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "gui-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 540000, - "usage_limit_field": "total_tokens", - "count_limit": 80, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 540000, - "usage_limit_field": "total_tokens", - "count_limit": 80, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-11-12T07:17:19.000+00:00", - "contextWindow": 256000, - "maxInputTokens": 254976, - "inferenceProvider": "bailian", - "name": "GUI-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2997010.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", - "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", - "docUrl": "https://help.aliyun.com/document_detail/2997010.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", - "docUrl": "https://help.aliyun.com/document_detail/2997010.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json deleted file mode 100644 index 69390930..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json +++ /dev/null @@ -1,60 +0,0 @@ -{ - "name": "一句话识别及翻译V1.0", - "description": "多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "gummy-chat-v1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-04T04:07:25.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "一句话识别及翻译V1.0", - "docUrl": "https://help.aliyun.com/document_detail/2866122.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "import requests\nfrom http import HTTPStatus\n\nimport dashscope\nfrom dashscope.audio.asr import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nr = requests.get(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\"\n)\nwith open(\"asr_example.wav\", \"wb\") as f:\n f.write(r.content)\n\ntranslator = TranslationRecognizerRealtime(\n model=\"gummy-chat-v1\",\n format=\"wav\",\n sample_rate=16000,\n translation_target_languages=[\"en\"],\n translation_enabled=True,\n callback=None,\n)\nresult = translator.call(\"asr_example.wav\")\nif not result.error_message:\n print(\"request id: \", result.request_id)\n print(\"transcription: \")\n for transcription_result in result.transcription_result_list:\n print(transcription_result.text)\n print(\"translation[en]: \")\n\n for translation_result in result.translation_result_list:\n print(translation_result.get_translation('en').text)\nelse:\n print(\"Error: \", result.error_message)", - "java": "import com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerParam;\nimport com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerRealtime;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranscriptionResult;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationRecognizerResultPack;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationResult;\n\nimport java.io.File;\nimport java.util.ArrayList;\n\npublic class Main {\n\n public static void main(String[] args) {\n String targetLanguage = \"en\";\n // 创建Recognition实例\n TranslationRecognizerRealtime translator = new TranslationRecognizerRealtime();\n // 创建RecognitionParam，请在实际使用中替换真实apiKey\n TranslationRecognizerParam param =\n TranslationRecognizerParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(\"gummy-chat-v1\")\n .format(\"wav\") // 'pcm'、'wav'、'mp3'、'opus'、'speex'、'aac'、'amr', you\n // can check the supported formats in the document\n .sampleRate(16000)\n .transcriptionEnabled(true)\n .sourceLanguage(\"auto\")\n .translationEnabled(true)\n .translationLanguages(new String[] {targetLanguage})\n .build();\n // 直接将结果保存到script.txt中\n TranslationRecognizerResultPack result = translator.call(param, new File(\"hello_world.wav\"));\n // 任务结束后关闭 websocket 连接\n translator.getDuplexApi().close(1000, \"bye\");\n if (result.getError() != null) {\n System.out.println(\"error: \" + result.getError());\n throw new RuntimeException(result.getError());\n } else {\n System.out.println(\"RequestId: \" + result.getRequestId());\n System.out.println(\"Transcription Results:\");\n ArrayList transcriptionResults = result.getTranscriptionResultList();\n for (int i = 0; i < transcriptionResults.size(); i++) {\n System.out.println(transcriptionResults.get(i).getText());\n }\n\n System.out.println(\"English Translation Results:\");\n ArrayList translationResultList = result.getTranslationResultList();\n for (int i = 0; i < translationResultList.size(); i++) {\n System.out.println(translationResultList.get(i).getTranslation(targetLanguage).getText());\n }\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json deleted file mode 100644 index eefec95c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json +++ /dev/null @@ -1,60 +0,0 @@ -{ - "name": "实时语音识别及翻译V1.0", - "description": "多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "gummy-realtime-v1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Audio-Translate" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-04T04:07:10.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "实时语音识别及翻译V1.0", - "docUrl": "https://help.aliyun.com/document_detail/2865393.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "import requests\nfrom http import HTTPStatus\n\nimport dashscope\nfrom dashscope.audio.asr import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nr = requests.get(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\"\n)\nwith open(\"asr_example.wav\", \"wb\") as f:\n f.write(r.content)\n\ntranslator = TranslationRecognizerRealtime(\n model=\"gummy-realtime-v1\",\n format=\"wav\",\n sample_rate=16000,\n translation_target_languages=[\"en\"],\n translation_enabled=True,\n callback=None,\n)\nresult = translator.call(\"asr_example.wav\")\nif not result.error_message:\n print(\"request id: \", result.request_id)\n print(\"transcription: \")\n for transcription_result in result.transcription_result_list:\n print(transcription_result.text)\n print(\"translation[en]: \")\n\n for translation_result in result.translation_result_list:\n print(translation_result.get_translation('en').text)\nelse:\n print(\"Error: \", result.error_message)", - "java": "import com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerParam;\nimport com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerRealtime;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranscriptionResult;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationRecognizerResultPack;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationResult;\n\nimport java.io.File;\nimport java.util.ArrayList;\n\npublic class Main {\n\n public static void main(String[] args) {\n String targetLanguage = \"en\";\n // 创建Recognition实例\n TranslationRecognizerRealtime translator = new TranslationRecognizerRealtime();\n // 创建RecognitionParam，请在实际使用中替换真实apiKey\n TranslationRecognizerParam param =\n TranslationRecognizerParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(\"gummy-realtime-v1\")\n .format(\"wav\") // 'pcm'、'wav'、'mp3'、'opus'、'speex'、'aac'、'amr', you\n // can check the supported formats in the document\n .sampleRate(16000)\n .transcriptionEnabled(true)\n .sourceLanguage(\"auto\")\n .translationEnabled(true)\n .translationLanguages(new String[] {targetLanguage})\n .build();\n // 直接将结果保存到script.txt中\n TranslationRecognizerResultPack result = translator.call(param, new File(\"hello_world.wav\"));\n // 任务结束后关闭 websocket 连接\n translator.getDuplexApi().close(1000, \"bye\");\n if (result.getError() != null) {\n System.out.println(\"error: \" + result.getError());\n throw new RuntimeException(result.getError());\n } else {\n System.out.println(\"RequestId: \" + result.getRequestId());\n System.out.println(\"Transcription Results:\");\n ArrayList transcriptionResults = result.getTranscriptionResultList();\n for (int i = 0; i < transcriptionResults.size(); i++) {\n System.out.println(transcriptionResults.get(i).getText());\n }\n\n System.out.println(\"English Translation Results:\");\n ArrayList translationResultList = result.getTranslationResultList();\n for (int i = 0; i < translationResultList.size(); i++) {\n System.out.println(translationResultList.get(i).getTranslation(targetLanguage).getText());\n }\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json index 2e70e68e..816195fa 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json @@ -12,7 +12,21 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。", "features": [ "model-experience" @@ -22,6 +36,22 @@ "message": "model not exist" }, "model": "happyhorse-1.1-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -45,7 +75,7 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T03:16:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", "category": "Visual", @@ -78,7 +108,7 @@ "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } @@ -103,6 +133,22 @@ "message": "model not exist" }, "model": "happyhorse-1.0-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -127,39 +173,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T17:07:07.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json deleted file mode 100644 index 8b69d7f3..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json +++ /dev/null @@ -1,178 +0,0 @@ -{ - "name": "HappyHorse-R2V", - "description": "HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。", - "features": [ - "model-experience" - ], - "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, - "model": "happyhorse-1.1-r2v", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-06-16T03:16:08.000+00:00", - "inferenceProvider": "bailian", - "name": "HappyHorse-1.1-R2V", - "docUrl": "https://help.aliyun.com/document_detail/3030778.html", - "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", - "docUrl": "https://help.aliyun.com/document_detail/3030778.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。", - "features": [ - "model-experience" - ], - "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, - "model": "happyhorse-1.0-r2v", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 10, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 10, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-26T12:42:22.000+00:00", - "inferenceProvider": "bailian", - "name": "HappyHorse-1.0-R2V", - "docUrl": "https://help.aliyun.com/document_detail/3030778.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", - "docUrl": "https://help.aliyun.com/document_detail/3030778.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json index edb5bce7..f3cc2fea 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json @@ -20,6 +20,22 @@ "message": "model not exist" }, "model": "happyhorse-1.1-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -43,45 +59,14 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T06:14:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } @@ -105,6 +90,22 @@ "message": "model not exist" }, "model": "happyhorse-1.0-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -127,44 +128,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T09:03:16.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json deleted file mode 100644 index 01a4bf23..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json +++ /dev/null @@ -1,88 +0,0 @@ -{ - "name": "HappyHorse-Video-Edit", - "description": "HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Video" - ] - }, - "description": "HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", - "features": [], - "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, - "model": "happyhorse-1.0-video-edit", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 10, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 10, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-26T07:51:50.000+00:00", - "inferenceProvider": "bailian", - "name": "HappyHorse-1.0-Video-Edit", - "docUrl": "https://help.aliyun.com/document_detail/3030779.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030779.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json deleted file mode 100644 index 05868c52..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "图像擦除补全", - "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "image-erase-completion", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-08-19T01:17:02.000+00:00", - "inferenceProvider": "bailian", - "name": "图像擦除补全", - "docUrl": "https://help.aliyun.com/document_detail/2840907.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json deleted file mode 100644 index e90a0e3b..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "人物实例分割", - "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "image-instance-segmentation", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-08-19T01:17:00.000+00:00", - "inferenceProvider": "bailian", - "name": "人物实例分割", - "docUrl": "https://help.aliyun.com/document_detail/2840906.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://image-instance-segmentation/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"image-instance-segmentation\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN01nC4QEU1x58LUeMjRL_!!6000000006391-49-tps-1590-1060.webp\"\n },\n \"parameters\":{\n }\n}'\n\ncurl -X GET \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\nhttps://image-instance-segmentation/api/v1/tasks/53950fb7-281a-4e60-xxxxxxxxxxxx" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json deleted file mode 100644 index e6182d40..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "图像画面扩展", - "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "image-out-painting", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-05-24T10:29:57.000+00:00", - "inferenceProvider": "bailian", - "name": "图像画面扩展", - "docUrl": "https://help.aliyun.com/document_detail/2796845.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json deleted file mode 100644 index 1646ab45..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json +++ /dev/null @@ -1,352 +0,0 @@ -{ - "name": "Kimi", - "description": "由月之暗面提供的Kimi系列模型的API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "K2.7 Code高速版与普通版是同一个模型，但输出速度约为普通版的 5-6 倍，常规编程场景下（取输入长度中位数）输出速度约 180 Token/s，短上下文场景可达 260 Token/s ，带来更极致的编程体验。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi/kimi-k2.7-code-highspeed", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 262144, - "latestOnlineAt": "2026-06-17T11:23:33.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 262144, - "offlineInfo": {}, - "inferenceProvider": "moonshot-ai", - "name": "kimi/kimi-k2.7-code-highspeed", - "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.7-code-highspeed\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.7-code-highspeed\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.7-code-highspeed\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Kimi K2.7 Code 是月之暗面 Kimi发布并开源的新一代编程专用模型，定位为 Kimi 迄今最智能的 Coding 模型。Kimi K2.7 Code 是一个以编码为中心的智能体模型（coding-focused agentic model），专为长程软件工程任务优化。它擅长跨多文件重构、功能实现、长会话调试等需要可靠指令遵循和端到端完成率的复杂工作流。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi/kimi-k2.7-code", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "VU", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 262144, - "latestOnlineAt": "2026-06-15T01:49:13.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 262144, - "offlineInfo": {}, - "inferenceProvider": "moonshot-ai", - "name": "kimi/kimi-k2.7-code", - "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.7-code\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.7-code\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Kimi K2.6 是 Kimi 最新最智能的模型，Kimi K2.6 的通用 Agent、代码、视觉理解等综合能力得到全面提升，其中在博士级难度的完整版人类最后的考试（Humanity’s Last Exam）、在考察模型真实软件工程能力的 SWE-Bench Pro、评估 Agent 深度检索能力的 DeepSearchQA 等基准测试中均取得行业领先的成绩，同时支持文本、图片与视频输入，思考与非思考模式，对话与 Agent 任务。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi/kimi-k2.6", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 262144, - "latestOnlineAt": "2026-04-26T08:54:45.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 262144, - "offlineInfo": {}, - "inferenceProvider": "moonshot-ai", - "name": "Kimi/Kimi K2.6", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Kimi K2.5 是 Kimi 在2026年最新推出的智能模型，在 Agent、代码、视觉理解及一系列通用智能任务上取得开源 SoTA 表现。同时 Kimi K2.5 也是 Kimi 迄今最全能的模型，原生的多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与 Agent 任务。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, - "model": "kimi/kimi-k2.5", - "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 262144, - "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 262144, - "offlineInfo": {}, - "inferenceProvider": "moonshot-ai", - "name": "Kimi/Kimi K2.5", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json deleted file mode 100644 index 4a04f994..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json +++ /dev/null @@ -1,297 +0,0 @@ -{ - "name": "可灵AI", - "description": "由可灵AI提供的高质量视频与图像生成及编辑模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "智能分镜可读懂剧本场景流转，自动调度机位和景别。原生多模态框架支持音画一致性。打破时长限制，多镜头故事创作更自由。", - "features": [], - "provider": "kling", - "limit": { - "message": "model not exist" - }, - "model": "kling/kling-v3-video-generation", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:28.000+00:00", - "inferenceProvider": "kling", - "name": "Kling Video 3.0", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026701.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text", - "Video" - ] - }, - "description": "新增“全能参考”，支持3-8秒视频或多图锚定角色元素。可匹配原声及口型驱动，实现角色本色呈现。视频一致性更强，表现更灵动。支持音画同步、智能分镜。", - "features": [], - "provider": "kling", - "limit": { - "message": "model not exist" - }, - "model": "kling/kling-v3-omni-video-generation", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:46.000+00:00", - "inferenceProvider": "kling", - "name": "Kling Video 3.0 Omni", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026701.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。", - "features": [], - "provider": "kling", - "limit": { - "message": "model not exist" - }, - "model": "kling/kling-v3-image-generation", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:34.000+00:00", - "inferenceProvider": "kling", - "name": "Kling Image 3.0", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026706.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。", - "features": [], - "provider": "kling", - "limit": { - "message": "model not exist" - }, - "model": "kling/kling-v3-omni-image-generation", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:39.000+00:00", - "inferenceProvider": "kling", - "name": "Kling Image 3.0 Omni", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026706.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json deleted file mode 100644 index a4387422..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json +++ /dev/null @@ -1,63 +0,0 @@ -{ - "name": "灵动人像LivePortrait-detect", - "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "liveportrait-detect", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-11-07T14:17:01.000+00:00", - "inferenceProvider": "bailian", - "name": "灵动人像LivePortrait-detect", - "docUrl": "https://help.aliyun.com/document_detail/2856727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json deleted file mode 100644 index 62fa8b89..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "灵动人像LivePortrait", - "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "liveportrait", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-11-07T14:17:00.000+00:00", - "inferenceProvider": "bailian", - "name": "灵动人像LivePortrait", - "docUrl": "https://help.aliyun.com/document_detail/2856730.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json deleted file mode 100644 index 43ee7b5c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json +++ /dev/null @@ -1,357 +0,0 @@ -{ - "name": "MiniMax文本模型", - "description": "由MiniMax提供的MiniMax-M系列文本模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Image", - "Text", - "Video" - ] - }, - "description": "MiniMax M3 凭借业界领先的 Coding 与 Agentic 能力、1M 超长上下文窗口以及原生多模态特性，可出色胜任企业级长文档理解、高质量内容生成、代码编写、Bug 修复及原生应用构建等任务；强大的 Agentic 能力端到端贯通工作流，原生多模态更带来流畅自然的图文混合交互体验。", - "features": [ - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/MiniMax-M3", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-06-01T02:18:01.000+00:00", - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "offlineInfo": {}, - "inferenceProvider": "mini-max", - "name": "MiniMax/MiniMax-M3", - "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"MiniMax/MiniMax-M3\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M3\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"MiniMax/MiniMax-M3\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3021620.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "M2.7 能够自行构建复杂 Agent Harness，并基于 Agent Teams、复杂 Skills、Tool Search tool 等能力，完成高度复杂的生产力任务。", - "features": [ - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/MiniMax-M2.7", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 20000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 20000000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-03-20T13:04:46.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 204800, - "offlineInfo": {}, - "inferenceProvider": "mini-max", - "name": "MiniMax/MiniMax-M2.7", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "智能体世界的SOTA，专为智能体2.0设计，将编码扩展到现实世界包括工作空间、娱乐和个人助理。模型亮点：全球SOTA开源编码与智能体模型；SWE-bench Pro和SWE-bench Verified得分高于Opus 4.6；在Excel、搜索与研究以及文档摘要方面的全球SOTA；未来工作空间的完美主力模型；闪电般快速：优化思维效率，100+ TPS，实现比 Opus 快 3 倍的速度；极致性价比，以支持始终在线的智能体。", - "features": [ - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/MiniMax-M2.5", - "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 204800, - "offlineInfo": {}, - "inferenceProvider": "mini-max", - "name": "MiniMax/MiniMax-M2.5", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "M2.1 的设计初衷在于打破“最顶级的 Agent 能力仅存在于闭源模型”的壁垒。我们在模型层面进行了针对性优化，显著提升了模型在代码生成、工具调用、复杂指令遵循及长程规划任务中的性能。从自动化进行多语言的软件开发，到执行多步骤的复杂办公工作流，MiniMax-M2.1 均表现出卓越的稳定性。我们致力于为开发者提供一个完全透明、可控且高可用的基础模型，以构建下一代自主智能体应用。", - "features": [ - "function-calling", - "cache" - ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax/MiniMax-M2.1", - "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 204800, - "offlineInfo": {}, - "inferenceProvider": "mini-max", - "name": "MiniMax/MiniMax-M2.1", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json deleted file mode 100644 index e3116a64..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "Paraformer语音识别-8k-v1", - "description": "Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-8k-v1", - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:17:24.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "mhttps://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Paraformer语音识别-8k-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-8k-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", - "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-8k-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json deleted file mode 100644 index 13d9ccb2..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json +++ /dev/null @@ -1,66 +0,0 @@ -{ - "name": "Paraformer语音识别-8k-v2", - "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-8k-v2", - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-19T11:29:28.000+00:00", - "inferenceProvider": "bailian", - "name": "Paraformer语音识别-8k-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-8k-v2',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", - "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-8k-v2\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json deleted file mode 100644 index cd175230..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "Paraformer语音识别-mtl-v1", - "description": "Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-mtl-v1", - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:18:59.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Paraformer语音识别-mtl-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-mtl-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", - "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-mtl-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json deleted file mode 100644 index b99160b0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json +++ /dev/null @@ -1,52 +0,0 @@ -{ - "name": "Paraformer实时语音识别-8k-v1", - "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-realtime-8k-v1", - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-06T11:36:50.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Paraformer实时语音识别-8k-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-8k-v1',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-8k-v1\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json deleted file mode 100644 index a3ab4c5e..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json +++ /dev/null @@ -1,42 +0,0 @@ -{ - "name": "Paraformer实时语音识别-8k-v2", - "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-realtime-8k-v2", - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-31T08:01:43.000+00:00", - "inferenceProvider": "bailian", - "name": "Paraformer实时语音识别-8k-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-8k-v2',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-8k-v2\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json deleted file mode 100644 index 1d4292c4..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "Paraformer实时语音识别-v1", - "description": "Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-realtime-v1", - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-06T11:36:01.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Paraformer实时语音识别-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-v1',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-v1\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json deleted file mode 100644 index 4b95da69..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json +++ /dev/null @@ -1,48 +0,0 @@ -{ - "name": "Paraformer实时语音识别-v2", - "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-realtime-v2", - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", - "name": "Paraformer实时语音识别-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-v2',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", - "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-v2\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json deleted file mode 100644 index ea6b8ea1..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "Paraformer语音识别-v1", - "description": "Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-v1", - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:20:05.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Paraformer语音识别-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", - "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json deleted file mode 100644 index d0e2040e..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json +++ /dev/null @@ -1,48 +0,0 @@ -{ - "name": "Paraformer语音识别-v2", - "description": "推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "paraformer-v2", - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", - "name": "Paraformer语音识别-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-v2',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", - "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-v2\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json deleted file mode 100644 index f484629a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json +++ /dev/null @@ -1,335 +0,0 @@ -{ - "name": "PixVerse C1", - "description": "由爱诗科技提供的PixVerse C系列视频大模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "C1是PixVerse在26年3月底推出的影视行业大模型，r2v（多主体参考生成视频）输入2-7张图像，智能融合不同主体，同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力和想象力、更接近影视专业水准的打斗动作和术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合多主体群像、多人对话、多人交互等复杂剧情，适合中景、全景镜头。\n如果输入了1张多宫格分镜图片（最高支持九宫格），则可以一键生成连续分镜长视频。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-c1-r2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-09T13:35:02.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-C1-r2v", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "C1是PixVerse在26年3月底推出的影视行业大模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-c1-t2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-09T13:34:47.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-C1-t2v", - "docUrl": "https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025608.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "C1是PixVerse在26年3月底推出的影视行业大模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-c1-kf2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-09T13:34:56.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-C1-kf2v", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025611.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "C1是PixVerse在26年3月底推出的影视行业大模型，it2v（图片生成视频）模型除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征。相比V6可增强提示词，拥有更强的想象力、更接近影视专业水准的打斗动作、术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合单人特写、单人独白、定格/慢动作、空镜转场等短时长镜头。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-c1-it2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-09T13:34:51.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-C1-it2v", - "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025609.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json deleted file mode 100644 index 02f27899..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json +++ /dev/null @@ -1,317 +0,0 @@ -{ - "name": "PixVerse V5.6", - "description": "由爱诗科技提供的PixVerse V系列视频大模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "输入文字描述，秒级生成与语义精准匹配的高质量视频，支持多种风格。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v5.6-t2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T06:58:48.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V5.6-t2v", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025608.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入2–7张图像，智能融合不同主体，保持风格统一与动作协调，轻松构建丰富叙事场景，提升内容可控性与创意自由度。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v5.6-r2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T06:43:04.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V5.6-r2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "在任意两张图片之间实现无缝转换，实现更流畅自然的场景过渡，打造视觉冲击力强的画面效果。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v5.6-kf2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T06:56:40.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V5.6-kf2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025611.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "上传任意图片，自由定制剧情、节奏与风格，生成生动连贯的视频。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v5.6-it2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-19T06:59:33.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V5.6-it2v", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025609.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json deleted file mode 100644 index 0740d773..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json +++ /dev/null @@ -1,345 +0,0 @@ -{ - "name": "PixVerse V6", - "description": "由爱诗科技提供的PixVerse V系列视频大模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "V6是PixVerse在26年3月底推出的新模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v6-t2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-01T07:48:41.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V6-t2v", - "docUrl": "○ https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025608.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "V6是PixVerse在26年3月底推出的新模型，r2v（多主体参考生成视频）模型全球排名第二，输入2-7张图像，智能融合不同主体，适合复杂的中景、远景视频镜头。同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力、更强的情绪表现力、和更流畅的高速运动画面。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v6-r2v", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-06-08T02:03:00.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V6-r2v", - "docUrl": "https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "V6是PixVerse在26年3月底推出的新模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v6-kf2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-01T07:48:31.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V6-kf2v", - "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025611.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "V6是PixVerse在26年3月底推出的新模型，it2v（图片生成视频）模型全球排名第二，it2v除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征，拥有更强的人物情绪、高速运动表现力。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", - "features": [], - "provider": "pixverse", - "limit": { - "message": "model not exist" - }, - "model": "pixverse/pixverse-v6-it2v", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-01T07:48:50.000+00:00", - "inferenceProvider": "pixverse", - "name": "PixVerse-V6-it2v", - "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025609.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json deleted file mode 100644 index 121259d4..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json +++ /dev/null @@ -1,108 +0,0 @@ -{ - "name": "QVQ-Max", - "description": "千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qvq-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-03-26T08:48:01.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 106496, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "QVQ-Max", - "docUrl": "https://help.aliyun.com/document_detail/2877996.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json deleted file mode 100644 index 1273bf44..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json +++ /dev/null @@ -1,110 +0,0 @@ -{ - "name": "Qwen-QVQ-Plus", - "description": "千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qvq-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-06-03T08:47:18.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 106496, - "offlineInfo": { - "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "QVQ-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2877996.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json deleted file mode 100644 index 52c4ecf2..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json +++ /dev/null @@ -1,116 +0,0 @@ -{ - "name": "Qwen-Audio-Realtime-Flash", - "description": "Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio", - "Text" - ], - "request_modality": [ - "Audio", - "Text" - ] - }, - "description": "千问实时语音对话大模型3.0 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音对话大模型3.0兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。极速版更注重极致的响应速度", - "features": [ - "function-calling" - ], - "provider": "qwen", - "model": "qwen-audio-3.0-realtime-flash", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "30", - "type": "audio_input_token", - "priceName": "输入：音频" - }, - { - "priceUnit": "每百万tokens", - "price": "3", - "type": "audio_text_input_token", - "priceName": "输入：文本" - }, - { - "priceUnit": "每百万tokens", - "price": "30", - "type": "audio_text_output_token", - "priceName": "输出：文本" - }, - { - "priceUnit": "每百万tokens", - "price": "100", - "type": "audio_output_token", - "priceName": "输出：文本+音频（输出的文本不计费）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Chatting" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:47.013+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, - "inferenceProvider": "aliyun-bailian", - "name": "千问实时语音对话大模型3.0（极速版）", - "docUrl": "https://help.aliyun.com/document_detail/3041584.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-flash\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", - "docUrl": "https://help.aliyun.com/document_detail/3041584.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json deleted file mode 100644 index e02c2e41..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json +++ /dev/null @@ -1,116 +0,0 @@ -{ - "name": "Qwen-Audio-Realtime-Plus", - "description": "Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio", - "Text" - ], - "request_modality": [ - "Audio", - "Text" - ] - }, - "description": "千问实时语音大模型 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音大模型 兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。标准版更注重高质量的回复结果。", - "features": [ - "function-calling" - ], - "provider": "qwen", - "model": "qwen-audio-3.0-realtime-plus", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "40", - "type": "audio_input_token", - "priceName": "输入：音频" - }, - { - "priceUnit": "每百万tokens", - "price": "5", - "type": "audio_text_input_token", - "priceName": "输入：文本" - }, - { - "priceUnit": "每百万tokens", - "price": "40", - "type": "audio_text_output_token", - "priceName": "输出：文本" - }, - { - "priceUnit": "每百万tokens", - "price": "150", - "type": "audio_output_token", - "priceName": "输出：文本+音频（输出的文本不计费）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 6, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Chatting" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:43.526+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, - "inferenceProvider": "aliyun-bailian", - "name": "千问实时语音大模型 （标准版）", - "docUrl": "https://help.aliyun.com/document_detail/3041584.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-plus\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", - "docUrl": "https://help.aliyun.com/document_detail/3041584.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json deleted file mode 100644 index f354c773..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json +++ /dev/null @@ -1,164 +0,0 @@ -{ - "name": "Qwen-Audio-TTS", - "description": "Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "qwen-audio-3.0-tts-plus是面向高质量语音生成场景打造的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，显著提升方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更准确地控制情绪、语气、角色、语速、音量和合成风格。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，进一步提升了音质、清晰度、分辨率和整体表现力。Plus 版本更强调合成效果和细节表现，适用于有更高音质、自然度和表现力要求的专业场景，如内容创作、有声书、影视配音、品牌声音设计和高品质语音服务。", - "features": [], - "provider": "qwen", - "model": "qwen-audio-3.0-tts-plus", - "prices": [ - { - "priceUnit": "每万字符", - "price": "1.4", - "type": "cosy_tts_number", - "priceName": "语音合成" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-14T09:19:15.113+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "qwen-audio-3.0-tts-plus", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "qwen-audio-3.0-tts-flash是面向实时交互场景优化的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，提升了方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，提升了音质、清晰度和整体表现力。Flash 版本重点优化实时合成体验，首包延时控制在 200ms 以内，适用于语音助手、实时对话、智能客服等低延迟交互场景。", - "features": [], - "provider": "qwen", - "model": "qwen-audio-3.0-tts-flash", - "prices": [ - { - "priceUnit": "每万字符", - "price": "1", - "type": "cosy_tts_number", - "priceName": "语音合成" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-14T09:47:36.084+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "qwen-audio-3.0-tts-flash", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json deleted file mode 100644 index 11d030a0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json +++ /dev/null @@ -1,109 +0,0 @@ -{ - "name": "Qwen-Coder-Plus", - "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-coder-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2024-11-11T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Coder-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json deleted file mode 100644 index 9d331b94..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json +++ /dev/null @@ -1,109 +0,0 @@ -{ - "name": "Qwen-Coder-Turbo", - "description": "Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-coder-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Coder-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json deleted file mode 100644 index 419b305c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json +++ /dev/null @@ -1,97 +0,0 @@ -{ - "name": "qwen-deep-research", - "description": "千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-deep-research", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1200000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1200000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-08-22T14:05:23.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "inferenceProvider": "bailian", - "name": "qwen-deep-research", - "docUrl": "https://help.aliyun.com/document_detail/2975991.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-deep-research\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-deep-research\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-deep-research\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-deep-research\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-deep-research\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-deep-research\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json deleted file mode 100644 index a5b2499e..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json +++ /dev/null @@ -1,99 +0,0 @@ -{ - "name": "Qwen-Doc-Turbo", - "description": "快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。", - "features": [ - "cache" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-doc-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-07-23T13:21:02.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 253952, - "inferenceProvider": "bailian", - "name": "Qwen-Doc-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2948885.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json deleted file mode 100644 index f4d5508c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json +++ /dev/null @@ -1,416 +0,0 @@ -{ - "name": "Qwen-Embedding", - "description": "基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量V4版本，是通义实验室基于Qwen3训练的多语言文本统一向量模型，相较V3版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升15%~40%；支持64~2048维用户自定义向量维度。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-v4", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1200000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1200000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-06-05T03:07:20.000+00:00", - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "通用文本向量-v4", - "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "category": "Embeddings", - "predictConfig": [ - { - "name": "topK" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v4\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v4\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-v3", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-07-12T09:44:51.000+00:00", - "inferenceProvider": "bailian", - "name": "通用文本向量-v3", - "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v3\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v3\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-v2", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:03:19.000+00:00", - "inferenceProvider": "bailian", - "name": "通用文本向量-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v2\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-v1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:02:12.000+00:00", - "inferenceProvider": "bailian", - "name": "通用文本向量-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-async-v2", - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:05:28.000+00:00", - "inferenceProvider": "bailian", - "name": "通用文本向量-async-v2", - "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v2\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text" - ] - }, - "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "text-embedding-async-v1", - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T09:04:41.000+00:00", - "inferenceProvider": "bailian", - "name": "通用文本向量-async-v1", - "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", - "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v1\",\ninput=input_texts\n)\nprint(resp)", - "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json deleted file mode 100644 index 6a62a6d2..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json +++ /dev/null @@ -1,101 +0,0 @@ -{ - "name": "Qwen-Flash-Character", - "description": "千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", - "features": [ - "model-experience", - "cache", - "web-search" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-flash-character", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-01-13T04:04:02.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 8000, - "inferenceProvider": "bailian", - "name": "Qwen-Flash-Character", - "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json deleted file mode 100644 index b9ac0286..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json +++ /dev/null @@ -1,139 +0,0 @@ -{ - "name": "Qwen-Flash", - "description": "Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 30, - "usage_limit": 5000000, - "usage_limit_field": "total_tokens", - "count_limit": 15000, - "usage_limit_period": 30, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 30, - "usage_limit": 5000000, - "usage_limit_field": "total_tokens", - "count_limit": 15000, - "usage_limit_period": 30, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "inferenceProvider": "bailian", - "name": "Qwen-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json deleted file mode 100644 index 1ee17628..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json +++ /dev/null @@ -1,83 +0,0 @@ -{ - "name": "Qwen-Image-2.0-Pro", - "description": "Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-2.0-pro", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "qwen-image-2.0-pro", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen-image-2.0-pro-2026-04-22", - "latestOnlineAt": "2026-04-22T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-2.0-Pro", - "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json deleted file mode 100644 index 2b28d8f3..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json +++ /dev/null @@ -1,83 +0,0 @@ -{ - "name": "Qwen-Image-2.0", - "description": "Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-2.0", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "qwen-image-2.0", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen-image-2.0-2026-03-03", - "latestOnlineAt": "2026-03-03T11:31:36.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-2.0", - "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json deleted file mode 100644 index dfdfec60..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json +++ /dev/null @@ -1,62 +0,0 @@ -{ - "name": "Qwen-Image-Edit-Max", - "description": "千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-edit-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-15T12:28:13.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-Edit-Max", - "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json deleted file mode 100644 index ec102b64..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json +++ /dev/null @@ -1,117 +0,0 @@ -{ - "name": "Qwen-Image-Edit-Plus", - "description": "千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-edit-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-10-30T09:10:49.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-Edit-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "千问系列首个图像编辑模型，成功将Qwen-Image的文本渲染能力拓展到编辑任务上。支持精准的中英双语文字编辑、视觉外观与语义双重编辑、具备强大的跨基准性能表现。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-edit", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-09-21T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-Edit", - "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://{Domain}/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json deleted file mode 100644 index 5f460769..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Qwen-Image-Max", - "description": "千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen-image-max-2025-12-30", - "latestOnlineAt": "2025-12-30T07:06:12.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-Max", - "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-max\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json deleted file mode 100644 index 55add199..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json +++ /dev/null @@ -1,156 +0,0 @@ -{ - "name": "Qwen-Image-Plus", - "description": "千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-23T10:49:31.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列首个图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-image", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "usage_limit": 1000000, - "usage_limit_field": "tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "usage_limit": 1000000, - "usage_limit_field": "tokens", - "count_limit": 2, - "usage_limit_period": 60, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-13T12:58:52.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-Image", - "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json deleted file mode 100644 index 27b295ec..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json +++ /dev/null @@ -1,192 +0,0 @@ -{ - "name": "Qwen-Long", - "description": "Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列上下文窗口最长，能力均衡且成本较低的模型，适合长文本分析、信息抽取、总结摘要和分类打标等任务。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-long-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 10000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 10, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "LATEST", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-03-19T02:45:13.000+00:00", - "contextWindow": 10000000, - "maxInputTokens": 10000000, - "inferenceProvider": "bailian", - "name": "Qwen-Long-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen-Long是在通义千问针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-long", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2024-05-20T14:57:26.000+00:00", - "contextWindow": 10000000, - "maxInputTokens": 10000000, - "inferenceProvider": "bailian", - "name": "Qwen-Long", - "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json deleted file mode 100644 index 89c3ed99..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json +++ /dev/null @@ -1,380 +0,0 @@ -{ - "name": "Qwen-Math-Plus", - "description": "Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问数学模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-math-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 3072, - "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", - "contextWindow": 4096, - "maxInputTokens": 3072, - "inferenceProvider": "bailian", - "name": "Qwen-Math-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列数学模型是专门用于数学解题的语言模型，推理效果好，模型性能优秀本模型为2024年9月19日快照版本，预计维护到下个版本发布后一个月（待定）。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-math-plus-0919", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "qwen-math-plus-2024-09-19", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 3072, - "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", - "contextWindow": 4096, - "maxInputTokens": 3072, - "inferenceProvider": "bailian", - "name": "Qwen-Math-Plus-2024-09-19", - "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列数学模型是专门用于数学解题的语言模型，推理效果好，模型性能优秀，本模型是动态更新版本，模型更新不会提前通知。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-math-plus-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "LATEST", - "maxOutputTokens": 3072, - "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", - "contextWindow": 4096, - "maxInputTokens": 3072, - "inferenceProvider": "bailian", - "name": "Qwen-Math-Plus-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问数学模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-math-plus-0816", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 20000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 20000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "qwen-math-plus-2024-08-16", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 3072, - "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", - "contextWindow": 4096, - "maxInputTokens": 3072, - "inferenceProvider": "bailian", - "name": "Qwen-Math-Plus-2024-08-16", - "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json deleted file mode 100644 index fc73582a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json +++ /dev/null @@ -1,109 +0,0 @@ -{ - "name": "Qwen-Math-Turbo", - "description": "Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列数学模型是专门用于数学解题的语言模型，推理速度快，成本低。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-math-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 3072, - "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", - "contextWindow": 4096, - "maxInputTokens": 3072, - "offlineInfo": { - "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Math-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json deleted file mode 100644 index b8bd3cb9..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json +++ /dev/null @@ -1,117 +0,0 @@ -{ - "name": "Qwen-Max", - "description": "千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 15, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 15, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2024-10-15T05:39:20.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen-Max", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json deleted file mode 100644 index 6c36cbab..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json +++ /dev/null @@ -1,79 +0,0 @@ -{ - "name": "Qwen-MT-Flash", - "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-mt-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 35000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 35000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "shortDescription": "基于Qwen3全面升级的轻量级文本翻译大模型", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-11-06T08:07:15.000+00:00", - "contextWindow": 16384, - "maxInputTokens": 8192, - "inferenceProvider": "bailian", - "name": "Qwen-MT-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json deleted file mode 100644 index d6a8ceda..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json +++ /dev/null @@ -1,81 +0,0 @@ -{ - "name": "Qwen-MT-Image", - "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-mt-image", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 1, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 1, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-22T09:53:46.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen-MT-Image", - "docUrl": "https://help.aliyun.com/document_detail/2977163.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json deleted file mode 100644 index edd7ca5f..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Qwen-MT-Lite", - "description": "基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-mt-lite", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-11-19T11:49:54.000+00:00", - "contextWindow": 16384, - "maxInputTokens": 8192, - "inferenceProvider": "bailian", - "name": "Qwen-MT-Lite", - "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json deleted file mode 100644 index bc356ae0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Qwen-MT-Plus", - "description": "基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-mt-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 25000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 25000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-07-22T06:16:44.000+00:00", - "contextWindow": 16384, - "maxInputTokens": 8192, - "inferenceProvider": "bailian", - "name": "Qwen-MT-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json deleted file mode 100644 index a9872aef..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Qwen-MT-Turbo", - "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-mt-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 35000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 35000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-07-22T06:16:54.000+00:00", - "contextWindow": 16384, - "maxInputTokens": 8192, - "inferenceProvider": "bailian", - "name": "Qwen-MT-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json deleted file mode 100644 index ddd80c4a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json +++ /dev/null @@ -1,178 +0,0 @@ -{ - "name": "Qwen-Omni-Turbo-Realtime", - "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-omni-turbo-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Omni" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 2048, - "latestOnlineAt": "2025-05-08T11:50:42.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen-Omni-Turbo-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问全新多模态理解生成大模型实时版，此版本为动态更新版本。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-omni-turbo-realtime-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Omni" - ], - "versionTag": "LATEST", - "maxOutputTokens": 2048, - "latestOnlineAt": "2025-05-08T12:29:07.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen-Omni-Turbo-Realtime-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json deleted file mode 100644 index d8bba37b..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json +++ /dev/null @@ -1,185 +0,0 @@ -{ - "name": "Qwen-Omni-Turbo", - "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", - "features": [ - "model-experience", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-omni-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 2048, - "latestOnlineAt": "2025-02-14T14:56:44.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen-Omni-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为动态更新版本。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-omni-turbo-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "versionTag": "LATEST", - "maxOutputTokens": 2048, - "latestOnlineAt": "2025-02-14T14:55:08.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen-Omni-Turbo-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json deleted file mode 100644 index fff3c32a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json +++ /dev/null @@ -1,102 +0,0 @@ -{ - "name": "Qwen-Plus-Character", - "description": "千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", - "features": [ - "model-experience", - "web-search", - "cache", - "structured-outputs" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-plus-character", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 120, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2025-03-20T09:16:40.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 32768, - "inferenceProvider": "bailian", - "name": "Qwen-Plus-Character", - "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json deleted file mode 100644 index ffe268ea..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json +++ /dev/null @@ -1,459 +0,0 @@ -{ - "name": "Qwen-Plus", - "description": "千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 30, - "usage_limit": 2500000, - "usage_limit_field": "total_tokens", - "count_limit": 15000, - "usage_limit_period": 30, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 30, - "usage_limit": 2500000, - "usage_limit_field": "total_tokens", - "count_limit": 15000, - "usage_limit_period": 30, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "inferenceProvider": "bailian", - "name": "Qwen-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。本模型是动态更新版本，模型更新不会提前通知。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-plus-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 30, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 7500, - "usage_limit_period": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 30, - "usage_limit": 200000, - "usage_limit_field": "total_tokens", - "count_limit": 7500, - "usage_limit_period": 10, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "LATEST", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-07-30T16:00:00.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 995904, - "inferenceProvider": "bailian", - "name": "Qwen-Plus-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-1125版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-plus-1220", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 150000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 150000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "qwen-plus-2024-12-20", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2024-12-26T12:24:51.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "inferenceProvider": "bailian", - "name": "Qwen-Plus-2024-12-20", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-2024-1220版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-plus-0112", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 150000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 150000, - "usage_limit_field": "total_tokens", - "count_limit": 5, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "qwen-plus-2025-01-12", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-01-15T11:28:32.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "inferenceProvider": "bailian", - "name": "Qwen-Plus-2025-01-12", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json deleted file mode 100644 index 6fa115a4..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json +++ /dev/null @@ -1,246 +0,0 @@ -{ - "name": "Qwen-Rerank", - "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-rerank", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 9000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 9000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-29T10:23:42.000+00:00", - "maxInputTokens": 120000, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Rerank", - "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-rerank", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 10, - "usage_limit": 5000000000, - "usage_limit_field": "total_tokens", - "count_limit": 900, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 10, - "usage_limit": 5000000000, - "usage_limit_field": "total_tokens", - "count_limit": 900, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 0, - "latestOnlineAt": "2025-10-21T08:21:26.000+00:00", - "contextWindow": 30000, - "maxInputTokens": 30000, - "inferenceProvider": "bailian", - "name": "千问3-Rerank", - "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "gte-rerank-v2", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 83000000, - "usage_limit_field": "total_tokens", - "count_limit": 84, - "usage_limit_period": 1, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 83000000, - "usage_limit_field": "total_tokens", - "count_limit": 84, - "usage_limit_period": 1, - "type": "model-default" - } - }, - "capabilities": [ - "TR" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 0, - "latestOnlineAt": "2025-03-20T08:32:00.000+00:00", - "contextWindow": 30000, - "maxInputTokens": 30000, - "inferenceProvider": "bailian", - "name": "深度文本重排序", - "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json deleted file mode 100644 index aa990dde..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json +++ /dev/null @@ -1,180 +0,0 @@ -{ - "name": "Qwen-TTS-Realtime", - "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-tts-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 7680, - "latestOnlineAt": "2025-07-16T06:02:09.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 512, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-TTS-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen-tts-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen-tts-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen-TTS实时模型是通义实验室千问模型中语音合成利器，始终与最新快照版能力相同。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。本模型是动态更新版本，模型更新不会提前通知。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-tts-realtime-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "versionTag": "LATEST", - "maxOutputTokens": 7680, - "latestOnlineAt": "2025-07-16T06:03:00.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 512, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-TTS-Realtime-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen-tts-realtime-latest',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen-tts-realtime-latest\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json deleted file mode 100644 index a0f026ff..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json +++ /dev/null @@ -1,128 +0,0 @@ -{ - "name": "Qwen-TTS", - "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-tts", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 7680, - "latestOnlineAt": "2025-04-20T08:27:59.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 512, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-TTS", - "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen-tts\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", - "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen-tts\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-tts-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "LATEST", - "maxOutputTokens": 7680, - "latestOnlineAt": "2025-06-25T16:00:00.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 512, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-TTS-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen-tts-latest\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", - "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen-tts-latest\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json deleted file mode 100644 index af0c0b42..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json +++ /dev/null @@ -1,143 +0,0 @@ -{ - "name": "Qwen-Turbo", - "description": "千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3系列Turbo模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-Turbo，达到同规模业界SOTA水平。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "cache", - "structured-outputs" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 15, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 12, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 15, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 12, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json deleted file mode 100644 index 21d91205..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json +++ /dev/null @@ -1,144 +0,0 @@ -{ - "name": "Qwen-VL-Embedding", - "description": "基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen2.5-vl-embedding", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 60000, - "usage_limit_field": "total_usage", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 60000, - "usage_limit_field": "total_usage", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "ME" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen2.5-VL-Embedding", - "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-embedding", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_usage", - "count_limit": 40, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 120000, - "usage_limit_field": "total_usage", - "count_limit": 40, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "ME" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", - "maxInputTokens": 32000, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Embedding", - "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "topK" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json deleted file mode 100644 index e57df8de..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json +++ /dev/null @@ -1,117 +0,0 @@ -{ - "name": "Qwen-VL-Max", - "description": "Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "千问VL-Max（qwen-vl-max），即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience", - "structured-outputs", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-vl-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-05-25T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-VL-Max", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json deleted file mode 100644 index 14c496eb..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json +++ /dev/null @@ -1,286 +0,0 @@ -{ - "name": "Qwen-VL-OCR", - "description": "Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", - "features": [ - "model-experience", - "batch" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-vl-ocr-latest", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "LATEST", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-09-22T16:00:00.000+00:00", - "contextWindow": 38192, - "maxInputTokens": 30000, - "inferenceProvider": "bailian", - "name": "QwenVL-OCR-Latest", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。本模型为2024年10月28日的快照版本。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-vl-ocr-1028", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "modelAlias": "qwen-vl-ocr-2024-10-28", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 4096, - "latestOnlineAt": "2024-11-14T13:51:21.000+00:00", - "contextWindow": 34096, - "maxInputTokens": 30000, - "inferenceProvider": "bailian", - "name": "QwenVL-OCR-2024-10-28", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", - "features": [ - "model-experience", - "batch" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen-vl-ocr", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen-vl-ocr-2025-11-20", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-11-19T16:00:00.000+00:00", - "contextWindow": 38192, - "maxInputTokens": 30000, - "inferenceProvider": "bailian", - "name": "QwenVL-OCR", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json deleted file mode 100644 index 9e9b6ac1..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json +++ /dev/null @@ -1,124 +0,0 @@ -{ - "name": "Qwen-VL-Plus", - "description": "Qwen-VL-Plus，即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "千问VL-Plus（qwen-vl-plus），即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience", - "structured-outputs", - "prefix-completion", - "cache", - "batch", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-vl-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-06-13T08:31:55.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "QwenVL-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "vl_high_resolution_images", - "key": "vl_high_resolution_images", - "default": false, - "tip": "是否提高输入图片的默认Token上限" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-plus\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-plus\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json deleted file mode 100644 index 526e2ad2..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json +++ /dev/null @@ -1,67 +0,0 @@ -{ - "name": "Qwen-声音设计", - "description": "Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-voice-design", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen-voice-design", - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-12T07:41:45.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Voice-Design", - "docUrl": "https://help.aliyun.com/document_detail/3000986.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", - "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", - "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json deleted file mode 100644 index b1ef359c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json +++ /dev/null @@ -1,60 +0,0 @@ -{ - "name": "Qwen-声音复刻", - "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen-voice-enrollment", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen-voice-enrollment", - "versionTag": "MAJOR", - "latestOnlineAt": "2025-11-27T05:44:15.000+00:00", - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen-Voice-Enrollment", - "docUrl": "https://help.aliyun.com/document_detail/2975034.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", - "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", - "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json deleted file mode 100644 index cc1c7767..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json +++ /dev/null @@ -1,95 +0,0 @@ -{ - "name": "Qwen2.5-开源模型", - "description": "Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "基于Qwen2.5训练的全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen2.5-omni-7b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 2048, - "latestOnlineAt": "2025-03-26T12:01:58.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 30720, - "inferenceProvider": "bailian", - "name": "Qwen2.5-Omni-7B", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json deleted file mode 100644 index d6052ad9..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json +++ /dev/null @@ -1,80 +0,0 @@ -{ - "name": "Qwen3-ASR-Flash-Filetrans", - "description": "Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-asr-flash-filetrans", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "count_limit": 100, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "count_limit": 100, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-11-17T13:12:02.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-ASR-Flash-Filetrans", - "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", - "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", - "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json deleted file mode 100644 index 9afac99e..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json +++ /dev/null @@ -1,78 +0,0 @@ -{ - "name": "Qwen3-ASR-Flash-Realtime", - "description": "Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-asr-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 20, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-10-27T10:00:46.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-ASR-Flash-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2989727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# example requires websocket-client library:\n# pip install websocket-client\n\nimport os\nimport time\nimport json\nimport threading\nimport base64\nimport websocket\nimport logging\nimport logging.handlers\nfrom datetime import datetime\n\nlogger = logging.getLogger(__name__)\nlogger.setLevel(logging.DEBUG)\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：API_KEY=\"sk-xxx\"\nAPI_KEY = os.environ.get(\"DASHSCOPE_API_KEY\")\nQWEN_MODEL = \"qwen3-asr-flash-realtime\"\n\nbaseUrl = \"wss://dashscope.aliyuncs.com/api-ws/v1/realtime\"\nurl = f\"{baseUrl}?model={QWEN_MODEL}\"\nprint(f\"Connecting to server: {url}\")\n\n# 注意： 如果是非vad模式，建议持续发送的音频时长累加不超过60s\nenableServerVad = True\n\nheaders = [\n \"Authorization: Bearer \" + API_KEY,\n \"OpenAI-Beta: realtime=v1\"\n]\n\ndef send_event(ws, event):\n logger.info(f\" Send event: {event['event_id']}, type={event['type']}\")\n ws.send(json.dumps(event))\n\ndef init_logger():\n formatter = logging.Formatter('%(asctime)s|%(levelname)s|%(message)s')\n\n filter = logging.handlers.RotatingFileHandler(\"omni_tester.log\", maxBytes = 100 * 1024 *1024, backupCount = 3)\n filter.setLevel(logging.DEBUG)\n filter.setFormatter(formatter)\n\n console = logging.StreamHandler()\n console.setLevel(logging.DEBUG)\n console.setFormatter(formatter)\n\n logger.addHandler(filter)\n logger.addHandler(console)\n\ndef on_open(ws):\n logger.info(\"Connected to server.\")\n\n # 会话更新事件\n event0 = {\n \"event_id\": \"event_123\",\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\"],\n \"input_audio_format\": \"pcm\",\n \"sample_rate\": 16000,\n \"input_audio_transcription\": {\n # 语种标识，可选，如果有明确的语种信息，建议设置\n \"language\": \"zh\",\n # 语料，可选，如果有语料，建议设置以增强识别效果\n # \"corpus\": {\n # \"text\": \"\"\n # }\n },\n \"turn_detection\": None\n }\n }\n event1 = {\n \"event_id\": \"event_123\",\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\"],\n \"input_audio_format\": \"pcm\",\n \"sample_rate\": 16000,\n \"input_audio_transcription\": {\n # 语种标识，可选，如果有明确的语种信息，建议设置\n \"language\": \"zh\",\n # 语料，可选，如果有语料，建议设置以增强识别效果\n # \"corpus\": {\n # \"text\": \"\"\n # }\n },\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.2,\n \"silence_duration_ms\": 800\n }\n }\n }\n\n global enableServerVad\n if enableServerVad:\n logger.info(f\"Sending event: {json.dumps(event1, indent=2)}\")\n ws.send(json.dumps(event1))\n else:\n logger.info(f\"Sending event: {json.dumps(event0, indent=2)}\")\n ws.send(json.dumps(event0))\n\ndef on_message(ws, message):\n try:\n data = json.loads(message)\n logger.info(f\"Received event: {json.dumps(data, ensure_ascii=False, indent=2)}\")\n except json.JSONDecodeError:\n logger.error(f\"Failed to parse message: {message}\")\n\ndef on_error(ws, error):\n logger.error(f\"Error: {error}\")\n\ndef on_close(ws, close_status_code, close_msg):\n logger.info(f\"Connection closed: {close_status_code} - {close_msg}\")\n\ndef send_audio(ws, local_audio_path):\n time.sleep(5)\n\n with open(local_audio_path, 'rb') as audio_file:\n logger.info(f\"文件读取开始: {datetime.now().strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]}\")\n while True:\n # 读取指定大小的二进制数据\n audio_data = audio_file.read(3200)\n if not audio_data:\n logger.info(f\"文件读取完毕: {datetime.now().strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]}\")\n global enableServerVad\n if enableServerVad is False:\n event = {\n \"event_id\": \"event_789\",\n \"type\": \"input_audio_buffer.commit\"\n }\n ws.send(json.dumps(event))\n break # 如果已达到文件结尾，则退出循环\n\n # 对读取的二进制数据进行 Base64 编码\n encoded_data = base64.b64encode(audio_data).decode('utf-8')\n #print(f\"读取数据：{len(audio_data)} 字节, 编码后: {len(encoded_data)} 字节\")\n\n eventd = {\n \"event_id\": \"event_\" + str(int(time.time() * 1000)),\n \"type\": \"input_audio_buffer.append\",\n \"audio\": encoded_data\n }\n ws.send(json.dumps(eventd))\n logger.info(f\"Sending audio event: {eventd['event_id']}\")\n\n # 模拟实时音频采集\n time.sleep(0.1)\n\n# 添加连接关闭处理函数\nws = websocket.WebSocketApp(\n url,\n header=headers,\n on_open=on_open,\n on_message=on_message,\n on_error=on_error,\n on_close=on_close\n)\n\ninit_logger()\nlogger.info(f\"Connecting to local WebSocket server at {url}...\")\n\n# 替换为待识别的音频文件路径\nlocal_audio_path = \"your_audio_file\"\nthread = threading.Thread(target=send_audio, args=(ws, local_audio_path))\nthread.start()\n\nws.run_forever()" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json deleted file mode 100644 index 0d5c8ec0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json +++ /dev/null @@ -1,54 +0,0 @@ -{ - "name": "Qwen3-ASR-Flash", - "description": "Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-asr-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "count_limit": 100, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "count_limit": 100, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "modelAlias": "qwen3-asr-flash", - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-08T05:39:02.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-ASR-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport dashscope\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json deleted file mode 100644 index 53b49766..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json +++ /dev/null @@ -1,106 +0,0 @@ -{ - "name": "Qwen3-Coder-30B-A3B-Instruct", - "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-coder-30b-a3b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-07-31T12:44:56.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 204800, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Coder-30B-A3B-Instruct", - "docUrl": "https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json deleted file mode 100644 index d9010d47..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json +++ /dev/null @@ -1,106 +0,0 @@ -{ - "name": "Qwen3-Coder-480B-A35B-Instruct", - "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-coder-480b-a35b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-07-22T13:29:39.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 204800, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Coder-480B-A35B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json deleted file mode 100644 index 71e5e762..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json +++ /dev/null @@ -1,108 +0,0 @@ -{ - "name": "Qwen3-Coder-Flash", - "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-coder-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "inferenceProvider": "bailian", - "name": "Qwen3-Coder-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json deleted file mode 100644 index 1d369815..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json +++ /dev/null @@ -1,114 +0,0 @@ -{ - "name": "Qwen3-Coder-Plus", - "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "cache" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-coder-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-07-22T13:29:17.000+00:00", - "contextWindow": 1000000, - "maxInputTokens": 997952, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Coder-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json deleted file mode 100644 index 3b4b1748..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json +++ /dev/null @@ -1,90 +0,0 @@ -{ - "name": "Qwen3-LiveTranslate-Flash-Realtime", - "description": "Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Image", - "Audio" - ] - }, - "description": "Qwen3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-livetranslate-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Audio-Translate" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2025-09-23T11:11:30.000+00:00", - "contextWindow": 53248, - "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-LiveTranslate-Flash-Realtime", - "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", - "docUrl": "https://help.aliyun.com/document_detail/2983281.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json deleted file mode 100644 index 73d436d0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json +++ /dev/null @@ -1,81 +0,0 @@ -{ - "name": "Qwen3-LiveTranslate-Flash", - "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Audio", - "Video" - ] - }, - "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-livetranslate-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-ASR" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2025-12-04T12:21:57.000+00:00", - "contextWindow": 53248, - "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-LiveTranslate-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2999748.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json deleted file mode 100644 index 95e6b638..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json +++ /dev/null @@ -1,290 +0,0 @@ -{ - "name": "Qwen3-Max", - "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "builtInToolMultiPrices": [], - "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-max", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 500, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-max-2026-01-23", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-09-23T11:23:19.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 258048, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Max", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - }, - "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "docUrl": "https://help.aliyun.com/document_detail/3016808.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "cache" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-max-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "qwen3-max-preview", - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2025-09-05T12:43:22.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 258048, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Max-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json deleted file mode 100644 index 98eec78f..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json +++ /dev/null @@ -1,80 +0,0 @@ -{ - "name": "Qwen3-Omni-30b-a3b-Captioner", - "description": "千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-omni-30b-a3b-captioner", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-16T16:38:17.000+00:00", - "contextWindow": 65536, - "maxInputTokens": 32768, - "inferenceProvider": "bailian", - "name": "Qwen3-Omni-30b-a3b-Captioner", - "docUrl": "https://help.aliyun.com/document_detail/2980468.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json deleted file mode 100644 index 50dfdaf6..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json +++ /dev/null @@ -1,102 +0,0 @@ -{ - "name": "Qwen3-Omni-Flash-Realtime", - "description": "Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-omni-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Omni" - ], - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-omni-flash-realtime-2025-12-01", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", - "contextWindow": 65536, - "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-Omni-Flash-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json deleted file mode 100644 index 494aeafe..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json +++ /dev/null @@ -1,107 +0,0 @@ -{ - "name": "Qwen3-Omni-Flash", - "description": "Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "description": "千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-omni-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "modelAlias": "qwen3-omni-flash", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-omni-flash-2025-12-01", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", - "contextWindow": 65536, - "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-Omni-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json deleted file mode 100644 index fd92af23..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json +++ /dev/null @@ -1,61 +0,0 @@ -{ - "name": "Qwen3-TTS-Flash-Realtime", - "description": "Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-flash-realtime", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-tts-flash-realtime-2025-11-27", - "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-TTS-Flash-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-flash-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-flash-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json deleted file mode 100644 index e1cf65b5..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json +++ /dev/null @@ -1,64 +0,0 @@ -{ - "name": "Qwen3-TTS-Flash", - "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-flash", - "versionTag": "MAJOR", - "shortDescription": "韵律拟人，低延迟，支持十种语言和国内多种方言输出", - "equivalentSnapshot": "qwen3-tts-flash-2025-11-27", - "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-TTS-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen3-tts-flash\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", - "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json deleted file mode 100644 index dd146c2e..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json +++ /dev/null @@ -1,80 +0,0 @@ -{ - "name": "qwen3-tts-instruct-flash-realtime", - "description": "通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-instruct-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-instruct-flash-realtime", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-21T07:33:55.000+00:00", - "inferenceProvider": "bailian", - "name": "qwen3-tts-instruct-flash-realtime", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-instruct-flash-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-instruct-flash-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json deleted file mode 100644 index 875e188b..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json +++ /dev/null @@ -1,87 +0,0 @@ -{ - "name": "Qwen3-TTS-Instruct-Flash", - "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-instruct-flash", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-instruct-flash", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-tts-instruct-flash-2026-01-26", - "latestOnlineAt": "2026-02-10T02:56:41.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-TTS-Instruct-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageHint", - "default": "" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "指令控制", - "key": "instructions", - "tip": "仅支持中英文，通过自然语言合成语音的语气、语速、情感及人物性格，需要具体客观的描述文字，如：\n· 请用非常激昂且高亢的语气说话，表现出获得重大成功后的狂喜与激动。\n· 语速请保持中等偏慢，语气要显得优雅、知性，给人以从容不迫的安心感。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json deleted file mode 100644 index 0c9eee16..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json +++ /dev/null @@ -1,61 +0,0 @@ -{ - "name": "Qwen3-TTS-VC-Realtime", - "description": "Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-vc-realtime-2026-01-15", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-vc-realtime-0115", - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-01-14T11:23:59.000+00:00", - "inferenceProvider": "bailian", - "name": "qwen3-tts-vc-realtime-2026-01-15", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "category": "Audio", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", - "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json deleted file mode 100644 index 0d15a100..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json +++ /dev/null @@ -1,60 +0,0 @@ -{ - "name": "Qwen3-TTS-VC", - "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-vc-2026-01-22", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-vc-0122", - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-02-10T03:00:44.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-TTS-VC-2026-01-22", - "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json deleted file mode 100644 index 518a47eb..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json +++ /dev/null @@ -1,81 +0,0 @@ -{ - "name": "Qwen3-TTS-VD-Realtime", - "description": "Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-vd-realtime-2026-01-15", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-vd-realtime-0115", - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-01-14T11:14:10.000+00:00", - "inferenceProvider": "bailian", - "name": "qwen3-tts-vd-realtime-2026-01-15", - "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json deleted file mode 100644 index f95a31a7..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json +++ /dev/null @@ -1,61 +0,0 @@ -{ - "name": "Qwen3-TTS-VD", - "description": "Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-tts-vd-2026-01-26", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 3, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "modelAlias": "qwen3-tts-vd-0126", - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2026-02-10T02:59:43.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen3-TTS-VD-2026-01-26", - "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json deleted file mode 100644 index 8f009242..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json +++ /dev/null @@ -1,145 +0,0 @@ -{ - "name": "Qwen3-VL-Flash", - "description": "Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 2500000, - "usage_limit_field": "total_tokens", - "count_limit": 250, - "usage_limit_period": 30, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 2500000, - "usage_limit_field": "total_tokens", - "count_limit": 250, - "usage_limit_period": 30, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "modelAlias": "qwen3-vl-plus", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-vl-flash-2026-01-22", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-10-14T06:52:12.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json deleted file mode 100644 index f662f875..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json +++ /dev/null @@ -1,140 +0,0 @@ -{ - "name": "Qwen3-VL-Plus", - "description": "Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "cache", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 5, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 250, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 5, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 250, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "modelAlias": "qwen3-vl-plus", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3-vl-plus-2025-12-19", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-01-25T16:00:00.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json index 5d0abd08..4a69a11c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。", "collectionTag": "qwen3.5", "features": [ @@ -32,13 +98,25 @@ "model": "qwen3.5-flash", "iconUrl": "", "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "start_time": 1781072616, + "usage_limit": 20000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, "model-default-actual": { "count_limit_period": 1, - "usage_limit": 10000000, + "start_time": 1781072616, + "usage_limit": 20000000, "usage_limit_field": "total_tokens", "count_limit": 500, + "end_time": 253370736000, "usage_limit_period": 60, - "type": "model-default" + "type": "user-spec" }, "model-default": { "count_limit_period": 1, @@ -49,10 +127,183 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.02", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", - "VU", - "TG" + "TG", + "VU" ], "modelAlias": "", "versionTag": "MAJOR", @@ -60,9 +311,10 @@ "latestOnlineAt": "2026-02-23T03:28:01.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "category": "Cost-optimized", "predictConfig": [ { "name": "system", @@ -125,23 +377,23 @@ "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json deleted file mode 100644 index 63a65f52..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json +++ /dev/null @@ -1,92 +0,0 @@ -{ - "name": "Qwen3.5-LiveTranslate-Flash-Realtime", - "description": "Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio", - "Text" - ], - "request_modality": [ - "Audio", - "Image" - ] - }, - "description": "Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。", - "collectionTag": "qwen3.5", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-livetranslate-flash-realtime", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Audio-Translate" - ], - "modelAlias": "qwen3.5-livetranslate-flash-realtime", - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-05-19T08:25:22.000+00:00", - "contextWindow": 53248, - "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3.5-LiveTranslate-Flash-Realtime", - "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", - "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", - "docUrl": "https://help.aliyun.com/document_detail/2983281.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json deleted file mode 100644 index 91c827b8..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json +++ /dev/null @@ -1,99 +0,0 @@ -{ - "name": "Qwen3.5-OCR", - "description": "Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Image" - ] - }, - "description": "Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景中）抽取上效果显著提升。", - "collectionTag": "qwen3.5", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-ocr", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-06-16T08:00:16.000+00:00", - "contextWindow": 65536, - "maxInputTokens": 49152, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "Qwen3.5-OCR", - "docUrl": "https://help.aliyun.com/document_detail/2860683.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json deleted file mode 100644 index 9954766f..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json +++ /dev/null @@ -1,101 +0,0 @@ -{ - "name": "Qwen3.5-Omni-Flash-Realtime", - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "collectionTag": "Qwen3.5", - "features": [ - "web-search", - "function-calling" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-omni-flash-realtime", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Omni" - ], - "modelAlias": "qwen3.5-omni-flash-realtime", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3.5-omni-flash-realtime-2026-03-15", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-03-30T03:54:16.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 196608, - "inferenceProvider": "bailian", - "name": "Qwen3.5-Omni-Flash-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json deleted file mode 100644 index 7a872831..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json +++ /dev/null @@ -1,103 +0,0 @@ -{ - "name": "Qwen3.5-Omni-Flash", - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "collectionTag": "Qwen3.5", - "features": [ - "web-search" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-omni-flash", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "modelAlias": "qwen3.5-omni-flash", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3.5-omni-flash-2026-03-15", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-03-30T03:58:51.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 196608, - "inferenceProvider": "bailian", - "name": "Qwen3.5-Omni-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json deleted file mode 100644 index e20db8f9..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json +++ /dev/null @@ -1,101 +0,0 @@ -{ - "name": "Qwen3.5-Omni-Plus-Realtime", - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", - "collectionTag": "Qwen3.5", - "features": [ - "web-search", - "function-calling" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-omni-plus-realtime", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Realtime-Omni" - ], - "modelAlias": "qwen3.5-omni-plus-realtime", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3.5-omni-plus-realtime-2026-03-15", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-03-30T03:54:11.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 196608, - "inferenceProvider": "bailian", - "name": "Qwen3.5-Omni-Plus-Realtime", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2880812.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json deleted file mode 100644 index 6adf2613..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json +++ /dev/null @@ -1,103 +0,0 @@ -{ - "name": "Qwen3.5-Omni-Plus", - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "collectionTag": "Qwen3.5", - "features": [ - "web-search" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-omni-plus", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "modelAlias": "qwen3.5-omni-plus", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3.5-omni-plus-2026-03-15", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-03-30T03:58:25.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 196608, - "inferenceProvider": "bailian", - "name": "Qwen3.5-Omni-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json index fc319d5d..992ce3f7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。", "collectionTag": "qwen3.5", "features": [ @@ -49,6 +115,179 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,7 +300,7 @@ "latestOnlineAt": "2026-02-15T09:15:31.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "predictConfig": [ @@ -126,23 +365,23 @@ "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json deleted file mode 100644 index b83e2fe2..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json +++ /dev/null @@ -1,582 +0,0 @@ -{ - "name": "Qwen3.5开源模型", - "description": "Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", - "collectionTag": "qwen3.5", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-27b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "trainingTypes": { - "sft": [ - "lora", - "full" - ] - }, - "inferenceProvider": "bailian", - "name": "Qwen3.5-27B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - }, - "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "docUrl": "https://help.aliyun.com/document_detail/3016808.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", - "collectionTag": "qwen3.5", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-397b-a17b", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 500000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-397B-A17B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - }, - "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "docUrl": "https://help.aliyun.com/document_detail/3016808.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", - "collectionTag": "qwen3.5", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-35b-a3b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-35B-A3B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - }, - "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "docUrl": "https://help.aliyun.com/document_detail/3016808.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。", - "collectionTag": "qwen3.5", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "web-search", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-122b-a10b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:28:11.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-122B-A10B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - }, - "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "docUrl": "https://help.aliyun.com/document_detail/3016808.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json index 33ca24e5..64364357 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。", "collectionTag": "qwen3.6", "features": [ @@ -48,6 +114,118 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -59,7 +237,7 @@ "latestOnlineAt": "2026-04-16T14:02:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "predictConfig": [ @@ -124,23 +302,23 @@ "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json index 79bc5b4f..be2db93d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json @@ -44,6 +44,70 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "11.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.9", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "90", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "18.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "TG" @@ -56,11 +120,11 @@ "maxInputTokens": 245760, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", @@ -126,17 +190,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json index 102483c2..337bf385 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。", "collectionTag": "qwen3.6", "features": [ @@ -31,13 +97,25 @@ }, "model": "qwen3.6-plus", "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "start_time": 1779415426, + "usage_limit": 8000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, "model-default-actual": { "count_limit_period": 1, - "usage_limit": 2500000, + "start_time": 1779415426, + "usage_limit": 8000000, "usage_limit_field": "total_tokens", "count_limit": 500, - "usage_limit_period": 30, - "type": "model-default" + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" }, "model-default": { "count_limit_period": 1, @@ -48,6 +126,122 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -60,7 +254,7 @@ "latestOnlineAt": "2026-04-01T11:55:34.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", @@ -126,23 +320,23 @@ "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json index 20e285d4..f716b6b9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。", "collectionTag": "qwen3.6", "features": [ @@ -28,6 +94,20 @@ "message": "model not exist" }, "model": "qwen3.6-35b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10.8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -57,7 +137,7 @@ "latestOnlineAt": "2026-04-16T14:01:43.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-35B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "predictConfig": [ @@ -122,24 +202,24 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-35b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } } @@ -169,6 +249,20 @@ "message": "model not exist" }, "model": "qwen3.6-27b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -198,7 +292,7 @@ "latestOnlineAt": "2026-04-22T12:50:52.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-27B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "predictConfig": [ @@ -263,17 +357,17 @@ "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json index e56dd891..9326bc08 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json @@ -11,7 +11,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。", "collectionTag": "qwen3.7", "features": [ @@ -27,6 +67,69 @@ "message": "model not exist" }, "model": "qwen3.7-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "discount": 0.5, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "discount": 0.5, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -56,7 +159,7 @@ "latestOnlineAt": "2026-05-21T06:37:11.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", @@ -111,22 +214,28 @@ 1, 32768 ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" } ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -141,7 +250,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型预览版，仅支持思考模式，开放纯文本模型能力供体验。主要优化面向用户的通用对话场景，例如知识问答、指令跟随、创意写作等。", "collectionTag": "qwen3.7", "features": [ @@ -155,6 +304,20 @@ "message": "model not exist" }, "model": "qwen3.7-max-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -183,76 +346,29 @@ "latestOnlineAt": "2026-05-19T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" }, "responsesAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json index c1fce0bc..a79d8241 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。", "collectionTag": "qwen3.7", "features": [ @@ -48,6 +114,144 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,7 +265,7 @@ "contextWindow": 1000000, "maxInputTokens": 991808, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", @@ -116,28 +320,34 @@ 1, 32768 ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" } ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json deleted file mode 100644 index b3a42bb4..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json +++ /dev/null @@ -1,2388 +0,0 @@ -{ - "name": "Qwen3开源模型", - "description": "Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-32b-thinking", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-10-21T06:26:30.000+00:00", - "contextWindow": 131072, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-32B-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-32b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-10-21T06:25:55.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-32B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-30b-a3b-thinking", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 126976, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-30B-A3B-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-30b-a3b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-30B-A3B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-8b-thinking", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 126976, - "trainingTypes": { - "sft": [ - "lora", - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-8B-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-8b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "trainingTypes": { - "sft": [ - "lora", - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-8B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-235b-a22b-thinking", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-23T10:58:07.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 126976, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-235B-A22B-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-235b-a22b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 1, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-23T10:53:31.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-235B-A22B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-next-80b-a3b-instruct", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-11T14:09:35.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Next-80B-A3B-Instruct", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。", - "collectionTag": "qwen3", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-next-80b-a3b-thinking", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-09-11T09:09:19.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 126976, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-Next-80B-A3B-Thinking", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-30b-a3b-instruct-2507", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-07-29T14:20:27.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "trainingTypes": { - "sft": [ - "lora", - "full" - ], - "cpt": [ - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-30B-A3B-Instruct-2507", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-30b-a3b-thinking-2507", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-07-30T08:38:09.000+00:00", - "contextWindow": 81920, - "maxInputTokens": 126976, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-30B-A3B-Thinking-2507", - "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-235b-a22b-thinking-2507", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-07-25T10:04:54.000+00:00", - "contextWindow": 131072, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-235B-A22B-Thinking-2507", - "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-235b-a22b-instruct-2507", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2025-07-22T13:30:21.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-235B-A22B-Instruct-2507", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-235b-a22b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 16384, - "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 129024, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-235B-A22B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-30b-a3b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 30, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 30, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 300, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-30B-A3B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-32b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 40, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 40, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "trainingTypes": { - "sft": [ - "lora", - "full" - ], - "dpo": [ - "lora", - "full" - ], - "cpt": [ - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-32B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-14B。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-14b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-04-28T07:37:07.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "trainingTypes": { - "sft": [ - "lora", - "full" - ], - "dpo": [ - "lora", - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-14B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。", - "collectionTag": "qwen3", - "features": [ - "model-experience", - "function-calling", - "structured-outputs", - "prefix-completion", - "fine-tuning" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-8b", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "trainingTypes": { - "sft": [ - "lora", - "full" - ], - "dpo": [ - "lora", - "full" - ] - }, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "Qwen3-8B", - "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Qwen3系列新一代代码生成模型，效果接近Qwen3-Coder-Plus兼具更优性能。模型重点优化仓库级别理解、支持多轮工具交互、提升对于agentic coding类工具的适配能力。", - "collectionTag": "qwen3", - "features": [ - "model-experience" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-coder-next", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "", - "versionTag": "SNAPSHOT", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-19T14:38:44.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 204800, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "通义千问3-Coder-Next", - "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json deleted file mode 100644 index 302d2793..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json +++ /dev/null @@ -1,108 +0,0 @@ -{ - "name": "Qwen-QwQ-Plus", - "description": "千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。", - "offlineAt": "2026-07-13T15:59:59.000+00:00", - "features": [ - "model-experience", - "function-calling", - "web-search", - "batch" - ], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwq-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2025-03-05T15:17:03.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" - } - }, - "inferenceProvider": "bailian", - "name": "QwQ-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/sambert.json b/skills/bailian-docs-llm-wiki/models/groups/sambert.json deleted file mode 100644 index 86b2dd73..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/sambert.json +++ /dev/null @@ -1,2586 +0,0 @@ -{ - "name": "Sambert语音合成", - "description": "提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiyue-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:17:01.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知悦", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiyue-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiyue-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiyuan-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-20T09:14:08.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知媛", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiyuan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiyuan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiying-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:17:43.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知颖", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiying-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiying-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiye-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:17:07.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知晔", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiye-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiye-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiya-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:17:55.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知雅", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiya-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiya-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhixiao-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知笑", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhixiao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhixiao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhixiang-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:13.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知祥", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhixiang-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhixiang-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiwei-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:04.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知薇", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiwei-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiwei-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiting-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:16.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知婷", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiting-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiting-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhistella-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:19:54.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知莎", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhistella-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhistella-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhishuo-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:18.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知硕", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhishuo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhishuo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhishu-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:05.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知树", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhishu-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhishu-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiru-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:22.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知茹", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiru-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiru-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiqian-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:08.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知倩", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiqian-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiqian-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiqi-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:25.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知琪", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiqi-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiqi-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhinan-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T08:10:47.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知楠", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhinan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhinan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhina-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:46.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知娜", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhina-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhina-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhimo-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:28.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知墨", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhiming-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:49.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知茗", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiming-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiming-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhimiao-emo-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:38.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知妙（多情感）", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimiao-emo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimiao-emo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhimao-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:34.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知猫", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhilun-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:12.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知伦", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhilun-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhilun-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhijing-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:41.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知婧", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhijing-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhijing-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhijia-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:14.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知佳", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhijia-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhijia-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhihao-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:45.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知浩", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhihao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhihao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhigui-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:21:15.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知柜", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhigui-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhigui-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhifei-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:49.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知飞", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhifei-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhifei-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhide-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:17.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知德", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhide-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhide-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhida-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:52.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知达", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhida-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhida-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-zhichu-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:21:12.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-知厨", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhichu-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhichu-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-waan-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:56.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Waan", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-waan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-waan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-perla-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:20.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Perla", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-perla-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-perla-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-indah-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:18:59.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Indah", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-indah-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-indah-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-hanna-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:21:09.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Hanna", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-hanna-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-hanna-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-eva-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:21:02.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Eva", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-eva-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-eva-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-donna-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:23.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Donna", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-donna-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-donna-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-clara-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:21:06.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Clara", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-clara-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-clara-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-cindy-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:26.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Cindy", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-cindy-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-cindy-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-camila-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:27.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Camila", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-camila-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-camila-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-cally-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:58.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Cally", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-cally-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-cally-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-brian-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:54.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Brian", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-brian-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-brian-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-betty-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:16:48.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Betty", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-betty-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-betty-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Text" - ] - }, - "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "sambert-beth-v1", - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:30.000+00:00", - "inferenceProvider": "bailian", - "name": "Sambert语音合成-Beth", - "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-beth-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", - "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-beth-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json deleted file mode 100644 index e4546f6c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "鞋靴模特", - "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "shoemodel-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-21T02:47:24.000+00:00", - "inferenceProvider": "bailian", - "name": "鞋靴模特", - "docUrl": "https://help.aliyun.com/document_detail/2804662.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json deleted file mode 100644 index 848f0974..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json +++ /dev/null @@ -1,376 +0,0 @@ -{ - "name": "SiliconFlow DeepSeek", - "description": "由硅基流动提供的DeepSeek系列模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V3.2 是一款兼具高计算效率与卓越推理和 Agent 性能的模型。其方法建立在三大关键技术突破之上：DeepSeek 稀疏注意力（DSA），一种高效的注意力机制，在保持模型性能的同时显著降低了计算复杂性，并特别针对长上下文场景进行了优化；可扩展的强化学习框架，通过该框架，模型性能可与 GPT-5 相媲美，其高算力版本在推理能力上可与 Gemini-3.0-Pro 匹敌；以及大规模 Agent 任务合成管线，旨在将推理能力整合到工具使用场景中，从而提高在复杂交互环境中的指令遵循和泛化能力。该模型在 2025 年国际数学奥林匹克（IMO）和国际信息学奥林匹克（IOI）中取得了金牌表现", - "features": [ - "function-calling", - "prefix-completion" - ], - "provider": "deepseek", - "limit": { - "message": "model not exist" - }, - "model": "siliconflow/deepseek-v3.2", - "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", - "contextWindow": 163840, - "maxInputTokens": 163840, - "inferenceProvider": "siliconflow", - "name": "SiliconFlow DeepSeek-V3.2", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V3.1-Terminus 是由深度求索（DeepSeek）发布的 V3.1 模型的更新版本，定位为混合智能体大语言模型。此次更新在保持模型原有能力的基础上，专注于修复用户反馈的问题并提升稳定性。它显著改善了语言一致性，减少了中英文混用和异常字符的出现。模型集成了“思考模式”（Thinking Mode）和“非思考模式”（Non-thinking Mode），用户可通过聊天模板灵活切换以适应不同任务。作为一个重要的优化，V3.1-Terminus 增强了代码智能体（Code Agent）和搜索智能体（Search Agent）的性能，使其在工具调用和执行多步复杂任务方面更加可靠", - "features": [ - "function-calling", - "prefix-completion" - ], - "provider": "deepseek", - "limit": { - "message": "model not exist" - }, - "model": "siliconflow/deepseek-v3.1-terminus", - "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", - "contextWindow": 163840, - "maxInputTokens": 163840, - "inferenceProvider": "siliconflow", - "name": "SiliconFlow DeepSeek-V3.1-Terminus", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "新版 DeepSeek-V3 （DeepSeek-V3-0324）与之前的 DeepSeek-V3-1226 使用同样的 base 模型，仅改进了后训练方法。新版 V3 模型借鉴 DeepSeek-R1 模型训练过程中所使用的强化学习技术，大幅提高了在推理类任务上的表现水平，在数学、代码类相关评测集上取得了超过 GPT-4.5 的得分成绩。此外该模型在工具调用、角色扮演、问答闲聊等方面也得到了一定幅度的能力提升。", - "features": [ - "function-calling", - "prefix-completion" - ], - "provider": "deepseek", - "limit": { - "message": "model not exist" - }, - "model": "siliconflow/deepseek-v3-0324", - "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 163840, - "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", - "contextWindow": 163840, - "maxInputTokens": 163840, - "inferenceProvider": "siliconflow", - "name": "SiliconFlow DeepSeek-V3-0324", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-R1-0528 是一款强化学习（RL）驱动的推理模型，解决了模型中的重复性和可读性问题。在 RL 之前，DeepSeek-R1 引入了冷启动数据，进一步优化了推理性能。它在数学、代码和推理任务中与 OpenAI-o1 表现相当，并且通过精心设计的训练方法，提升了整体效果。", - "features": [ - "function-calling", - "prefix-completion" - ], - "provider": "deepseek", - "limit": { - "message": "model not exist" - }, - "model": "siliconflow/deepseek-r1-0528", - "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 50000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", - "contextWindow": 163840, - "maxInputTokens": 163840, - "offlineInfo": {}, - "inferenceProvider": "siliconflow", - "name": "SiliconFlow DeepSeek-R1-0528", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3014912.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json deleted file mode 100644 index 3896d924..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json +++ /dev/null @@ -1,70 +0,0 @@ -{ - "name": "语音识别热词", - "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "speech-biasing", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "ASR" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "语音识别热词", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json deleted file mode 100644 index 58eca4da..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json +++ /dev/null @@ -1,103 +0,0 @@ -{ - "name": "StepFun推理模型", - "description": "由阶跃星辰StepFun提供的Step系列推理模型API服务", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "Step 3.7 Flash 是阶跃星辰最新推出的生产级 Agent 高效率 Flash 模型，专为 Agent、Coding、Search 与多模态工作流打造，在速度、成本、执行可靠性与复杂任务完成能力之间实现了更优平衡。具备多模态感知与执行、视觉搜索与工具增强、高可靠工具调用与编排，以及 Agent 生态兼容优化等核心能力。", - "features": [ - "function-calling", - "structured-outputs", - "cache", - "batch" - ], - "provider": "stepfun", - "limit": { - "message": "model not exist" - }, - "model": "stepfun/step-3.7-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 2000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 2000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 262144, - "latestOnlineAt": "2026-05-25T06:20:45.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 262144, - "offlineInfo": {}, - "inferenceProvider": "stepfun", - "name": "stepfun/step-3.7-flash", - "docUrl": "https://help.aliyun.com/document_detail/3036697.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 1, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"stepfun/step-3.7-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"stepfun/step-3.7-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"stepfun/step-3.7-flash\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", - "docUrl": "https://help.aliyun.com/document_detail/3036697.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json deleted file mode 100644 index 115a1609..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json +++ /dev/null @@ -1,97 +0,0 @@ -{ - "name": "意图分类模型", - "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "tongyi-intent-detect-v3", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 20, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2024-12-12T11:33:25.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 8192, - "inferenceProvider": "bailian", - "name": "意图分类模型", - "docUrl": "https://help.aliyun.com/document_detail/2861138.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016807.html" - } - }, - "dashscope": { - "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json deleted file mode 100644 index 1510e8ce..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json +++ /dev/null @@ -1,95 +0,0 @@ -{ - "name": "通义晓蜜-对话分析-flash", - "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "tongyi-xiaomi-analysis-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-01-09T03:23:03.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 28672, - "inferenceProvider": "bailian", - "name": "通义晓蜜-对话分析-flash", - "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" - } - }, - "dashscope": { - "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json deleted file mode 100644 index 706099a7..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json +++ /dev/null @@ -1,95 +0,0 @@ -{ - "name": "通义晓蜜-对话分析-pro", - "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", - "features": [], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "tongyi-xiaomi-analysis-pro", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 600, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-01-09T03:31:49.000+00:00", - "contextWindow": 32768, - "maxInputTokens": 28672, - "inferenceProvider": "bailian", - "name": "通义晓蜜-对话分析-pro", - "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" - } - }, - "dashscope": { - "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json deleted file mode 100644 index a467552c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json +++ /dev/null @@ -1,172 +0,0 @@ -{ - "name": "Tripo", - "description": "AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "3D-Generation" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。", - "features": [ - "function-calling", - "structured-outputs", - "batch" - ], - "provider": "tripo", - "limit": { - "message": "model not exist" - }, - "model": "Tripo/Tripo-P1.0", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "3D-generation" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T04:07:42.000+00:00", - "inferenceProvider": "tripo", - "name": "Tripo-P1.0", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "3D-Generation" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。", - "features": [ - "function-calling", - "batch", - "structured-outputs" - ], - "provider": "tripo", - "limit": { - "message": "model not exist" - }, - "model": "Tripo/Tripo-H3.1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "3D-generation" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T04:08:07.000+00:00", - "inferenceProvider": "tripo", - "name": "Tripo-H3.1", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json deleted file mode 100644 index a7fcb0ab..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json +++ /dev/null @@ -1,660 +0,0 @@ -{ - "name": "Vanchin DeepSeek", - "description": "由快手万擎提供的DeepSeek系列模型API服务。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V4系列是强大的混合专家（MoE）语言模型，包含DeepSeek-V4-Pro（1.6T总参数，49B激活参数）。支持高达100万（1M）token的上下文长度，是在超过32T高质量多样化token上预训练的开源模型。", - "features": [ - "function-calling", - "structured-outputs", - "cache" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-v4-pro", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "12", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "24", - "type": "output_token", - "priceName": "输出" - }, - { - "priceUnit": "每百万tokens", - "price": "1", - "type": "input_token_cache", - "priceName": "输入（缓存命中）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 300000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 393216, - "latestOnlineAt": "2026-05-29T08:10:54.000+00:00", - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "offlineInfo": {}, - "inferenceProvider": "vanchin", - "name": "vanchin/deepseek-v4-pro", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v4-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V3.2 是一款实现了高计算效率与卓越推理及代理（Agent）性能完美协调的模型。该模型建立在 DeepSeek-V3 的基础之上，通过引入 DeepSeek 稀疏注意力（DSA）、可扩展的强化学习框架以及大规模代理任务合成流水线等关键技术突破，推动了开源大语言模型的前沿发展。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-v3.2-think", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "2", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "3", - "type": "output_token", - "priceName": "输出" - }, - { - "priceUnit": "每百万tokens", - "price": "0.2", - "type": "input_token_cache", - "priceName": "输入（缓存命中）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 600000, - "usage_limit_field": "total_tokens", - "count_limit": 30, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-04-07T09:00:00.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "inferenceProvider": "vanchin", - "name": "Vanchin/DeepSeek-V3.2-think", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.2-think\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V3.1-Terminus 是 DeepSeek-V3.1 的更新版本，旨在保持模型原有核心能力的同时，针对用户反馈的问题进行了修复和优化。该版本的模型结构与 DeepSeek-V3 保持一致，并在特定领域进行了显著增强。", - "features": [ - "function-calling", - "cache" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-v3.1-terminus", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "4", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "12", - "type": "output_token", - "priceName": "输出" - }, - { - "priceUnit": "每百万tokens", - "price": "1.6", - "type": "input_token_cache", - "priceName": "输入（缓存命中）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-04-07T08:57:08.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 32768, - "inferenceProvider": "vanchin", - "name": "Vanchin/DeepSeek-V3.1-Terminus", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-V3 由深度求索（DeepSeek）于 2024 年 12 月发布，是目前开源社区领先的混合专家（MoE）语言模型：总参数 671B，每个 token 仅激活 37B 参数。模型在 14.8 万亿高质量 tokens 上完成预训练，原生支持 128k 上下文。通过创新的无辅助损失负载均衡策略、多头潜在注意力（MLA）架构和 FP8 混合精度训练。", - "features": [ - "structured-outputs", - "prefix-completion", - "function-calling", - "cache" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-v3", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "2", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "8", - "type": "output_token", - "priceName": "输出" - }, - { - "priceUnit": "每百万tokens", - "price": "0.8", - "type": "input_token_cache", - "priceName": "输入（缓存命中）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 16384, - "latestOnlineAt": "2026-04-07T08:56:49.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 131072, - "inferenceProvider": "vanchin", - "name": "Vanchin/DeepSeek-V3", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "DeepSeek-R1 是深度求索于 2025 年 1 月开源的 6710 亿参数混合专家（MoE）推理模型，推理时仅激活 370 亿参数。作为首个通过纯强化学习（无监督微调）训练的千亿级模型，实现了链式思维（CoT）的自然涌现。模型在 RL 前加入冷启动数据解决了 R1-Zero 的重复和混语问题，在数学、代码、推理任务上达到 OpenAI o1 水平。", - "features": [ - "function-calling", - "prefix-completion", - "cache" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-r1", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "4", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "16", - "type": "output_token", - "priceName": "输出" - }, - { - "priceUnit": "每百万tokens", - "price": "1.6", - "type": "input_token_cache", - "priceName": "输入（缓存命中）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Reasoning", - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-04-07T08:56:31.000+00:00", - "contextWindow": 131072, - "maxInputTokens": 98304, - "inferenceProvider": "vanchin", - "name": "Vanchin/DeepSeek-R1", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-r1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "DeepSeek-OCR以 “探索视觉 - 文本压缩边界” 为核心目标，从大语言模型（LLM）视角重新定义视觉编码器的功能定位，为文档识别、图像转文本等高频场景提供了兼顾精度与效率的全新解决方案。", - "features": [ - "structured-outputs" - ], - "provider": "deepseek", - "model": "vanchin/deepseek-ocr", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "0.216", - "type": "input_token", - "priceName": "输入" - }, - { - "priceUnit": "每百万tokens", - "price": "0.216", - "type": "output_token", - "priceName": "输出" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 50, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "VU", - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 8192, - "latestOnlineAt": "2026-04-07T08:56:06.000+00:00", - "contextWindow": 8192, - "maxInputTokens": 8192, - "inferenceProvider": "vanchin", - "name": "Vanchin/DeepSeek-OCR", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"vanchin/deepseek-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\"\n }\n ]\n }\n ]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\",\n },\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\",\n },\n ],\n }\n ],\n)\n\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nasync function main() {\n const completion = await openai.chat.completions.create({\n model: 'vanchin/deepseek-ocr',\n messages: [\n {\n role: 'user',\n content: [\n {\n type: 'image_url',\n image_url: {\n url: 'https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg',\n detail: 'high',\n },\n },\n {\n type: 'text',\n text: 'Read all the text in the image.',\n },\n ],\n },\n ],\n });\n\n console.log(completion.choices[0].message.content);\n}\n\nmain();", - "docUrl": "https://help.aliyun.com/document_detail/3027089.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json deleted file mode 100644 index d9aa0e65..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "视频风格重绘", - "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Video" - ] - }, - "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "video-style-transform", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", - "inferenceProvider": "bailian", - "name": "视频风格重绘", - "docUrl": "https://help.aliyun.com/document_detail/2846319.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json deleted file mode 100644 index b12ee860..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json +++ /dev/null @@ -1,66 +0,0 @@ -{ - "name": "声动人像VideoRetalk", - "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Video", - "Audio" - ] - }, - "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "videoretalk", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-12-10T06:23:15.000+00:00", - "inferenceProvider": "bailian", - "name": "声动人像VideoRetalk", - "docUrl": "https://help.aliyun.com/document_detail/2860466.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json deleted file mode 100644 index deec3f50..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json +++ /dev/null @@ -1,378 +0,0 @@ -{ - "name": "Vidu AI生图", - "description": "由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，对中英文字的精准渲染、UI/图表等设计细节的像素级还原，适合制作海报、信息图等。", - "features": [], - "provider": "vidu", - "model": "vidu/vidu-image_reference2image", - "prices": [ - { - "priceUnit": "每张", - "price": "0.625", - "type": "image_type_1k", - "priceName": "图片生成（1K）" - }, - { - "priceUnit": "每张", - "price": "1", - "type": "image_type_2k", - "priceName": "图片生成（2K）" - }, - { - "priceUnit": "每张", - "price": "1.46875", - "type": "image_type_4k", - "priceName": "图片生成（4K）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:27:07.000+00:00", - "inferenceProvider": "vidu", - "name": "Vidu-image_reference2image", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/vidu-image_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，主打高速高质与低成本，成本比Pro降低约50%。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-fast_reference2image", - "prices": [ - { - "priceUnit": "每张", - "price": "0.46875", - "type": "image_type_1k", - "priceName": "图片生成（1K）" - }, - { - "priceUnit": "每张", - "price": "0.78125", - "type": "image_type_2k", - "priceName": "图片生成（2K）" - }, - { - "priceUnit": "每张", - "price": "1.09375", - "type": "image_type_4k", - "priceName": "图片生成（4K）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:27:39.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-fast_reference2image", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq3-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，擅长处理复杂逻辑，具备超强上下文一致性和工业级稳定性。适合专业设计、漫剧制作等。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-pro_reference2image", - "prices": [ - { - "priceUnit": "每张", - "price": "0.9375", - "type": "image_type_1k", - "priceName": "图片生成（1K）" - }, - { - "priceUnit": "每张", - "price": "0.9375", - "type": "image_type_2k", - "priceName": "图片生成（2K）" - }, - { - "priceUnit": "每张", - "price": "1.71875", - "type": "image_type_4k", - "priceName": "图片生成（4K）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:28:16.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Pro_reference2image", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-pro_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，语义理解能力大幅提升，支持更多风格。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-fast_reference2image", - "prices": [ - { - "priceUnit": "每张", - "price": "0.28125", - "type": "image_type_1k", - "priceName": "图片生成（1K）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:27:49.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-fast_reference2image", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3045893.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json deleted file mode 100644 index c01401c5..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json +++ /dev/null @@ -1,1917 +0,0 @@ -{ - "name": "Vidu AI生视频", - "description": "由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "ViduQ3-Ad是面向广告行业的专用模型，主打\"营销级切镜+智能运镜+直出音效\"三大能力，上传商品图即可生成16秒广告视频，降低广告视频的创作门槛与制作成本。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-ad_reference2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.75", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.90625", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:26:50.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Ad_reference2video", - "docUrl": "https://help.aliyun.com/document_detail/3045891.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-ad_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "ViduQ3-Drama 是面向精品剧/AI漫剧生产的专用模型，主打\"一致性+动效+细节美学+性价比\"四大维度的全面升级，效果更稳、情绪更真、动态更强。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-drama_reference2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:30:16.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Drama_reference2video", - "docUrl": "https://help.aliyun.com/document_detail/3045892.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-drama_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ3-Pro-fast生成速度更快，性价比更高；较 ViduQ2-Pro-fast 生成时长 10 秒扩展至 16 秒 ，可实现更复杂的镜头切换与叙事逻辑。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-pro-fast_img2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.46875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 300, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-07-09T07:29:54.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Pro-fast_img2video", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html", - "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-pro-fast_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ3-Pro图生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-pro_img2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.9375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:47:00.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-pro_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "输入一段文本，生成视频。ViduQ3-Pro文生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-pro_text2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.9375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:57.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Pro_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-pro_text2video\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"960*528\",\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\nimport os\n\n# 以下为北京地域URL\ndashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(api_key=api_key,\n model='vidu/viduq3-turbo_text2video',\n prompt='一只小猫在月光下奔跑',\n size='960*528',\n duration=5,\n resolution='540P',\n watermark=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Text2Video {\n\n static {\n // 以下为北京地域url\n Constants.baseHttpApiUrl = \"https://dashscope.aliyuncs.com/api/v1\";\n }\n\n // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\n public static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n /**\n * Create a video compositing task and wait for the task to complete.\n */\n public static void text2Video() throws ApiException, NoApiKeyException, InputRequiredException {\n VideoSynthesis vs = new VideoSynthesis();\n VideoSynthesisParam param =\n VideoSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"vidu/viduq3-turbo_text2video\")\n .prompt(\"一只小猫在月光下奔跑\")\n .size(\"960*528\")\n .resolution(\"540P\")\n .duration(5)\n .watermark(true)\n .build();\n System.out.println(\"please wait...\");\n VideoSynthesisResult result = vs.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n text2Video();\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3026197.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入首帧图、尾帧图与文本描述，生成视频。ViduQ3-Pro首尾帧生视频是旗舰级音视频原生模型。支持长达16秒的音画同步生成，实现多镜头自由切换，精准把控节奏、情绪与叙事连贯性。参数量领先，画质、人物一致性及情绪表现卓越，达电影级标准。适用于广告（电商、TVC、效果投放）、漫剧、真人剧及游戏等专业生产场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-pro_start-end2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.9375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:33.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-pro_start-end2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026199.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ3-Turbo图生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-turbo_img2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.4375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:46.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-turbo_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "输入一段文本，生成视频。ViduQ3-Turbo文生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-turbo_text2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.4375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:51.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Turbo_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-turbo_text2video\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"960*528\",\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\nimport os\n\n# 以下为北京地域URL\ndashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(api_key=api_key,\n model='vidu/viduq3-turbo_text2video',\n prompt='一只小猫在月光下奔跑',\n size='960*528',\n duration=5,\n resolution='540P',\n watermark=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Text2Video {\n\n static {\n // 以下为北京地域url\n Constants.baseHttpApiUrl = \"https://dashscope.aliyuncs.com/api/v1\";\n }\n\n // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\n public static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n /**\n * Create a video compositing task and wait for the task to complete.\n */\n public static void text2Video() throws ApiException, NoApiKeyException, InputRequiredException {\n VideoSynthesis vs = new VideoSynthesis();\n VideoSynthesisParam param =\n VideoSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"vidu/viduq3-turbo_text2video\")\n .prompt(\"一只小猫在月光下奔跑\")\n .size(\"960*528\")\n .resolution(\"540P\")\n .duration(5)\n .watermark(true)\n .build();\n System.out.println(\"please wait...\");\n VideoSynthesisResult result = vs.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n text2Video();\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3026197.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入首帧图、尾帧图与文本描述，生成视频。ViduQ3-Turbo首尾帧生视频是高性能加速版模型。生成效率极高，兼具优质画质与动态表现，尤其在打斗场面、情绪渲染及语义理解上表现出色。性价比突出，适合图片社交、AI陪伴及特效素材等泛娱乐场景。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-turbo_start-end2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.4375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:28.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-turbo_start-end2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026199.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ2-Pro图生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-pro_img2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.15625", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.34375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.71875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:18.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-pro_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入首帧图、尾帧图与文本描述，生成视频。ViduQ2-Pro首尾帧生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-pro_start-end2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.15625", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.34375", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.71875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:02.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-pro_start-end2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026199.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text", - "Video" - ] - }, - "description": "输入参考视频、图片与文本描述，生成视频。ViduQ2-Pro参考生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-pro_reference2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:23.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Pro_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-pro_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ2-Turbo图生视频是极速生成引擎。720P 5s视频最快仅需19秒，1080P 5s视频约27秒。人物动作与表情自然逼真，真实感强，在打斗等高动态场景中效果出色，运动幅度大。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-turbo_img2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.0875", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.46875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:08.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-turbo_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "输入一段文本，生成视频。ViduQ2文生视频是精准指令遵循与细腻情感捕捉模型。具备卓越的剧情控制力，能深刻理解并表现微表情变化；镜头语言丰富，运镜流畅，画面张力十足。广泛适用于影视动漫、广告电商、短剧及文旅等行业。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2_text2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.1125", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.21875", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:21.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2_text2video\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"960*528\",\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\nimport os\n\n# 以下为北京地域URL\ndashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(api_key=api_key,\n model='vidu/viduq3-turbo_text2video',\n prompt='一只小猫在月光下奔跑',\n size='960*528',\n duration=5,\n resolution='540P',\n watermark=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam;\nimport com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Text2Video {\n\n static {\n // 以下为北京地域url\n Constants.baseHttpApiUrl = \"https://dashscope.aliyuncs.com/api/v1\";\n }\n\n // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\n public static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n /**\n * Create a video compositing task and wait for the task to complete.\n */\n public static void text2Video() throws ApiException, NoApiKeyException, InputRequiredException {\n VideoSynthesis vs = new VideoSynthesis();\n VideoSynthesisParam param =\n VideoSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"vidu/viduq3-turbo_text2video\")\n .prompt(\"一只小猫在月光下奔跑\")\n .size(\"960*528\")\n .resolution(\"540P\")\n .duration(5)\n .watermark(true)\n .build();\n System.out.println(\"please wait...\");\n VideoSynthesisResult result = vs.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n text2Video();\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3026197.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入首帧图、尾帧图与文本描述，生成视频。ViduQ2-Turbo首尾帧生视频是极速生成引擎。720P 5s视频最快仅需19秒，1080P 5s视频约27秒。人物动作与表情自然逼真，真实感强，在打斗等高动态场景中效果出色，运动幅度大。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-turbo_start-end2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.0875", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.25", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.46875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:46:14.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-turbo_start-end2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"540P\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026199.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "输入参考图片与文本描述，生成视频。ViduQ2参考生视频是精准指令遵循与细腻情感捕捉模型。具备卓越的剧情控制力，能深刻理解并表现微表情变化；镜头语言丰富，运镜流畅，画面张力十足。广泛适用于影视动漫、广告电商、短剧及文旅等行业。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2_reference2video", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.21875", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.28125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.71875", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-03-26T13:45:51.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入1-7张参考图片与文本描述，生成视频。ViduQ3支持智能切镜，多机位一致性更出色。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3_reference2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.625", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T08:35:14.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入1-7张参考图片与文本描述，生成视频。ViduQ3-Turbo生成速度快，性价比高。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-turbo_reference2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.15625", - "type": "video_ratio_540p", - "priceName": "视频生成（540P）" - }, - { - "priceUnit": "每秒", - "price": "0.3125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.40625", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T08:33:07.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-Turbo_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-turbo_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入1-7张参考图片与文本描述，生成视频。ViduQ3-mix画面质感强，均衡性好。ViduQ3参考生视频为剧而生，万物可参，声画同出：6大特效（粒子 / 流体 / 动力学 / 运镜 / 转场 / 光影）、5大音效（环境 / 动态 / 氛围 / 拟音 / 情绪）、4大场景（短剧 / 漫剧 / 影视剧 / 广告）全面融合，轻松驾驭短剧、漫剧、广告等多元创作。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq3-mix_reference2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.78125", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.9375", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-28T02:40:03.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ3-mix_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq3-mix_reference2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"size\": \"1280*720\",\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026200.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "输入图片与文本描述，生成视频。ViduQ2-Pro-fast价格触底、效果稳定，生成速度较turbo提高2-3倍。ViduQ2图生视频是全球首创「万物可参考」视频模型。支持特效、表情、纹理、动作、人物、场景等六大维度参考，实现编辑全面进化。通过可控式增、删、改，达成精细化视频编辑，专为漫剧、短剧、影视制作打造的生产级创作引擎。", - "features": [], - "provider": "vidu", - "model": "vidu/viduq2-pro-fast_img2video", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.1", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.2", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T12:09:25.000+00:00", - "inferenceProvider": "vidu", - "name": "ViduQ2-Pro-fast_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"vidu/viduq2-pro-fast_img2video\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"duration\": 5,\n \"resolution\": \"720P\",\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3026198.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json deleted file mode 100644 index b603e699..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "虚拟模特V2", - "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "virtualmodel-v2", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-25T15:18:50.000+00:00", - "inferenceProvider": "bailian", - "name": "虚拟模特V2", - "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json deleted file mode 100644 index 943af02a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json +++ /dev/null @@ -1,44 +0,0 @@ -{ - "name": "大模型声音复刻及声音设计", - "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Audio" - ], - "request_modality": [ - "Audio" - ] - }, - "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "voice-enrollment", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 10, - "type": "model-default" - } - }, - "capabilities": [ - "TTS" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "大模型声音复刻及声音设计", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [] - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json deleted file mode 100644 index 582f6265..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json +++ /dev/null @@ -1,396 +0,0 @@ -{ - "name": "Wan-Image", - "description": "指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现", - "collectionTag": "wan2.7", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.7-image", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-01T03:48:10.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.7-Image", - "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3026980.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "万相2.7-图像生成与编辑旗舰版模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现。", - "collectionTag": "wan2.7", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.7-image-pro", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-01T05:32:16.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.7-Image-Pro", - "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3026980.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image", - "Text" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。", - "collectionTag": "wan2.6", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.6-image", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-15T11:55:32.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.6-Image", - "docUrl": "https://help.aliyun.com/document_detail/3001143.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。", - "collectionTag": "wan2.5", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.5-i2i-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-23T13:38:14.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.5-I2I-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2982258.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-imageedit", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-25T07:22:03.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-ImageEdit", - "docUrl": "https://help.aliyun.com/document_detail/2868981.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json deleted file mode 100644 index f06cfe9b..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json +++ /dev/null @@ -1,1108 +0,0 @@ -{ - "name": "Wan-I2V", - "description": "图片生成视频内容，稳定保持图像主体、风格和文字等细节信息", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Audio", - "Image", - "Text" - ] - }, - "description": "万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。", - "collectionTag": "wan2.7", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.7-i2v", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-03T03:21:23.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.7-I2V", - "docUrl": "https://help.aliyun.com/document_detail/3025059.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3025059.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Image", - "Audio", - "Text" - ] - }, - "description": "万相2.6-图生视频-Flash，生成更快更高性价比。智能分镜调度支持多镜头叙事，多人稳定对话，更自然真实音色，最高支持15秒时长生成", - "collectionTag": "Wan2.6", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.6-i2v-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-15T07:18:54.000+00:00", - "inferenceProvider": "bailian", - "name": "wan2.6-I2V-flash", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "category": "Wan", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Image", - "Text", - "Audio" - ] - }, - "description": "万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成", - "collectionTag": "wan2.6", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.6-i2v", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-03T13:03:01.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.6-I2V", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Audio" - ] - }, - "description": "万相2.5-图生视频-Preview，全新升级技术架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。", - "collectionTag": "wan2.5", - "features": [ - "model-experience", - "fine-tuning" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.5-i2v-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-19T08:44:48.000+00:00", - "trainingTypes": { - "sft": [ - "lora" - ] - }, - "inferenceProvider": "bailian", - "name": "Wan2.5-I2V-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "全新升级的万相2.2图生视频，视频品质更高。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-i2v-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-I2V-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text" - ] - }, - "description": "全新升级的万相2.2-首尾帧生视频，生成速度更快。优化视频动态稳定性与成功率，更强大的指令遵循能力，两张图片生成丝滑过度视频。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-kf2v-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-12T05:38:00.000+00:00", - "trainingTypes": { - "sft": [ - "lora" - ] - }, - "inferenceProvider": "bailian", - "name": "Wan2.2-KF2V-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Video", - "Image" - ] - }, - "description": "wan2.2-animate-move图生动作是一款角色动画生成模型，用户只需上传一张角色照片和一段参考表演视频，即可将视频中的表情和动作迁移到图片角色上，生成高保真的动画视频。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-animate-move", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 1 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 1 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "Wan2.2-Animate-Move", - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-19T04:02:30.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-Animate-Move", - "docUrl": "https://help.aliyun.com/document_detail/2981852.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Video", - "Image" - ] - }, - "description": "wan2.2-animate-mix视频换人是一款角色替换的模型产品，上传一张角色照片与一段表演视频，即可将原视频中的角色精准替换为照片中的角色，完整保留原始视频的场景、光照和色调等环境细节。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-animate-mix", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 1 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 1 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-19T04:02:28.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-Animate-Mix", - "docUrl": "https://help.aliyun.com/document_detail/2982219.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-i2v-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", - "trainingTypes": { - "sft": [ - "lora" - ] - }, - "inferenceProvider": "bailian", - "name": "Wan2.2-I2V-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [], - "request_modality": [ - "Image" - ] - }, - "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v-detect", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-S2V-Detect", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", - "inferenceProvider": "bailian", - "name": "通义万相2.2-数字人-S2V", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相2.1-首尾帧-Plus，两张图片生成丝滑过度视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成画面细节更丰富。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-kf2v-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-04-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-KF2V-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相2.1-图生视频-Plus，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，视频质量更高。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-i2v-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-20T03:30:02.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-I2V-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相2.1-图生视频-Turbo，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成速度更快。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-i2v-turbo", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-02-27T02:24:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-I2V-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json deleted file mode 100644 index bbadbc33..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json +++ /dev/null @@ -1,306 +0,0 @@ -{ - "name": "Wan-R2V", - "description": "参考视频中的人或物，精准保持形象和声音，支持多参考合拍", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Audio", - "Image", - "Text", - "Video" - ] - }, - "description": "Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。", - "collectionTag": "wan2.7", - "features": [], - "provider": "wan", - "model": "wan2.7-r2v", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.6", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "1", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-03T03:21:08.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.7-R2V", - "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-r2v\",\n \"input\": {\n \"prompt\": \"视频2抱着图片3在咖啡厅里弹奏一支舒缓的美式乡村民谣，视频1笑着看着视频2\",\n \"media\": [\n {\n \"type\": \"reference_video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/hfugmr/wan-r2v-role1.mp4\"\n },\n {\n \"type\": \"reference_video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qigswt/wan-r2v-role2.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": false,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3001146.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Image", - "Video", - "Text" - ] - }, - "description": "万相2.6-参考生视频-Flash，生成更快性价比更高。支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍", - "features": [], - "provider": "wan", - "model": "wan2.6-r2v-flash", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "0.5", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - }, - { - "priceUnit": "每秒", - "price": "0.15", - "type": "720P_no_audio", - "priceName": "视频生成（720P 无声）" - }, - { - "priceUnit": "每秒", - "price": "0.25", - "type": "1080P_no_audio", - "priceName": "视频生成（1080P 无声）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-29T10:15:01.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.6-R2V-Flash", - "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-r2v-flash\",\n \"input\": {\n \"prompt\": \"character1在沙发上开心地看电影\",\n \"reference_urls\":[\"https://cdn.wanx.aliyuncs.com/static/demo-wan26/vace.mp4\"]\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"shot_type\":\"multi\"\n }\n}'\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Image", - "Video", - "Text" - ] - }, - "description": "万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍。提醒：当使用视频进行参考时，输入视频也会计入费用，详见模型计费文档。", - "collectionTag": "wan2.6", - "features": [], - "provider": "wan", - "model": "wan2.6-r2v", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.6", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "1", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-15T16:08:49.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.6-R2V", - "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-r2v\",\n \"input\": {\n \"prompt\": \"character1在沙发上开心地看电影\",\n \"reference_urls\":[\"https://cdn.wanx.aliyuncs.com/static/demo-wan26/vace.mp4\"]\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"shot_type\":\"multi\"\n }\n}'\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json deleted file mode 100644 index 71e37806..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json +++ /dev/null @@ -1,702 +0,0 @@ -{ - "name": "Wan-T2I", - "description": "文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。", - "collectionTag": "wan2.6", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.6-t2i", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 1, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 1, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-15T08:05:15.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.6-T2I", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.6-t2i\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"prompt_extend\": true,\n \"watermark\": false,\n \"n\": 2,\n \"size\": \"1280*1280\"\n }\n}'" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。", - "collectionTag": "wan2.5", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.5-t2i-preview", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-19T08:44:37.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.5-T2I-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "全新升级的万相2.2文生图，更丰富的画面细节。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。", - "collectionTag": "wan2.2", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-t2i-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-T2I-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "全新升级的万相2.2文生图，更快的生成速度。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。", - "collectionTag": "wan2.2", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-t2i-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-T2I-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.1-文生图-Plus，更丰富的画面细节，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-t2i-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-08T16:09:10.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-T2I-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.1-文生图-Turbo，更快的生成速度，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-t2i-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-T2I-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", - "inferenceProvider": "bailian", - "name": "wanx-t2i", - "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "style", - "key": "style", - "default": "", - "tip": "输出风格" - }, - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "种子值", - "range": [ - 1, - 4294967289 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-v1-0521", - "capabilities": [ - "IG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", - "inferenceProvider": "bailian", - "name": "万相-文本生成图像-2024-05-21", - "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Wan2.0-T2I-Turbo，更擅长质感人像和创意设计画作生成，在图像美观度、真实感、艺术性上全面升级，支持最大200万像素生成，支持智能提示词改写等。", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.0-t2i-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-20T03:29:28.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.0-T2I-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json deleted file mode 100644 index bdaeaa25..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json +++ /dev/null @@ -1,586 +0,0 @@ -{ - "name": "Wan-T2V", - "description": "文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Audio", - "Text" - ] - }, - "description": "Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。", - "collectionTag": "wan2.7", - "features": [ - "model-experience" - ], - "provider": "wan", - "model": "wan2.7-t2v", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.6", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "1", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-03T03:21:57.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.7-T2V", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-t2v\",\n \"input\": {\n \"prompt\": \"一段紧张刺激的侦探追查故事，展现电影级叙事能力。第1个镜头[0-3秒] 全景：雨夜的纽约街头，霓虹灯闪烁，一位身穿黑色风衣的侦探快步行走。 第2个镜头[3-6秒] 中景：侦探进入一栋老旧建筑，雨水打湿了他的外套，门在他身后缓缓关闭。 第3个镜头[6-9秒] 特写：侦探的眼神坚毅专注，远处传来警笛声，他微微皱眉思考。 第4个镜头[9-12秒] 中景：侦探在昏暗走廊中小心前行，手电筒照亮前方。 第5个镜头[12-15秒] 特写：侦探发现关键线索，脸上露出恍然大悟的表情。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"prompt_extend\": true,\n \"watermark\": true,\n \"duration\": 15\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Text", - "Audio" - ] - }, - "description": "万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量", - "collectionTag": "wan2.6", - "features": [ - "model-experience" - ], - "provider": "wan", - "model": "wan2.6-t2v", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.6", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "1", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - }, - { - "priceUnit": "每秒", - "price": "0.6", - "discount": 0.5, - "type": "720P_batch", - "priceName": "视频生成（720P Batch Chat）" - }, - { - "priceUnit": "每秒", - "price": "1", - "discount": 0.5, - "type": "1080P_batch", - "priceName": "视频生成（1080P Batch Chat）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-12-03T13:03:19.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.6-T2V", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-t2v\",\n \"input\": {\n \"prompt\": \"一幅史诗级可爱的场景。一只小巧可爱的卡通小猫将军，身穿细节精致的金色盔甲，头戴一个稍大的头盔，勇敢地站在悬崖上。他骑着一匹虽小但英勇的战马，说：”青海长云暗雪山，孤城遥望玉门关。黄沙百战穿金甲，不破楼兰终不还。“。悬崖下方，一支由老鼠组成的、数量庞大、无穷无尽的军队正带着临时制作的武器向前冲锋。这是一个戏剧性的、大规模的战斗场景，灵感来自中国古代的战争史诗。远处的雪山上空，天空乌云密布。整体氛围是“可爱”与“霸气”的搞笑和史诗般的融合。\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video", - "Audio" - ], - "request_modality": [ - "Text", - "Audio" - ] - }, - "description": "万相2.5-文生视频-Preview，全新升级模型架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。", - "collectionTag": "wan2.5", - "features": [ - "model-experience" - ], - "provider": "wan", - "model": "wan2.5-t2v-preview", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.3", - "type": "video_ratio_480p", - "priceName": "视频生成（480P）" - }, - { - "priceUnit": "每秒", - "price": "0.6", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - }, - { - "priceUnit": "每秒", - "price": "1", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-09-19T06:10:15.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.5-T2V-Preview", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wan2.5-t2v-preview',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "全新升级的万相2.2文生视频，视频品质更高。可稳定生成大幅度复杂运动，支持影视级画面表现与控制，更强大的指令遵循能力，实现物理世界还原。", - "collectionTag": "wan2.2", - "features": [], - "provider": "wan", - "model": "wan2.2-t2v-plus", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.14", - "type": "video_ratio_480p", - "priceName": "视频生成（480P）" - }, - { - "priceUnit": "每秒", - "price": "0.7", - "type": "video_ratio_1080p", - "priceName": "视频生成（1080P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.2-T2V-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wan2.2-t2v-plus',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.1-文生视频-Plus，一句话生成视频。视频品质更高，支持大幅度复杂运动、现实物理规律还原、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升。", - "features": [], - "provider": "wan", - "model": "wanx2.1-t2v-plus", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.7", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-08T16:13:12.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.1-T2V-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wanx2.1-t2v-plus',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相2.1-文生视频-Turbo，一句话生成视频。生成速度更快，支持大幅度复杂运动、现实物理规律还原、丰富的艺术风格和影视级画面质感，指令遵循能力进一步提升。", - "features": [], - "provider": "wan", - "model": "wanx2.1-t2v-turbo", - "prices": [ - { - "priceUnit": "每秒", - "price": "0.24", - "type": "video_ratio_480p", - "priceName": "视频生成（480P）" - }, - { - "priceUnit": "每秒", - "price": "0.24", - "type": "video_ratio_720p", - "priceName": "视频生成（720P）" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", - "inferenceProvider": "aliyun-bailian", - "name": "Wan2.1-T2V-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wanx2.1-t2v-turbo',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json deleted file mode 100644 index 0d3e9cb0..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json +++ /dev/null @@ -1,96 +0,0 @@ -{ - "name": "Wan-VideoEdit", - "description": "通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Image", - "Text", - "Video" - ] - }, - "description": "Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。", - "collectionTag": "wan2.7", - "features": [ - "model-experience" - ], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.7-videoedit", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 5, - "type": "model-default", - "async_user_concurrency_limit": 5 - } - }, - "capabilities": [ - "VG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-03T03:19:53.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.7-VideoEdit", - "docUrl": "https://help.aliyun.com/document_detail/3021842.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长", - "key": "duration", - "tip": "可以选择与输入视频时长相同，或者指定不大于输入视频的时长" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3021842.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json deleted file mode 100644 index a4634e68..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "图像背景生成", - "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-background-generation-v2", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-03-22T03:33:37.000+00:00", - "inferenceProvider": "bailian", - "name": "图像背景生成", - "docUrl": "https://help.aliyun.com/document_detail/2712497.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json deleted file mode 100644 index e6b46d90..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "创意海报生成", - "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-poster-generation-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-21T02:49:20.000+00:00", - "inferenceProvider": "bailian", - "name": "创意海报生成", - "docUrl": "https://help.aliyun.com/document_detail/2807172.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json deleted file mode 100644 index 96a6474d..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json +++ /dev/null @@ -1,67 +0,0 @@ -{ - "name": "万相-涂鸦作画", - "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-sketch-to-image-lite", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-11T11:21:09.000+00:00", - "inferenceProvider": "bailian", - "name": "万相-涂鸦作画", - "docUrl": "https://help.aliyun.com/document_detail/2712498.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json deleted file mode 100644 index a1c224b1..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "人像风格重绘", - "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-style-repaint-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-03-22T03:32:59.000+00:00", - "inferenceProvider": "bailian", - "name": "人像风格重绘", - "docUrl": "https://help.aliyun.com/document_detail/2712493.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json deleted file mode 100644 index c8356a15..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "虚拟模特", - "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-virtualmodel", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-25T15:19:20.000+00:00", - "inferenceProvider": "bailian", - "name": "虚拟模特", - "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json deleted file mode 100644 index 564e557a..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json +++ /dev/null @@ -1,67 +0,0 @@ -{ - "name": "万相-图像局部重绘", - "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Image" - ] - }, - "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-x-painting", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-05-28T10:45:39.000+00:00", - "inferenceProvider": "bailian", - "name": "万相-图像局部重绘", - "docUrl": "https://help.aliyun.com/document_detail/2797051.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json deleted file mode 100644 index 5e744220..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json +++ /dev/null @@ -1,85 +0,0 @@ -{ - "name": "Wan2.1-VACE-Plus", - "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Video" - ], - "request_modality": [ - "Text", - "Image", - "Video" - ] - }, - "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx2.1-vace-plus", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "type": "model-default", - "async_user_concurrency_limit": 2 - } - }, - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-05-13T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.1-VACE-Plus", - "docUrl": "https://help.aliyun.com/document_detail/2922183.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json deleted file mode 100644 index af883d86..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json +++ /dev/null @@ -1,65 +0,0 @@ -{ - "name": "WordArt锦书-文字变形", - "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "wordart-semantic", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T08:30:26.000+00:00", - "inferenceProvider": "bailian", - "name": "WordArt锦书-文字变形", - "docUrl": "https://help.aliyun.com/document_detail/2712513.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json deleted file mode 100644 index 540893be..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json +++ /dev/null @@ -1,66 +0,0 @@ -{ - "name": "WordArt锦书-文字纹理生成", - "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text", - "Image" - ] - }, - "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", - "features": [], - "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "wordart-texture", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-04-09T08:29:05.000+00:00", - "inferenceProvider": "bailian", - "name": "WordArt锦书-文字纹理生成", - "docUrl": "https://help.aliyun.com/document_detail/2712510.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json deleted file mode 100644 index f2a2f453..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json +++ /dev/null @@ -1,89 +0,0 @@ -{ - "name": "MiMo文本模型", - "description": "由小米MiMo提供的MiMo文本模型API服务", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiMo-V2.5-Pro 是小米发布的最新旗舰模型。与前代模型相比，它在通用智能体能力、复杂软件工程以及长程任务等方面都有显著提升，在 ClawEval、GDPVal 和 SWE-bench Pro 等基准测试中均位列前茅。它能够独立且完全自主地完成需要人类专家耗时数天甚至数周的专业任务，涉及上千次工具调用。其高达 100 万 token 的上下文长度，非常适合集成到各种智能体框架中使用。", - "features": [ - "function-calling", - "structured-outputs", - "cache" - ], - "provider": "xiaomi", - "limit": { - "message": "model not exist" - }, - "model": "xiaomi/mimo-v2.5-pro", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 1000000, - "usage_limit_field": "total_tokens", - "count_limit": 100, - "usage_limit_period": 6, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-05-18T06:32:14.000+00:00", - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "offlineInfo": {}, - "inferenceProvider": "xiaomi", - "name": "xiaomi/mimo-v2.5-pro", - "docUrl": "https://help.aliyun.com/document_detail/3033942.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json deleted file mode 100644 index 5ad27363..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json +++ /dev/null @@ -1,76 +0,0 @@ -{ - "name": "Z-Image-Turbo", - "description": "Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。", - "features": [ - "model-experience" - ], - "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "z-image-turbo", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 1, - "count_limit": 2, - "type": "model-default" - } - }, - "capabilities": [ - "IG" - ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-12-18T06:43:39.000+00:00", - "inferenceProvider": "bailian", - "name": "Z-Image-Turbo", - "docUrl": "https://help.aliyun.com/document_detail/3002354.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1024*1024", - "tip": "请先选择输出分辨率，再选择输出宽高比" - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"z-image-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"film grain, analog film texture, soft film lighting, Kodak Portra 400 style, cinematic grainy texture, photorealistic details, subtle noise, (film grain:1.2)。采用近景特写镜头拍摄的东亚年轻女性，呈现户外雪地场景。她体型纤瘦，呈站立姿势，身体微微向右侧倾斜，头部抬起看向画面上方，姿态自然放松。她的面部是典型东亚长相，肤色白皙，脸颊带有自然的红润感，五官清秀：眼睛是深棕色，眼型偏圆，眼神略带惊讶地望向上方，眼白部分可见；眉毛是深黑色，形状自然弯长；鼻子小巧挺直，嘴唇涂有红色口红，唇瓣微张，表情带着轻微的惊讶或好奇。她的头发是深黑色长直发，发丝被风吹得略显凌乱，部分垂在脸颊两侧，头顶佩戴一顶深灰色的头盔，头盔边缘露出少量发丝。服装是蓝白拼接的厚重外套，外套材质看起来是毛绒与布料结合，显得温暖厚实，适合雪地环境。背景是被白雪覆盖的户外场景，远处可见模糊的树木轮廓，天空是明亮的浅蓝色，带有少量白云，光线是强烈的自然日光，照亮人物面部与头发，形成清晰的光影，色调以蓝、白、黑为主，整体风格清新自然。画面顶部有黑色提示框，内有“Press esc to exit full screen”的白色文字。镜头的近景视角放大了人物的表情与细节，营造出户外雪地的真实氛围。\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"prompt_extend\": false,\n \"size\": \"1120*1440\"\n }\n}'" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json deleted file mode 100644 index fe07e72c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json +++ /dev/null @@ -1,264 +0,0 @@ -{ - "name": "智谱GLM系列文本模型", - "description": "由智谱提供的GLM系列文本模型API服务", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "智谱原厂直供，最新旗舰模型。GLM-5.2 是智谱迄今能力最强的开源模型，支持真正可用的 1M 上下文，并在长程任务中继续保持领先。", - "features": [ - "function-calling", - "structured-outputs", - "prefix-completion", - "cache" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "ZHIPU/GLM-5.2", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-06-16T02:16:54.000+00:00", - "contextWindow": 1048576, - "maxInputTokens": 1048576, - "offlineInfo": {}, - "inferenceProvider": "zhipu-ai", - "name": "ZHIPU/GLM-5.2", - "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3026315.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "GLM-5.1 是智谱最新旗舰模型，代码能力大大增强，长程任务显著提升，能够在单次任务中持续、自主地工作长达 8 小时，完成从规划、执行到迭代优化的完整闭环，交付工程级成果。\n在综合能力与 Coding 能力上，GLM-5.1 整体表现对齐 Claude Opus 4.6，并在长程自主执行、复杂工程优化与真实开发场景中展现出更强的持续工作能力，是构建 Autonomous Agent 与长程 Coding Agent 的理想基座。", - "features": [ - "function-calling", - "structured-outputs", - "cache", - "prefix-completion" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "ZHIPU/GLM-5.1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-05-18T06:33:45.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 204800, - "offlineInfo": {}, - "inferenceProvider": "zhipu-ai", - "name": "ZHIPU/GLM-5.1", - "docUrl": "https://help.aliyun.com/zh/model-studio/glm-zhipu", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3026315.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "智谱新一代旗舰基座,面向AgenticEngineering,实现从代码到工程的范式跃迁,擅长复杂系统工程与长程Agent任务。", - "features": [ - "function-calling", - "cache", - "structured-outputs", - "prefix-completion" - ], - "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "ZHIPU/GLM-5", - "iconUrl": "", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 3000000, - "usage_limit_field": "total_tokens", - "count_limit": 200, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "TG", - "Reasoning" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "maxOutputTokens": 131072, - "latestOnlineAt": "2026-03-26T10:47:50.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 204800, - "offlineInfo": {}, - "inferenceProvider": "zhipu-ai", - "name": "ZHIPU/GLM-5", - "docUrl": "https://help.aliyun.com/document_detail/3026315.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], - "samples": { - "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3026315.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/index.json b/skills/bailian-docs-llm-wiki/models/index.json index eec19c59..35e5abff 100644 --- a/skills/bailian-docs-llm-wiki/models/index.json +++ b/skills/bailian-docs-llm-wiki/models/index.json @@ -1,227 +1,19 @@ { "updatedAt": "2026-07-15", - "totalFamilies": 169, - "totalModels": 382, + "totalFamilies": 11, + "totalModels": 26, "capabilityDistribution": { - "TG": 35, - "IG": 30, - "VG": 25, - "TTS": 16, - "Reasoning": 14, - "ASR": 12, - "VU": 9, - "Realtime-ASR": 7, - "Multimodal-Omni": 5, - "Realtime-Omni": 4, - "Realtime-Audio-Translate": 3, - "ME": 2, - "Realtime-Chatting": 2, - "Realtime-Text-to-Speech": 2, - "TR": 2, - "3D-generation": 1 + "Reasoning": 6, + "VG": 2, + "TG": 2, + "VU": 1 }, "providerDistribution": { - "qwen": 100, - "qwen-domain-model": 34, - "wan": 13, - "happyhorse": 4, - "mini-max": 3, - "deepseek": 3, - "zhipu-ai": 3, - "pixverse": 3, - "moonshot-ai": 2, - "vidu": 2, - "kling": 1, - "stepfun": 1, - "tripo": 1, - "xiaomi": 1 + "qwen": 8, + "happyhorse": 2, + "deepseek": 1 }, "families": [ - { - "slug": "Kimi-K2", - "name": "Kimi", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "VU", - "Reasoning" - ], - "providers": [ - "moonshot-ai" - ], - "itemCount": 5, - "items": [ - "kimi-k2-thinking", - "kimi-k2.5", - "kimi-k2.6", - "kimi-k2.7-code", - "Moonshot-Kimi-K2-Instruct" - ], - "maxContextWindow": 262144 - }, - { - "slug": "MiniMax-M2.1", - "name": "MiniMax", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "mini-max" - ], - "itemCount": 2, - "items": [ - "MiniMax-M2.1", - "MiniMax-M2.5" - ], - "maxContextWindow": 204800 - }, - { - "slug": "MiniMax-speech-market-place", - "name": "MiniMax-Speech系列语音模型", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "mini-max" - ], - "itemCount": 4, - "items": [ - "MiniMax/speech-02-hd", - "MiniMax/speech-02-turbo", - "MiniMax/speech-2.8-hd", - "MiniMax/speech-2.8-turbo" - ] - }, - { - "slug": "aitryon-parsing-v1", - "name": "AI试衣OutfitAnyone-图片分割", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "aitryon-parsing-v1" - ] - }, - { - "slug": "aitryon-plus", - "name": "AI试衣-Plus版", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "aitryon-plus" - ] - }, - { - "slug": "aitryon-refiner", - "name": "AI试衣OutfitAnyone-图片精修", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "aitryon-refiner" - ] - }, - { - "slug": "aitryon", - "name": "AI试衣-基础版", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "aitryon" - ] - }, - { - "slug": "animate-anyone-detect-gen2", - "name": "舞动人像AnimateAnyone-detect", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "animate-anyone-detect-gen2" - ] - }, - { - "slug": "animate-anyone-gen2", - "name": "舞动人像AnimateAnyone", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "animate-anyone-gen2" - ] - }, - { - "slug": "animate-anyone-template-gen2", - "name": "舞动人像AnimateAnyone-template", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "animate-anyone-template-gen2" - ] - }, - { - "slug": "cosyvoice", - "name": "CosyVoice大模型", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen", - "qwen-domain-model" - ], - "itemCount": 7, - "items": [ - "cosyvoice-clone-v1", - "cosyvoice-v1", - "cosyvoice-v2", - "cosyvoice-v3-flash", - "cosyvoice-v3-plus", - "cosyvoice-v3.5-flash", - "cosyvoice-v3.5-plus" - ] - }, { "slug": "deepseek", "name": "DeepSeek", @@ -251,2657 +43,180 @@ "maxContextWindow": 1000000 }, { - "slug": "embedding", - "name": "通义多模态向量", - "primaryCapability": "ME", - "capabilities": [ - "ME" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 3, - "items": [ - "multimodal-embedding-v1", - "tongyi-embedding-vision-flash", - "tongyi-embedding-vision-plus" - ] - }, - { - "slug": "emo-detect-v1", - "name": "悦动人像EMO-detect", + "slug": "happyhorse-i2v", + "name": "HappyHorse-I2V", "primaryCapability": "VG", "capabilities": [ "VG" ], "providers": [ - "qwen" + "happyhorse" ], - "itemCount": 1, + "itemCount": 2, "items": [ - "emo-detect-v1" + "happyhorse-1.0-i2v", + "happyhorse-1.1-i2v" ] }, { - "slug": "emo-v1", - "name": "悦动人像EMO", + "slug": "happyhorse-t2v", + "name": "HappyHorse-T2V", "primaryCapability": "VG", "capabilities": [ "VG" ], "providers": [ - "qwen" + "happyhorse" ], - "itemCount": 1, + "itemCount": 2, "items": [ - "emo-v1" + "happyhorse-1.0-t2v", + "happyhorse-1.1-t2v" ] }, { - "slug": "emoji-detect-v1", - "name": "表情包Emoji-detect", - "primaryCapability": "VG", + "slug": "qwen3.5-flash", + "name": "Qwen3.5-Flash", + "primaryCapability": "Reasoning", "capabilities": [ - "VG" + "Reasoning", + "TG", + "VU" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "emoji-detect-v1" - ] - }, - { - "slug": "emoji-v1", - "name": "表情包Emoji", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" + "qwen3.5-flash" ], - "itemCount": 1, - "items": [ - "emoji-v1" - ] + "maxContextWindow": 1000000 }, { - "slug": "facechain-facedetect", - "name": "FaceChain人物图像检测", - "primaryCapability": "IG", + "slug": "qwen3.5-plus", + "name": "Qwen3.5-Plus", + "primaryCapability": "TG", "capabilities": [ - "IG" + "TG", + "Reasoning", + "VU" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "facechain-facedetect" - ] - }, - { - "slug": "facechain-generation", - "name": "FaceChain人物写真生成", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" + "qwen3.5-plus" ], - "itemCount": 1, - "items": [ - "facechain-generation" - ] + "maxContextWindow": 1000000 }, { - "slug": "farui-plus", - "name": "通义法睿-Plus-32K", - "primaryCapability": "TG", + "slug": "qwen3.6-flash", + "name": "Qwen3.6-Flash", + "primaryCapability": "Reasoning", "capabilities": [ + "Reasoning", + "VU", "TG" ], "providers": [ - "qwen-domain-model" + "qwen" ], "itemCount": 1, "items": [ - "farui-plus" + "qwen3.6-flash" ], - "maxContextWindow": 12000 + "maxContextWindow": 1000000 }, { - "slug": "fun-asr-flash", - "name": "Fun-ASR-Flash", - "primaryCapability": "ASR", + "slug": "qwen3.6-max", + "name": "Qwen3.6-Max", + "primaryCapability": "Reasoning", "capabilities": [ - "ASR" + "Reasoning", + "TG" ], "providers": [ - "qwen-domain-model" + "qwen" ], "itemCount": 1, "items": [ - "fun-asr-flash-2026-06-15" - ] - }, - { - "slug": "fun-asr-realtime", - "name": "Fun-ASR实时语音识别", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 2, - "items": [ - "fun-asr-flash-8k-realtime", - "fun-asr-realtime" - ] - }, - { - "slug": "fun-asr", - "name": "Fun-ASR语音识别", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen-domain-model" + "qwen3.6-max-preview" ], - "itemCount": 2, - "items": [ - "fun-asr", - "fun-asr-mtl" - ] + "maxContextWindow": 262144 }, { - "slug": "fun-music", - "name": "音乐生成", - "primaryCapability": "TTS", + "slug": "qwen3.6-plus", + "name": "Qwen3.6-Plus", + "primaryCapability": "Reasoning", "capabilities": [ - "TTS" + "Reasoning", + "VU", + "TG" ], "providers": [ "qwen" ], - "itemCount": 2, + "itemCount": 1, "items": [ - "fun-music-preview", - "fun-music-v1" - ] + "qwen3.6-plus" + ], + "maxContextWindow": 1000000 }, { - "slug": "glm-4.5", - "name": "GLM", - "primaryCapability": "TG", + "slug": "qwen3.6", + "name": "Qwen3.6开源模型", + "primaryCapability": "VU", "capabilities": [ + "VU", "TG", "Reasoning" ], "providers": [ - "zhipu-ai" + "qwen" ], - "itemCount": 7, + "itemCount": 2, "items": [ - "glm-4.5", - "glm-4.5-air", - "glm-4.6", - "glm-4.7", - "glm-5", - "glm-5.1", - "glm-5.2" + "qwen3.6-27b", + "qwen3.6-35b-a3b" ], - "maxContextWindow": 1048576 + "maxContextWindow": 262144 }, { - "slug": "glm-fast", - "name": "GLM-5.2-Fast", - "primaryCapability": "TG", + "slug": "qwen3.7-max", + "name": "Qwen3.7-Max", + "primaryCapability": "Reasoning", "capabilities": [ + "Reasoning", "TG" ], "providers": [ - "zhipu-ai" + "qwen" ], - "itemCount": 1, + "itemCount": 2, "items": [ - "glm-5.2-fast-preview" + "qwen3.7-max", + "qwen3.7-max-preview" ], - "maxContextWindow": 1048576 + "maxContextWindow": 1000000 }, { - "slug": "gui-plus", - "name": "GUI-Plus", - "primaryCapability": "VU", + "slug": "qwen3.7-plus", + "name": "Qwen3.7-Plus", + "primaryCapability": "TG", "capabilities": [ + "TG", + "Reasoning", "VU" ], "providers": [ - "qwen-domain-model" + "qwen" ], "itemCount": 1, "items": [ - "gui-plus" - ], - "maxContextWindow": 256000 - }, - { - "slug": "gummy-chat-v1", - "name": "一句话识别及翻译V1.0", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "gummy-chat-v1" - ] - }, - { - "slug": "gummy-realtime-v1", - "name": "实时语音识别及翻译V1.0", - "primaryCapability": "Realtime-Audio-Translate", - "capabilities": [ - "Realtime-Audio-Translate" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "gummy-realtime-v1" - ] - }, - { - "slug": "happyhorse-i2v", - "name": "HappyHorse-I2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "happyhorse" - ], - "itemCount": 2, - "items": [ - "happyhorse-1.0-i2v", - "happyhorse-1.1-i2v" - ] - }, - { - "slug": "happyhorse-r2v", - "name": "HappyHorse-R2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "happyhorse" - ], - "itemCount": 2, - "items": [ - "happyhorse-1.0-r2v", - "happyhorse-1.1-r2v" - ] - }, - { - "slug": "happyhorse-t2v", - "name": "HappyHorse-T2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "happyhorse" - ], - "itemCount": 2, - "items": [ - "happyhorse-1.0-t2v", - "happyhorse-1.1-t2v" - ] - }, - { - "slug": "happyhorse-video-edit", - "name": "HappyHorse-Video-Edit", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "happyhorse" - ], - "itemCount": 1, - "items": [ - "happyhorse-1.0-video-edit" - ] - }, - { - "slug": "image-erase-completion", - "name": "图像擦除补全", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "image-erase-completion" - ] - }, - { - "slug": "image-instance-segmentation", - "name": "人物实例分割", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "image-instance-segmentation" - ] - }, - { - "slug": "image-out-painting", - "name": "图像画面扩展", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "image-out-painting" - ] - }, - { - "slug": "kimi-models-market-place", - "name": "Kimi", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "providers": [ - "moonshot-ai" - ], - "itemCount": 4, - "items": [ - "kimi/kimi-k2.5", - "kimi/kimi-k2.6", - "kimi/kimi-k2.7-code", - "kimi/kimi-k2.7-code-highspeed" - ], - "maxContextWindow": 262144 - }, - { - "slug": "kling-models-market-place", - "name": "可灵AI", - "primaryCapability": "VG", - "capabilities": [ - "VG", - "IG" - ], - "providers": [ - "kling" - ], - "itemCount": 4, - "items": [ - "kling/kling-v3-image-generation", - "kling/kling-v3-omni-image-generation", - "kling/kling-v3-omni-video-generation", - "kling/kling-v3-video-generation" - ] - }, - { - "slug": "liveportrait-detect", - "name": "灵动人像LivePortrait-detect", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "liveportrait-detect" - ] - }, - { - "slug": "liveportrait", - "name": "灵动人像LivePortrait", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "liveportrait" - ] - }, - { - "slug": "minimax-models-market-place", - "name": "MiniMax文本模型", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "providers": [ - "mini-max" - ], - "itemCount": 4, - "items": [ - "MiniMax/MiniMax-M2.1", - "MiniMax/MiniMax-M2.5", - "MiniMax/MiniMax-M2.7", - "MiniMax/MiniMax-M3" - ], - "maxContextWindow": 1048576 - }, - { - "slug": "paraformer-8k-v1", - "name": "Paraformer语音识别-8k-v1", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-8k-v1" - ] - }, - { - "slug": "paraformer-8k-v2", - "name": "Paraformer语音识别-8k-v2", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-8k-v2" - ] - }, - { - "slug": "paraformer-mtl-v1", - "name": "Paraformer语音识别-mtl-v1", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-mtl-v1" - ] - }, - { - "slug": "paraformer-realtime-8k-v1", - "name": "Paraformer实时语音识别-8k-v1", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-realtime-8k-v1" - ] - }, - { - "slug": "paraformer-realtime-8k-v2", - "name": "Paraformer实时语音识别-8k-v2", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-realtime-8k-v2" - ] - }, - { - "slug": "paraformer-realtime-v1", - "name": "Paraformer实时语音识别-v1", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-realtime-v1" - ] - }, - { - "slug": "paraformer-realtime-v2", - "name": "Paraformer实时语音识别-v2", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-realtime-v2" - ] - }, - { - "slug": "paraformer-v1", - "name": "Paraformer语音识别-v1", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-v1" - ] - }, - { - "slug": "paraformer-v2", - "name": "Paraformer语音识别-v2", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "paraformer-v2" - ] - }, - { - "slug": "pixverse-c1-market-place", - "name": "PixVerse C1", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "pixverse" - ], - "itemCount": 4, - "items": [ - "pixverse/pixverse-c1-it2v", - "pixverse/pixverse-c1-kf2v", - "pixverse/pixverse-c1-r2v", - "pixverse/pixverse-c1-t2v" - ] - }, - { - "slug": "pixverse-market-place", - "name": "PixVerse V5.6", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "pixverse" - ], - "itemCount": 4, - "items": [ - "pixverse/pixverse-v5.6-it2v", - "pixverse/pixverse-v5.6-kf2v", - "pixverse/pixverse-v5.6-r2v", - "pixverse/pixverse-v5.6-t2v" - ] - }, - { - "slug": "pixverse-v6-market-place", - "name": "PixVerse V6", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "pixverse" - ], - "itemCount": 4, - "items": [ - "pixverse/pixverse-v6-it2v", - "pixverse/pixverse-v6-kf2v", - "pixverse/pixverse-v6-r2v", - "pixverse/pixverse-v6-t2v" - ] - }, - { - "slug": "qvq-max", - "name": "QVQ-Max", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qvq-max" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qvq-plus", - "name": "Qwen-QVQ-Plus", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qvq-plus" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-audio-realtime-flash", - "name": "Qwen-Audio-Realtime-Flash", - "primaryCapability": "Realtime-Chatting", - "capabilities": [ - "Realtime-Chatting" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-audio-3.0-realtime-flash" - ], - "maxContextWindow": 8192 - }, - { - "slug": "qwen-audio-realtime-plus", - "name": "Qwen-Audio-Realtime-Plus", - "primaryCapability": "Realtime-Chatting", - "capabilities": [ - "Realtime-Chatting" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-audio-3.0-realtime-plus" - ], - "maxContextWindow": 8192 - }, - { - "slug": "qwen-audio-tts", - "name": "Qwen-Audio-TTS", - "primaryCapability": "Realtime-Text-to-Speech", - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-audio-3.0-tts-flash", - "qwen-audio-3.0-tts-plus" - ] - }, - { - "slug": "qwen-coder-plus", - "name": "Qwen-Coder-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-coder-plus" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-coder-turbo", - "name": "Qwen-Coder-Turbo", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-coder-turbo" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-deep-research", - "name": "qwen-deep-research", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-deep-research" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen-doc-turbo", - "name": "Qwen-Doc-Turbo", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-doc-turbo" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen-embedding", - "name": "Qwen-Embedding", - "primaryCapability": "TR", - "capabilities": [ - "TR" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 6, - "items": [ - "text-embedding-async-v1", - "text-embedding-async-v2", - "text-embedding-v1", - "text-embedding-v2", - "text-embedding-v3", - "text-embedding-v4" - ] - }, - { - "slug": "qwen-flash-character", - "name": "Qwen-Flash-Character", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-flash-character" - ], - "maxContextWindow": 8192 - }, - { - "slug": "qwen-flash", - "name": "Qwen-Flash", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-flash" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen-image-2.0-pro", - "name": "Qwen-Image-2.0-Pro", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-image-2.0-pro" - ] - }, - { - "slug": "qwen-image-2.0", - "name": "Qwen-Image-2.0", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-image-2.0" - ] - }, - { - "slug": "qwen-image-edit-max", - "name": "Qwen-Image-Edit-Max", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-image-edit-max" - ] - }, - { - "slug": "qwen-image-edit", - "name": "Qwen-Image-Edit-Plus", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-image-edit", - "qwen-image-edit-plus" - ] - }, - { - "slug": "qwen-image-max", - "name": "Qwen-Image-Max", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-image-max" - ] - }, - { - "slug": "qwen-image-plus", - "name": "Qwen-Image-Plus", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-image", - "qwen-image-plus" - ] - }, - { - "slug": "qwen-long", - "name": "Qwen-Long", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 2, - "items": [ - "qwen-long", - "qwen-long-latest" - ], - "maxContextWindow": 10000000 - }, - { - "slug": "qwen-math-plus", - "name": "Qwen-Math-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 4, - "items": [ - "qwen-math-plus", - "qwen-math-plus-0816", - "qwen-math-plus-0919", - "qwen-math-plus-latest" - ], - "maxContextWindow": 4096 - }, - { - "slug": "qwen-math-turbo", - "name": "Qwen-Math-Turbo", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-math-turbo" - ], - "maxContextWindow": 4096 - }, - { - "slug": "qwen-max", - "name": "Qwen-Max", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-max" - ], - "maxContextWindow": 32768 - }, - { - "slug": "qwen-mt-flash", - "name": "Qwen-MT-Flash", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-mt-flash" - ], - "maxContextWindow": 16384 - }, - { - "slug": "qwen-mt-image", - "name": "Qwen-MT-Image", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-mt-image" - ] - }, - { - "slug": "qwen-mt-lite", - "name": "Qwen-MT-Lite", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-mt-lite" - ], - "maxContextWindow": 16384 - }, - { - "slug": "qwen-mt-plus", - "name": "Qwen-MT-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-mt-plus" - ], - "maxContextWindow": 16384 - }, - { - "slug": "qwen-mt-turbo", - "name": "Qwen-MT-Turbo", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-mt-turbo" - ], - "maxContextWindow": 16384 - }, - { - "slug": "qwen-omni-turbo-realtime", - "name": "Qwen-Omni-Turbo-Realtime", - "primaryCapability": "Realtime-Omni", - "capabilities": [ - "Realtime-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-omni-turbo-realtime", - "qwen-omni-turbo-realtime-latest" - ], - "maxContextWindow": 32768 - }, - { - "slug": "qwen-omni-turbo", - "name": "Qwen-Omni-Turbo", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-omni-turbo", - "qwen-omni-turbo-latest" - ], - "maxContextWindow": 32768 - }, - { - "slug": "qwen-plus-character", - "name": "Qwen-Plus-Character", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen-plus-character" - ], - "maxContextWindow": 32768 - }, - { - "slug": "qwen-plus", - "name": "Qwen-Plus", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 4, - "items": [ - "qwen-plus", - "qwen-plus-0112", - "qwen-plus-1220", - "qwen-plus-latest" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen-rerank", - "name": "Qwen-Rerank", - "primaryCapability": "TR", - "capabilities": [ - "TR" - ], - "providers": [ - "qwen", - "qwen-domain-model" - ], - "itemCount": 3, - "items": [ - "gte-rerank-v2", - "qwen3-rerank", - "qwen3-vl-rerank" - ], - "maxContextWindow": 30000 - }, - { - "slug": "qwen-tts-realtime", - "name": "Qwen-TTS-Realtime", - "primaryCapability": "Realtime-Text-to-Speech", - "capabilities": [ - "Realtime-Text-to-Speech" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-tts-realtime", - "qwen-tts-realtime-latest" - ], - "maxContextWindow": 8192 - }, - { - "slug": "qwen-tts", - "name": "Qwen-TTS", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen-tts", - "qwen-tts-latest" - ], - "maxContextWindow": 8192 - }, - { - "slug": "qwen-turbo", - "name": "Qwen-Turbo", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-turbo" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-vl-embedding", - "name": "Qwen-VL-Embedding", - "primaryCapability": "ME", - "capabilities": [ - "ME" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen2.5-vl-embedding", - "qwen3-vl-embedding" - ] - }, - { - "slug": "qwen-vl-max", - "name": "Qwen-VL-Max", - "primaryCapability": "VU", - "capabilities": [ - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-vl-max" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-vl-ocr", - "name": "Qwen-VL-OCR", - "primaryCapability": "VU", - "capabilities": [ - "VU" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 3, - "items": [ - "qwen-vl-ocr", - "qwen-vl-ocr-1028", - "qwen-vl-ocr-latest" - ], - "maxContextWindow": 38192 - }, - { - "slug": "qwen-vl-plus", - "name": "Qwen-VL-Plus", - "primaryCapability": "VU", - "capabilities": [ - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-vl-plus" - ], - "maxContextWindow": 131072 - }, - { - "slug": "qwen-voice-design", - "name": "Qwen-声音设计", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-voice-design" - ] - }, - { - "slug": "qwen-voice-enrollment", - "name": "Qwen-声音复刻", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen-voice-enrollment" - ] - }, - { - "slug": "qwen2.5", - "name": "Qwen2.5-开源模型", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen2.5-omni-7b" - ], - "maxContextWindow": 32768 - }, - { - "slug": "qwen3-asr-flash-filetrans", - "name": "Qwen3-ASR-Flash-Filetrans", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-asr-flash-filetrans" - ] - }, - { - "slug": "qwen3-asr-flash-realtime", - "name": "Qwen3-ASR-Flash-Realtime", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-asr-flash-realtime" - ] - }, - { - "slug": "qwen3-asr-flash", - "name": "Qwen3-ASR-Flash", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-asr-flash" - ] - }, - { - "slug": "qwen3-coder-30b-a3b-instruct", - "name": "Qwen3-Coder-30B-A3B-Instruct", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-coder-30b-a3b-instruct" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3-coder-480b-a35b-instruct", - "name": "Qwen3-Coder-480B-A35B-Instruct", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-coder-480b-a35b-instruct" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3-coder-flash", - "name": "Qwen3-Coder-Flash", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-coder-flash" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3-coder-plus", - "name": "Qwen3-Coder-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-coder-plus" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3-livetranslate-flash-realtime", - "name": "Qwen3-LiveTranslate-Flash-Realtime", - "primaryCapability": "Realtime-Audio-Translate", - "capabilities": [ - "Realtime-Audio-Translate" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-livetranslate-flash-realtime" - ], - "maxContextWindow": 53248 - }, - { - "slug": "qwen3-livetranslate-flash", - "name": "Qwen3-LiveTranslate-Flash", - "primaryCapability": "Realtime-ASR", - "capabilities": [ - "Realtime-ASR" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3-livetranslate-flash" - ], - "maxContextWindow": 53248 - }, - { - "slug": "qwen3-max", - "name": "Qwen3-Max", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen3-max", - "qwen3-max-preview" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3-omni-30b-a3b-captioner", - "name": "Qwen3-Omni-30b-a3b-Captioner", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-omni-30b-a3b-captioner" - ], - "maxContextWindow": 65536 - }, - { - "slug": "qwen3-omni-flash-realtime", - "name": "Qwen3-Omni-Flash-Realtime", - "primaryCapability": "Realtime-Omni", - "capabilities": [ - "Realtime-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-omni-flash-realtime" - ], - "maxContextWindow": 65536 - }, - { - "slug": "qwen3-omni-flash", - "name": "Qwen3-Omni-Flash", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-omni-flash" - ], - "maxContextWindow": 65536 - }, - { - "slug": "qwen3-tts-flash-realtime", - "name": "Qwen3-TTS-Flash-Realtime", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-flash-realtime" - ] - }, - { - "slug": "qwen3-tts-flash", - "name": "Qwen3-TTS-Flash", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-flash" - ] - }, - { - "slug": "qwen3-tts-instruct-flash-realtime", - "name": "qwen3-tts-instruct-flash-realtime", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-instruct-flash-realtime" - ] - }, - { - "slug": "qwen3-tts-instruct-flash", - "name": "Qwen3-TTS-Instruct-Flash", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-instruct-flash" - ] - }, - { - "slug": "qwen3-tts-vc-realtime", - "name": "Qwen3-TTS-VC-Realtime", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-vc-realtime-2026-01-15" - ] - }, - { - "slug": "qwen3-tts-vc", - "name": "Qwen3-TTS-VC", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-vc-2026-01-22" - ] - }, - { - "slug": "qwen3-tts-vd-realtime", - "name": "Qwen3-TTS-VD-Realtime", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-vd-realtime-2026-01-15" - ] - }, - { - "slug": "qwen3-tts-vd", - "name": "Qwen3-TTS-VD", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-tts-vd-2026-01-26" - ] - }, - { - "slug": "qwen3-vl-flash", - "name": "Qwen3-VL-Flash", - "primaryCapability": "VU", - "capabilities": [ - "VU", - "Reasoning" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-vl-flash" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3-vl-plus", - "name": "Qwen3-VL-Plus", - "primaryCapability": "VU", - "capabilities": [ - "VU", - "Reasoning" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3-vl-plus" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.5-flash", - "name": "Qwen3.5-Flash", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-flash" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3.5-livetranslate-flash-realtime", - "name": "Qwen3.5-LiveTranslate-Flash-Realtime", - "primaryCapability": "Realtime-Audio-Translate", - "capabilities": [ - "Realtime-Audio-Translate" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "qwen3.5-livetranslate-flash-realtime" - ], - "maxContextWindow": 53248 - }, - { - "slug": "qwen3.5-ocr", - "name": "Qwen3.5-OCR", - "primaryCapability": "VU", - "capabilities": [ - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-ocr" - ], - "maxContextWindow": 65536 - }, - { - "slug": "qwen3.5-omni-flash-realtime", - "name": "Qwen3.5-Omni-Flash-Realtime", - "primaryCapability": "Realtime-Omni", - "capabilities": [ - "Realtime-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-omni-flash-realtime" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.5-omni-flash", - "name": "Qwen3.5-Omni-Flash", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-omni-flash" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.5-omni-plus-realtime", - "name": "Qwen3.5-Omni-Plus-Realtime", - "primaryCapability": "Realtime-Omni", - "capabilities": [ - "Realtime-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-omni-plus-realtime" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.5-omni-plus", - "name": "Qwen3.5-Omni-Plus", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-omni-plus" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.5-plus", - "name": "Qwen3.5-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-plus" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3.5", - "name": "Qwen3.5开源模型", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 4, - "items": [ - "qwen3.5-122b-a10b", - "qwen3.5-27b", - "qwen3.5-35b-a3b", - "qwen3.5-397b-a17b" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.6-flash", - "name": "Qwen3.6-Flash", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.6-flash" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3.6-max", - "name": "Qwen3.6-Max", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.6-max-preview" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.6-plus", - "name": "Qwen3.6-Plus", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "VU", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.6-plus" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3.6", - "name": "Qwen3.6开源模型", - "primaryCapability": "VU", - "capabilities": [ - "VU", - "TG", - "Reasoning" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen3.6-27b", - "qwen3.6-35b-a3b" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwen3.7-max", - "name": "Qwen3.7-Max", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 2, - "items": [ - "qwen3.7-max", - "qwen3.7-max-preview" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3.7-plus", - "name": "Qwen3.7-Plus", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.7-plus" - ], - "maxContextWindow": 1000000 - }, - { - "slug": "qwen3", - "name": "Qwen3开源模型", - "primaryCapability": "VU", - "capabilities": [ - "VU", - "Reasoning", - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 20, - "items": [ - "qwen3-14b", - "qwen3-235b-a22b", - "qwen3-235b-a22b-instruct-2507", - "qwen3-235b-a22b-thinking-2507", - "qwen3-30b-a3b", - "qwen3-30b-a3b-instruct-2507", - "qwen3-30b-a3b-thinking-2507", - "qwen3-32b", - "qwen3-8b", - "qwen3-coder-next", - "qwen3-next-80b-a3b-instruct", - "qwen3-next-80b-a3b-thinking", - "qwen3-vl-235b-a22b-instruct", - "qwen3-vl-235b-a22b-thinking", - "qwen3-vl-30b-a3b-instruct", - "qwen3-vl-30b-a3b-thinking", - "qwen3-vl-32b-instruct", - "qwen3-vl-32b-thinking", - "qwen3-vl-8b-instruct", - "qwen3-vl-8b-thinking" - ], - "maxContextWindow": 262144 - }, - { - "slug": "qwq-plus", - "name": "Qwen-QwQ-Plus", - "primaryCapability": "Reasoning", - "capabilities": [ - "Reasoning" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwq-plus" - ], - "maxContextWindow": 131072 - }, - { - "slug": "sambert", - "name": "Sambert语音合成", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 43, - "items": [ - "sambert-beth-v1", - "sambert-betty-v1", - "sambert-brian-v1", - "sambert-cally-v1", - "sambert-camila-v1", - "sambert-cindy-v1", - "sambert-clara-v1", - "sambert-donna-v1", - "sambert-eva-v1", - "sambert-hanna-v1", - "sambert-indah-v1", - "sambert-perla-v1", - "sambert-waan-v1", - "sambert-zhichu-v1", - "sambert-zhida-v1", - "sambert-zhide-v1", - "sambert-zhifei-v1", - "sambert-zhigui-v1", - "sambert-zhihao-v1", - "sambert-zhijia-v1", - "sambert-zhijing-v1", - "sambert-zhilun-v1", - "sambert-zhimao-v1", - "sambert-zhimiao-emo-v1", - "sambert-zhiming-v1", - "sambert-zhimo-v1", - "sambert-zhina-v1", - "sambert-zhinan-v1", - "sambert-zhiqi-v1", - "sambert-zhiqian-v1", - "sambert-zhiru-v1", - "sambert-zhishu-v1", - "sambert-zhishuo-v1", - "sambert-zhistella-v1", - "sambert-zhiting-v1", - "sambert-zhiwei-v1", - "sambert-zhixiang-v1", - "sambert-zhixiao-v1", - "sambert-zhiya-v1", - "sambert-zhiye-v1", - "sambert-zhiying-v1", - "sambert-zhiyuan-v1", - "sambert-zhiyue-v1" - ] - }, - { - "slug": "shoemodel-v1", - "name": "鞋靴模特", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "shoemodel-v1" - ] - }, - { - "slug": "siliconflow-models", - "name": "SiliconFlow DeepSeek", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning" - ], - "providers": [ - "deepseek" - ], - "itemCount": 4, - "items": [ - "siliconflow/deepseek-r1-0528", - "siliconflow/deepseek-v3-0324", - "siliconflow/deepseek-v3.1-terminus", - "siliconflow/deepseek-v3.2" - ], - "maxContextWindow": 163840 - }, - { - "slug": "speech-biasing", - "name": "语音识别热词", - "primaryCapability": "ASR", - "capabilities": [ - "ASR" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "speech-biasing" - ] - }, - { - "slug": "stepfun-models-market-place", - "name": "StepFun推理模型", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "stepfun" - ], - "itemCount": 1, - "items": [ - "stepfun/step-3.7-flash" - ], - "maxContextWindow": 262144 - }, - { - "slug": "tongyi-intent-detect-v3", - "name": "意图分类模型", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "tongyi-intent-detect-v3" - ], - "maxContextWindow": 8192 - }, - { - "slug": "tongyi-xiaomi-analysis-flash", - "name": "通义晓蜜-对话分析-flash", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "tongyi-xiaomi-analysis-flash" - ], - "maxContextWindow": 32768 - }, - { - "slug": "tongyi-xiaomi-analysis-pro", - "name": "通义晓蜜-对话分析-pro", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "tongyi-xiaomi-analysis-pro" - ], - "maxContextWindow": 32768 - }, - { - "slug": "tripo-models-market-place", - "name": "Tripo", - "primaryCapability": "3D-generation", - "capabilities": [ - "3D-generation" - ], - "providers": [ - "tripo" - ], - "itemCount": 2, - "items": [ - "Tripo/Tripo-H3.1", - "Tripo/Tripo-P1.0" - ] - }, - { - "slug": "vanchin-models-market-place", - "name": "Vanchin DeepSeek", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning", - "VU" - ], - "providers": [ - "deepseek" - ], - "itemCount": 6, - "items": [ - "vanchin/deepseek-ocr", - "vanchin/deepseek-r1", - "vanchin/deepseek-v3", - "vanchin/deepseek-v3.1-terminus", - "vanchin/deepseek-v3.2-think", - "vanchin/deepseek-v4-pro" - ], - "maxContextWindow": 1048576 - }, - { - "slug": "video-style-transform", - "name": "视频风格重绘", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "video-style-transform" - ] - }, - { - "slug": "videoretalk", - "name": "声动人像VideoRetalk", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "videoretalk" - ] - }, - { - "slug": "vidu-image-models-market-place", - "name": "Vidu AI生图", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "vidu" - ], - "itemCount": 4, - "items": [ - "vidu/vidu-image_reference2image", - "vidu/viduq2-fast_reference2image", - "vidu/viduq2-pro_reference2image", - "vidu/viduq3-fast_reference2image" - ] - }, - { - "slug": "vidu-models-market-place", - "name": "Vidu AI生视频", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "vidu" - ], - "itemCount": 20, - "items": [ - "vidu/viduq2_reference2video", - "vidu/viduq2_text2video", - "vidu/viduq2-pro_img2video", - "vidu/viduq2-pro_reference2video", - "vidu/viduq2-pro_start-end2video", - "vidu/viduq2-pro-fast_img2video", - "vidu/viduq2-turbo_img2video", - "vidu/viduq2-turbo_start-end2video", - "vidu/viduq3_reference2video", - "vidu/viduq3-ad_reference2video", - "vidu/viduq3-drama_reference2video", - "vidu/viduq3-mix_reference2video", - "vidu/viduq3-pro_img2video", - "vidu/viduq3-pro_start-end2video", - "vidu/viduq3-pro_text2video", - "vidu/viduq3-pro-fast_img2video", - "vidu/viduq3-turbo_img2video", - "vidu/viduq3-turbo_reference2video", - "vidu/viduq3-turbo_start-end2video", - "vidu/viduq3-turbo_text2video" - ] - }, - { - "slug": "virtualmodel-v2", - "name": "虚拟模特V2", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "virtualmodel-v2" - ] - }, - { - "slug": "voice-enrollment", - "name": "大模型声音复刻及声音设计", - "primaryCapability": "TTS", - "capabilities": [ - "TTS" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "voice-enrollment" - ] - }, - { - "slug": "wan-image-edit", - "name": "Wan-Image", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 5, - "items": [ - "wan2.5-i2i-preview", - "wan2.6-image", - "wan2.7-image", - "wan2.7-image-pro", - "wanx2.1-imageedit" - ] - }, - { - "slug": "wan-image-to-video", - "name": "Wan-I2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "wan" - ], - "itemCount": 14, - "items": [ - "wan2.2-animate-mix", - "wan2.2-animate-move", - "wan2.2-i2v-flash", - "wan2.2-i2v-plus", - "wan2.2-kf2v-flash", - "wan2.2-s2v", - "wan2.2-s2v-detect", - "wan2.5-i2v-preview", - "wan2.6-i2v", - "wan2.6-i2v-flash", - "wan2.7-i2v", - "wanx2.1-i2v-plus", - "wanx2.1-i2v-turbo", - "wanx2.1-kf2v-plus" - ] - }, - { - "slug": "wan-reference-to-video", - "name": "Wan-R2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "wan" - ], - "itemCount": 3, - "items": [ - "wan2.6-r2v", - "wan2.6-r2v-flash", - "wan2.7-r2v" - ] - }, - { - "slug": "wan-text-to-image", - "name": "Wan-T2I", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 9, - "items": [ - "wan2.2-t2i-flash", - "wan2.2-t2i-plus", - "wan2.5-t2i-preview", - "wan2.6-t2i", - "wanx-v1", - "wanx-v1-0521", - "wanx2.0-t2i-turbo", - "wanx2.1-t2i-plus", - "wanx2.1-t2i-turbo" - ] - }, - { - "slug": "wan-text-to-video", - "name": "Wan-T2V", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "wan" - ], - "itemCount": 6, - "items": [ - "wan2.2-t2v-plus", - "wan2.5-t2v-preview", - "wan2.6-t2v", - "wan2.7-t2v", - "wanx2.1-t2v-plus", - "wanx2.1-t2v-turbo" - ] - }, - { - "slug": "wan-video-edit", - "name": "Wan-VideoEdit", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wan2.7-videoedit" - ] - }, - { - "slug": "wanx-background-generation-v2", - "name": "图像背景生成", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-background-generation-v2" - ] - }, - { - "slug": "wanx-poster-generation-v1", - "name": "创意海报生成", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-poster-generation-v1" - ] - }, - { - "slug": "wanx-sketch-to-image-lite", - "name": "万相-涂鸦作画", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-sketch-to-image-lite" - ] - }, - { - "slug": "wanx-style-repaint-v1", - "name": "人像风格重绘", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-style-repaint-v1" - ] - }, - { - "slug": "wanx-virtualmodel", - "name": "虚拟模特", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-virtualmodel" - ] - }, - { - "slug": "wanx-x-painting", - "name": "万相-图像局部重绘", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx-x-painting" - ] - }, - { - "slug": "wanx2.1-vace-plus", - "name": "Wan2.1-VACE-Plus", - "primaryCapability": "VG", - "capabilities": [ - "VG" - ], - "providers": [ - "wan" - ], - "itemCount": 1, - "items": [ - "wanx2.1-vace-plus" - ] - }, - { - "slug": "wordart-semantic", - "name": "WordArt锦书-文字变形", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "wordart-semantic" - ] - }, - { - "slug": "wordart-texture", - "name": "WordArt锦书-文字纹理生成", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "wordart-texture" - ] - }, - { - "slug": "xiaomi-models-market-place", - "name": "MiMo文本模型", - "primaryCapability": "TG", - "capabilities": [ - "TG" - ], - "providers": [ - "xiaomi" - ], - "itemCount": 1, - "items": [ - "xiaomi/mimo-v2.5-pro" - ], - "maxContextWindow": 1048576 - }, - { - "slug": "z-image-turbo", - "name": "Z-Image-Turbo", - "primaryCapability": "IG", - "capabilities": [ - "IG" - ], - "providers": [ - "qwen-domain-model" - ], - "itemCount": 1, - "items": [ - "z-image-turbo" - ] - }, - { - "slug": "zhipu-models-market-place", - "name": "智谱GLM系列文本模型", - "primaryCapability": "TG", - "capabilities": [ - "TG", - "Reasoning" - ], - "providers": [ - "zhipu-ai" - ], - "itemCount": 3, - "items": [ - "ZHIPU/GLM-5", - "ZHIPU/GLM-5.1", - "ZHIPU/GLM-5.2" + "qwen3.7-plus" ], - "maxContextWindow": 1048576 + "maxContextWindow": 1000000 } ] } diff --git a/skills/bailian-docs-llm-wiki/models/index.md b/skills/bailian-docs-llm-wiki/models/index.md index 7fc0ab8b..6ed87f9e 100644 --- a/skills/bailian-docs-llm-wiki/models/index.md +++ b/skills/bailian-docs-llm-wiki/models/index.md @@ -1,6 +1,6 @@ # 百炼模型市场索引 -> 自动生成 · 共 169 个模型家族 · 382 个主干模型 · 更新于 2026-07-15 +> 自动生成 · 共 11 个模型家族 · 26 个主干模型 · 更新于 2026-07-15 **机器查询走结构化文件**： @@ -11,252 +11,12 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[].slug`。 -## 文本生成 `TG` — 35 个家族 - -- [GLM](groups/glm-4.5.json) — GLM是由智谱提供的开源模型。 - - 模型：`glm-4.5`, `glm-4.5-air`, `glm-4.6`, `glm-4.7`, `glm-5`, `glm-5.1`, `glm-5.2` -- [GLM-5.2-Fast](groups/glm-fast.json) — GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过… - - 模型：`glm-5.2-fast-preview` -- [Kimi](groups/Kimi-K2.json) — Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。 - - 模型：`kimi-k2-thinking`, `kimi-k2.5`, `kimi-k2.6`, `kimi-k2.7-code`, `Moonshot-Kimi-K2-Instruct` -- [Kimi](groups/kimi-models-market-place.json) — 由月之暗面提供的Kimi系列模型的API服务。 - - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed` -- [MiMo文本模型](groups/xiaomi-models-market-place.json) — 由小米MiMo提供的MiMo文本模型API服务 - - 模型：`xiaomi/mimo-v2.5-pro` -- [MiniMax文本模型](groups/minimax-models-market-place.json) — 由MiniMax提供的MiniMax-M系列文本模型API服务。 - - 模型：`MiniMax/MiniMax-M2.1`, `MiniMax/MiniMax-M2.5`, `MiniMax/MiniMax-M2.7`, `MiniMax/MiniMax-M3` -- [Qwen-Coder-Plus](groups/qwen-coder-plus.json) — 千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。 - - 模型：`qwen-coder-plus` -- [Qwen-Coder-Turbo](groups/qwen-coder-turbo.json) — Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。 - - 模型：`qwen-coder-turbo` -- [qwen-deep-research](groups/qwen-deep-research.json) — 千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。 - - 模型：`qwen-deep-research` -- [Qwen-Doc-Turbo](groups/qwen-doc-turbo.json) — 快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。 - - 模型：`qwen-doc-turbo` -- [Qwen-Flash-Character](groups/qwen-flash-character.json) — 千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 - - 模型：`qwen-flash-character` -- [Qwen-Long](groups/qwen-long.json) — Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服… - - 模型：`qwen-long`, `qwen-long-latest` -- [Qwen-Math-Plus](groups/qwen-math-plus.json) — Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。 - - 模型：`qwen-math-plus`, `qwen-math-plus-0816`, `qwen-math-plus-0919`, `qwen-math-plus-latest` -- [Qwen-Math-Turbo](groups/qwen-math-turbo.json) — Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。 - - 模型：`qwen-math-turbo` -- [Qwen-Max](groups/qwen-max.json) — 千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。 - - 模型：`qwen-max` -- [Qwen-MT-Flash](groups/qwen-mt-flash.json) — 基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 - - 模型：`qwen-mt-flash` -- [Qwen-MT-Lite](groups/qwen-mt-lite.json) — 基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 - - 模型：`qwen-mt-lite` -- [Qwen-MT-Plus](groups/qwen-mt-plus.json) — 基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 - - 模型：`qwen-mt-plus` -- [Qwen-MT-Turbo](groups/qwen-mt-turbo.json) — 基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 - - 模型：`qwen-mt-turbo` -- [Qwen-Plus-Character](groups/qwen-plus-character.json) — 千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 - - 模型：`qwen-plus-character` -- [Qwen3-Coder-30B-A3B-Instruct](groups/qwen3-coder-30b-a3b-instruct.json) — 基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 - - 模型：`qwen3-coder-30b-a3b-instruct` -- [Qwen3-Coder-480B-A35B-Instruct](groups/qwen3-coder-480b-a35b-instruct.json) — 基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 - - 模型：`qwen3-coder-480b-a35b-instruct` -- [Qwen3-Coder-Flash](groups/qwen3-coder-flash.json) — 基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 - - 模型：`qwen3-coder-flash` -- [Qwen3-Coder-Plus](groups/qwen3-coder-plus.json) — 基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 - - 模型：`qwen3-coder-plus` -- [Qwen3-Max](groups/qwen3-max.json) — 千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 - - 模型：`qwen3-max`, `qwen3-max-preview` -- [Qwen3.5-Plus](groups/qwen3.5-plus.json) — Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 - - 模型：`qwen3.5-plus` -- [Qwen3.7-Plus](groups/qwen3.7-plus.json) — Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真… - - 模型：`qwen3.7-plus` -- [SiliconFlow DeepSeek](groups/siliconflow-models.json) — 由硅基流动提供的DeepSeek系列模型API服务。 - - 模型：`siliconflow/deepseek-r1-0528`, `siliconflow/deepseek-v3-0324`, `siliconflow/deepseek-v3.1-terminus`, `siliconflow/deepseek-v3.2` -- [StepFun推理模型](groups/stepfun-models-market-place.json) — 由阶跃星辰StepFun提供的Step系列推理模型API服务 - - 模型：`stepfun/step-3.7-flash` -- [Vanchin DeepSeek](groups/vanchin-models-market-place.json) — 由快手万擎提供的DeepSeek系列模型API服务。 - - 模型：`vanchin/deepseek-ocr`, `vanchin/deepseek-r1`, `vanchin/deepseek-v3`, `vanchin/deepseek-v3.1-terminus`, `vanchin/deepseek-v3.2-think`, `vanchin/deepseek-v4-pro` -- [意图分类模型](groups/tongyi-intent-detect-v3.json) — 意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果… - - 模型：`tongyi-intent-detect-v3` -- [智谱GLM系列文本模型](groups/zhipu-models-market-place.json) — 由智谱提供的GLM系列文本模型API服务 - - 模型：`ZHIPU/GLM-5`, `ZHIPU/GLM-5.1`, `ZHIPU/GLM-5.2` -- [通义晓蜜-对话分析-flash](groups/tongyi-xiaomi-analysis-flash.json) — 通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。 - - 模型：`tongyi-xiaomi-analysis-flash` -- [通义晓蜜-对话分析-pro](groups/tongyi-xiaomi-analysis-pro.json) — 通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。 - - 模型：`tongyi-xiaomi-analysis-pro` -- [通义法睿-Plus-32K](groups/farui-plus.json) — 通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分… - - 模型：`farui-plus` - -## 图像生成 `IG` — 30 个家族 - -- [AI试衣-Plus版](groups/aitryon-plus.json) — aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服… - - 模型：`aitryon-plus` -- [AI试衣-基础版](groups/aitryon.json) — aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。 - - 模型：`aitryon` -- [AI试衣OutfitAnyone-图片分割](groups/aitryon-parsing-v1.json) — 图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。 - - 模型：`aitryon-parsing-v1` -- [AI试衣OutfitAnyone-图片精修](groups/aitryon-refiner.json) — 图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。 - - 模型：`aitryon-refiner` -- [FaceChain人物写真生成](groups/facechain-generation.json) — 基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。 - - 模型：`facechain-generation` -- [FaceChain人物图像检测](groups/facechain-facedetect.json) — 对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。 - - 模型：`facechain-facedetect` -- [Qwen-Image-2.0](groups/qwen-image-2.0.json) — Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模… - - 模型：`qwen-image-2.0` -- [Qwen-Image-2.0-Pro](groups/qwen-image-2.0-pro.json) — Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系… - - 模型：`qwen-image-2.0-pro` -- [Qwen-Image-Edit-Max](groups/qwen-image-edit-max.json) — 千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。 - - 模型：`qwen-image-edit-max` -- [Qwen-Image-Edit-Plus](groups/qwen-image-edit.json) — 千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 - - 模型：`qwen-image-edit`, `qwen-image-edit-plus` -- [Qwen-Image-Max](groups/qwen-image-max.json) — 千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。 - - 模型：`qwen-image-max` -- [Qwen-Image-Plus](groups/qwen-image-plus.json) — 千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。 - - 模型：`qwen-image`, `qwen-image-plus` -- [Qwen-MT-Image](groups/qwen-mt-image.json) — 专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。 - - 模型：`qwen-mt-image` -- [Vidu AI生图](groups/vidu-image-models-market-place.json) — 由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。 - - 模型：`vidu/vidu-image_reference2image`, `vidu/viduq2-fast_reference2image`, `vidu/viduq2-pro_reference2image`, `vidu/viduq3-fast_reference2image` -- [Wan-Image](groups/wan-image-edit.json) — 指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等 - - 模型：`wan2.5-i2i-preview`, `wan2.6-image`, `wan2.7-image`, `wan2.7-image-pro`, `wanx2.1-imageedit` -- [Wan-T2I](groups/wan-text-to-image.json) — 文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现 - - 模型：`wan2.2-t2i-flash`, `wan2.2-t2i-plus`, `wan2.5-t2i-preview`, `wan2.6-t2i`, `wanx-v1`, `wanx-v1-0521`, `wanx2.0-t2i-turbo`, `wanx2.1-t2i-plus`, `wanx2.1-t2i-turbo` -- [WordArt锦书-文字变形](groups/wordart-semantic.json) — WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。 - - 模型：`wordart-semantic` -- [WordArt锦书-文字纹理生成](groups/wordart-texture.json) — WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海… - - 模型：`wordart-texture` -- [Z-Image-Turbo](groups/z-image-turbo.json) — Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双… - - 模型：`z-image-turbo` -- [万相-图像局部重绘](groups/wanx-x-painting.json) — 万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局… - - 模型：`wanx-x-painting` -- [万相-涂鸦作画](groups/wanx-sketch-to-image-lite.json) — 万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、… - - 模型：`wanx-sketch-to-image-lite` -- [人像风格重绘](groups/wanx-style-repaint-v1.json) — 人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。 - - 模型：`wanx-style-repaint-v1` -- [人物实例分割](groups/image-instance-segmentation.json) — 人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。 - - 模型：`image-instance-segmentation` -- [创意海报生成](groups/wanx-poster-generation-v1.json) — 创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。 - - 模型：`wanx-poster-generation-v1` -- [图像擦除补全](groups/image-erase-completion.json) — 图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计… - - 模型：`image-erase-completion` -- [图像画面扩展](groups/image-out-painting.json) — 图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意… - - 模型：`image-out-painting` -- [图像背景生成](groups/wanx-background-generation-v2.json) — 图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。 - - 模型：`wanx-background-generation-v2` -- [虚拟模特](groups/wanx-virtualmodel.json) — 虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如… - - 模型：`wanx-virtualmodel` -- [虚拟模特V2](groups/virtualmodel-v2.json) — 虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如… - - 模型：`virtualmodel-v2` -- [鞋靴模特](groups/shoemodel-v1.json) — 鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新… - - 模型：`shoemodel-v1` - -## 视频生成 `VG` — 25 个家族 - -- [HappyHorse-I2V](groups/happyhorse-i2v.json) — HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。 - - 模型：`happyhorse-1.0-i2v`, `happyhorse-1.1-i2v` -- [HappyHorse-R2V](groups/happyhorse-r2v.json) — HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 - - 模型：`happyhorse-1.0-r2v`, `happyhorse-1.1-r2v` -- [HappyHorse-T2V](groups/happyhorse-t2v.json) — HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 - - 模型：`happyhorse-1.0-t2v`, `happyhorse-1.1-t2v` -- [HappyHorse-Video-Edit](groups/happyhorse-video-edit.json) — HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 - - 模型：`happyhorse-1.0-video-edit` -- [PixVerse C1](groups/pixverse-c1-market-place.json) — 由爱诗科技提供的PixVerse C系列视频大模型API服务。 - - 模型：`pixverse/pixverse-c1-it2v`, `pixverse/pixverse-c1-kf2v`, `pixverse/pixverse-c1-r2v`, `pixverse/pixverse-c1-t2v` -- [PixVerse V5.6](groups/pixverse-market-place.json) — 由爱诗科技提供的PixVerse V系列视频大模型API服务。 - - 模型：`pixverse/pixverse-v5.6-it2v`, `pixverse/pixverse-v5.6-kf2v`, `pixverse/pixverse-v5.6-r2v`, `pixverse/pixverse-v5.6-t2v` -- [PixVerse V6](groups/pixverse-v6-market-place.json) — 由爱诗科技提供的PixVerse V系列视频大模型API服务。 - - 模型：`pixverse/pixverse-v6-it2v`, `pixverse/pixverse-v6-kf2v`, `pixverse/pixverse-v6-r2v`, `pixverse/pixverse-v6-t2v` -- [Vidu AI生视频](groups/vidu-models-market-place.json) — 由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。 - - 模型：`vidu/viduq2_reference2video`, `vidu/viduq2_text2video`, `vidu/viduq2-pro_img2video`, `vidu/viduq2-pro_reference2video`, `vidu/viduq2-pro_start-end2video`, `vidu/viduq2-pro-fast_img2video`, `vidu/viduq2-turbo_img2video`, `vidu/viduq2-turbo_start-end2video`, `vidu/viduq3_reference2video`, `vidu/viduq3-ad_reference2video`, `vidu/viduq3-drama_reference2video`, `vidu/viduq3-mix_reference2video`, `vidu/viduq3-pro_img2video`, `vidu/viduq3-pro_start-end2video`, `vidu/viduq3-pro_text2video`, `vidu/viduq3-pro-fast_img2video`, `vidu/viduq3-turbo_img2video`, `vidu/viduq3-turbo_reference2video`, `vidu/viduq3-turbo_start-end2video`, `vidu/viduq3-turbo_text2video` -- [Wan-I2V](groups/wan-image-to-video.json) — 图片生成视频内容，稳定保持图像主体、风格和文字等细节信息 - - 模型：`wan2.2-animate-mix`, `wan2.2-animate-move`, `wan2.2-i2v-flash`, `wan2.2-i2v-plus`, `wan2.2-kf2v-flash`, `wan2.2-s2v`, `wan2.2-s2v-detect`, `wan2.5-i2v-preview`, `wan2.6-i2v`, `wan2.6-i2v-flash`, `wan2.7-i2v`, `wanx2.1-i2v-plus`, `wanx2.1-i2v-turbo`, `wanx2.1-kf2v-plus` -- [Wan-R2V](groups/wan-reference-to-video.json) — 参考视频中的人或物，精准保持形象和声音，支持多参考合拍 - - 模型：`wan2.6-r2v`, `wan2.6-r2v-flash`, `wan2.7-r2v` -- [Wan-T2V](groups/wan-text-to-video.json) — 文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循 - - 模型：`wan2.2-t2v-plus`, `wan2.5-t2v-preview`, `wan2.6-t2v`, `wan2.7-t2v`, `wanx2.1-t2v-plus`, `wanx2.1-t2v-turbo` -- [Wan-VideoEdit](groups/wan-video-edit.json) — 通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等 - - 模型：`wan2.7-videoedit` -- [Wan2.1-VACE-Plus](groups/wanx2.1-vace-plus.json) — 万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。 - - 模型：`wanx2.1-vace-plus` -- [可灵AI](groups/kling-models-market-place.json) — 由可灵AI提供的高质量视频与图像生成及编辑模型。 - - 模型：`kling/kling-v3-image-generation`, `kling/kling-v3-omni-image-generation`, `kling/kling-v3-omni-video-generation`, `kling/kling-v3-video-generation` -- [声动人像VideoRetalk](groups/videoretalk.json) — VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。 - - 模型：`videoretalk` -- [悦动人像EMO](groups/emo-v1.json) — EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。 - - 模型：`emo-v1` -- [悦动人像EMO-detect](groups/emo-detect-v1.json) — EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 - - 模型：`emo-detect-v1` -- [灵动人像LivePortrait](groups/liveportrait.json) — LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。 - - 模型：`liveportrait` -- [灵动人像LivePortrait-detect](groups/liveportrait-detect.json) — LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 - - 模型：`liveportrait-detect` -- [视频风格重绘](groups/video-style-transform.json) — 视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡… - - 模型：`video-style-transform` -- [舞动人像AnimateAnyone](groups/animate-anyone-gen2.json) — AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。 - - 模型：`animate-anyone-gen2` -- [舞动人像AnimateAnyone-detect](groups/animate-anyone-detect-gen2.json) — AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 - - 模型：`animate-anyone-detect-gen2` -- [舞动人像AnimateAnyone-template](groups/animate-anyone-template-gen2.json) — AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。 - - 模型：`animate-anyone-template-gen2` -- [表情包Emoji](groups/emoji-v1.json) — 表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。 - - 模型：`emoji-v1` -- [表情包Emoji-detect](groups/emoji-detect-v1.json) — 表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 - - 模型：`emoji-detect-v1` - -## 语音合成 `TTS` — 16 个家族 - -- [CosyVoice大模型](groups/cosyvoice.json) — 基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。 - - 模型：`cosyvoice-clone-v1`, `cosyvoice-v1`, `cosyvoice-v2`, `cosyvoice-v3-flash`, `cosyvoice-v3-plus`, `cosyvoice-v3.5-flash`, `cosyvoice-v3.5-plus` -- [MiniMax-Speech系列语音模型](groups/MiniMax-speech-market-place.json) — 由MiniMax提供的MiniMax-Speech系列语音模型API服务。 - - 模型：`MiniMax/speech-02-hd`, `MiniMax/speech-02-turbo`, `MiniMax/speech-2.8-hd`, `MiniMax/speech-2.8-turbo` -- [Qwen-TTS](groups/qwen-tts.json) — 千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。 - - 模型：`qwen-tts`, `qwen-tts-latest` -- [Qwen-声音复刻](groups/qwen-voice-enrollment.json) — 千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复… - - 模型：`qwen-voice-enrollment` -- [Qwen-声音设计](groups/qwen-voice-design.json) — Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11… - - 模型：`qwen-voice-design` -- [Qwen3-TTS-Flash](groups/qwen3-tts-flash.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量… - - 模型：`qwen3-tts-flash` -- [Qwen3-TTS-Flash-Realtime](groups/qwen3-tts-flash-realtime.json) — Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输… - - 模型：`qwen3-tts-flash-realtime` -- [Qwen3-TTS-Instruct-Flash](groups/qwen3-tts-instruct-flash.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中… - - 模型：`qwen3-tts-instruct-flash` -- [qwen3-tts-instruct-flash-realtime](groups/qwen3-tts-instruct-flash-realtime.json) — 通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文I… - - 模型：`qwen3-tts-instruct-flash-realtime` -- [Qwen3-TTS-VC](groups/qwen3-tts-vc.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模… - - 模型：`qwen3-tts-vc-2026-01-22` -- [Qwen3-TTS-VC-Realtime](groups/qwen3-tts-vc-realtime.json) — Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的… - - 模型：`qwen3-tts-vc-realtime-2026-01-15` -- [Qwen3-TTS-VD](groups/qwen3-tts-vd.json) — Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数… - - 模型：`qwen3-tts-vd-2026-01-26` -- [Qwen3-TTS-VD-Realtime](groups/qwen3-tts-vd-realtime.json) — Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数… - - 模型：`qwen3-tts-vd-realtime-2026-01-15` -- [Sambert语音合成](groups/sambert.json) — 提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。 - - 模型：`sambert-beth-v1`, `sambert-betty-v1`, `sambert-brian-v1`, `sambert-cally-v1`, `sambert-camila-v1`, `sambert-cindy-v1`, `sambert-clara-v1`, `sambert-donna-v1`, `sambert-eva-v1`, `sambert-hanna-v1`, `sambert-indah-v1`, `sambert-perla-v1`, `sambert-waan-v1`, `sambert-zhichu-v1`, `sambert-zhida-v1`, `sambert-zhide-v1`, `sambert-zhifei-v1`, `sambert-zhigui-v1`, `sambert-zhihao-v1`, `sambert-zhijia-v1`, `sambert-zhijing-v1`, `sambert-zhilun-v1`, `sambert-zhimao-v1`, `sambert-zhimiao-emo-v1`, `sambert-zhiming-v1`, `sambert-zhimo-v1`, `sambert-zhina-v1`, `sambert-zhinan-v1`, `sambert-zhiqi-v1`, `sambert-zhiqian-v1`, `sambert-zhiru-v1`, `sambert-zhishu-v1`, `sambert-zhishuo-v1`, `sambert-zhistella-v1`, `sambert-zhiting-v1`, `sambert-zhiwei-v1`, `sambert-zhixiang-v1`, `sambert-zhixiao-v1`, `sambert-zhiya-v1`, `sambert-zhiye-v1`, `sambert-zhiying-v1`, `sambert-zhiyuan-v1`, `sambert-zhiyue-v1` -- [大模型声音复刻及声音设计](groups/voice-enrollment.json) — 大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。 大模型声音设计使用FunAudioGen-VD模型… - - 模型：`voice-enrollment` -- [音乐生成](groups/fun-music.json) — 百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。 - - 模型：`fun-music-preview`, `fun-music-v1` - -## 推理 `Reasoning` — 14 个家族 +## 推理 `Reasoning` — 6 个家族 - [DeepSeek](groups/deepseek.json) — DeepSeek是由深度求索提供的开源模型，包含 V3.1、V3、R1以及基于Qwen2.5系列蒸馏的大语言模型。 - 模型：`deepseek-r1`, `deepseek-r1-0528`, `deepseek-r1-distill-qwen-1.5b`, `deepseek-r1-distill-qwen-14b`, `deepseek-r1-distill-qwen-32b`, `deepseek-r1-distill-qwen-7b`, `deepseek-v3`, `deepseek-v3.1`, `deepseek-v3.2`, `deepseek-v3.2-exp`, `deepseek-v4-flash`, `deepseek-v4-pro` -- [MiniMax](groups/MiniMax-M2.1.json) — MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。 - - 模型：`MiniMax-M2.1`, `MiniMax-M2.5` -- [QVQ-Max](groups/qvq-max.json) — 千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 - - 模型：`qvq-max` -- [Qwen-Flash](groups/qwen-flash.json) — Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 - - 模型：`qwen-flash` -- [Qwen-Plus](groups/qwen-plus.json) — 千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。 - - 模型：`qwen-plus`, `qwen-plus-0112`, `qwen-plus-1220`, `qwen-plus-latest` -- [Qwen-QVQ-Plus](groups/qvq-plus.json) — 千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 - - 模型：`qvq-plus` -- [Qwen-QwQ-Plus](groups/qwq-plus.json) — 千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（… - - 模型：`qwq-plus` -- [Qwen-Turbo](groups/qwen-turbo.json) — 千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。 - - 模型：`qwen-turbo` - [Qwen3.5-Flash](groups/qwen3.5-flash.json) — Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 - 模型：`qwen3.5-flash` -- [Qwen3.5开源模型](groups/qwen3.5.json) — Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。 - - 模型：`qwen3.5-122b-a10b`, `qwen3.5-27b`, `qwen3.5-35b-a3b`, `qwen3.5-397b-a17b` - [Qwen3.6-Flash](groups/qwen3.6-flash.json) — Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉… - 模型：`qwen3.6-flash` - [Qwen3.6-Max](groups/qwen3.6-max.json) — Qwen3.6原生Max模型，相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提… @@ -266,133 +26,21 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Qwen3.7-Max](groups/qwen3.7-max.json) — Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自… - 模型：`qwen3.7-max`, `qwen3.7-max-preview` -## 语音识别 `ASR` — 12 个家族 - -- [Fun-ASR-Flash](groups/fun-asr-flash.json) — 百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词… - - 模型：`fun-asr-flash-2026-06-15` -- [Fun-ASR语音识别](groups/fun-asr.json) — 通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。 - - 模型：`fun-asr`, `fun-asr-mtl` -- [Paraformer语音识别-8k-v1](groups/paraformer-8k-v1.json) — Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。 - - 模型：`paraformer-8k-v1` -- [Paraformer语音识别-8k-v2](groups/paraformer-8k-v2.json) — Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。 - - 模型：`paraformer-8k-v2` -- [Paraformer语音识别-mtl-v1](groups/paraformer-mtl-v1.json) — Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话… - - 模型：`paraformer-mtl-v1` -- [Paraformer语音识别-v1](groups/paraformer-v1.json) — Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 - - 模型：`paraformer-v1` -- [Paraformer语音识别-v2](groups/paraformer-v2.json) — 推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）… - - 模型：`paraformer-v2` -- [Qwen3-ASR-Flash](groups/qwen3-asr-flash.json) — Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精… - - 模型：`qwen3-asr-flash` -- [Qwen3-ASR-Flash-Filetrans](groups/qwen3-asr-flash-filetrans.json) — Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音… - - 模型：`qwen3-asr-flash-filetrans` -- [Qwen3-Omni-30b-a3b-Captioner](groups/qwen3-omni-30b-a3b-captioner.json) — 千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音… - - 模型：`qwen3-omni-30b-a3b-captioner` -- [一句话识别及翻译V1.0](groups/gummy-chat-v1.json) — 多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 - - 模型：`gummy-chat-v1` -- [语音识别热词](groups/speech-biasing.json) — 热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行… - - 模型：`speech-biasing` - -## 视觉理解 `VU` — 9 个家族 - -- [GUI-Plus](groups/gui-plus.json) — GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作… - - 模型：`gui-plus` -- [Qwen-VL-Max](groups/qwen-vl-max.json) — Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。 - - 模型：`qwen-vl-max` -- [Qwen-VL-OCR](groups/qwen-vl-ocr.json) — Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 - - 模型：`qwen-vl-ocr`, `qwen-vl-ocr-1028`, `qwen-vl-ocr-latest` -- [Qwen-VL-Plus](groups/qwen-vl-plus.json) — Qwen-VL-Plus，即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。 - - 模型：`qwen-vl-plus` -- [Qwen3-VL-Flash](groups/qwen3-vl-flash.json) — Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识… - - 模型：`qwen3-vl-flash` -- [Qwen3-VL-Plus](groups/qwen3-vl-plus.json) — Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与… - - 模型：`qwen3-vl-plus` -- [Qwen3.5-OCR](groups/qwen3.5-ocr.json) — Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。 - - 模型：`qwen3.5-ocr` -- [Qwen3.6开源模型](groups/qwen3.6.json) — Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。 - - 模型：`qwen3.6-27b`, `qwen3.6-35b-a3b` -- [Qwen3开源模型](groups/qwen3.json) — Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。 - - 模型：`qwen3-14b`, `qwen3-235b-a22b`, `qwen3-235b-a22b-instruct-2507`, `qwen3-235b-a22b-thinking-2507`, `qwen3-30b-a3b`, `qwen3-30b-a3b-instruct-2507`, `qwen3-30b-a3b-thinking-2507`, `qwen3-32b`, `qwen3-8b`, `qwen3-coder-next`, `qwen3-next-80b-a3b-instruct`, `qwen3-next-80b-a3b-thinking`, `qwen3-vl-235b-a22b-instruct`, `qwen3-vl-235b-a22b-thinking`, `qwen3-vl-30b-a3b-instruct`, `qwen3-vl-30b-a3b-thinking`, `qwen3-vl-32b-instruct`, `qwen3-vl-32b-thinking`, `qwen3-vl-8b-instruct`, `qwen3-vl-8b-thinking` - -## 实时语音识别 `Realtime-ASR` — 7 个家族 +## 视频生成 `VG` — 2 个家族 -- [Fun-ASR实时语音识别](groups/fun-asr-realtime.json) — 通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/… - - 模型：`fun-asr-flash-8k-realtime`, `fun-asr-realtime` -- [Paraformer实时语音识别-8k-v1](groups/paraformer-realtime-8k-v1.json) — Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。 - - 模型：`paraformer-realtime-8k-v1` -- [Paraformer实时语音识别-8k-v2](groups/paraformer-realtime-8k-v2.json) — 推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服… - - 模型：`paraformer-realtime-8k-v2` -- [Paraformer实时语音识别-v1](groups/paraformer-realtime-v1.json) — Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。 - - 模型：`paraformer-realtime-v1` -- [Paraformer实时语音识别-v2](groups/paraformer-realtime-v2.json) — 推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支… - - 模型：`paraformer-realtime-v2` -- [Qwen3-ASR-Flash-Realtime](groups/qwen3-asr-flash-realtime.json) — Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实… - - 模型：`qwen3-asr-flash-realtime` -- [Qwen3-LiveTranslate-Flash](groups/qwen3-livetranslate-flash.json) — Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，… - - 模型：`qwen3-livetranslate-flash` - -## 全模态 `Multimodal-Omni` — 5 个家族 - -- [Qwen-Omni-Turbo](groups/qwen-omni-turbo.json) — 千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 - - 模型：`qwen-omni-turbo`, `qwen-omni-turbo-latest` -- [Qwen2.5-开源模型](groups/qwen2.5.json) — Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。 - - 模型：`qwen2.5-omni-7b` -- [Qwen3-Omni-Flash](groups/qwen3-omni-flash.json) — Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互… - - 模型：`qwen3-omni-flash` -- [Qwen3.5-Omni-Flash](groups/qwen3.5-omni-flash.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… - - 模型：`qwen3.5-omni-flash` -- [Qwen3.5-Omni-Plus](groups/qwen3.5-omni-plus.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… - - 模型：`qwen3.5-omni-plus` - -## 实时全模态 `Realtime-Omni` — 4 个家族 - -- [Qwen-Omni-Turbo-Realtime](groups/qwen-omni-turbo-realtime.json) — 千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。 - - 模型：`qwen-omni-turbo-realtime`, `qwen-omni-turbo-realtime-latest` -- [Qwen3-Omni-Flash-Realtime](groups/qwen3-omni-flash-realtime.json) — Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文… - - 模型：`qwen3-omni-flash-realtime` -- [Qwen3.5-Omni-Flash-Realtime](groups/qwen3.5-omni-flash-realtime.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话… - - 模型：`qwen3.5-omni-flash-realtime` -- [Qwen3.5-Omni-Plus-Realtime](groups/qwen3.5-omni-plus-realtime.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话… - - 模型：`qwen3.5-omni-plus-realtime` - -## 实时音频翻译 `Realtime-Audio-Translate` — 3 个家族 - -- [Qwen3-LiveTranslate-Flash-Realtime](groups/qwen3-livetranslate-flash-realtime.json) — Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言… - - 模型：`qwen3-livetranslate-flash-realtime` -- [Qwen3.5-LiveTranslate-Flash-Realtime](groups/qwen3.5-livetranslate-flash-realtime.json) — Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐… - - 模型：`qwen3.5-livetranslate-flash-realtime` -- [实时语音识别及翻译V1.0](groups/gummy-realtime-v1.json) — 多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 - - 模型：`gummy-realtime-v1` - -## 多模态嵌入 `ME` — 2 个家族 - -- [Qwen-VL-Embedding](groups/qwen-vl-embedding.json) — 基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。 - - 模型：`qwen2.5-vl-embedding`, `qwen3-vl-embedding` -- [通义多模态向量](groups/embedding.json) — 基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。 - - 模型：`multimodal-embedding-v1`, `tongyi-embedding-vision-flash`, `tongyi-embedding-vision-plus` - -## Realtime-Chatting `Realtime-Chatting` — 2 个家族 - -- [Qwen-Audio-Realtime-Flash](groups/qwen-audio-realtime-flash.json) — Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和… - - 模型：`qwen-audio-3.0-realtime-flash` -- [Qwen-Audio-Realtime-Plus](groups/qwen-audio-realtime-plus.json) — Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和… - - 模型：`qwen-audio-3.0-realtime-plus` - -## 实时语音合成 `Realtime-Text-to-Speech` — 2 个家族 - -- [Qwen-Audio-TTS](groups/qwen-audio-tts.json) — Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，… - - 模型：`qwen-audio-3.0-tts-flash`, `qwen-audio-3.0-tts-plus` -- [Qwen-TTS-Realtime](groups/qwen-tts-realtime.json) — Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。 - - 模型：`qwen-tts-realtime`, `qwen-tts-realtime-latest` +- [HappyHorse-I2V](groups/happyhorse-i2v.json) — HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。 + - 模型：`happyhorse-1.0-i2v`, `happyhorse-1.1-i2v` +- [HappyHorse-T2V](groups/happyhorse-t2v.json) — HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + - 模型：`happyhorse-1.0-t2v`, `happyhorse-1.1-t2v` -## 翻译 `TR` — 2 个家族 +## 文本生成 `TG` — 2 个家族 -- [Qwen-Embedding](groups/qwen-embedding.json) — 基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度… - - 模型：`text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` -- [Qwen-Rerank](groups/qwen-rerank.json) — 基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。 - - 模型：`gte-rerank-v2`, `qwen3-rerank`, `qwen3-vl-rerank` +- [Qwen3.5-Plus](groups/qwen3.5-plus.json) — Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 + - 模型：`qwen3.5-plus` +- [Qwen3.7-Plus](groups/qwen3.7-plus.json) — Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真… + - 模型：`qwen3.7-plus` -## 3D 生成 `3D-generation` — 1 个家族 +## 视觉理解 `VU` — 1 个家族 -- [Tripo](groups/tripo-models-market-place.json) — AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。 - - 模型：`Tripo/Tripo-H3.1`, `Tripo/Tripo-P1.0` +- [Qwen3.6开源模型](groups/qwen3.6.json) — Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。 + - 模型：`qwen3.6-27b`, `qwen3.6-35b-a3b` diff --git a/skills/bailian-docs-llm-wiki/models/models.jsonl b/skills/bailian-docs-llm-wiki/models/models.jsonl index f292e3b7..adcdab2f 100644 --- a/skills/bailian-docs-llm-wiki/models/models.jsonl +++ b/skills/bailian-docs-llm-wiki/models/models.jsonl @@ -1,28 +1,3 @@ -{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} -{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} -{"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","family":"aitryon-refiner","familyName":"AI试衣OutfitAnyone-图片精修","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796663.html","detailPath":"groups/aitryon-refiner.json"} -{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} -{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} -{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} -{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} -{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} {"model":"deepseek-r1","name":"DeepSeek-R1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_batch","unit":"每百万tokens","price":"2"},{"type":"output_token_batch","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} @@ -35,348 +10,17 @@ {"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} -{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} -{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} -{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} -{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} -{"model":"facechain-facedetect","name":"FaceChain人物图像检测","family":"facechain-facedetect","familyName":"FaceChain人物图像检测","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712507.html","detailPath":"groups/facechain-facedetect.json"} -{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} -{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} -{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} -{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} -{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} -{"model":"glm-4.5","name":"GLM-4.5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-4.5-air","name":"GLM-4.5-Air","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-4.6","name":"GLM-4.6","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["Reasoning","TG"],"features":["model-experience","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-4.7","name":"GLM-4.7","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5","name":"GLM-5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.1","name":"GLM-5.1","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":202745,"maxInputTokens":202745,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} -{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} -{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} -{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} -{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} -{"model":"image-erase-completion","name":"图像擦除补全","family":"image-erase-completion","familyName":"图像擦除补全","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840907.html","detailPath":"groups/image-erase-completion.json"} -{"model":"image-instance-segmentation","name":"人物实例分割","family":"image-instance-segmentation","familyName":"人物实例分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840906.html","detailPath":"groups/image-instance-segmentation.json"} -{"model":"image-out-painting","name":"图像画面扩展","family":"image-out-painting","familyName":"图像画面扩展","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796845.html","detailPath":"groups/image-out-painting.json"} -{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} -{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} -{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} -{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} -{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} -{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} -{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} -{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} -{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} -{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} -{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} -{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} -{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} -{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} -{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} -{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} -{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} -{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} -{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} -{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} -{"model":"qwen-deep-research","name":"qwen-deep-research","family":"qwen-deep-research","familyName":"qwen-deep-research","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975991.html","detailPath":"groups/qwen-deep-research.json"} -{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} -{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} -{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} -{"model":"qwen-flash","name":"Qwen-Flash","family":"qwen-flash","familyName":"Qwen-Flash","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-flash.json"} -{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} -{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} -{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} -{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-max","name":"Qwen-Image-Max","family":"qwen-image-max","familyName":"Qwen-Image-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-max.json"} -{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} -{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} -{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} -{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} -{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} -{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} -{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} -{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} -{"model":"qwen-plus","name":"Qwen-Plus","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","batch"],"contextWindow":1000000,"maxInputTokens":995904,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} -{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} -{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-plus","name":"QwenVL-Plus","family":"qwen-vl-plus","familyName":"Qwen-VL-Plus","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-plus.json"} -{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} -{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} -{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} -{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} -{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} -{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} -{"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","family":"qwen3-coder-30b-a3b-instruct","familyName":"Qwen3-Coder-30B-A3B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls","detailPath":"groups/qwen3-coder-30b-a3b-instruct.json"} -{"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","family":"qwen3-coder-480b-a35b-instruct","familyName":"Qwen3-Coder-480B-A35B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-480b-a35b-instruct.json"} -{"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","family":"qwen3-coder-flash","familyName":"Qwen3-Coder-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-flash.json"} -{"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","family":"qwen3-coder-plus","familyName":"Qwen3-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-plus.json"} -{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} -{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} -{"model":"qwen3-max","name":"Qwen3-Max","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} -{"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","cache"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} -{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} -{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} -{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} -{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} -{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} -{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} -{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} -{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} -{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} -{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} -{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} -{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","family":"qwen3-vl-flash","familyName":"Qwen3-VL-Flash","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-flash.json"} -{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","family":"qwen3-vl-plus","familyName":"Qwen3-VL-Plus","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-plus.json"} -{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} -{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} -{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} -{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} -{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} -{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} -{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} +{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","TG","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"user-spec":{"count_limit":500,"count_limit_period":1,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} {"model":"qwen3.5-plus","name":"Qwen3.5-Plus","family":"qwen3.5-plus","familyName":"Qwen3.5-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-plus.json"} -{"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} -{"model":"qwen3.5-27b","name":"Qwen3.5-27B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","fine-tuning"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} -{"model":"qwen3.5-35b-a3b","name":"Qwen3.5-35B-A3B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} -{"model":"qwen3.5-397b-a17b","name":"Qwen3.5-397B-A17B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} {"model":"qwen3.6-flash","name":"Qwen3.6-Flash","family":"qwen3.6-flash","familyName":"Qwen3.6-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-flash.json"} {"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","family":"qwen3.6-max","familyName":"Qwen3.6-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":245760,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-max.json"} -{"model":"qwen3.6-plus","name":"Qwen3.6-Plus","family":"qwen3.6-plus","familyName":"Qwen3.6-Plus","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-plus.json"} -{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} -{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.6-plus","name":"Qwen3.6-Plus","family":"qwen3.6-plus","familyName":"Qwen3.6-Plus","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"user-spec":{"count_limit":500,"count_limit_period":1,"usage_limit":8000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":8000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-plus.json"} +{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3"},{"type":"output_token","unit":"每百万tokens","price":"18"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"10.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"6"},{"type":"output_token_batch","unit":"每百万tokens","price":"18"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"15"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"1.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"12"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} {"model":"qwen3.7-plus","name":"Qwen3.7-Plus","family":"qwen3.7-plus","familyName":"Qwen3.7-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search","structured-outputs"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-plus.json"} -{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} -{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"shoemodel-v1","name":"鞋靴模特","family":"shoemodel-v1","familyName":"鞋靴模特","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2804662.html","detailPath":"groups/shoemodel-v1.json"} -{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"speech-biasing","name":"语音识别热词","family":"speech-biasing","familyName":"语音识别热词","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/speech-biasing.json"} -{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} -{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} -{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} -{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} -{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} -{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} -{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["VU","TG"],"features":["structured-outputs"],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.216"},{"type":"output_token","unit":"每百万tokens","price":"0.216"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["structured-outputs","prefix-completion","function-calling","cache"],"contextWindow":131072,"maxInputTokens":131072,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":131072,"maxInputTokens":32768,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} -{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} -{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.625"},{"type":"image_type_2k","unit":"每张","price":"1"},{"type":"image_type_4k","unit":"每张","price":"1.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} -{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.28125"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} -{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.9375"},{"type":"image_type_2k","unit":"每张","price":"0.9375"},{"type":"image_type_4k","unit":"每张","price":"1.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} -{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.46875"},{"type":"image_type_2k","unit":"每张","price":"0.78125"},{"type":"image_type_4k","unit":"每张","price":"1.09375"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} -{"model":"vidu/viduq2_reference2video","name":"ViduQ2_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.21875"},{"type":"video_ratio_720p","unit":"每秒","price":"0.28125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2_text2video","name":"ViduQ2_text2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.1125"},{"type":"video_ratio_720p","unit":"每秒","price":"0.21875"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-pro_img2video","name":"ViduQ2-Pro_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.15625"},{"type":"video_ratio_720p","unit":"每秒","price":"0.34375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-pro_reference2video","name":"ViduQ2-Pro_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.78125"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-pro_start-end2video","name":"ViduQ2-Pro_start-end2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.15625"},{"type":"video_ratio_720p","unit":"每秒","price":"0.34375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-pro-fast_img2video","name":"ViduQ2-Pro-fast_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.1"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-turbo_img2video","name":"ViduQ2-Turbo_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.0875"},{"type":"video_ratio_720p","unit":"每秒","price":"0.25"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq2-turbo_start-end2video","name":"ViduQ2-Turbo_start-end2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.0875"},{"type":"video_ratio_720p","unit":"每秒","price":"0.25"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3_reference2video","name":"ViduQ3_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_720p","unit":"每秒","price":"0.625"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.78125"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-ad_reference2video","name":"ViduQ3-Ad_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.75"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.90625"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045891.html","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-drama_reference2video","name":"ViduQ3-Drama_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_1080p","unit":"每秒","price":"0.875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045892.html","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-mix_reference2video","name":"ViduQ3-mix_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.78125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.9375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-pro_img2video","name":"ViduQ3-Pro_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_720p","unit":"每秒","price":"0.78125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.9375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-pro_start-end2video","name":"ViduQ3-Pro_start-end2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_720p","unit":"每秒","price":"0.78125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.9375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-pro_text2video","name":"ViduQ3-Pro_text2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_720p","unit":"每秒","price":"0.78125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.9375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-pro-fast_img2video","name":"ViduQ3-Pro-fast_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026198.html","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-turbo_img2video","name":"ViduQ3-Turbo_img2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.4375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-turbo_reference2video","name":"ViduQ3-Turbo_reference2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.15625"},{"type":"video_ratio_720p","unit":"每秒","price":"0.3125"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.40625"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-turbo_start-end2video","name":"ViduQ3-Turbo_start-end2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.4375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.4375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} -{"model":"virtualmodel-v2","name":"虚拟模特V2","family":"virtualmodel-v2","familyName":"虚拟模特V2","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/virtualmodel-v2.json"} -{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","family":"voice-enrollment","familyName":"大模型声音复刻及声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/voice-enrollment.json"} -{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-r2v","name":"Wan2.6-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} -{"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} -{"model":"wan2.7-r2v","name":"Wan2.7-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} -{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.6-t2v","name":"Wan2.6-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.7-t2v","name":"Wan2.7-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.24"},{"type":"video_ratio_720p","unit":"每秒","price":"0.24"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} -{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} -{"model":"wanx-poster-generation-v1","name":"创意海报生成","family":"wanx-poster-generation-v1","familyName":"创意海报生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807172.html","detailPath":"groups/wanx-poster-generation-v1.json"} -{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} -{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} -{"model":"wanx-virtualmodel","name":"虚拟模特","family":"wanx-virtualmodel","familyName":"虚拟模特","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/wanx-virtualmodel.json"} -{"model":"wanx-x-painting","name":"万相-图像局部重绘","family":"wanx-x-painting","familyName":"万相-图像局部重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2797051.html","detailPath":"groups/wanx-x-painting.json"} -{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} -{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} -{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} -{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} -{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} -{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md index 70e4cdb2..683ee951 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md @@ -40,6 +40,12 @@ API概述 获取对话摘要、标题生成、关键词、字段信息抽取、问题及解决方案、服务质检、代办事项、满意度、情绪检测、QA抽取、用户画像、标签分类等对话分析结果，应用调用支持 HTTP 调用来完成客户的响应。 +[CreateTask](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask) + +通过上传离线任务数据进行通义晓蜜CCAI-对话分析 + +通过创建离线异步任务，进行对话分析。应用调用支持 HTTP 调用来完成客户的响应。 + [GetTaskResult](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-gettaskresult) 通过任务ID获取离线任务分析结果 @@ -109,17 +115,3 @@ API概述 语音文件实时分析 对进行语音文件进行实时对话分析。应用调用支持 HTTPS 调用来完成客户的响应。 - -## 其他 - -API - -标题 - -API概述 - -[CreateTask](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask) - -通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -通过创建离线异步任务，进行对话分析。应用调用支持 HTTP 调用来完成客户的响应。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md index f5d39958..82fecbfe 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md @@ -36,7 +36,7 @@ OpenAPI 错误码发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更。 @@ -70,7 +70,7 @@ OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 @@ -230,7 +230,7 @@ OpenAPI 名称 操作 -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) 新增 OpenAPI。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md index 811877bb..fb6d78c0 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md @@ -136,7 +136,7 @@ none anytrans:BatchTranslateForHtml -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) none diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md deleted file mode 100644 index ae184fa3..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md +++ /dev/null @@ -1,179 +0,0 @@ -# 使用指南 - -本篇文档主要介绍通义晓蜜对话Agent中对话Agent构建和可视化流程技能编排的使用指南。 - -## **1\. 开通产品** - -- 路径：[阿里云百炼-应用广场-通义晓蜜对话Agent](https://bailian.console.aliyun.com/?spm=5176.29619931.J__Z58Z6CX7MY__Ll8p1ZOR.1.74cd521cS5IvE8#/app/app-market/beebot)。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935960.png) - -- 进入晓蜜对话Agent应用操作控制台，在右上角点击**免费开通**。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935961.png) - -- 进入通义晓蜜对话Agent的开通界面，开通后，按实际调用量收费，即按日生成账单在阿里云账户余额中扣除。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932056.png) - -## **2\.** Agent构建 - -### **2.1 创建应用** - -点击**创建应用**，输入应用名称和应用描述，即可完成基础的Agent创建。 - -- 应用名称：按照业务需求填写Agent的名称； - -- 应用描述：按照业务应用场景添加Agent的描述。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935962.png) - -创建完成后的操作控制台： - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6538449471/p965131.png) - -### **2.2 配置界面** - -#### **1\. 人设** - -- 定义：在机器人空间，每个机器人可以添加人设，即通过提示词的方式调试机器人，可以在人设中对机器人的角色、工作流、技能、限制要求等内容在人设中作对应的定义要求。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932089.png) - -- 选择模板：您可在**选择模板**窗口，插入通用模板，按照模板要求输入对应的人设限制。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932111.png) - -- 优化：当您编辑完人设后，可以点击**优化**，模型会根据您的输入，优化人设内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932123.png) - - -#### **2\. 机器人开场白** - -- 定义：指您打开机器人对话框时，机器人发起对话的开场白。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932121.png) - - -#### **3\. 知识库（流程技能、高频问答知识）** - -- 定义：可绑定当前Agent对应的流程技能和高频问答知识。流程技能的具体的配置步骤可参考《[3\. 可视化流程技能编排](#a7e7986fc4k9q)》，高频问答知识的配置步骤可参考《[4\. 高频问答知识的配置](#d0b14121834k6)》。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965135.png) - - -#### **4\. 更多设置** - -- 模型：当前内置**通义晓蜜大模型**，其中您可更改温度系数和上下文轮次来修改模型 - -- 通识知识： - -- 回复语言：指机器人回复的语言 - - - 与用户语言相同：指机器人回复的语言根据您提问的语言进行回复； - - - 仅中文：指无论您用什么语言提问，机器人都用中文回复； - - - 仅英文：指无论您用什么语言提问，机器人都用英文回复。 - -- 安全： - - - 安全拦截：当系统检查到输入和输出内容涉及到“答案敏感词”时，自动回复“敏感词话术”； - - - 安全预设话术：用户问题包含“用户敏感词”或模型生成回复包含敏感内容时，机器人自动使用敏感回复话术进行回复。默认安全预设话术为：“您说的这个问题我不能回答，您可以尝试询问其他问题”。 - -- 模型生成异常：当系统发生异常或服务超时情况下，机器人兜底回复话术。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932146.png) - -### **2.3 调用量界面** - -在调用量界面可以查看token的消耗量。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932315.png) - -### **2.4 API界面** - -对话Agent的调试信息会同步到API接口中，您可复制当前代码进行调用。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932320.png) - -## **3\.** 可视化流程技能编排 - -### **3.1 新建流程** - -1. 点击**知识库管理>流程**或者**+绑定流程**，打开流程列表窗口。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932331.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932343.png) - -2. 点击**新建流程**，打开新建流程创建。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932344.png) - -3. 按照要求填写对应流程**名称**和**描述**，点击**确认**，即可创建空白流程 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932352.png) - -4. 找到新建的流程，点击**编辑**，进入流程编排页面。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932353.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932356.png) - -5. 新建开启分支，进入流程画布，在开始节点后，添加“分支”设置触发进入流程后继分支的条件。具体的操作步骤可查看下面的小视频： - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931138.gif) - -6. 根据对话逻辑选择节点编排对话流程。 - - ![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931139.png) - - -**重要** - -流程编排完成后，点击**测试**，机器人将自动进行流程完整性检测，如果错误，请根据错误提示优化流程。具体情况可参考如下小视频： - -![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931140.gif) - -### **3.2 流程调试** - -1. 调试前需对环境进行设置 - - -- 服务模拟：流程内的 API 插件会直接使用 mock 值进行返回，适用于 API 还没有准备好的情况。 - -- 随路参数：在用户发送问题时，同时带给机器人的外部参数，如电话接通时，可以将用户呼入号码以随路参数传递给机器人，后续在 API 插件调用时可以使用该参数。 - - -![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931141.png) - -2. 对话调试 - - 直接进行对话，机器人回复后，可以点击**生成完成**查看机器人输出的内容，针对参数收集可以查看到机器人收集到的具体参数信息。 - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931142.gif) - - -## **4**. 高频问答知识的配置 - -### **4.1 新建高频问答知识** - -1. 点击**知识库管理>高频问答**或者**+绑定高频问答**，打开高频问答列表窗口。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965149.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965150.png) - -2. 点击**创建高频问答库**，打开创建窗口，填写高频问答库名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965152.png) - -3. 找到新建的高频问答库，点击编辑，进入添加高频问答知识的页面。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965154.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965155.png) - -4. 点击**新增高频问答**或者**导入高频问答**，在高频问答库中添加对应的高频问题，按照要求填写问题、答案类型、问题答案、生效时间、相似问法，点击**提交**，即可新增成功。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965159.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965160.png) - - 新增成功后：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965161.png) - - -### **4.2 绑定高频问答知识库及测试效果** - -1. 点击**绑定**，即可将对应知识库绑定到机器人上。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965162.png) - - -绑定成功展示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965165.png) - -2. 在左侧对话测试窗测试效果，如下图所示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965166.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md deleted file mode 100644 index 7ef2855d..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md +++ /dev/null @@ -1,117 +0,0 @@ -# API概览 - -## **API标准及多语言预置SDK** - -本产品（`ContactCenterAI/2024-06-03`）的OpenAPI采用[ROA](https://help.aliyun.com/zh/sdk/product-overview/roa-mechanism)签名风格。我们已经为开发者封装了常见编程语言的SDK，开发者可通过[下载SDK](https://api.aliyun.com/api-tools/sdk/ContactCenterAI?version=2024-06-03)直接调用本产品OpenAPI而无需关心技术细节。如果现有SDK不能满足使用需求，可通过签名机制进行自签名对接。由于自签名细节非常复杂，需花费 5个工作日左右。因此建议加入我们的服务钉钉群（147535001692），在专家指导下进行签名对接。 - -在使用API前，您需要准备好身份账号及访问密钥（AccessKey），才能有效通过客户端工具（SDK、CLI等）访问API。细节请参见[获取AccessKey](https://help.aliyun.com/zh/ram/user-guide/create-an-accesskey-pair)。 - -## **自定义签名场景** - -若您的业务场景有特殊需求，需通过自签名方式对接 API，建议优先咨询我们的技术支持团队（服务钉钉群：147535001692），获取专业指导以确保高效接入。 - -## **账号与安全准备** - -阿里云账号具备对所有资源的完全管理权限。一旦 AccessKey 泄露，所有相关资源都将面临未经授权访问的风险。为确保安全，建议创建一个仅具备 API 访问权限的[RAM用户](https://help.aliyun.com/zh/ram/user-guide/create-a-ram-user)并配置其 AccessKey，同时基于最小权限原则 (PoLP) 配置 RAM 策略。仅在明确需要阿里云账号权限的特定场景下，才使用阿里云账号。 - -## API目录 - -API - -标题 - -API概述 - -[RunCompletion](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletion) - -通过模版ID调用通义晓蜜CCAI-对话分析AIO应用 - -支持调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[RunCompletionMessage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletionmessage) - -使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -支持以Message协议格式调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[AnalyzeConversation](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation) - -通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -获取对话摘要、标题生成、关键词、字段信息抽取、问题及解决方案、服务质检、代办事项、满意度、情绪检测、QA抽取、用户画像、标签分类等对话分析结果，应用调用支持 HTTP 调用来完成客户的响应。 - -[GetTaskResult](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-gettaskresult) - -通过任务ID获取离线任务分析结果 - -通过任务ID获取离线任务对话分析结果。应用调用支持 HTTPS调用来完成客户的响应。 - -[CreateTask](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask) - -通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -通过创建离线异步任务，进行对话分析。应用调用支持 HTTP 调用来完成客户的响应。 - -[AnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeimage) - -图片内容分析 - -通过通义晓蜜CCAI-对话分析AIO应用进行图片内容分析。具体包括以下场景：水印检测。应用调用支持 HTTP 调用来完成客户的响应。 - -[GeneralAnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-generalanalyzeimage) - -通用图片分析 - -通用图片分析。 - -## 热词管理 - -API - -标题 - -API概述 - -[CreateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createvocab) - -创建热词 - -将一组语音热词上传到服务端，并获取返回热词ID。 - -[UpdateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-updatevocab) - -修改热词 - -根据词表的ID可以更新对应的词表信息，包括词表名称、词表描述信息、词表的词和权重。 - -[ListVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-listvocab) - -获取热词列表 - -列举指定业务空间下的热词列表信息。 - -[DeleteVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-deletevocab) - -删除热词 - -根据词表的ID删除对应的词表。 - -[GetVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-getvocab) - -获取热词 - -根据词表的ID获取对应的词表信息。 - -## 不推荐或白名单开放 - -API - -标题 - -API概述 - -[AnalyzeAudioSync](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeaudiosync) - -语音文件实时分析 - -对进行语音文件进行实时对话分析。应用调用支持 HTTPS 调用来完成客户的响应。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md deleted file mode 100644 index fe169264..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md +++ /dev/null @@ -1,141 +0,0 @@ -# 字段信息抽取最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行字段信息抽取的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的信息抽取能力，对客服和用户的对话记录（文本、录音文件）进行理解、识别、抽取，如抽取客服工单中的字段信息，如客户所在的地区信息、年龄、日期时间、办理事项等，提升工单填写效率。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行字段信息抽取，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和属性描述，进行属性抽取。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4924970771/CAEQURiBgICV77fjnRkiIDlmNTk4MWEyZDJhYzQyZjBhNjY0ZjVjYmRlNjc1MzA54811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - - - -com.alibaba - -fastjson - -2.0.58 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // 抽取字段名称和字段描述 - List fieldList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field1 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field1.setName("问题类型"); - field1.setDesc("客户咨询的问题类型"); - fieldList.add(field1); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field2 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field2.setName("公司名称"); - field2.setDesc("客服所属的保险公司名称"); - fieldList.add(field2); - - request.setFields(fieldList); - - // fields表示属性抽取任务 - request.setResultTypes(Arrays.asList("fields")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md deleted file mode 100644 index ac8dcf19..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md +++ /dev/null @@ -1,138 +0,0 @@ -# 客服服务质检最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行客服服务质检的最佳实践。 - -## **应用场景** - -通过通义晓蜜CCAI-AIO的服务质检能力，分析客服和用户的对话记录（文本、录音文件）、发现客服的服务质量问题，进而提升客服服务效率、服务规范，提升客户体验。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行服务质检，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和质检项，进行对话分析。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0924970771/CAEQURiBgIDOp7TjnRkiIDUzYjVkNGQ4OTdjZDQ1MjliYWVmMjk5MjEzNzczYTNm4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection serviceInspection = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection(); - List inspectionContents = new ArrayList<>(); - - // 质检项定义 - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content1 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content1.setTitle("客服是否过度承诺"); - content1.setContent("客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。"); - inspectionContents.add(content1); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content2 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content2.setTitle("客户情绪是否正向"); - content2.setContent("分析对话内容，输出用户在对话中表现出的情绪，详细要求：a. 当客户表现出负面情绪时，判定为消极；b. 当客户表现中积极情绪时，判定为积极；c. 如果客户文本没有明显的消极或积极情感色彩，则判定为中性。"); - inspectionContents.add(content2); - - - serviceInspection.setInspectionContents(inspectionContents); - serviceInspection.setInspectionIntroduction("请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等"); - serviceInspection.setSceneIntroduction("保险销售场景"); - - request.setServiceInspection(serviceInspection); - // service_inspection表示服务质检任务 - request.setResultTypes(Arrays.asList("service_inspection")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md deleted file mode 100644 index fabe48ad..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md +++ /dev/null @@ -1,118 +0,0 @@ -# 摘要生成（含摘要/标题/关键词）最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行摘要总结的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的总结摘要能力自动提取文档中的重要信息，如对话摘要总结、标题生成、关键词等，从而提高工作效率，减少人工处理成本。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行摘要总结，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息和对话内容，进行摘要总结。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8924970771/CAEQURiBgMCMwbTpnRkiIGY5ODI0YzFiZjA5MTQ2MjhhNGUzYTVmNThjYWQyODdl4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // summary 表示总结摘要任务 - // keywords 表示关键词抽取 - // title 表示抽取标题 - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md deleted file mode 100644 index 592b3229..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md +++ /dev/null @@ -1,563 +0,0 @@ -# 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - -- 关于任务类型，请参见[通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation)中resultTypes字段的描述。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取workspaceId和appId** - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935587.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935588.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么领取五险一金呢").build()); - - List resultTypes=new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue=AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.analyzeConversationWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - AnalyzeConversationResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - List sentenceList = new ArrayList<>(); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问您想咨询五险一金哪方面的问题呢"); - sentenceList.add(sentences1); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("怎么查询账户呢"); - sentenceList.add(sentences2); - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-1111"); - request.setDialogue(dialogue); - - request.setSceneName("中国移动"); - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": false, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - events, err := util.ReadAsString(resp["body"].(io.ReadCloser)) - - if err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - - fmt.Println(tea.StringValue(events)) - - return nil -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` - -## 异步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么查询账户呢").build()); - - List resultTypes = new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue = AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture x = client.analyzeConversation(completionParam); - AnalyzeConversationResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL***********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - - } -} -``` - -## 同步流式调用 - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": true, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - eventChan, errChan := util.ReadAsSSE(resp["body"].(io.ReadCloser)) - for { - select { - case event, ok := <-eventChan: - if !ok { - return nil - } - fmt.Println("-------------------------------------") - fmt.Printf("Event ID: %s, Event name: %s, Data: %s\n", event.ID, event.Event, event.Data) - case err, ok := <-errChan: - if ok && err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - } - } - -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md deleted file mode 100644 index ed6cafa1..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md +++ /dev/null @@ -1,225 +0,0 @@ -# 通过通义晓蜜CCAI-对话分析AIO应用进行图片分析 - -本文向您介绍一个通过通义晓蜜CCAI-对话分析AIO应用进行图片分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流失调用 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - AnalyzeImageRequest request = new AnalyzeImageRequest(); - request.setStream(false); - request.setResultTypes(Arrays.asList("watermark")); - List imageList = new ArrayList<>(); - imageList.add("http://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - request.setImageUrls(imageList); - AnalyzeImageResponse response=client.analyzeImage(workspaceId,appId,request); - System.out.println(response); - } -} -``` - -## 异步非流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(false).imageUrls(imageList).build(); - CompletableFuture x = client.analyzeImage(request); - AnalyzeImageResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL**********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(true).imageUrls(imageList).build(); - ResponseIterable x = client.analyzeImageWithResponseIterable(request); - ResponseIterator iterator = x.iterator(); - String lastTxt = ""; - while (iterator.hasNext()) { - AnalyzeImageResponseBody event = iterator.next(); - lastTxt = event.getText(); - System.out.println(JSON.toJSONString(event)); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md deleted file mode 100644 index 7dcbd086..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md +++ /dev/null @@ -1,197 +0,0 @@ -# 通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -本文向您介绍一个通过上传离线任务数据进行通义晓蜜CCAI-对话分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935590.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935591.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击**应用实践**。 - -2. 在应用实践列表中找到点击通义晓蜜CCAI-对话分析AIO的**立即查看**。 - -3. 点击上方**我的应用**，展示应用卡片列表。 - -4. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 创建语音任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - - request.setTaskType("audio"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskRequest.CreateTaskRequestTranscription transcription=new CreateTaskRequest.CreateTaskRequestTranscription(); - transcription.setFileName("***.mkv"); - transcription.setVoiceFileUrl("https://***.oss-cn-beijing.aliyuncs.com/****/***.mkv"); - request.setTranscription(transcription); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - -} -``` - -## 创建文本任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - CreateTaskRequest.CreateTaskRequestDialogue dialogue = new CreateTaskRequest.CreateTaskRequestDialogue(); - - List sentences = new ArrayList<>(); - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences1 = new CreateTaskRequest.CreateTaskRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问有什么事，你什么性别,胖不胖"); - sentences.add(sentences1); - - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences2 = new CreateTaskRequest.CreateTaskRequestDialogueSentences (); - sentences2.setRole("user"); - sentences2.setText("我要买保险，我是男的，很瘦"); - sentences.add(sentences2); - dialogue.setSentences(sentences); - dialogue.setSessionId("sessionId-01"); - - request.setDialogue(dialogue); - - request.setTaskType("text"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - - -} -``` - -## 获取任务结果 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - String taskId = "*****-****-****-*****-****"; - GetTaskResultRequest request = new GetTaskResultRequest(); - request.setTaskId(taskId); - GetTaskResultResponse response = client.getTaskResult(request); - System.out.println(JSONObject.toJSONString(response)); - } -} -``` - -## **相关文档** - -关于任务类型，请参[通过上传离线任务数据进行通义晓蜜CCAI-对话分析](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask)见中resultTypes字段的描述。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 75449a68..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,385 +0,0 @@ -# 通过原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - - 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 接口入参位置 - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935585.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935586.png) - -1. 访问[应用广场应用实践](https://bailian.console.aliyun.com/#/app-market/lightApplication)页面，选择**通义晓蜜CCAI-对话分析AIO**，单击**立即查看**。 - -2. 单击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## **异步流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.runCompletionMessageWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionMessageResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture generateCompletionResponseCompletableFuture = client.runCompletionMessage(completionParam); - RunCompletionMessageResponse generateCompletionResponse = generateCompletionResponseCompletableFuture.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getText())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getFinishReason())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getRequestId())); - } -} -``` - -Python - -``` -import asyncio - -from alibabacloud_contactcenterai20240603.client import Client - -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import (RunCompletionRequestDialogueSentences, - RunCompletionRequestDialogue, - RunCompletionRequestFields, - RunCompletionRequest) -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -async def run_async_sse(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = await client.run_completion_message_async(workSpace, appId, request) - body = response.body - print(body) - -if __name__ == '__main__': - asyncio.run(run_async_sse()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - //Prompt - RunCompletionMessageRequest request = new RunCompletionMessageRequest(); - List messageList = new ArrayList<>(); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message1 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message1.setRole("system").setContent("You are a helpful assistant."); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message2 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message2.setRole("user").setContent("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话\\n客户:你自己要号手机\\n客户:31\\n客服:他是1公分\\n客户:那年后，我看\\n客户:后来点开看\\n客户:嗯，要不是以前的\\n客户:这不看你看\\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\\n客服:你在这跟到后面来\\n客服:这里不是大多少天都会员了吗\\n客户:那你不是大于多少天都结了吗\\n客服:嗯\\n客户:嗯\\n客服:我们有1个\\n客户:就是你这个\\n客服:他说你这里面也按照\\n客户:他有多少是蒙细别的吗\\n客户:这3天之后\\n客服:当前执行的\\n客服:这些，其实你这个也是\\n客户:那你这个颜色\\n客服:按照他给了预期了，是不是结多少天\\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\\n客服:那是什么问题\\n客服:你3天的时候关内\\n客户:你要是原来那个\\n客户:你基本上就那你要看\\n客服:你要是原来的，你，你基本上都要不要\\n客服:5天\\n客户:对的\\n客户:啊，关于\\n客户:15新啊\\n客户:上面录的\\n\\n```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。"); - messageList.add(message1); - messageList.add(message2); - - request.setMessages(messageList); - request.setStream(false); - request.setModelCode("tyxmTurbo"); - - RunCompletionMessageResponse responseBody = client.runCompletionMessage(workspaceId, appId, request); - RunCompletionMessageResponseBody body = responseBody.getBody(); - System.out.println(JSON.toJSONString(body)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client - -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = client.run_completion_message(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 9fa17554..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,404 +0,0 @@ -# 通过模板ID调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用SDK的安装、使用及注意事项。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## **接口入参位置** - -### **workspaceId** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -1. 访问**应用广场**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -### **templateIds** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 点击**管理**进入对应的应用卡片。 - -4. 点击**自定义指令**模板，切换为**专业构建模式**。 - -5. 点击右上方**指令模板管理**。 - -6. **自定义模板**列表中**模板ID**为入参中templateIds。 - -7. 如果还未创建自定义模板，请直接点击右上角**保存指令模板**按钮。 - - -## **安装SDK** - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## 异步流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).modelCode("tyxmTurbo").dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - ResponseIterable x = client.runCompletionWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - CompletableFuture x = client.runCompletion(completionParam); - RunCompletionResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -Python - -``` -import asyncio -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -async def run_async(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", - "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = await client.run_completion_async(workSpace, appId, request) - body = response.body - print(body) -if __name__ == '__main__': - asyncio.run(run_async()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - RunCompletionRequest request = new RunCompletionRequest(); - //对话信息 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto1 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto1.setRole("user").setText("我要办理信用卡"); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto2 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto2.setRole("agent").setText("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息"); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - //属性填充 - List fieldList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestFields field1 = new RunCompletionRequest.RunCompletionRequestFields(); - field1.setName("姓名").setDesc("用户的名称"); - RunCompletionRequest.RunCompletionRequestFields field2 = new RunCompletionRequest.RunCompletionRequestFields(); - field2.setName("信用卡号").setDesc("用户的信用卡号"); - fieldList.add(field1); - fieldList.add(field2); - RunCompletionRequest.RunCompletionRequestDialogue dialogue = new RunCompletionRequest.RunCompletionRequestDialogue(); - dialogue.setSessionId("session_01_asdfasdfasd").setSentences(sentenceDTOList); - //构建请求参数 - request.setDialogue(dialogue).setStream(false).setModelCode("tyxmTurbo").setFields(fieldList).setTemplateIds(Arrays.asList(templateId)); - RunCompletionResponse runCompletionResponse = client.runCompletion(workspaceId, appId, request); - RunCompletionResponseBody responseBody = runCompletionResponse.getBody(); - System.out.println(JSON.toJSONString(responseBody)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = client.run_completion(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md deleted file mode 100644 index 3f99a056..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md +++ /dev/null @@ -1,67 +0,0 @@ -# 产品概述 - -本文档介绍通义晓蜜CCAI-对话分析AIO产品能力及产品优势。 - -## **什么是通义晓蜜CCAI-对话分析AIO** - -对话分析AIO，即对话分析all-in-one API，是基于深度调优的对话大模型， 为营销服类产品提供智能化升级所需的生成式摘要总结、质检、分析等能力的官方应用。 - -- **面向对象：**开发者、自研企业、传统呼叫中心采购者及友商等。 - -- **核心能力：**为客户提供多种规格专属模型（支持切换），并提供多指令精准执行、自带最佳实践的API及贴合业务场景的被集成方式。 - -- **服务形式：**通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式。 - -- **核心价值：**免去企业SFT、探索prompt 、应用最佳实践开发的成本，使企业专注在业务逻辑的实现。 - -- **付费模式：**按调用次数后付费。 - - -## **产品能力** - -1. **多指令精准执行** - - -- **总结摘要：**根据对话内容，记录通话中最核心的信息。 - -- **信息抽取：**根据指令，抽取并组织关键信息。 - -- **质检分析：**按照指令进行检测，例如情绪、敏感词等 - -- **多指令任务：**生成标题、生成关键词、生成摘要、维度检测、信息抽取等。 - - -2. **多模型选择** - - -- 模型配置：平台提供通义晓蜜-Turbo、通义晓蜜-Plus两种模型规格。 - - -3. **多指令模板** - - -- **指令模板：**平台提供官方预置常用指令模板并支持管理自定义模板。 - - -4. **调试信息** - - -- SaaS管理：提供SaaS调试窗，用于验证效果并最终提供API服务。 - -- API服务：通过API服务输出给客户 - - -**5.数据报表** - -- 调用数据：提供按时、按日调用数据报表。 - -- 账单数据：请前往[费用与成本](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)查询。 - - -## **产品优势** - -- 结合大模型特性打造新能力，将过去多个业务场景凝聚成一个个通用、轻量、易集成的能力。 - -- 企业对于一次服务记录/工单所需要的所有质检、分析、处理的动作，可通过一次调用完成，得到结构化的结果。 - -- 提供通话工单分类（支持级联分类）、自定义信息抽取、自定义质检规则并检出能力。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md deleted file mode 100644 index 8b9c7fd7..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md +++ /dev/null @@ -1,38 +0,0 @@ -# 如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量 - -本文档介绍如何对应用进行编辑、删除等管理，如何进行API调用、如何查看调用量。 - -## **我的应用** - -路径：[应用广场](https://bailian.console.aliyun.com/#/app-market)→ 点击[应用实践](https://bailian.console.aliyun.com/?tab=app#/app-market/lightApplication)→ 找到**通义晓蜜CCAI-对话分析AIO**，选择[查看详情](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)，即可进入**我的应用** **。** 在**应用广场**页面，找到**通义晓蜜CCAI-对话分析AIO**应用卡片，单击**立即查看**。 - -### 3.1 应用复制、删除、修改 - -- **应用复制：**在**我的应用**中，可以点击应用右上角选择复制应用，对该应用进行复制。应用复制：在**我的应用**页面，单击目标应用卡片右上角的菜单图标，选择**复制应用**，即可复制该应用。应用复制成功后，在**我的应用**列表中可看到新增的复制应用卡片，其名称自动追加"副本"后缀。每张应用卡片底部提供**管理**、**调用量**、**API调用**三个操作入口。 - -- **应用修改：**同时可以点击应用右上角选择修改应用，对该应用名称进行修改。应用修改：在**我的应用**中，点击应用右上角的三点菜单，选择**修改应用**，即可对该应用信息进行修改。应用修改：可以在**编辑应用名称**对话框中修改**应用名称**，名称最多支持50个字符，修改完成后单击**确定**保存。 - -- **应用删除：**如果需要删除该应用，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。应用删除：同样可以点击应用右上角选择**删除应用**，对该应用进行删除。应用删除：点击应用右上角选择删除应用，系统弹出**确认删除**对话框，提示"删除已发布应用可能会对您的线上业务产生影响，您确定删除吗？"，单击**确认删除**完成删除操作，单击**取消**可取消操作。 - - -### 3.2 API调用 - -- 应用API：应用提供标准化API接口，用于向客户系统输出对话分析结果。支持集成官方预置Agent模板或用户自定义模板，客户可通过API获取结构化响应，并自主设计前端展示样式。 - - - 当通过**对话分析Agent创建方式**创建应用时，使用自定义指令的API调用： - - 在**对话分析Agent**的自定义指令页面右上角，单击**API示例**按钮，页面右侧展开API调用信息面板，依次包含：获取AccessKeyID和AccessKeySecret、获取Workspace ID和App ID、安装SDK（支持Java、Python、Go，Java通过Maven dependency引入）以及代码示例（支持Java、Curl、Python、Go）。代码示例中展示了`CcAiPassSync`类的调用逻辑，包括设置`workspaceId`与`appId`常量、构造`RunCompletionRequest`请求及组装对话内容。左侧**指令信息**区域可引入`${dialogue}`和`${fields}`两个变量，**模型配置**选择所需模型后，单击底部**测试（command + enter）**按钮即可在中间**效果测试**区域查看调用结果。 - - - 通过**自定义创建方式**的API调用：单击页面顶部的**API示例**按钮，右侧面板展示API调用所需信息。单击**获取API-KEY、APP-ID和Workspace ID**蓝色链接获取AccessKey ID、AccessKey Secret、Workspace ID和App ID。安装SDK时，在Maven项目的pom.xml中添加依赖`com.aliyun:alibabacloud-ccai-passthrough-sync`（版本3.3.3）。代码示例使用Java类`CcAiPassSync`，设置`workspaceId`和`appId`参数，构造`RunCompletionRequest`对象，传入包含user和agent角色的对话句子列表进行调用。 - - -### **3.3 调用量** - -- 调用次数按输入和输出token总数计量，以2000 tokens 为一个计量单位，输入与输出总token数小于等于2000 tokens为1次调用，大于2000小于等于4000 tokens 为2次调用，向上取整，以此类推。 - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -在**调用量**页签中，可通过**所属空间**、**选择模型**、**应用**、**数据来源**下拉筛选条件，以及时间范围（**昨天**、**近3天**、**近七天**、**近15天**或自定义日期范围）查看对应的调用次数统计折线图。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md deleted file mode 100644 index b33e5bc8..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md +++ /dev/null @@ -1,60 +0,0 @@ -# 如何基于自定义方式创建应用 - -本文档介绍了如何通过自定义方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**，再点击**创建应用**。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入应用调试界面。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**自定义创建**：通过编写自定义指令（Prompt）来构建应用逻辑，用户可使用内置或自定义指令模板进行测试，适合有一定经验的人员。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -## **应用配置** - -进入已经创建完成的应用中进行配置。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009398.png) - -- **模型配置：**通义晓蜜Plus 和 Turbo 模型仅支持语音与文本分析；当分析对象为图片时，系统将默认使用通义晓蜜VL模型，该模型不可手动选择。 - -- **指令信息：**通过编写指令信息来配置对应的任务、格式、要求等，来完成对应分析任务。 - - - **变量配置：**若需要在对话过程中引用更多变量可以在此配置，在指令编辑器中输入 `/` 可触发变量自动补全，选择后插入对应变量引用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009509.png) - - - **选择指令模板：**同时可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009498.png) - - - **保存指令模板：**在编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。选择‘覆盖已有指令模板’，需从下拉列表中选择目标模板名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009505.png) - - - **指令优化：**对编写完成的指令信息进行AI优化。![指令优化示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009511.png) - -- **分析对象类型** - - 分析对象类型可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3624933771/p1059909.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - -- **字段信息：**当需要获取到对话内容的字段信息时，可以使用“信息抽取预置模板”、“多指令模板”创建指令任务，同时也需要引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![字段信息示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - -- 点击**“测试”**按钮**，**查看测试结果![测试结果示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009528.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md deleted file mode 100644 index 85bc7021..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md +++ /dev/null @@ -1,140 +0,0 @@ -# 如何进行基于对话分析Agent方式创建应用 - -本文档介绍了如何通过对话分析Agent方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**基于对话分析Agent创建**：通过预置最佳实践示例或上传对话数据，体验大模型生成式摘要、总结、服务质检等全场景应用能力。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -### 基于对话分析Agent创建 - -**说明** - -注意：当选择基于对话分析Agent创建方式，只有选择自**定义指令**\-**专业构建模式**可以用图片分析，其他方式无法对图片进行分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006405.png) - -进入已经创建完成的应用中后，可以选择**对话分析Agent**、**自定义指令**方式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914644.png) - -1. #### **选择对话分析Agent方式**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914645.png) - - **对话分析维度：**可以根据实际业务需求的需要通过对话分析Agent来选择对应的维度。 - - - **标准指令：**已预置标准prompt，可快速生成理想结果。可选值为【标题、摘要、关键词、Q&A、问题解决方案】，至少选择一个选项。 - - - **高级指令：**服务质检、标签分类等高级指令有一定的业务属性，该部分指令在示例通话中已预置标准prompt，若自行上传通话数据且对结果有一定要求，建议在预置指令模板→专业模式中编辑自定义指令进行调试。可选值为【服务质检、关键信息、标签分类】。 - - - **分析对象类型：**根据自己业务需要分析的数据类型选择【纯文本、语音】。当选择**纯文本**时，可以上传不超过15000字的文本内容，同时我们还提供了行业对话示例来进行测试；当选择**语音**时，支持单个不超过40MB的WAV或MP3格式文件，上传完成后会自动将其转译为文本信息内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9479940671/p1008496.png) - - **信息内容：**需要按照以下格式编写对话信息，可以通过使用已经提供的行业对话示例。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可直接插入内置的行业对话示例文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914647.png) - - - **点击“测试”按钮，查看输出结果：**测试出来的结果生成的指令与我选择的对话分析维度相对应。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914651.png) - -2. #### **选择自定义模板方式** - - 可以选择简单构建模式&专业构建模式。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914696.png) - - - ##### **简单构建模式说明** ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914697.png) - - - **模型配置**：通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **指令类型：**可以根据实际业务需求分析的全部维度，根据选择的标准指令与高级指令，在指令信息中编辑指令的prompt。 - - - 标准指令：可以选择【标题、摘要、关键词、Q&A、问题解决方案】，并在下方指令信息中展示出系统内置prompt，可以对其进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850108.png) - - - 高级指令：可以选择【服务质检、关键信息、标签分类】，并在下面指令信息中展示对应配置，进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057646.png) - - - **服务质检：**对通话中存在的客户情绪、敏感词、服务质量等内容进行检测。可根据业务需要质检的场景和质检项。 - - - **质检项（名称+描述）：**设置服务质检中需要质检的名称与描述。 - - - **关键信息（名称+类型）：**提取通话中配置的关键信息，通过名称与类型配置。 - - - **标签分类（名称+描述）：**对通话内容进行分类定义，标签长度不超过10个字。 - - - **分析对象类型：**需要按照以下格式编写对话信息，可选择纯文本、语音两种方式。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可以选择行业对话示例进行插入内置对话文本。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850123.png) - - - **点击“测试”按钮，查看测试结果。**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914710.png) - - - ##### 专业构建模式说明![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914716.png) - - **说明** - - 在模式切换时，当前模式所编辑的内容不再生效，对话内容将会重置。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057648.png) - - - **模型配置：**通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **选择指令模板：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006412.png) - - - **保存指令模板：**在自定义编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006414.png) - - - **指令模板管理：**可以查看预置模板和自定义保存的模板指令。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914728.png) - - - **变量配置：**除了内置${field}、${dialogue}两个变量以外，如需在分析过程中引用更多变量，可以完成变量配置。测试数据可以作为测试过程中的模拟数据，临时使用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939566.png) - - - 在编辑完成后，即可在指令信息中插入，使用“/”进行插入保存的变量。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939584.png) - - - **分析对象类型：**可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057649.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，同时可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - - - **字段信息：**当使用“信息抽取预置模板”、“多指令模板”创建指令任务，可引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - - - 点击**“测试”**按钮**，**查看输出结果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939588.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md deleted file mode 100644 index b3b43117..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md +++ /dev/null @@ -1,48 +0,0 @@ -# 热词组配置管理与使用 - -为提升语音转译的准确性，您可以在语音质检分析场景中使用热词组。本文档将介绍其配置与使用方法。 - -## **热词配置** - -热词组仅对离线/实时语音质检分析场景生效，用于提升语音转译的准确性。 - -### **1.热词组管理** - -- 进入热词组管理的路径： - - - 路径1：进入[**通义晓蜜CCAI-对话分析AIO**](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)后，点击我的应用，可在界面中看到**热词组管理**按钮。 - - - 路径2：通过进入具体应用的配置页面，点击**选择热词组**时进行添加对应的热词组。 - - **说明** - - 目前热词组支持通过自定义创建的应用；通过Agent创建的应用，需在‘专业构建模式’下才能找到并配置热词组。 - - 点击后右侧弹出 **选择热词组** 侧边栏，顶部提示"热词组可以提高语音转译准确性，使用时仅对语音质检分析生效"。可通过搜索框查找已有热词组，或单击 **新建热词组** 创建新的热词组，列表中勾选所需热词组即可完成关联。 - -- 在热词组管理界面，点击‘**新建热词组**’，填写名称和热词后，点击‘**确定**’完成创建。单击**新建热词组**按钮打开新建弹窗，在弹窗中填写热词组名称并通过**添加热词**逐条添加热词。 - - - 热词组名称：不超过10个字。 - - - 热词：每个热词组不超过128个热词。 - - - 热词：填入对应需要提高准确率的热词，重要限制：当前版本热词仅支持纯汉字，不支持英文、数字及特殊字符。包含非汉字的热词将无法生效。 - - - 权重：取值范围为1到5之间的整数，默认值：2；如果效果不明显可以适当增加权重，但是当权重较大时可能会引起负面效果，导致其他词语识别不准确。 - - - 操作：删除已添加的热词。 - -- 创建成功后，新的热词组将显示在列表中，并支持编辑或删除。在已创建的热词组名称右侧，单击齿轮图标，可在下拉菜单中选择**编辑**或**删除**该热词组。 - - -### **2.热词组在应用中的使用** - -进入具体应用的设置，选择并绑定所需的热词组。 - -**说明** - -每个应用仅支持绑定一个热词组。 - -1. 选择对应的热词组：单击**选择热词组**按钮，在弹出的浮层中单击目标热词组对应的**选用**按钮即可完成绑定。 - -2. 选择完成后可以查看绑定情况。在**自定义指令**配置页面下方的**热词组**区域，单击**选择热词组**按钮，将已创建的热词组添加到当前指令配置中。已添加的热词组会以标签形式展示，可单击删除图标移除。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md deleted file mode 100644 index c64ccbc5..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md +++ /dev/null @@ -1,74 +0,0 @@ -# 知识库的使用 - -采用检索增强生成（RAG）技术，根据用户上传的外部信息源检索相关信息，然后将检索到的内容整合到用户输入中，从而帮助大模型生成更准确的回答。 - -## **知识库** - -知识库仅对自定义指令的专业模式或者自定义创建的应用中的纯文本和语音对象分析场景生效，用于提升大模型生成回答的准确性。 - -### **1.知识库配置** - -1. **知识库配置的开启** - - 在自定义创建应用中开启**知识**开关后可以添加文档、表格、图片类型知识用于辅助分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059009.png) - -2. **知识库的调用方式** - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059025.png) - - 1. **必定调用：**每一轮的用户输入都会进入知识库检索与联网搜索，适用于高频知识问答场景。 - - 2. **智能调用：**智能判断用户的输入信息是否进入知识库检索或者联网搜索，适合灵活的对话场景。 - -3. **可使用的知识说明**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059033.png) - - 1. **文档知识：**使用知识库文件类数据构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 2. **表格：**使用数据中心表格类、数据库类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 3. **图片：**使用数据中心图片类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - -4. **知识库其他配置：** - - 1. **知识库过滤：**开启后，将通过配置判断prompt的方式引入大模型判断，对文档和表格召回结果进行二次智能过滤。 - - 2. **对话摘要总结：**开启后，可通过大模型对其对话内容进行概括总结。 - -5. **知识库的选择** - - 开启知识库后，可以在对应的文档、表格、图片知识类目中点击“+”按钮，右侧弹出对应知识库列表，可以选择已存的知识库进行的进行添加。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059059.png) - -6. **知识库的相似度阈值与权重** - - 选择对应的知识库后，支持配置相似度阈值与权重来控制筛选检索结果和知识库召回顺序。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059337.png) - - 1. **知识库的相似度阈值：** - - 阈值范围在0.01～1，默认为0.2，可以根据场景需要进行调整，注意：只有语义相似度得分高于此值的文本才会被召回。若此值设置得过高，将导致知识库丢弃所有相关的文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059377.png) - - 2. **知识库的权重：** - - 权重范围在0.5～2.默认为1，可以根据场景需要进行调整，在多路召回时，若多个知识库召回的文本切片相似度分数相同，系统将优先返回权重更高的知识库中的文本切片。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059385.png) - -7. **知识库的创建** - - 点击**创建新知识库**按钮，跳转到知识库创建的页面中，具体操作可参考[创建和使用知识库](https://help.aliyun.com/zh/model-studio/rag-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059090.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059088.png) - - **说明** - - 目前通义晓蜜CCAI-AIO产品所绑定的知识库仅支持文档搜索、数据查询、图片问答，不支持音视频搜索。 - - -### **2.知识库在应用中的使用** - -- 根据两种创建应用的方式区分： - - - 在**基于对话分析Agent创建方式**中的使用：仅支持在自定义指令中的**专业构建模式**中使用，**分析对象类型**需要选择纯文本、语音。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059324.png) - - - 在**自定义创建方式**中的使用：**分析对象类型**选择纯文本、语音即可使用。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md deleted file mode 100644 index 3e231b8b..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md +++ /dev/null @@ -1,85 +0,0 @@ -# 语音对话机器人操作指南 - -本文档介绍通义晓蜜CCAI-语音对话机器人在阿里云百炼控制台如何操作。 - -## **1\. 创建应用** - -- 路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000694.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用描述：**自定义填入应用实际使用描述说明。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000698.png) - - -## **2.机器人配置** - -机器人配置目前使用的为prompt构建模式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914777.png) - -- **模型选择：**通义晓蜜-Plus、通义晓蜜-Max、通义晓蜜-Turbo。 - -- **指令信息：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了通用场景、服务满意度调研、家电上门安装预约、游戏福利推送介绍四种模板。同时支持自定义指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000715.png) - -- **变量配置：**可在指令信息中通过**${xxx}**样式进行插入。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000733.png) - -- **指令配置：**如果当前参考的流程话术中存在特殊指令#\[...\]，请在回复中添加#\[...\]，目前支持传入挂机指令#\[HangUp\]。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000747.png) - -- **语音配置：**进行音色选择配置，配置完成后可在机器人呼叫时运用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000739.png) - - - 音色模板：可选择大模型音色，如：龙小夏V2、龙小夏等。 - - - 音量：范围为0～100，值越大声音越响亮。 - - - 语速：范围为-500～500，值越大语速越快。 - - - 音调：范围为-500～500，值越大音调越高昂。 - -- **高级配置：**对机器人的其他能力进行配置。 - - - 静默超时：自定义配置时长，范围在1～60秒，当对话过程中用户回复超过配置的静默时长后播报静默话术。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000740.png) - -- **点击“呼叫”按钮，查看测试结果。** - - **说明** - - 目前通过控制台测试时不支持变量传入。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000746.png) - -- **发布机器人：**点击**发布**按钮，当页面提出**发布成功**，即表示为成功发布。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000757.png) - - -## **3.我的应用** - -路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -### **3.1 调用量** - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000755.png) - -### 3.2 应用修改、删除 - -- **应用修改：**进入我的应用后，可以点击应用右上角选择修改应用，对该应用名称进行修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000750.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000751.png) - -- **删除应用：**进入我的应用后，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000752.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000753.png) - - -### **3.3API调用** - -应用API：通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000756.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md index d03c071a..cc6f0110 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md @@ -19,7 +19,7 @@ ### **步骤一：创建函数（FC）** -1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/fc/user-guide/creating-an-event-function)。 +1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/creating-an-event-function)。 - 创建视频理解提交任务函数（建议函数名称：SubmitVideoAnalysisTask）。 @@ -49,7 +49,7 @@ pip3 install alibabacloud_endpoint_util alibabacloud_tea_openapi alibabacloud_quanmiaolightapp20240801 -t . ``` - - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/fc/user-guide/create-a-custom-layer-1)。 + - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/fc/create-a-custom-layer-1)。 3. 您可自行优化调整代码中的入参：比如`prompt`模板、`modelId`等。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md index b5b89f69..49fb108e 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-optimization.md @@ -129,11 +129,11 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 - **在数据管理页面编辑标签**：对于已上传的文件，可单击右侧的**标签**进行编辑，相关API是[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag)。 - 目前阿里云百炼支持以下两种方式使用标签： + 目前阿里云百炼支持以下方式使用标签： - [通过API调用阿里云百炼应用](https://help.aliyun.com/zh/model-studio/application-calling-guide#4100253b7chc3)时，可以在请求参数`tags`中指定标签。 - - 在调试应用时设置标签（本方式仅适用于[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)）。 + - 在调试应用时设置标签（本方式仅适用于[智能体应用（Agent 1.0）](https://help.aliyun.com/zh/model-studio/single-agent-application)）。 > 此处设置仅应用于该智能体后续的用户问答。 @@ -141,6 +141,42 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 切换至**召回策略**页签，在**标签过滤**中选择需要应用的标签（如示例中的『硬件』『智能手机』）即可。 + - 在新版智能体中，通过系统提示词引导智能体按标签进行过滤。 + + 1. **为知识库文档设置标签**：在管理知识库时，为各文件添加标签。 + + 例如，百炼电脑的标签为`bailian_pc`，百炼系列手机产品介绍的标签为`bailian_mobile`，百炼电脑服务及优惠政策的标签为`bailian_service`和`bailian_pc`。 + + 2. **在系统提示词中引导智能体**：在智能体的系统提示词中，清晰地说明各标签的含义与使用规则。例如： + + ``` + 你是阿里云百炼产品的智能客服，请回答与阿里云百炼产品相关的问题。 + + 当问题比较宽泛时，检索整个知识库。 + 当问题比较明确时，使用知识库标签进行定向检索。标签映射关系如下： + 1. 与百炼手机相关：bailian_mobile + 2. 与百炼电脑相关：bailian_pc + 3. 与百炼产品相关：bailian_mobile 或 bailian_pc + 4. 与百炼电脑服务及优惠政策相关：bailian_service 和 bailian_pc + ``` + + 3. 智能体会识别用户意图，匹配对应的标签，并仅从带有这些标签的文件中检索内容。匹配逻辑包含以下几种模式： + + - **单标签匹配**：对于问题“百炼有哪些手机产品？”，智能体会检索所有带有`bailian_mobile`标签的文件。 + + - **多标签“或”逻辑**：对于问题“百炼有哪些产品？”，智能体会检索所有带有`bailian_mobile`或`bailian_pc`标签的文件。 + + - **多标签“与”逻辑**：对于问题“百炼电脑的服务政策是什么？”，智能体仅检索同时带有`bailian_pc`和`bailian_service`标签的文件。 + + + 您可以在知识库交互卡片中查看标签匹配逻辑，并通过调整提示词来优化匹配效果。匹配逻辑的格式如下： + + - **单标签匹配**：`[{"tags":["bailian_mobile"]}]` + + - **多标签“或”逻辑**：`[{"tags":["bailian_mobile","bailian_pc"]}]` + + - **多标签“与”逻辑：**`[{"tags":["bailian_service"]},{"tags":["bailian_pc"]}]` + 2. **典型问题****：**知识库中含多个结构内容相同/相近的文件，如文件A和文件B的内容都包含“功能概述”章节，只希望在A文件的“功能概述”中检索。 @@ -302,7 +338,7 @@ RAG（Retrieval Augmented Generation，检索增强生成）是一种结合了 从下方示意图可以看到，目标知识库中实际与用户提示词相关，需要返回的文本切片总共有7个（下图左侧，已用绿色标出），但由于已经超出了当前设定的最大召回片段数K，因此包含优势5（超长待机）和优势6（拍照清晰）的文本切片被舍弃，没有提供给大模型。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8526439771/CAEQURiBgMCatMinohkiIDY3YTVhOWY2MjNjMjRkYzc5NTU1ZmVhNGQ2MGQ5ODc24762899_20250109142407.621.svg) + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6954804871/CAEQURiBgMCatMinohkiIDY3YTVhOWY2MjNjMjRkYzc5NTU1ZmVhNGQ2MGQ5ODc24762899_20250109142407.621.svg) 由于 RAG 本身无法判断需要多少个文本切片才能给出“完整”的答案，因此即使最终提供的文本切片有遗漏，随后大模型仍然会基于缺失的文本切片生成不完整的回答。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md index eabfd600..2673fe86 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md @@ -79,7 +79,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "model": "voice-enrollment", "input": { "action": "create_voice", - "target_model": "qwen-audio-3.0-tts-plus", + "target_model": "qwen-audio-3.0-tts-flash", "prefix": "myvoice", "url": "https://your-audio-url.wav", "language_hints": ["zh"] @@ -181,7 +181,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -189,7 +189,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围（因模型而异）： -- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-flash： - zh：中文 @@ -297,7 +297,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -307,7 +307,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 @@ -320,7 +320,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ ``` { "output": { - "voice_id": "qwen-audio-3.0-tts-plus-myvoice-xxxxxx" + "voice_id": "qwen-audio-3.0-tts-flash-myvoice-xxxxxx" }, "usage": { "count": 1 @@ -702,7 +702,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "output": { "voice_list": [ { - "voice_id": "qwen-audio-3.0-tts-plus-myvoice-xxxxxx", + "voice_id": "qwen-audio-3.0-tts-flash-myvoice-xxxxxx", "gmt_create": "2024-12-11 13:38:02", "gmt_modified": "2024-12-11 13:38:02", "status": "OK" @@ -873,7 +873,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "output": { "gmt_create": "2024-12-11 13:38:02", "resource_link": "https://yourAudioFileUrl", - "target_model": "qwen-audio-3.0-tts-plus", + "target_model": "qwen-audio-3.0-tts-flash", "gmt_modified": "2024-12-11 13:38:02", "status": "OK" }, diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md index 68d6d5fc..be92e944 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md @@ -289,7 +289,7 @@ List **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -297,7 +297,7 @@ List 取值范围（因模型而异）： -- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-flash： - zh：中文 @@ -374,7 +374,7 @@ Float **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -404,7 +404,7 @@ boolean **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 @@ -431,7 +431,7 @@ public class Main { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; String apiKey = System.getenv("DASHSCOPE_API_KEY"); - String targetModel = "qwen-audio-3.0-tts-plus"; + String targetModel = "qwen-audio-3.0-tts-flash"; String prefix = "myvoice"; String fileUrl = "https://your-audio-file-url"; String cloneModelName = "voice-enrollment"; @@ -506,7 +506,7 @@ import org.slf4j.LoggerFactory; public class Main { public static String apiKey = System.getenv("DASHSCOPE_API_KEY"); // 如果您没有配置环境变量，请在此处用您的API-KEY进行替换 - private static String voiceId = "qwen-audio-3.0-tts-plus-myvoice-xxx"; // 请按实际情况进行替换 + private static String voiceId = "qwen-audio-3.0-tts-flash-myvoice-xxx"; // 请按实际情况进行替换 private static final Logger logger = LoggerFactory.getLogger(Main.class); public static void main(String[] args) @@ -535,7 +535,7 @@ import org.slf4j.LoggerFactory; public class Main { public static String apiKey = System.getenv("DASHSCOPE_API_KEY"); // 如果您没有配置环境变量，请在此处用您的API-KEY进行替换 private static String fileUrl = "https://your-audio-file-url"; // 请按实际情况进行替换 - private static String voiceId = "qwen-audio-3.0-tts-plus-myvoice-xxx"; // 请按实际情况进行替换 + private static String voiceId = "qwen-audio-3.0-tts-flash-myvoice-xxx"; // 请按实际情况进行替换 private static final Logger logger = LoggerFactory.getLogger(Main.class); public static void main(String[] args) @@ -562,7 +562,7 @@ import org.slf4j.LoggerFactory; public class Main { public static String apiKey = System.getenv("DASHSCOPE_API_KEY"); // 如果您没有配置环境变量，请在此处用您的API-KEY进行替换 - private static String voiceId = "qwen-audio-3.0-tts-plus-myvoice-xxx"; // 请按实际情况进行替换 + private static String voiceId = "qwen-audio-3.0-tts-flash-myvoice-xxx"; // 请按实际情况进行替换 private static final Logger logger = LoggerFactory.getLogger(Main.class); public static void main(String[] args) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md index d23268bd..14b73cd2 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md @@ -112,7 +112,7 @@ List\[str\] **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -120,7 +120,7 @@ List\[str\] 取值范围（因模型而异）： -- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-flash： - zh：中文 @@ -199,7 +199,7 @@ float **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -213,7 +213,7 @@ bool **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 @@ -391,7 +391,7 @@ service = VoiceEnrollmentService() # 避免频繁调用。每次调用都会创建新音色，达到配额上限后将无法创建。 voice_id = service.create_voice( - target_model='qwen-audio-3.0-tts-plus', + target_model='qwen-audio-3.0-tts-flash', prefix='myvoice', url='https://your-audio-file-url' # language_hints=['zh'], @@ -427,7 +427,7 @@ from dashscope.audio.tts_v2 import VoiceEnrollmentService dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" service = VoiceEnrollmentService() -voice_id = 'qwen-audio-3.0-tts-plus-myvoice-xxxxxxxx' +voice_id = 'qwen-audio-3.0-tts-flash-myvoice-xxxxxxxx' voice_details = service.query_voice(voice_id=voice_id) @@ -444,7 +444,7 @@ dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co service = VoiceEnrollmentService() service.update_voice( - voice_id='qwen-audio-3.0-tts-plus-myvoice-xxxxxxxx', + voice_id='qwen-audio-3.0-tts-flash-myvoice-xxxxxxxx', url='https://your-new-audio-file-url' ) print(f"Update submitted. Request ID: {service.get_last_request_id()}") @@ -458,6 +458,6 @@ from dashscope.audio.tts_v2 import VoiceEnrollmentService dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" service = VoiceEnrollmentService() -service.delete_voice(voice_id='qwen-audio-3.0-tts-plus-myvoice-xxxxxxxx') +service.delete_voice(voice_id='qwen-audio-3.0-tts-flash-myvoice-xxxxxxxx') print(f"Deletion submitted. Request ID: {service.get_last_request_id()}") ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md index f04270dd..6eb49640 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md @@ -18,6 +18,22 @@ **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) +qwen3.7-text-embedding + +2560、2,048、1,536、1,024（默认）、768、512、256 + +20 + +128,000 + +0.0005元 + +中文、英语、西班牙语、法语、葡萄牙语、印尼语、日语、韩语、德语、俄罗斯语等201种主流语种与方言 + +各100万Token + +有效期：百炼开通后90天内 + text-embedding-v4 > 属于[Qwen3-Embedding](https://qwenlm.github.io/zh/blog/qwen3-embedding/)系列 @@ -34,10 +50,6 @@ text-embedding-v4 中文、英语、西班牙语、法语、葡萄牙语、印尼语、日语、韩语、德语、俄罗斯语等100+主流语种及多种编程语言 -各100万Token - -有效期：百炼开通后90天内 - text-embedding-v3 1,024（默认）、768、512、256、128或64 @@ -78,9 +90,11 @@ text-embedding-v1 ## 公共云 -**使用SDK调用时需配置的base\_url：**`https://dashscope.aliyuncs.com/compatible-mode/v1` +**使用SDK调用时需配置的base\_url：**`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -**使用HTTP方式调用时需配置的endpoint：**`POST https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings` +**使用HTTP方式调用时需配置的endpoint：**`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ### **请求体** @@ -94,7 +108,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), # 如果您没有配置环境变量，请在此处用您的API Key进行替换 - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" # 百炼服务的base_url + # 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" ) completion = client.embeddings.create( @@ -120,7 +135,8 @@ public class Main { // 创建客户端，使用环境变量中的API密钥 OpenAIClient client = OpenAIOkHttpClient.builder() .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .baseUrl("https://dashscope.aliyuncs.com/compatible-mode/v1") + // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + .baseUrl("https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") .build(); // 创建向量化请求参数 @@ -148,7 +164,7 @@ public class Main { ## curl ``` -curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -169,7 +185,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), # 如果您没有配置环境变量，请在此处用您的API Key进行替换 - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" # 百炼服务的base_url + # 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" ) completion = client.embeddings.create( @@ -199,7 +216,8 @@ public class Main { // 创建客户端，使用环境变量中的API密钥 OpenAIClient client = OpenAIOkHttpClient.builder() .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .baseUrl("https://dashscope.aliyuncs.com/compatible-mode/v1") + // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + .baseUrl("https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") .build(); // 创建输入字符串列表 @@ -246,7 +264,7 @@ public class Main { ## curl ``` -curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -272,7 +290,8 @@ from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), # 如果您没有配置环境变量，请在此处用您的API Key进行替换 - base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" # 百炼服务的base_url + # 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" ) # 确保将 'texts_to_embedding.txt' 替换为您自己的文件名或路径 with open('texts_to_embedding.txt', 'r', encoding='utf-8') as f: @@ -303,7 +322,8 @@ public class Main { // 创建客户端，使用环境变量中的API密钥 OpenAIClient client = OpenAIOkHttpClient.builder() .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .baseUrl("https://dashscope.aliyuncs.com/compatible-mode/v1") + // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 + .baseUrl("https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") .build(); // 确保将 'texts_to_embedding.txt' 替换为您自己的文件名或绝对路径 @@ -350,7 +370,7 @@ public class Main { ``` FILE_CONTENT=$(cat texts_to_embedding.txt | jq -Rs .) -curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -369,6 +389,12 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ 输入待处理的文本。可以是字符串（string）、字符串列表（array）或文件（file）。不同模型版本支持的文本长度和批量大小不同，具体如下： +- **qwen3.7-text-embedding 模型：** + + - **输入为字符串**：最长支持 **128,000** Token。 + + - **输入为字符串列表或文件**：最多支持 **20** 条（行），每条（行）最长支持 **128,000** Token。 + - **text-embedding-v3 / v4 模型：** - **输入为字符串**：最长支持 **8,192** Token。 @@ -384,7 +410,7 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ **dimensions** `_integer_` **可选** -指定的向量维度，必须为以下值之一：2048（仅适用于`text-embedding-v4`）、1536（仅适用于`text-embedding-v4`）1024、768、512、256、128 或 64，默认值为1024。 +指定的向量维度，必须为以下值之一：`2560（仅适用于qwen3.7-text-embedding）、`2048（仅适用于`text-embedding-v4`）、1536（仅适用于`text-embedding-v4`）1024、768、512、256、128 或 64，默认值为1024。 **encoding\_format** `_string_` **可选** @@ -482,9 +508,11 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ ## 公共云 -**使用SDK调用时需配置的base\_url：**https://dashscope.aliyuncs.com/api/v1 +**使用SDK调用时需配置的base\_url：**https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1 + +**使用HTTP方式调用时需配置的endpoint：**POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding -**使用HTTP方式调用时需配置的endpoint：**POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ### **请求体** @@ -495,6 +523,8 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/embeddings' \ ``` import dashscope from http import HTTPStatus +# 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" resp = dashscope.TextEmbedding.call( model="text-embedding-v4", @@ -515,12 +545,15 @@ import com.alibaba.dashscope.embeddings.TextEmbeddingParam; import com.alibaba.dashscope.embeddings.TextEmbeddingResult; import com.alibaba.dashscope.exception.ApiException; import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.Constants; /** * 千问文本向量模型调用示例 */ public final class Main { public static void main(String[] args) { + // 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; try { // 构建请求参数 TextEmbeddingParam param = TextEmbeddingParam @@ -547,7 +580,7 @@ public final class Main { ## curl ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -571,6 +604,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text- ``` import dashscope from http import HTTPStatus +# 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" DASHSCOPE_MAX_BATCH_SIZE = 10 @@ -611,11 +646,14 @@ import com.alibaba.dashscope.embeddings.TextEmbeddingResult; import com.alibaba.dashscope.exception.ApiException; import com.alibaba.dashscope.exception.NoApiKeyException; import com.alibaba.dashscope.embeddings.TextEmbeddingResultItem; +import com.alibaba.dashscope.utils.Constants; public final class Main { private static final int DASHSCOPE_MAX_BATCH_SIZE = 10; public static void main(String[] args) { + // 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; List inputs = Arrays.asList( "风急天高猿啸哀", "渚清沙白鸟飞回", @@ -664,7 +702,7 @@ public final class Main { ## curl ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -691,6 +729,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text- ``` from http import HTTPStatus from dashscope import TextEmbedding +# 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" with open('texts_to_embedding.txt', 'r', encoding='utf-8') as f: resp = TextEmbedding.call( @@ -716,9 +756,12 @@ import com.alibaba.dashscope.embeddings.TextEmbeddingParam; import com.alibaba.dashscope.embeddings.TextEmbeddingResult; import com.alibaba.dashscope.exception.ApiException; import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.Constants; public final class Main { public static void main(String[] args) { + // 以下为华北2（北京）地域的配置，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; try (BufferedReader reader = new BufferedReader(new FileReader("<文件所来自的内容根的路径>"))) { StringBuilder content = new StringBuilder(); String line; @@ -753,7 +796,7 @@ public final class Main { ``` FILE_CONTENT=$(cat texts_to_embedding.txt | jq -Rs .) -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ @@ -776,6 +819,12 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text- 输入待处理的文本。可以是字符串（string）、字符串列表（array）或文件（file）。不同模型版本支持的文本长度和批量大小不同，具体如下： +- **qwen3.7-text-embedding 模型：** + + - **输入为字符串**：最长支持 **128,000** Token。 + + - **输入为字符串列表或文件**：最多支持 **20** 条（行），每条（行）最长支持 **128,000** Token。 + - **text-embedding-v3 / v4 模型：** - **输入为字符串**：最长支持 **8,192** Token。 @@ -799,13 +848,13 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text- > 通过 HTTP 调用时，请将 **dimension** 放入parameters对象中。 -指定的向量维度，必须为以下值之一：2048（仅适用于`text-embedding-v4`）、1536（仅适用于`text-embedding-v4`）1024、768、512、256、128 或 64，默认值为1024。 +指定的向量维度，必须为以下值之一：`2560（仅适用于qwen3.7-text-embedding）、`2048（仅适用于`text-embedding-v4`）、1536（仅适用于`text-embedding-v4`）1024、768、512、256、128 或 64，默认值为1024。 **output\_type** `_string_` **可选** > 通过 HTTP 调用时，请将 **output\_type** 放入parameters对象中。 -用户指定输出离散向量表示只适用于`text-embedding-v3`与`text-embedding-v4`模型，取值在dense、sparse、dense&sparse之间，默认取dense，只输出连续向量。 +用户指定输出离散向量表示只适用于`qwen3.7-text-embedding、``text-embedding-v3`与`text-embedding-v4`模型，取值在dense、sparse、dense&sparse之间，默认取dense，只输出连续向量。 **instruct** `_string_` **可选** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md index c781b70e..833a58c2 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md @@ -469,6 +469,10 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( - vidu/viduq3-drama\_reference2video：取值为\[2, 15\]之间的整数。默认值为`5`。 + - 当分镜数量过多或明显少于duration时，模型可能自动调整时长以保障故事完整性，实际视频时长可能多于或少于duration设定值。 + + - 计费按实际输出时长计算。 + - vidu/viduq3-mix\_reference2video：取值为\[1, 16\]之间的整数。默认值为`5`。 - vidu/viduq3\_reference2video：取值为\[1, 16\]之间的整数。默认值为`5`。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md index a5e04187..7a0b36c6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md @@ -1262,7 +1262,7 @@ cosyvoice-v3.5-plus ](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/cosyvoice-v3.5-plus) -华北2（北京）新加坡 +华北2（北京） 模型 ID`cosyvoice-v3.5-plus` @@ -1270,12 +1270,6 @@ Request URL`wss://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing? API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`cosyvoice-v3.5-plus` - -Request URL`wss://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api-ws/v1/inference` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - [ MiniMax/speech-2.8-hd diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md index d8d8a8c7..48d7cea0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md @@ -8507,19 +8507,27 @@ Tripo/Tripo-P1.0 **限流条件（超出任一数值时触发限流）** -**每秒钟调用次数（RPS）** +**每分钟调用次数（RPM）** **每分钟消耗Token数（TPM）/作业数** > **仅输入Token** +qwen3.7-text-embedding + +中国内地 + +1,800 + +1,000,000 + text-embedding-v1 > 用[Batch API](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)调用服务时，不受限流限制。 中国内地 -30 +1,800 1,200,000 @@ -8529,7 +8537,7 @@ text-embedding-v2 中国内地 -30 +1,800 1,200,000 @@ -8539,7 +8547,7 @@ text-embedding-v3 中国内地 -30 +1,800 1,200,000 @@ -8549,7 +8557,7 @@ text-embedding-v4 中国内地 -30 +1,800 1,200,000 @@ -8557,7 +8565,7 @@ text-embedding-async-v1 中国内地 -1 +60 当前用户在系统通用文本向量异步作业排队中和运行中的作业数量不超过50个。 @@ -8567,7 +8575,7 @@ text-embedding-async-v2 中国内地 -1 +60 当前用户在系统通用文本向量异步作业排队中和运行中的作业数量不超过50个。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tripo-3d-generation-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tripo-3d-generation-guide.md index d1154f93..52bc6b70 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tripo-3d-generation-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tripo-3d-generation-guide.md @@ -4,7 +4,7 @@ **重要** -本文档仅适用于“中国内地（北京）”地域，且必须使用该地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 +本文档仅适用于华北2（北京）地域，且必须使用该地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 ## **快速开始** @@ -24,8 +24,10 @@ #### **步骤1：创建任务获取任务ID** +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -47,7 +49,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -136,7 +138,7 @@ Tripo/Tripo-P1.0 > 上图为效果展示，实际产物为[3D文件](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260508/smfsql/red_car.glb) ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -168,7 +170,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 上图为效果展示，实际产物为[3D文件](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260507/lqsfqk/tripo-image-to-3d-result.glb) ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -200,7 +202,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 上图为效果展示，实际产物为[3D文件](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260507/qoomsm/tripo-multi-image-to-3d-result.glb) ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md index a23c8c47..65c53e25 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md @@ -22,7 +22,7 @@ OpenAI gpt-4o-tts、Google Chirp 3 ElevenLabs Multilingual v3 -`qwen-audio-3.0-tts-plus`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） +`qwen-audio-3.0-tts-flash`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） ## 标准语音合成还是自定义音色？ @@ -54,7 +54,7 @@ ElevenLabs Multilingual v3 `qwen-audio-3.0-tts-plus`、`MiniMax/speech-2.8-hd` -`qwen-audio-3.0-tts-plus`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） +`qwen-audio-3.0-tts-flash`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） - **使用标准语音合成**：当内置音色库能满足需求，希望快速上手、无需额外配置时。 @@ -89,7 +89,7 @@ ElevenLabs Multilingual v3 推荐模型 -`qwen-audio-3.0-tts-plus`、`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` +`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` `cosyvoice-v3.5-plus`、`cosyvoice-v3.5-flash` @@ -145,7 +145,7 @@ Qwen-Audio-TTS WebSocket / HTTP -支持 +不支持 不支持 @@ -205,7 +205,7 @@ HTTP WebSocket / HTTP -支持 +不支持 不支持 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md index 82abe4ac..8e525acd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md @@ -232,7 +232,6 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 - deepseek-v4-pro @@ -246,7 +245,7 @@ deepseek-v4-flash 256K -不支持缓存 +0.2（缓存命中部分按 20% 折算容量） 无阶梯（1.0） @@ -254,6 +253,14 @@ Kimi-K2.6 256K +0.2（缓存命中部分按 20% 折算容量） + +无阶梯（1.0） + +Qwen3.6-flash-2026-04-16 + +256K + 不支持缓存 无阶梯（1.0） diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md index 590cc2b9..a108688f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md @@ -25,6 +25,22 @@ **7月** +**日期** + +**功能模块** + +**功能点** + +**功能说明** + +7月14日 + +平台功能 + +GLM-5.2 Fast mode 模式降价通知 + +GLM-5.2 Fast mode 模式降价通知[了解详情](https://www.aliyun.com/notice/118443) + 7月13日 平台功能 @@ -41,14 +57,6 @@ 部分老旧模型下线通知[了解详情](https://www.aliyun.com/notice/118434) -**日期** - -**功能模块** - -**功能点** - -**功能说明** - 7月9日 平台功能 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md index b87f831a..2f0e655e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md @@ -16346,6 +16346,14 @@ Tripo/Tripo-P1.0 有效期：阿里云百炼开通后90天内 +qwen3.7-text-embedding + +中国内地 + +0.5元 + +100万Token + text-embedding-v4 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 diff --git a/skills/bailian-docs-llm-wiki/raw/test/test.md b/skills/bailian-docs-llm-wiki/raw/test/test.md new file mode 100644 index 00000000..0b2a42c2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/test/test.md @@ -0,0 +1 @@ +# Test doc diff --git a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md index 81251fc1..030ac4d8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md @@ -1,87 +1,65 @@ # 3d generation -百炼平台基于 Tripo 模型提供 3D 资产生成能力，支持文生 3D、单图生 3D 与多图生 3D 三种输入方式，产出带贴图的 PBR 材质 GLB 模型或无贴图基础模型。由于生成耗时较长，API 采用[异步调用](../concepts/async-invocation.md)，整体流程为「创建任务 → 轮询获取结果」。详细接口与参数见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 适用范围 - -- 仅适用于**华北2（北京）**地域，且必须使用该地域的 [API Key](../concepts/api-key.md)。 -- 需先在百炼控制台模型市场搜索「Tripo」并开通服务、完成授权，再配置好 [API Key](../concepts/api-key.md) 环境变量。具体开通与配置步骤见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 调用流程 - -API 仅支持[异步调用](../concepts/async-invocation.md)，包含两个步骤： - -1. **创建任务**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` -2. **轮询查询结果**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` - -创建任务时必须携带 `X-DashScope-Async: enable` 请求头，否则会报错 `current user api does not support synchronous calls`。成功创建后返回 `task_id`，有效期 24 小时，**请勿重复创建任务**，轮询获取即可。 - -轮询建议间隔约 15 秒，任务状态流转为 `PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED` / `FAILED`。查询接口默认 RPS 为 20，如需更高频查询或事件通知建议配置异步任务回调。 - -## 支持的模型 - -| 模型名 | 定位 | 输出面数 | 对应官方 API 版本 | -| --- | --- | --- | --- | -| `Tripo/Tripo-H3.1` | 高精度生成 | 最高 200 万面 | `v3.1-20260211` | -| `Tripo/Tripo-P1.0` | 专业生成，速度更快 | 最高 2 万面 | `P1-20260311` | - -## 输入方式 - -`input` 中 `prompt`、`image`、`images` 三者**互斥**，只能选其一，同时传多个将报错。 - -- **文生 3D**：`prompt` 必填，支持中英文等多语言，每个字符计 1 个字符，最大 1024 字符。 -- **单图生 3D**：`image` 必填，传入单张图像公网 URL。图像格式限 JPEG/PNG，宽高范围 [20, 6000] 像素（建议边长大于 256），文件不超过 20MB，支持 HTTP/HTTPS。 -- **多图生 3D**：`images` 必填，数组长度固定为 4，对应视角顺序为**前、左、后、右**；不需要的视角传空对象 `{}`。实际有效图片数为 2~4 张，多张图像的分辨率和宽高比不要求一致。每个对象含 `type`（`jpeg` 或 `png`）与 `file_token`（公网 URL）字段。 - -## 关键参数（parameters） - -| 参数 | 适用模型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `texture_quality` | 全部 | `standard` | 贴图质量，可选 `standard`（标清）/`detailed`（高清） | -| `geometry_quality` | `Tripo/Tripo-H3.1` | `standard` | 几何精度，`standard` 最高 150 万面，`ultra` 最高 200 万面 | -| `pbr` | 全部 | `true` | 是否生成 PBR 材质模型。设为 `true` 时强制启用贴图，返回 `pbr_model_url` | -| `texture` | 全部 | `true` | 是否生成贴图。生成无贴图模型需**同时**将 `texture` 和 `pbr` 设为 `false`，返回 `base_model_url` | - -## 响应与产物 - -成功响应的 `output.results` 仅在 `task_status` 为 `SUCCEEDED` 时返回，包含以下字段： - -- `pbr_model_url`：PBR 材质模型（GLB）下载 URL，当 `pbr` 为 `true`（默认）时返回。 -- `base_model_url`：无贴图基础模型（GLB）下载 URL，当 `texture` 与 `pbr` 均为 `false` 时返回。 -- `rendered_image_url`：3D 模型预览渲染图（1 张）URL。 - -> **注意**：以上下载链接有效期均为 **2 小时**，请及时下载。 - -`usage` 字段记录任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）、生成数量 `count`、贴图质量与几何精度，仅对成功结果计数。任务状态 `task_status` 的完整枚举为 `PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `CANCELED` / `UNKNOWN`，其中 `UNKNOWN` 表示任务不存在或超过 24 小时有效期。更多响应字段说明见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 限制与注意事项 - -- 仅限北京地域 [API Key](../concepts/api-key.md) 调用，地域不匹配将无法使用。 -- `task_id` 查询有效期 24 小时，超时返回 `UNKNOWN` 且无法再查询。 -- 查询接口默认 RPS 限制为 20，建议通过异步任务回调获取更高频通知。 -- 产物下载链接有效期仅 2 小时。 -- 调用失败时响应中会返回 `code` 与 `message`，可参照百炼错误码文档排查。 +百炼平台的 3D 生成能力基于 Tripo 模型，支持文生 3D、单图生 3D 和多图生 3D 三种输入模式，输出带 PBR 材质或无贴图的 GLB 格式模型及预览渲染图。该能力为异步任务，需通过 `task_id` 轮询获取结果，**仅在华北2（北京）地域可用**。详细接口规范与行为约束请参考 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 + +## 支持的模型/功能 + +- **模型列表**： + - `Tripo/Tripo-H3.1`：高精度生成，最高支持 200 万面，对应 Tripo 官方 API 版本 `v3.1-20260211`； + - `Tripo/Tripo-P1.0`：专业级快速生成，最高 2 万面，对应版本 `P1-20260311`。 +- **输入模式**（三者互斥）： + - 文生 3D：通过 `prompt` 字段传入文本描述； + - 单图生 3D：通过 `image` 字段传入单张公网 URL 图像； + - 多图生 3D：通过 `images` 数组传入 4 张按「前、左、后、右」顺序排列的图像（空视角用 `{}` 占位），实际有效图数为 2–4 张。 +- **输出类型**： + - 默认返回 `pbr_model_url`（带 PBR 材质的 GLB）； + - 若显式设置 `"texture": false, "pbr": false`，则返回 `base_model_url`（无贴图基础模型）； + - 始终返回 `rendered_image_url`（单张预览图）。 + +> **注意**：文档中 `images` 数组长度固定为 4，但示例中存在传入 2 张图 + 2 个 `{}` 的写法，与“实际有效图数为 2~4 张”的说明一致；而部分旧文档曾误述为“必须填满 4 张”，此表述已过时，请以 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中的当前定义为准。 + +## 关键参数 + +| 参数 | 类型 | 是否必填 | 说明 | +|------|------|----------|------| +| `model` | string | 必填 | 固定为 `Tripo/Tripo-H3.1` 或 `Tripo/Tripo-P1.0` | +| `input.prompt` / `input.image` / `input.images` | string / string / array | 条件必填 | 三者仅选其一；`images` 数组长度恒为 4，空视角用 `{}` | +| `parameters.texture_quality` | string | 可选 | `standard`（默认）或 `detailed`；仅对带贴图输出生效 | +| `parameters.geometry_quality` | string | 可选 | 仅 `Tripo/Tripo-H3.1` 支持；`standard`（≤150 万面）或 `ultra`（≤200 万面） | +| `parameters.pbr` | boolean | 可选 | 默认 `true`；设为 `false` 时需同步设 `texture: false` 才能获得无贴图模型 | +| `parameters.texture` | boolean | 可选 | 默认 `true`；与 `pbr` 联动，详见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) | + +## 使用方式 + +1. **前置准备**： + - 在[百炼控制台（北京地域）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)开通 Tripo 服务； + - 配置环境变量 `DASHSCOPE_API_KEY`（仅限北京地域 API Key）。 + +2. **创建任务**（POST）： + - Endpoint：`https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + - 请求头必须包含：`Content-Type: application/json`、`Authorization: Bearer `、`X-DashScope-Async: enable` + - 成功响应含 `task_id`（有效期 24 小时），**禁止重复提交相同任务**。 + +3. **轮询结果**（GET）： + - Endpoint：`https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` + - 建议间隔 ≥15 秒；状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED`/`FAILED`； + - `SUCCEEDED` 时 `output.results` 返回 `pbr_model_url`、`base_model_url`（按参数配置）和 `rendered_image_url`，所有 URL 有效期均为 2 小时。 + +## 限制和注意事项 + +- **地域限制**：API 仅支持华北2（北京）地域，跨地域调用将失败； +- **输入限制**： + - `prompt` 最长 1024 字符； + - 单图 `image` 或 `images[i].file_token` 必须为公网可访问的 HTTP/HTTPS URL，格式为 JPEG/PNG，分辨率 [20, 6000] 像素，单文件 ≤20MB； +- **任务生命周期**： + - `task_id` 有效期严格为 24 小时，超时后查询返回 `task_status: UNKNOWN`； + - 成功结果中的 URL（如 `pbr_model_url`）有效期仅 2 小时，需及时下载； +- **错误处理**： + - 缺少 `X-DashScope-Async: enable` 头将报错 `current user api does not support synchronous calls`； + - 错误码详情见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中引用的错误码文档。 ## 来源文档 - [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md index 3669a20d..fe0f5b4e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md @@ -1,150 +1,63 @@ # application call -阿里云百炼平台提供两套 API 来调用智能体和工作流应用：**OpenAI 兼容的 Responses API** 和 **DashScope API**。两者均支持同步/[异步调用](../concepts/async-invocation.md)、多轮对话、[流式输出](../concepts/streaming.md)等核心能力，开发者可根据生态兼容性和功能需求选择合适的接入方式。调用前需先获取 APP ID（以及子[业务空间](../concepts/workspace.md)场景下的 Workspace ID）和 [API Key](../concepts/api-key.md)。 - -## 前置准备 - -### 获取凭证 - -通过 API 调用应用时，必须提供 **APP ID** 来指定目标应用。如果应用位于子[业务空间](../concepts/workspace.md)，还需提供 **Workspace ID**。详细获取方式参见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 - -- **APP ID**：在控制台「应用管理」页面的应用卡片上复制。 -- **Workspace ID**：在调用子[业务空间](../concepts/workspace.md)下的应用或特定地域（德国、华北2、新加坡、日本）的模型时必须提供，可通过控制台右上角图标查看。 - -> **注意**：目前只能通过控制台手动获取 APP ID 和 Workspace ID，不支持通过 API 或 CLI 查询。 - -### 其他前提 - -- 已获取 [API Key](../concepts/api-key.md) 并配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已创建并发布百炼应用（智能体或工作流）。 -- 如使用 SDK 调用，需安装对应的 SDK（OpenAI SDK 或 [DashScope SDK](../concepts/dashscope-sdk.md)）。 - -## 两套 API 对比 - -| 维度 | Responses API（OpenAI 兼容） | DashScope API | -|------|---------------------------|---------------| -| Endpoint | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK | OpenAI Python/Java SDK | DashScope Python/Java SDK | -| 多轮对话 | 通过 `input` 数组传递完整历史消息 | 通过 `session_id` 或 `messages` 维护上下文 | -| [异步调用](../concepts/async-invocation.md) | 设置 `background=true` | 暂不支持（仅 Responses API 提供） | -| 适用地域 | 仅华北2（北京） | 仅华北2（北京） | - -## Responses API（OpenAI 兼容模式） - -### 同步调用 - -适用于需要即时获取结果的实时交互场景。完整参数说明参见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 - -**核心请求参数：** - -| 参数 | 类型 | 必选 | 说明 | -|------|------|------|------| -| `input` | string / array | 是 | 请求输入，可为简单字符串或包含多轮对话历史的消息数组 | -| `stream` | boolean | 否 | 是否[流式输出](../concepts/streaming.md)，默认 `false` | -| `background` | boolean | 否 | 是否异步执行，默认 `false` | - -**Python 示例（单轮对话）：** - -```python -from openai import OpenAI -import os - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url=f'https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/' -) - -response = client.responses.create(input="你是谁？") -print(response.model_dump_json(indent=2)) -``` - -**多轮对话**需在 `input` 中传递完整的消息历史数组，每条消息包含 `role`（user/assistant/system）和 `content` 字段。 - -### [多模态](../concepts/multimodal.md)输入 - -Responses API 支持在 `content` 数组中混合多种输入类型： - -- **图像输入**：通过 `input_image` 类型传入图片 URL。[智能体应用](../concepts/agent-application.md)需选用通义千问 VL 系列模型并将文件处理方式设为「自定义处理」。 -- **文件输入**：通过 `input_file` 类型传入文件 URL。仅[智能体应用](../concepts/agent-application.md)支持，需配置「全文引用」或「切片检索」处理方式。 - -### [异步调用](../concepts/async-invocation.md) - -对于耗时较长的任务（如生成报告、多步骤工具调用），可设置 `background=true` 开启异步模式，避免请求超时。详细流程参见 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 - -核心流程： - -1. **创建任务**：请求中设置 `background=true`，API 立即返回任务 ID。 -2. **轮询状态**：通过 `client.responses.retrieve(task_id)` 定期查询任务状态。 -3. **处理结果**：当状态变为 `completed`、`failed` 或 `cancelled` 时获取最终结果。 - -> **注意**：异步任务暂不支持[流式输出](../concepts/streaming.md)（`stream=true`）。 - -### [流式输出](../concepts/streaming.md) - -设置 `stream=true` 可边生成边输出，适用于需要实时展示生成内容的场景。若应用类型为工作流，需在结束节点或流程输出节点中启用「[流式输出](../concepts/streaming.md)」开关并重新发布。 - -## DashScope API - -DashScope API 提供更全面的功能支持，适合需要深度集成百炼平台能力的场景。支持 Python、Java、PHP、Node.js、C#、Go 等多种语言的 HTTP 调用，以及 Python/Java SDK。详细参数参见 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 - -**Python 示例：** - -```python -from dashscope import Application -import os - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -**多轮对话**通过 `session_id` 维护上下文：首次请求无需传入，响应中会返回 `session_id`；后续请求携带该值即可延续对话。`session_id` 在最后一次请求后 1 小时内有效。 - -新版[智能体应用](../concepts/agent-application.md)（Agent 2.0）的调用方式与上述基本一致，参见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 - -## 参数传递 - -### 自定义参数 - -工作流应用中定义的自定义参数，通过请求体中的 `biz_params` 传递，参数名和类型需与应用内配置保持一致。 - -```python -response = await client.responses.create( - input="你好", - extra_body={"biz_params": {"city": "北京"}}, - background=True -) -``` - -### 插件参数 - -智能体或工作流中配置的插件工具参数，同样通过 `biz_params` 传递。工作流需在开始节点创建自定义参数并将其传入插件节点的输入参数中。 - -## 限制与注意事项 - -- 两套 API 目前均**仅适用于华北2（北京）地域**。 -- Responses API 的多轮对话暂不支持基于 `pre_response_id` 或 `conversation_id` 的上下文功能，需每次传递完整对话历史。 -- [异步调用](../concepts/async-invocation.md)仅 Responses API 支持，且不能与[流式输出](../concepts/streaming.md)同时使用。 -- RAM 子账号查看[业务空间](../concepts/workspace.md)管理页面需要超级管理员权限（`AliyunBailianFullAccess` 或 `AliyunBailianControlFullAccess`）。 -- [API Key](../concepts/api-key.md) 不建议硬编码到代码中，应通过环境变量配置以降低泄露风险。 +`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可通过同步、异步或流式方式发起请求，支持文本、图像、文件等多模态输入，并可复用 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)或原生 DashScope SDK。调用前需准备 APP ID、Workspace ID（如适用）及有效的 API Key。 + +## 支持的模型/功能 + +- **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流应用，详见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) 和 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 +- **多模态输入**： + - 图像：需选用通义千问 VL 系列模型，并在应用中配置为“自定义处理”（智能体）或设置 `imageList` 入参（工作流）[同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； + - 文件：仅智能体应用支持，需启用“全文引用”或“切片检索”文件处理方式； + - 音频/视频：当前仅支持作为 URL 传入（如 `file_url`），由模型节点解析内容。 +- **会话管理**： + - DashScope API 通过 `session_id` 维护上下文，有效期为最后一次请求后 1 小时； + - OpenAI 兼容模式（Responses API）暂不支持 `pre_response_id` 或 `conversation_id`，需在每次请求中传递完整对话历史。 + +> **注意**：文档 3 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 3 明确限定为“新版智能体应用”，而文档 5 泛指“智能体与工作流应用”。实际调用时，工作流应用应优先参考文档 5；若使用新版智能体，文档 3 提供更精确的参数说明。 + +## 关键参数 + +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `app_id` | string | 是 | 应用唯一标识，从[应用管理](https://bailian.console.aliyun.com/#/app-center)页面获取。 | +| `workspace_id` | string | 否（按需） | 业务空间唯一标识，子业务空间或德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域下必须提供，详见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 | +| `input` / `prompt` | string 或 array | 是 | 核心输入：
- DashScope API 使用 `prompt` 字符串（单轮）或 `messages` 数组（多轮）；
- Responses API 使用 `input`，支持字符串（单轮）或消息对象数组（含 `role`, `content`，支持 `input_text`/`input_image`/`input_file`）。 | +| `stream` | boolean | 否 | 仅 Responses API 支持。`true` 启用[流式输出](../concepts/streaming-output.md)；工作流应用需在结束节点启用“[流式输出](../concepts/streaming-output.md)”开关并重新发布。 | +| `background` | boolean | 否 | 仅 Responses API 支持。`true` 切换为异步模式，立即返回任务 ID；异步任务不支持 `stream=true`。 | +| `biz_params` | object | 否 | 仅 Responses API 异步调用支持，用于传递工作流/智能体中预设的自定义参数（如 `{"city": "北京"}`）。 | + +## 使用方式 + +### 1. 接口地址 +- **DashScope API（推荐用于新版智能体/工作流）**： + `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` +- **Responses API（OpenAI 兼容，支持同步/异步/流式）**： + 同步：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` + 异步：同上，但请求体含 `"background": true` + +### 2. 认证方式 +- 所有请求均需在 Header 中携带 `Authorization: Bearer ${DASHSCOPE_API_KEY}`。 +- API Key 需通过[密钥管理](https://bailian.console.aliyun.com/?tab=app#/api-key)获取并配置为环境变量 `DASHSCOPE_API_KEY`。 + +### 3. 代码示例（核心场景） +- **同步调用（文本）**：见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) 的 Python/curl 示例。 +- **多轮对话（DashScope）**：首次调用不传 `session_id`，后续请求携带响应中的 `output.session_id`。 +- **异步调用**：设置 `background=true` 获取 `task_id`，再轮询 `GET /responses/{task_id}` 查询状态（详见 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)）。 + +## 限制和注意事项 + +- **地域限制**：所有文档均明确标注“仅适用于华北2（北京）地域”，其他地域（如德国、新加坡）需配合 `workspace_id` 使用对应 Base URL，且部分功能可能受限。 +- **凭证获取**：APP ID 和 Workspace ID **仅支持控制台手动获取**，不提供 API 或 CLI 查询接口 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **权限要求**：查询全部 Workspace ID 需主账号或具备 `AliyunBailianFullAccess` 权限的 RAM 子账号；普通子账号仅能查看已加入的业务空间。 +- **超时与重试**：同步调用默认超时时间较短，耗时任务（如复杂工作流）务必使用异步模式；异步任务轮询间隔建议 ≥2 秒。 +- **流式限制**：异步调用 (`background=true`) 与[流式输出](../concepts/streaming-output.md) (`stream=true`) **互斥**，二者不可同时启用。 ## 来源文档 - [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) - [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) -- [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) +- [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md index 9fde96a5..46235c73 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md @@ -1,231 +1,109 @@ # application component api reference -百炼平台应用组件 API（`bailian/2023-12-29`）提供了数据连接、知识库、Prompt 模板、长期记忆等核心能力的 OpenAPI 接口，采用 ROA 签名风格。开发者可通过阿里云百炼 SDK 直接调用，也可使用自签名方式对接。所有接口均需传入 `WorkspaceId`（[业务空间](../concepts/workspace.md) ID），RAM 子账号需要先获取对应权限策略并加入[业务空间](../concepts/workspace.md)后才能调用。 +本 API 参考文档面向开发者，系统性地描述了百炼平台 Application Component（应用组件）层提供的核心 OpenAPI 能力，覆盖数据连接（原应用数据）、知识库、Prompt 模板等关键功能模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，推荐通过官方 SDK 调用以简化鉴权与请求构造。详细接入方式与安全要求请参见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 -## 服务接入点与鉴权 +## 支持的模型/功能 -当前支持两个地域的接入点： +Application Component API 主要提供三类能力： -| 地域 | 地域 ID | 公网接入地址 | VPC 接入地址 | -|------|---------|-------------|-------------| -| 华北2（北京） | cn-beijing | bailian.cn-beijing.aliyuncs.com | bailian-vpc.cn-beijing.aliyuncs.com | -| 新加坡 | ap-southeast-1 | bailian.ap-southeast-1.aliyuncs.com | bailian-vpc.ap-southeast-1.aliyuncs.com | +- **数据连接管理**：支持类目（Category）、文件（File）、表格（Table）、连接器（Connector）的全生命周期操作，包括创建、查询、更新、删除及解析设置管理。例如，`AddCategory` 用于构建分类体系，`ApplyFileUploadLease` + `AddFile` 组合实现文件上传与入库，`ChangeParseSetting` 可为不同文件类型（如 `.pdf`, `.jpg`）指定专用解析器（如 `DOCMIND_LLM_VERSION`, `DASH_QWEN_VL_PARSER`）。具体支持的解析器类型可通过 `GetAvailableParserTypes` 接口动态查询，详见 [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md)。 +- **知识库（RAG Index）管理**：覆盖知识库的创建（`CreateIndex`）、提交构建（`SubmitIndexJob`）、追加文档（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）、查询（`ListIndices`, `ListIndexDocuments`）、更新（`UpdateIndex`）及删除（`DeleteIndex`）全流程。同时支持细粒度操作，如切片（Chunk）的增删改查（`ListChunks`, `UpdateChunk`, `DeleteChunk`）和监控（`GetIndexMonitor`）。 +- **Prompt 工程支持**：提供 Prompt 模板的创建（`CreatePromptTemplate`）与获取（`GetPromptTemplate`）能力，支持变量占位符（如 `${theme}`），便于在应用中复用标准化提示词。 -调用前需准备 AccessKey，建议使用 RAM 用户而非主账号以降低安全风险。RAM 权限策略的 RamCode 为 `sfm`，授权粒度为操作级。大多数写操作需要 `AliyunBailianDataFullAccess` 策略，部分只读接口（如 DescribeFile、GetIndexJobStatus）也支持 `AliyunBailianDataReadOnlyAccess`。详见[授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 +> **注意**：`CreateIndex` 接口仅初始化作业，必须调用 `SubmitIndexJob` 才能真正触发知识库构建；而 `Retrieve` 接口的响应延迟较高，需合理配置客户端超时与重试策略。 -## 数据连接（原应用数据） +## 关键参数 -数据连接相关 API 用于管理类目、文件、解析设置、表格和连接器，是构建知识库的数据基础。 +- **通用路径参数**：几乎所有接口均需 `WorkspaceId`（业务空间 ID），用于隔离资源。其值可在控制台业务空间详情页获取，或通过 `ListIndices` 等接口返回结果中提取。 +- **身份认证参数**：所有请求必须携带有效的 AccessKey ID/Secret，并按 ROA 规范签名。强烈建议使用 [阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29) 自动处理，避免手动签名错误。相关安全准备细节见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 +- **核心业务参数**： + - 类目/文件操作：`CategoryId`（来自 `AddCategory` 返回）、`FileId`（来自 `AddFile` 返回）、`ConnectorId`（来自 `AddConnector` 返回）。 + - 知识库操作：`IndexId`（来自 `CreateIndex` 返回）、`JobId`（来自 `SubmitIndexJob` 返回）、`PipelineId`（同 `IndexId`，用于切片操作）。 + - 文件解析：`Parser`（在 `AddFile` 中指定，如 `AUTO_SELECT` 或 `DOCMIND_LLM_VERSION`）。 + - 分页与过滤：`MaxResults`/`NextToken`（列表接口）、`DocumentStatus`（知识库文件状态过滤）、`IndexName`（知识库名称模糊查询）。 -### 类目管理 +## 使用方式 -| API | 说明 | 限流 | 幂等性 | -|-----|------|------|--------| -| AddCategory | 在[业务空间](../concepts/workspace.md)中新建类目，每空间最多 500 个 | 5 次/秒 | 否 | -| ListCategory | 查询类目列表，支持分页 | 5 次/秒 | 是 | -| DeleteCategory | 永久删除指定类目 | 5 次/秒 | 是 | +1. **环境准备**：确保已创建具备最小权限的 RAM 用户，并授予 `AliyunBailianDataFullAccess`（读写）或 `AliyunBailianDataReadOnlyAccess`（只读）策略。具体授权模型详见 [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 +2. **服务接入**：根据地域选择对应接入点，例如华北2（北京）的公网地址为 `bailian.cn-beijing.aliyuncs.com`，VPC 地址为 `bailian-vpc.cn-beijing.aliyuncs.com`。完整列表见 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)。 +3. **典型流程示例（构建知识库）**： + - 调用 `AddCategory` 创建类目； + - 调用 `ApplyFileUploadLease` 获取租约； + - 将文件上传至租约地址； + - 调用 `AddFile` 导入文件至该类目； + - 调用 `CreateIndex` 初始化知识库； + - 调用 `SubmitIndexJob` 启动构建； + - 调用 `GetIndexJobStatus` 轮询任务状态直至完成； + - 调用 `Retrieve` 进行检索。 -> **注意**：当前不支持通过 API 查询或新增数据表，数据表操作请通过控制台完成。 +## 限制和注意事项 -### 文件管理 - -文件上传采用两步流程：先调用 ApplyFileUploadLease 获取上传租约，使用返回的 URL 上传文件后，再调用 AddFile 将文件导入百炼。也可通过 AddFilesFromAuthorizedOss 直接从已授权的 OSS Bucket 导入。详见[ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| ApplyFileUploadLease | 申请上传租约（知识库文件或会话交互文件） | 10 次/秒 | -| AddFile | 将临时存储文件导入数据连接 | 10 次/秒 | -| AddFilesFromAuthorizedOss | 从已授权 OSS Bucket 批量导入文件 | 5 次/秒 | -| DescribeFile | 查询文件基本信息（名称、类型、状态等） | 10 次/秒 | -| ListFile | 分页查询指定类目下的文件列表 | 5 次/秒 | -| UpdateFileTag | 更新单个文件的标签 | 5 次/秒 | -| BatchUpdateFileTag | 批量更新文件标签 | 5 次/秒 | -| DeleteFile | 删除单个文件 | 5 次/秒 | -| DeleteFiles | 批量删除文件 | 5 次/秒 | - -AddFile 接口的 `Parser` 参数支持以下解析器类型： -- `DOCMIND`（智能文档解析） -- `DOCMIND_DIGITAL`（电子文档解析） -- `DOCMIND_LLM_VERSION`（大模型文档解析） -- `DASH_QWEN_VL_PARSER`（Qwen VL 解析） -- `DOCMIND_LLM_VERSION_MEDIA`（音视频解析） -- `AUTO_SELECT`（自动选择解析器） - -### 解析设置 - -| API | 说明 | -|-----|------| -| GetParseSettings | 获取类目的解析设置 | -| GetAvailableParserTypes | 获取指定文件支持的解析器类型列表 | -| ChangeParseSetting | 修改类目的解析设置 | - -### 表格与连接器 - -| API | 说明 | -|-----|------| -| AddTable | 添加表格 | -| UpdateTableFromAuthorizedOss | 从已授权 OSS Bucket 更新表格 | -| AddConnector | 新增连接器 | -| GetConnector | 获取连接器信息（当前仅支持文件连接器） | -| UpdateConnector | 编辑连接器名称和描述 | - -连接器的 `StorageType` 支持 `OSS_CUSTOM`（自有 OSS 存储）和 `OSS_PLATFORM`（平台 OSS 存储）。 - -## Prompt 工程 - -Prompt 模板 API 支持对 Prompt 模板的完整 CRUD 操作。模板内容支持变量占位符（如 `${theme}`），系统会自动提取变量列表。详见[CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md)。 - -| API | 方法 | 说明 | -|-----|------|------| -| CreatePromptTemplate | POST | 创建模板（暂不支持文生图模板） | -| GetPromptTemplate | GET | 按模板 ID 获取详情 | -| UpdatePromptTemplate | PATCH | 增量更新模板名称或内容 | -| DeletePromptTemplate | DELETE | 按模板 ID 删除 | -| ListPromptTemplates | GET | 分页查询模板列表，支持按名称和类型（System/Custom）过滤 | - -## 知识库 - -知识库 API 是百炼 RAG 能力的核心，覆盖知识库的创建、数据导入、检索、文件与切片管理全流程。 - -### 知识库生命周期 - -创建知识库的典型流程为：CreateIndex -> SubmitIndexJob -> 轮询 GetIndexJobStatus 直到完成。详见[CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| CreateIndex | 创建知识库（非结构化或结构化），不具幂等性 | 10 次/秒 | -| SubmitIndexJob | 提交知识库创建任务，必须在 CreateIndex 后调用 | 10 次/秒 | -| SubmitIndexAddDocumentsJob | 向已有知识库追加文件（不支持数据查询/图片问答类） | 10 次/秒 | -| GetIndexJobStatus | 查询任务状态，调用间隔建议 5 秒以上 | - | -| UpdateIndex | 更新知识库配置（名称、描述、检索参数等） | - | -| ListIndices | 分页查询[业务空间](../concepts/workspace.md)下的知识库列表 | 10 次/秒 | -| DeleteIndex | 永久删除知识库（不可逆，不删除源文件） | 10 次/秒 | -| GetIndexMonitor | 获取知识库监控数据 | - | - -> **注意**：CreateIndex 仅初始化知识库，必须后续调用 SubmitIndexJob 才能完成创建，否则将得到空知识库。CreateIndex 不具幂等性，重复调用会创建多个同名知识库。 - -UpdateIndex 支持调整检索参数： -- `DenseSimilarityTopK`：向量检索 Top K，范围 [0-100]，默认 100 -- `SparseSimilarityTopK`：关键词检索 Top K，范围 [0-100]，默认 100 -- 两者之和不超过 200 -- `RerankMinScore`：排序最低分数，范围 [0-1] -- `PipelineCommercialType`：知识库规格（standard / enterprise） - -### 知识库检索 - -Retrieve 接口用于在指定知识库中检索信息，支持通过百炼 SDK（AccessKey 鉴权）或 Spring AI Alibaba（API-Key 鉴权）调用。接口具有幂等性，但因包含复杂检索逻辑，响应时间可能较长，建议合理设置超时和重试策略。详见[Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md)。 - -### 文件与切片管理 - -| API | 说明 | -|-----|------| -| ListIndexFileDetails | 查询知识库中文件的详细信息，支持按状态和名称过滤 | -| ListIndexDocuments | 查询知识库中文件的概要信息 | -| DeleteIndexDocument | 从知识库中删除指定文件 | -| ListChunks | 查询文件的切片列表（文档搜索类查指定文件，数据查询类查全部） | -| UpdateChunk | 修改切片内容和标题（仅支持文档搜索类知识库） | -| DeleteChunk | 删除指定切片 | - -文件导入状态包括：`RUNNING`（构建中）、`FINISH`（成功）、`INSERT_ERROR`（导入失败）、`PARSE_FAILED`（解析失败）、`DOC_PARSING`（解析中）、`DELETED`（已删除）。 - -## 长期记忆 - -长期记忆 API 用于管理智能体的记忆能力，包括记忆体（Memory）和记忆片段（MemoryNode）两个层级。 - -| API | 说明 | -|-----|------| -| CreateMemory | 创建长期记忆体 | -| GetMemory | 获取记忆体详情 | -| UpdateMemory | 更新记忆体 | -| DeleteMemory | 删除记忆体 | -| ListMemories | 查询记忆体列表 | -| CreateMemoryNode | 创建记忆片段 | -| GetMemoryNode | 获取记忆片段详情 | -| UpdateMemoryNode | 更新记忆片段 | -| DeleteMemoryNode | 删除记忆片段 | -| ListMemoryNodes | 查询记忆片段列表 | - -## 其他 - -| API | 说明 | -|-----|------| -| ApplyTempStorageLease | 申请临时文件上传许可 | -| GetAlipayTransferStatus | 查询支付宝打赏状态 | -| GetAlipayUrl | 获取支付宝打赏 URL | - -## 通用注意事项 - -- 所有接口均需 `WorkspaceId` 路径参数，获取方式参见[业务空间](../concepts/workspace.md)文档 -- 建议使用官方 SDK 调用而非自签名，自签名对接复杂度高（约需 5 个工作日） -- 分页查询使用 `NextToken` / `MaxResults` 模式（部分接口使用 `PageNumber` / `PageSize`） -- 各接口限流频率为 5-15 次/秒不等，触发限流后需等待后重试 -- 版本变更历史可查看[版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md)，近期变更包括 CreateIndex 入参调整、UpdateIndex 新增、GetIndexMonitor 新增等 +- **限流规则**：各接口有独立 QPS 限制，例如 `AddCategory`/`ListCategory`/`DeleteCategory` 为 5 次/秒，`ApplyFileUploadLease`/`AddFile`/`DescribeFile` 为 10 次/秒，`ListIndexDocuments` 为 15 次/秒。超出将返回 429 错误，需实现退避重试逻辑。 +- **幂等性**：`ListCategory`, `DescribeFile`, `ListFile`, `GetIndexJobStatus`, `Retrieve`, `DeleteIndex`, `UpdateChunk`, `DeleteChunk` 等接口具有幂等性；而 `AddCategory`, `AddFile`, `CreateIndex`, `SubmitIndexJob` 等不具备，重复调用可能产生冗余资源。 +- **功能边界**： + - 数据表（Table）的创建与删除不支持 API，必须通过控制台操作（见 `AddTable` 和 `DeleteFile` 文档说明）。 + - `DeleteFile` 仅删除应用数据中的文件，不影响已构建的知识库；反之，`DeleteIndexDocument` 仅删除知识库中的索引，不影响原始文件。 + - `UpdateIndex` 的 `PipelineCommercialCu` 参数仅对旗舰版（`enterprise`）知识库生效，标准版（`standard`）传入将被忽略。 +- **版本兼容性**：API 行为可能随版本变更，例如 `DescribeFile` 在 2026-01-15 发生了返回结构变更，`CreateIndex` 在 2026-03-27 和 2026-03-30 均有入参调整。开发者应关注 [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) 并及时适配。 ## 来源文档 - [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) -- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) +- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) -- [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) -- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) +- [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) -- [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) +- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [ListFile - 文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) -- [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) -- [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) +- [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) -- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) -- [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) +- [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) +- [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) +- [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) +- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [ChangeParseSetting - 修改类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) -- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) +- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [AddConnector - 新增连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) -- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) -- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) -- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) -- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) -- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) -- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - [CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) +- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - [Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) -- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [ListIndexDocuments - 查询知识库下的文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) -- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [UpdateIndex - 更新知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) -- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) +- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) +- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) -- [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) +- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) +- [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - [DeleteChunk - 删除切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) +- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetIndexMonitor - 获取知识库监控数据](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) -- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) +- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) +- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) +- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) +- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) +- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [CreateMemory - 创建长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) -- [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) +- [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md index c7a00663..4c45ce38 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md @@ -1,37 +1,53 @@ # file management api -文件管理 API 用于管理上传至百炼平台的文件，覆盖上传、查询、列举和删除等基础操作。它是使用需要文件输入的模型能力（如文档解析、多模态理解、批量任务等）的前置步骤，开发者需先将文件上传到平台并获取文件标识，再在后续调用中引用。详见 [文件管理](../../raw/model-api-reference/file-management-api.md)。 +文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询详情、列举已上传文件及删除文件。该 API 与模型调用解耦，不参与推理过程，仅用于文件资源的元数据与二进制内容管理。所有操作均需通过 `Authorization: Bearer ` 认证，并遵循平台统一的错误响应格式（详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)）。 -## 核心功能 +## 支持的模型/功能 -根据 [文件管理](../../raw/model-api-reference/file-management-api.md) 的说明，该 API 提供以下针对平台文件的操作： +- **功能范围**：当前仅支持通用文件托管，**不绑定任何特定大模型**；上传后的文件可被 `qwen-vl-plus`、`qwen2-audio` 等多模态模型在请求中通过 `file_id` 引用（如 `messages[0].image.file_id`），但文件管理 API 本身不执行模型推理。 +- **操作类型**：`POST /v1/files`（上传）、`GET /v1/files/{file_id}`（查询）、`GET /v1/files`（列举）、`DELETE /v1/files/{file_id}`（删除）。 +- 注意：`qwen2-audio` 模型虽支持音频文件输入，但其文件上传必须经由本 API 完成，不可直传至 `/v1/chat/completions` —— 此限制在 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 中明确说明。 -- **上传（Upload）**：将本地文件上传至百炼平台，上传成功后返回文件标识，供后续模型调用引用。 -- **查询（Retrieve）**：根据文件标识查询单个文件的元信息与状态。 -- **列举（List）**：列出账号下已上传的文件集合，便于管理与清理。 -- **删除（Delete）**：移除不再需要的文件，释放存储资源。 +## 关键参数 -## 使用方式 +| 参数 | 位置 | 类型 | 必填 | 说明 | +|------|------|------|------|------| +| `file` | form-data | binary | 是 | 文件二进制流，支持 `image/*`, `audio/*`, `text/plain`, `application/pdf` 等常见 MIME 类型 | +| `purpose` | form-data | string | 否 | 取值为 `"batch"` 或 `"vision"`（默认 `"batch"`）；`"vision"` 用于图像类多模态模型（如 `qwen-vl-plus`），影响后续 token 计费逻辑 | +| `file_id` | path | string | 是（查询/删除时） | 由平台生成的唯一文件标识符，长度为 24 位十六进制字符串 | -典型的使用流程是"先上传、再引用、后清理"： +> **注意**：文档中曾提及 `purpose=embedding` 选项，但该值已在 v2.3.0 版本后废弃，实际调用将返回 `400 Bad Request`；请以 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 当前版本为准。 -1. 通过上传操作把文件送入平台，拿到文件标识。 -2. 在需要文件输入的模型 API 调用中传入该标识。 -3. 使用完毕后按需删除文件。 +## 使用方式 -关于各操作的具体请求参数、返回字段和调用示例，请以 [文件管理](../../raw/model-api-reference/file-management-api.md) 的原始文档为准。 +1. **上传文件**： + ```bash + curl -X POST "https://dashscope.aliyuncs.com/api/v1/files" \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -F "file=@/path/to/image.jpg" \ + -F "purpose=vision" + ``` + 成功响应包含 `id`, `filename`, `size`, `purpose`, `status="uploaded"`。 + +2. **在模型请求中引用**： + 将返回的 `file_id` 填入消息内容，例如： + ```json + { + "model": "qwen-vl-plus", + "messages": [{"role": "user", "content": [{"type": "image_url", "image_url": {"file_id": "xxx"}}]}] + } + ``` ## 限制和注意事项 -- 上传前建议先确认目标模型或能力所支持的文件类型与大小限制。 -- 文件标识是后续调用的关键，请妥善保存；文件被删除后其标识将失效。 -- 列举与删除操作影响的是账号级别的文件资源，批量清理时请谨慎确认。 - -> **注意**：本页仅概述文件管理 API 的能力范围，具体的接口路径、鉴权方式、参数细节与配额限制可能随平台更新而变化，实际集成时请以原始文档最新版本为准。 +- 单文件大小上限为 **100 MB**（PDF/音频）或 **20 MB**（图像），超出将返回 `413 Payload Too Large`； +- 每个 API Key 默认最多存储 **10,000 个文件**，超限时需先删除旧文件； +- 已删除文件不可恢复，且 `file_id` 不会复用； +- 文件上传后立即可用，但元数据同步可能存在秒级延迟，建议上传后等待 `status="uploaded"` 再引用； +- 所有文件默认保留 **90 天**，无访问行为的文件可能被系统自动清理（具体策略参见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)）。 ## 来源文档 - [文件管理](../../raw/model-api-reference/file-management-api.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md index 451b4707..b963ca1b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md @@ -1,130 +1,52 @@ # frameworks -阿里云百炼支持通过主流开源框架集成其大模型应用与云端[知识库](../concepts/knowledge-base.md)能力。当前官方文档覆盖两类框架：基于 Python 的 LlamaIndex，用于构建 RAG 应用；以及基于 Java 的 Spring AI Alibaba，用于集成百炼智能体/[工作流](../concepts/workflow.md)应用并检索百炼[知识库](../concepts/knowledge-base.md)。两者均以 [API Key](../concepts/api-key.md) 鉴权，复用百炼的数据管理与模型推理能力。 +阿里云百炼平台提供多种主流 AI 开发框架的集成支持，帮助开发者快速构建 RAG 应用、智能体/工作流应用及知识库检索服务。当前主要通过 LlamaIndex 和 Spring AI Alibaba 两大框架实现与百炼能力的对接，覆盖云端知识库管理、大模型调用、文档切分与重排、流式响应等关键能力。所有集成均依赖百炼统一的 API Key 认证机制，并需配合控制台创建的应用或知识库资源使用。 -## 支持的框架与功能 +## 支持的模型/功能 -| 框架 | 语言 | 主要能力 | -| --- | --- | --- | -| LlamaIndex | Python 3.9+ | 读取本地文件上传到百炼应用数据、构建云端[知识库](../concepts/knowledge-base.md)、构建检索引擎与 RAG 应用 | -| Spring AI Alibaba | Java（Spring Boot 3.x，JDK 17+） | 调用百炼[智能体应用](../concepts/agent-application.md)/[工作流](../concepts/workflow.md)应用（流式与非流式）、检索百炼知识库 | +- **RAG 场景**：通过 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 支持基于云端知识库的端到端 RAG 构建，包括文档上传（`.txt`/`.docx`/`.pdf`）、默认智能切分、官方向量嵌入（不可自定义）、检索引擎构建与问答生成。 +- **智能体与工作流应用集成**：通过 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) 支持调用已发布的**智能体应用**和**工作流应用**，支持非流式与流式响应，并可获取 `docReferences` 和 `thoughts` 等结构化输出。 +- **知识库直接检索**：通过 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) 提供 `DashScopeDocumentRetriever`，支持按知识库名称（`INDEX_NAME`）检索上下文片段，并自动注入提示词模板交由大模型（默认 `qwen-max`）生成回答。 -- LlamaIndex 路线将知识库部署在云端，使用默认的智能文档切分与官方向量模型，**不支持**自定义文档切分方式或自定义嵌入模型。如需本地知识库或灵活切分，应改用本地知识库方案，详见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 -- Spring AI Alibaba 的应用集成**仅支持**[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用两类，需提前在百炼控制台创建并获取应用 ID。 +> **注意**：LlamaIndex 方案明确声明“不支持自定义文档切分方式或自定义嵌入模型”，而 Spring AI Alibaba 的知识库检索方案未提及切分/嵌入控制能力，二者在知识库底层处理粒度上存在差异，实际选型时应以业务是否需要定制化预处理为准。 -## 前提条件 +## 关键参数 -1. 开通阿里云百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -2. 将 [API Key](../concepts/api-key.md) 配置到环境变量，避免硬编码泄露： - - LlamaIndex：按百炼通用约定配置。 - - Spring AI Alibaba 应用集成：推荐变量名 `DASHSCOPE_API_KEY`，应用 ID 用 `APP_ID`，子[业务空间](../concepts/workspace.md)用 `WORKSPACE_ID`。 - - Spring AI Alibaba 知识库检索：推荐变量名 `AI_DASHSCOPE_API_KEY`，子[业务空间](../concepts/workspace.md)用 `AI_DASHSCOPE_WORKSPACE_ID`。 -3. 若应用或知识库创建在子[业务空间](../concepts/workspace.md)，需额外获取[业务空间](../concepts/workspace.md) ID 并配置对应环境变量。 +| 参数名 | 来源框架 | 说明 | 示例值 | +|--------|----------|------|--------| +| `model_name` | LlamaIndex | 设置生成回答所用的大模型 | `"qwen-max"`（见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)） | +| `APP_ID` | Spring AI Alibaba（应用集成） | 智能体或工作流应用的唯一 ID | `app-xxxxxx` | +| `DASHSCOPE_API_KEY` | Spring AI Alibaba（应用集成） | 百炼 API Key 环境变量名（推荐） | — | +| `AI_DASHSCOPE_API_KEY` | Spring AI Alibaba（知识库检索） | 百炼 API Key 环境变量名（知识库场景专用） | — | +| `INDEX_NAME` | Spring AI Alibaba（知识库检索） | 待检索知识库的名称（需提前在控制台创建） | `"测试知识库"` | +| `WORKSPACE_ID` / `AI_DASHSCOPE_WORKSPACE_ID` | Spring AI Alibaba | 子业务空间 ID（仅当应用或知识库部署在子空间时必需） | `ws-xxxxxx` | +| `similarity_top_k`, `similarity_cutoff`, `top_n` | LlamaIndex | 检索结果数量、相似度阈值、重排后返回数 | `5`, `0.4`, `1` | -> **注意**：Spring AI Alibaba 两篇文档对 [API Key](../concepts/api-key.md) 环境变量名约定不一致（应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`）。两者均为约定俗成，可按工程实际统一，关键是 `application.yml` 中 `${...}` 占位符与实际变量名一致。 +## 使用方式 -## LlamaIndex：构建 RAG 应用 +- **LlamaIndex 集成**： + 1. 安装 `llama-index` 及 `llama-index-readers-dashscope` 等依赖； + 2. 使用 `DashScopeCloudIndex.from_documents()` 构建云端知识库； + 3. 调用 `index.as_query_engine()` 并配置 `node_postprocessors`（如 `SimilarityPostprocessor` + `DashScopeRerank`）启用过滤与重排； + 4. 通过 `query_engine.query()` 发起 RAG 查询。 -### 方案概览 +- **Spring AI Alibaba（应用集成）**： + 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖； + 2. 在 `application.yml` 中配置 `spring.ai.dashscope.agent.app-id` 和 `api-key`； + 3. 注入 `DashScopeAgent`，调用 `.call()`（非流式）或 `.stream()`（流式）方法传入 `Prompt`。 -1. 读取本地文件（`.txt`、`.docx`、`.pdf` 等非结构化数据）并上传到云端，构建云端知识库。 -2. 基于云端知识库构建检索引擎，接收用户提问、检索相关文本片段，与提问合并后送入大模型生成回答；检索不到相关内容时返回报错信息。 +- **Spring AI Alibaba（知识库检索）**： + 1. 同样引入 `spring-ai-alibaba-starter-dashscope`； + 2. 配置 `spring.ai.dashscope.api-key`（注意变量名区别）； + 3. 构建 `DashScopeDocumentRetriever` 并绑定至 `ChatClient` 的 `DocumentRetrievalAdvisor`； + 4. 通过 `chatClient.prompt().user(...).stream().chatResponse()` 触发带上下文的生成。 -完整流程与示例代码参见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +## 限制和注意事项 -### 关键参数 - -构建检索引擎时需手动调整以下参数： - -- `Settings.llm = DashScope(model_name="qwen-max")`：生成回答时调用的大模型，可传 `qwen-max` 等模型名称。 -- `similarity_top_k`：相似度最高的检索结果数（示例为 5）。 -- `similarity_cutoff`：过滤检索结果的最低相似度阈值（示例为 0.4）。 -- `top_n`：重排后返回语义相关度最高的结果数（示例为 1）。 - -检索引擎默认结果可能不满足需求，可通过 `node_postprocessors` 做后处理： -- `SimilarityPostprocessor(similarity_cutoff=...)`：过滤低于阈值的检索结果。 -- `DashScopeRerank(top_n=..., model="gte-rerank")`：对检索结果重排，返回最相关结果。 -- `response_mode="tree_summarize"`：响应聚合方式。 - -### 使用方式 - -1. 下载示例包 `llamaindex_cloud_rag.zip` 并解压，`docs/` 内为示例业务文件（可替换），`create_cloud_index.py` 用于建库，`rag.py` 用于运行 RAG 应用。 -2. `pip install -r requirements.txt` 安装依赖。 -3. `python create_cloud_index.py`：将 `docs/` 文件上传到百炼应用数据并创建云端知识库（示例知识库名 `my_first_index`）。 -4. `python rag.py`：读取已创建的云端知识库，启动本地交互式 RAG 应用；输入问题回车得到回答，输入 `q` 退出。 - -> **注意**：本方案使用百炼云端智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活控制请改用本地知识库方案。 - -## Spring AI Alibaba：集成大模型应用 - -### 环境要求 - -- Spring Boot 3.x -- JDK 17 或更高版本 - -### 依赖与配置 - -在 `pom.xml` 中添加 `spring-ai-alibaba-starter-dashscope`（示例版本 `1.0.0.2`）及 `spring-boot-starter-web` 等依赖。`application.yml` 配置示例： - -```yaml -spring: - ai: - dashscope: - agent: - app-id: ${APP_ID} - api-key: ${DASHSCOPE_API_KEY} - # workspace-id: ${WORKSPACE_ID} # 子业务空间时启用 -``` - -### 调用方式 - -通过 `DashScopeAgent` 调用百炼大模型应用，支持两种模式： - -- **非流式调用**：`agent.call(new Prompt(message, DashScopeAgentOptions.builder().withAppId(appId).build()))`，返回 `ChatResponse`，可从 `output` 元数据中取出 `docReferences`（文档引用）与 `thoughts`（思考过程）。 -- **流式调用**：`agent.stream(...)` 返回 `Flux`，构造 `DashScopeAgent` 时可设置 `sessionId`、`incrementalOutput`（增量输出）、`hasThoughts`（返回思考）等选项，接口 `produces="text/event-stream"`。 - -工程入口为标准 `@SpringBootApplication`，启动后可用 Postman 等工具访问 `/ai/bailian/agent/call` 或 `/ai/bailian/agent/stream` 测试。完整代码与示例工程参见[使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 - -## Spring AI Alibaba：检索百炼知识库 - -### 环境要求 - -- JDK 17 或更高版本 -- Spring Boot 3 GA 或更高版本 - -### 使用方式 - -1. 从 Spring AI Alibaba examples 仓库下载 `bailian-rag-knowledge` 示例（需整个 examples 目录以保证结构与依赖完整）。 -2. 配置 `AI_DASHSCOPE_API_KEY`（及可选的 `AI_DASHSCOPE_WORKSPACE_ID`）。 -3. 通过 `DashScopeDocumentRetriever` 检索百炼知识库： - -```java -DocumentRetriever retriever = new DashScopeDocumentRetriever(dashscopeApi, - DashScopeDocumentRetrieverOptions.builder().withIndexName(INDEX_NAME).build()); - -this.chatClient = builder - .defaultAdvisors(new DocumentRetrievalAdvisor(retriever, retrievalSystemTemplate)) - .build(); -``` - -- `INDEX_NAME` 为待检索知识库名称，**需提前在百炼控制台创建**。 -- 检索到的文本切片与原始问题一并提交给大模型生成回答，默认模型 `qwen-max`，可通过 `DashScopeChatOptions.builder().withModel("qwen-plus").build()` 切换。 -- 建议使用系统提示词模板约束模型"仅依据上下文回答，答案不在上下文中则告知无法回答"。 - -详情参见[通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md)。 - -## [计费](../concepts/billing.md)与错误处理 - -- 百炼应用本身不收费，但通过应用调用模型会产生模型推理（调用）费用。 -- 通用错误码参见百炼[错误信息](https://help.aliyun.com/zh/model-studio/error-code)文档。 - -## 限制与注意事项 - -- LlamaIndex 云端方案不支持自定义文档切分与嵌入模型；本地需可访问公网，文件上传与生成回答均需等待。 -- Spring AI Alibaba 应用集成仅支持[智能体应用](../concepts/agent-application.md)与工作流应用，其他应用类型不在支持范围。 -- 知识库检索需提前创建好知识库并获取其名称；检索默认[业务空间](../concepts/workspace.md)知识库无需配置 `workspace-id`。 -- 子[业务空间](../concepts/workspace.md)场景必须配置对应的[业务空间](../concepts/workspace.md) ID 环境变量，否则会鉴权或定位失败。 -- [API Key](../concepts/api-key.md) 一律通过环境变量注入，切勿硬编码到源码或配置文件中。 +- **LlamaIndex 方案限制**：仅支持 `.txt`/`.docx`/`.pdf` 文件上传；知识库必须部署在云端；不支持自定义切分逻辑与嵌入模型；文件上传依赖公网访问能力。详见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **Spring AI Alibaba 应用集成限制**：**仅支持智能体应用和工作流应用**，不支持直接调用基础大模型 API 或知识库 API；`DashScopeAgent` 不提供对检索过程的细粒度控制（如 top-k、重排器选择），其检索行为由应用内部逻辑决定。 +- **环境变量命名不一致**：Spring AI Alibaba 文档中，应用集成要求 `DASHSCOPE_API_KEY`，而知识库检索要求 `AI_DASHSCOPE_API_KEY` —— 二者不可混用，否则初始化失败。> **注意**：该差异已在两篇 Spring AI Alibaba 文档中明确体现，属设计约定，非过时信息，但需开发者严格区分场景配置。 +- **计费说明**：所有框架调用最终均产生模型推理费用（按 token 计费），百炼应用本身不单独收费。具体计费项参见官方文档。 ## 来源文档 @@ -133,21 +55,3 @@ this.chatClient = builder - [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md index 5c8b9da6..ffe62825 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md @@ -1,73 +1,80 @@ # image generation -阿里云百炼平台提供覆盖文生图、图像编辑、图像翻译及一系列创意工具的图像生成 API，涵盖千问（Qwen-Image）、万相（Wan/Wanx）、Z-Image、可灵（Kling）、Vidu 等模型家族。这些接口统一通过 DashScope 网关调用，既支持 HTTP，也支持 DashScope Python/Java SDK，可满足文生图、图生图、局部重绘、扩图、背景生成、虚拟模特、AI 试衣、创意海报等多样化场景。 +百炼平台提供丰富的图像生成与编辑能力，涵盖文生图、图生图、局部重绘、风格迁移、背景生成、AI试衣等20余种专业场景。所有模型均通过统一的HTTP API或DashScope SDK调用，支持同步与异步两种模式，适用于从快速原型验证到高并发生产环境的各类需求。 -## 支持的模型与功能 +## 支持的模型/功能 -按能力大致可分为四类： +平台当前提供三大类图像模型能力： -- **通用文生图**：千问 `qwen-image-*` 系列擅长复杂文本渲染与图文混排；万相 `wan2.6-t2i` / `wan2.5-t2i-preview` / `wan2.2-t2i-*` / `wanx2.1-t2i-*` 系列支持写实与多种艺术风格；`z-image-turbo` 为轻量快速生图模型，支持中英文字渲染。详见 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) 与 [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md)。 -- **图像编辑 / 生成一体**：`qwen-image-edit-*`、`wan2.7-image*`、`wan2.6-image`、`wan2.5-i2i-preview`、`wanx2.1-imageedit` 支持单图编辑、多图融合、图文混排、局部重绘、去水印、扩图、超分、上色、线稿生图等。参见 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) 与 [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md)。 -- **第三方模型**：可灵 `kling/kling-v3-*`（文生图、参考图生图、分镜组图）、Vidu `vidu/*`（参考生图、文生图、图片编辑，最多 14 张参考图）。 -- **创意与电商工具**：人像风格重绘、虚拟模特、鞋靴模特、图像画面扩展（扩图）、创意海报生成、人物实例分割、AI 试衣 OutfitAnyone、图像背景生成、图像擦除补全、人物写真 FaceChain、创意文字 WordArt 锦书、千问图像翻译（Qwen-MT-Image）。 +- **通用文生图与编辑**：包括千问系列（`qwen-image-*`、`qwen-image-edit-*`）、万相系列（`wan2.7-image-*`、`wan2.6-t2i`、`wanx2.1-t2i-*`）、Z-Image（`z-image-turbo`）和可灵（`kling/kling-v3-*`）。其中 `qwen-image-2.0-pro` 和 `wan2.7-image-pro` 为当前推荐主力模型，分别在文字渲染精度与4K高清输出上具备优势 [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md)。 + +- **垂直场景专用模型**：覆盖电商与设计工作流，如虚拟模特（`virtualmodel-v2`）、鞋靴模特（`shoemodel-v1`）、创意海报生成（`wanx-poster-generation-v1`）、图像背景生成（`wanx-background-generation-v2`）、人物实例分割（`image-instance-segmentation`）及图像擦除补全（`image-erase-completion`）。这些模型多为地域限定（仅华北2北京），且部分处于免费体验阶段 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -## 调用方式 +- **创意工具与辅助能力**：包括涂鸦作画（`wanx-sketch-to-image-lite`）、人像风格重绘（`wanx-style-repaint-v1`）、AI试衣（`aitryon-plus`）、FaceChain人物写真、WordArt锦书文字艺术等。其中 FaceChain 需先完成人物形象训练再生成写真，而 WordArt 锦书则专注于汉字纹理与变形 [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md)。 -绝大多数图像模型处理耗时较长（通常 1-2 分钟），因此接口以**异步**为主，流程分两步： +> **注意**：文档中存在模型命名与能力描述不一致的情况。例如，`wan2.6-t2i` 在 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) 中明确标注为“支持HTTP同步调用”，但同系列 `wan2.5-t2i-preview` 及更早版本则“不支持HTTP同步调用”；而 `wan2.7-image-pro` 在 [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) 中声明“仅文生图场景支持4K分辨率”，但未说明组图生成的最高分辨率限制。开发者应以实际调用返回的 `400 Bad Request` 错误码及官方控制台模型详情页为准。 -1. **创建任务获取任务 ID**：提交请求，必须设置请求头 `X-DashScope-Async: enable`（缺失会报错 `current user api does not support synchronous calls`），返回 `task_id`。 -2. **根据任务 ID 轮询结果**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`，任务成功后返回图像 URL，**有效期 24 小时**，`task_id` 有效期同样为 24 小时。**请勿重复创建任务**，轮询即可。 - -新一代模型（如 `wan2.6-image`、`wan2.7-image`、`z-image-turbo`）新增 **HTTP 同步调用**，一次请求即可返回结果，接口路径为 `.../aigc/multimodal-generation/generation`，请求体采用 `messages` 结构（`content` 内含 `text` / `image`）。 +## 关键参数 -> **注意**：不同模型的接口路径并不统一。文生图 V1/涂鸦/局部重绘等旧模型走 `.../text2image/image-synthesis` 或 `.../image2image/image-synthesis`；可灵、Vidu 走 `.../image-generation/generation`；虚拟模特、鞋靴模特走 `.../virtualmodel/generation`；扩图走 `.../image2image/out-painting`。调用前请以对应模型文档为准。 +所有图像API均通过 `parameters` 对象传递核心控制参数，常见字段如下： -## 关键参数 +- `size` / `resolution` / `aspect_ratio`：控制输出尺寸。格式多样，如 `"1024*1024"`（万相V1/V2）、`"2K"`（万相2.7）、`"1k"`（可灵、Vidu）、`"1:1"`（可灵、Vidu）。总像素范围普遍为 `512×512` 至 `2048×2048`，`wan2.7-image-pro` 文生图支持 `4K`（`3840×2160`）[万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md)。 +- `n`：生成图片张数，取值范围因模型而异：`1–6`（千问系列）、`1–9`（可灵）、`1–4`（创意海报、鞋靴模特）。 +- `watermark`：布尔值，控制是否添加平台水印，默认 `true`，部分模型（如 `wan2.7-image-pro`）支持设为 `false`。 +- `prompt_extend`：启用智能提示词扩展，返回优化后的提示词及推理过程，会增加响应时间（Z-Image、万相2.6等支持）。 +- `style_index` / `style_ref_url`：用于人像风格重绘，前者指定预置风格索引，后者传入自定义风格参考图。 +- `X-DashScope-Async`：**必选请求头**，异步调用必须设为 `"enable"`；缺失将报错 `"current user api does not support synchronous calls"`。 -- `model`（必选）：模型名，如 `wanx2.1-t2i-turbo`、`qwen-mt-image`、`wan2.6-image` 等。 -- `input`：输入内容。文生图用 `prompt`，可附 `negative_prompt`；新协议模型用 `messages`；图像编辑/参考类用 `images`、`image_url`、`base_image_url`、`mask_image_url` 等。 -- `parameters`：`size`（如 `1024*1024`、`1K`/`2K`/`4K`）、`n`（生成张数）、`aspect_ratio`、`resolution`、`watermark`、`prompt_extend`（智能扩写/思考）、`thinking_mode` 等，按模型不同而异。 +## 使用方式 -分辨率与张数因模型而异：千问 Pro 系列总像素 512\*512～2048\*2048、可生成 1-6 张；`qwen-image-max` 固定 1 张；万相 2.6 总像素在 1280\*1280～1440\*1440；可灵支持 1k/2k/4k、1-9 张。 +### 调用模式 +- **同步调用**：适用于耗时较短（通常 < 15s）的模型，如 `z-image-turbo`、`wan2.6-t2i`、`qwen-image-*`（Pro/Plus系列默认同步）。一次HTTP POST即可返回结果，无需轮询。 +- **异步调用**：适用于耗时较长（1–2分钟）的模型，如万相V1、局部重绘、虚拟模特、背景生成等。流程分两步： + 1. `POST /api/v1/services/.../generation` 创建任务，获取 `task_id`； + 2. `GET /api/v1/tasks/{task_id}` 轮询状态，直至 `task_status == "SUCCEEDED"` 后获取 `output.results[].url`。 -## 计费、限流与注意事项 +### 地域与域名 +- 华北2（北京）、新加坡、美国（弗吉尼亚）地域拥有独立API Key与请求地址，**不可混用**。 +- 强烈建议迁移至业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），以获得更高性能与稳定性；旧域名（`dashscope.aliyuncs.com`）仍兼容但非最优 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)。 -- **计费**：仅对模型**成功生成的输出图片**计费，输入图片及处理失败不计费、不占免费额度。免费额度开通后自动发放，有效期 90 天，主账号与 RAM 子账号共享。计费与限流详情见 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **限流**：主账号与 RAM 子账号共用任务下发 QPS 与同时处理中任务数量限制。 -- **地域隔离**：华北2（北京）、新加坡、美国（弗吉尼亚）等地域拥有**独立的 API Key 与请求地址，不可混用**，跨地域调用会导致鉴权失败或报错。百炼推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），现有域名仍可用。 -- **输入图片要求**：图片 URL 必须公网可访问，否则报 `BadRequest.InputDownloadFailed`；URL 不能包含中文字符；常见格式限制为 JPG/PNG/JPEG/BMP/WEBP，分辨率多要求 [512, 4096] 像素、大小不超过 10MB。 +### 认证与环境 +- 必须配置 `Authorization: Bearer $DASHSCOPE_API_KEY` 请求头。 +- 推荐通过环境变量管理API Key，并使用DashScope SDK（Python/Java）简化调用逻辑。 -> **注意**：部分创意工具模型（如 `wanx-x-painting` 图像局部重绘、`wanx-virtualmodel`/`virtualmodel-v2` 虚拟模特、`shoemodel-v1` 鞋靴模特、`wanx-poster-generation-v1` 创意海报、`image-erase-completion` 图像擦除补全、`image-instance-segmentation` 人物实例分割）当前**仅提供免费体验，免费额度用完后不可调用且不支持付费**。官方建议迁移到 [千问-图像编辑](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)或万相 2.1 等替代方案。 +## 限制和注意事项 -> **注意**：文生图 V1 版（`wanx-v1`）已被 V2 版全面替代，且仅适用于华北2（北京）地域；千问图像翻译（`qwen-mt-image`）同样仅在华北2（北京）可用，且不支持非中/英语种之间的直接互译（如日译韩）。新项目应优先选择推荐的新版模型。 +- **免费额度与计费**：所有模型均提供500张免费额度（有效期90天），主账号与RAM子账号共享。超出后按模型单价计费（如 `wanx-v1`: 0.16元/张，`wanx-style-repaint-v1`: 0.12元/张），仅对**成功生成的输出图片**收费 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 +- **图片URL要求**：输入图片URL必须公网可访问、无中文路径、支持HTTP/HTTPS。若下载失败，错误码为 `BadRequest.InputDownloadFailed`，需检查链接有效性或上传至OSS等云存储 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 +- **限流策略**：主账号与RAM子账号共用QPS/RPS限制（常见为2 QPS），同时处理中任务数上限为1–5个，超限将返回 `429 Too Many Requests`。 +- **模型可用性**：部分模型（如 `wanx-x-painting`、`wanx-virtualmodel`、`shoemodel-v1`）当前仅限免费体验，额度用尽后不可调用且不支持付费，文档已明确提示替代方案 [图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md)。 ## 来源文档 - [常见问题](../../raw/model-api-reference/image-generation/image-faq.md) -- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) +- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) +- [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) -- [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) +- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) -- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-涂鸦作画API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) +- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) - [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) +- [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) -- [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) -- [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) -- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) +- [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) - [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) +- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [创意文字WordArt锦书](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) -- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md index 658992ce..4b190623 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md @@ -1,57 +1,41 @@ # knowledge -百炼平台「知识检索与问答」相关的 HTTP REST API 概览，提供跨知识库语义检索与基于知识库的智能问答两个接口。这两个接口属于 DashScope 应用网关体系，通过 [API Key](../concepts/api-key.md) Bearer 鉴权调用，与 `CreateIndex`、`Retrieve` 等 OpenAPI RPC 接口不同。详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +knowledge 是百炼平台提供的知识检索与问答能力，通过统一的应用网关 API 提供语义检索和基于知识库的流式问答服务。该能力不依赖底层 OpenAPI（如 `CreateIndex` 等 RPC 接口），而是面向应用层提供标准化 HTTP REST 接口，适用于 RAG 场景下的快速集成。详细设计与行为请参考 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 -## 接口列表 +## 支持的模型/功能 -| 接口 | 描述 | 路径 | -| --- | --- | --- | -| 知识检索 | 跨多个知识库执行联合语义检索，返回按相关性排序的切片 | `POST /api/v1/indices/knowledge/search` | -| 知识问答 | 基于知识库的智能问答，通过 SSE [流式输出](../concepts/streaming-output.md)，依次返回规划、工具调用、生成三个阶段 | `POST /api/v2/apps/knowledge/chat` | +- **知识检索**：跨多个知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），适用于召回阶段。 +- **知识问答**：端到端智能问答，支持 SSE 流式响应，输出包含规划（planning）、工具调用（tool calling）和生成（generation）三个逻辑阶段，需配合已部署的知识库与应用配置使用。 + > **注意**：知识问答接口 `/api/v2/apps/knowledge/chat` 的三阶段输出行为与部分旧版文档描述的“单次生成”存在差异，以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 中的 SSE 分阶段说明为准。 -## 鉴权与 Base URL +## 关键参数 -所有请求须在请求头携带 `Authorization: Bearer `，并使用[业务空间](../concepts/workspace.md) ID 拼接的 Base URL： +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `workspaceId` | string | 是 | 业务空间 ID，用于构造 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`），非用户 UID 或 Project ID。获取路径见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 | +| `Authorization` | header | 是 | `Bearer `，API Key 需在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 申请。 | +| `top_k`（检索） | integer | 否 | 检索返回切片数量，默认 5，最大 100。 | +| `stream`（问答） | boolean | 否 | 是否启用 SSE 流式响应，默认 `true`；设为 `false` 将返回完整 JSON 响应（非流式）。 | -``` -https://{workspaceId}.cn-beijing.maas.aliyuncs.com -``` +## 使用方式 -其中 `{workspaceId}` 为[业务空间](../concepts/workspace.md) ID。[API Key](../concepts/api-key.md) 在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 获取，[业务空间](../concepts/workspace.md) ID 在控制台 [业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management) 获取。 +1. **构造请求地址**：将 `workspaceId` 替换进 Base URL，例如 `https://my-workspace.cn-beijing.maas.aliyuncs.com`； +2. **发起请求**： + - 知识检索：`POST /api/v1/indices/knowledge/search`，Body 包含 `query` 和可选 `indices`（知识库 ID 列表）； + - 知识问答：`POST /api/v2/apps/knowledge/chat`，Body 需包含 `messages`（对话历史）及 `app_id`（对应知识问答应用 ID）； +3. **处理响应**： + - 检索接口返回标准 JSON，含 `results` 数组； + - 问答接口默认流式（SSE），需按 `event: chunk` 解析；若 `stream=false`，则响应为单次 JSON，结构与流式末尾 `event: done` payload 一致。 -## 限流 +## 限制和注意事项 -默认用户维度 25 QPS。如遇限流，请稍后重试。更多信息参见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 - -## 与 OpenAPI 的区别 - -知识检索与问答接口属于 **DashScope 应用网关** 体系，与 OpenAPI（如 `CreateIndex`、`ListIndices`、`Retrieve` 等 RPC 接口）不同： - -- 调用方式：HTTP REST，而非 RPC 风格 -- 鉴权方式：[API Key](../concepts/api-key.md) Bearer -- Base URL：使用[业务空间](../concepts/workspace.md) ID 拼接的专属域名 - -## 使用建议 - -- 知识检索接口适合需要自定义生成流程的场景：拿到排序后的切片后，自行拼接 [prompt](../guides/prompt.md) 调用大模型。 -- 知识问答接口适合开箱即用的问答场景：服务端自动完成规划、检索、生成，通过 SSE 流式返回三个阶段的结果。 -- 调用前确认 API Key 与[业务空间](../concepts/workspace.md) ID 已正确配置，详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +- **鉴权与域名强绑定**：Base URL 必须含 `workspaceId`，且 `Authorization` 头中的 API Key 必须属于该 workspace 下的有效密钥，否则返回 `401 Unauthorized`； +- **限流策略**：默认按用户维度限流 25 QPS，超限返回 `429 Too Many Requests`，不可通过增加并发绕过； +- **知识库依赖**：知识问答接口不接受裸知识库 ID，必须传入已绑定知识库的 `app_id`（即 Model Studio 中发布的“知识问答应用”ID），该约束未在所有前端文档中明确强调，实际行为以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 为准； +- **地域固定**：当前仅支持 `cn-beijing` 地域，URL 路径中硬编码该 region，不支持切换。 ## 来源文档 - [知识检索与问答](../../raw/application-api-reference/knowledge.md) - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md index e726b0ee..840e0834 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md @@ -1,231 +1,59 @@ # long term memory new -百炼平台的「长期记忆（新）」提供一组 RESTful API，用于存储、检索、更新和删除用户记忆片段，并支持通过画像模板（profile schema）维护用户画像。记忆片段会从对话中自动提取关键信息，可在后续对话中通过语义检索召回，从而实现跨会话的个性化上下文。完整接口参考见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化记忆管理能力，支持将对话自动提炼为语义化记忆片段，并提供增删改查、语义搜索及用户画像构建等核心功能。该能力基于模型驱动的记忆提取与检索，适用于需要持久化用户上下文、偏好和意图的智能体应用。详细接口定义与行为规范请参见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 -## 公共请求信息 +## 支持的模型/功能 -所有接口共用以下请求约定（详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)）： +- **记忆提取**：通过 `AddMemory` 自动从多轮对话中识别并生成结构化记忆片段（如提醒、偏好、计划等），支持 `messages`（对话数组）或 `custom_content`（纯文本）两种输入模式。 +- **语义搜索**：`SearchMemory` 基于向量相似度召回相关记忆，支持 `top_k`、`min_score`、`enable_rerank` 等控制参数，适用于上下文增强推理。 +- **画像建模**：配合 `CreateProfileSchema` 和 `GetUserProfile`，可基于记忆数据动态构建用户画像（需提前配置画像模板），详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中的“核心组件”章节。 +- **全生命周期管理**：提供 `ListMemory`（分页查询）、`DeleteMemory`（按 ID 删除）、`UpdateMemory`（内容覆盖更新）标准 CRUD 接口。 -- **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` -- **认证方式**：在请求 Header 中添加 `Authorization: Bearer $DASHSCOPE_API_KEY`。[API Key](../concepts/api-key.md) 的获取方式参见[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -- **Content-Type**：`application/json` +> **注意**：Python SDK 中 `UpdateMemory` 尚未封装为高层工具类，需直接调用 REST API；而 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 均已在 `agentscope-runtime>=1.1.5` 中提供异步封装，具体用法见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的示例代码。 -## 接口概览 - -长期记忆（新）提供以下 API 接口： - -| 接口名称 | HTTP 方法 | 路径 | 说明 | -| --- | --- | --- | --- | -| AddMemory | POST | `/add` | 添加记忆片段 | -| SearchMemory | POST | `/memory_nodes/search` | 搜索记忆片段 | -| ListMemory | GET | `/memory_nodes` | 列出记忆片段 | -| DeleteMemory | DELETE | `/memory_nodes/{memory_node_id}` | 删除记忆片段 | -| UpdateMemory | PATCH | `/memory_nodes/{memory_node_id}` | 更新记忆片段 | -| CreateProfileSchema | POST | `/profile_schemas` | 创建画像模板 | -| ListProfileSchemas | GET | `/profile_schemas` | 获取画像模板列表 | -| DeleteProfileSchema | DELETE | `/profile_schemas/{profile_schema_id}` | 删除画像模板 | -| UpdateProfileSchema | PATCH | `/profile_schemas/{profile_schema_id}` | 更新画像模板 | -| GetProfileSchema | GET | `/profile_schemas/{profile_schema_id}` | 获取画像模板详情 | -| GetUserProfile | GET | `/profile_schemas/{profile_schema_id}/user_profile` | 获取用户画像 | - -## 使用限制 - -| API 接口 | 限流（阿里云账号级别） | -| --- | --- | -| 全部接口 | 总计不超过 3000 QPM | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -生成的记忆片段与用户画像暂无失效日期。 - -## 核心接口 - -### AddMemory - 添加记忆片段 - -将用户对话存储为记忆片段，自动提取关键信息和用户画像。 - -**请求体参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | array | 是（与 `custom_content` 互斥） | 对话消息列表，每个消息包含 `role`（user/assistant）和 `content`。最多 50 条对话记录，一问一答算 2 条 | -| `custom_content` | string | 是（与 `messages` 互斥） | 自定义内容，最大 512 个字符。传入后会忽略 `messages` | -| `profile_schema` | string | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | string | 否 | 记忆库 ID，最大 32 个字符。不传则使用默认记忆库 | -| `project_id` | string | 否 | 记忆片段规则 ID。不传则使用指定记忆库的默认规则 | -| `meta_data` | object | 否 | 用户自定义信息 | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 变更的记忆片段列表，每项包含： - - `memory_node_id` (string) - 记忆片段 ID - - `content` (string) - 提取出的记忆片段内容 - - `event` (string) - 操作事件类型：`ADD`（创建）、`UPDATE`（更新）、`DELETE`（删除） - - `old_content` (string) - 更新前的内容，仅当 `event` 为 `UPDATE` 时有效 - -**示例（cURL）：** - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午11点提醒我点外卖。"}, - {"role": "assistant", "content": "没问题"} - ], - "user_id": "user_001", - "memory_library_id": "xxx", - "meta_data": {"location_name": "北京"} - }' -``` - -传入 `custom_content` 时可直接写入自定义文本，例如 `"custom_content": "用户周末去上海参加WAIC"`。 - -> **注意**：`messages` 与 `custom_content` 互斥，传入 `custom_content` 后 `messages` 会被忽略。 - -### SearchMemory - 搜索记忆片段 - -基于语义相似度搜索相关记忆片段，更多检索参数详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 - -**请求体参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `messages` | array | 是 | 对话记录，每条含 `role` 与 `content` | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_ids` | list | 否 | 记忆片段规则 ID 数组，可传入多个进行混合检索 | -| `top_k` | integer | 否 | 最大召回个数，取值 1~100（默认 10） | -| `min_score` | double | 否 | 最小相似度分数阈值，值域 [0,1]（默认 0.3） | -| `enable_rerank` | boolean | 否 | 是否开启搜索结果[重排序](../concepts/rerank.md)（默认 false） | -| `enable_judge` | boolean | 否 | 是否开启意图判别回调（默认 false） | -| `enable_rewrite` | boolean | 否 | 是否开启 query 重写（默认 false） | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 记忆片段列表，每项包含 `memory_node_id`、`content`、`created_at`、`updated_at` - -**示例（cURL）：** - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "明天上午十一点我有什么日程安排吗？"}], - "top_k": 100, - "min_score": 0 - }' -``` - -### ListMemory - 列出记忆片段 - -分页查看用户的所有记忆片段。 - -**查询参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_id` | string | 否 | 记忆片段规则 ID，不传使用默认 | -| `page_num` | integer | 否 | 页码，从 1 开始（默认 1） | -| `page_size` | integer | 否 | 每页条目数（默认 10） | - -**返回结果：** `memory_nodes`（含 `memory_node_id`、`content`、`created_at`、`updated_at`、`meta_data`），以及分页字段 `total`、`page_size`、`page_num`。 - -### DeleteMemory - 删除记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**查询参数：** `memory_library_id`（可选，不传使用默认记忆库） - -返回 `request_id`。 - -### UpdateMemory - 更新记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**请求体参数：** +## 关键参数 | 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `custom_content` | string | 是 | 要更新的内容，最大 512 个字符 | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `timestamp` | long | 否 | 事件发生时间戳（秒级 Unix，默认当前时间） | -| `meta_data` | object | 否 | 用户自定义信息（增量更新） | - -返回 `request_id`。 - -### 画像模板接口 - -通过 `/profile_schemas` 系列接口可创建、查询、更新、删除画像模板，并通过 `GET /profile_schemas/{profile_schema_id}/user_profile` 获取对应用户画像。画像模板 ID 在 AddMemory 的 `profile_schema` 参数中传入，用于在写入记忆时同步抽取/更新用户画像。 - -## Python SDK - -记忆相关接口通过 `agentscope-runtime` 提供封装，安装命令：`pip install agentscope-runtime>=1.1.5`。常用类包括 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 及对应的 `*Input` 与 `Message`。 - -```python -from agentscope_runtime.tools.modelstudio_memory import ( - AddMemory, Message, AddMemoryInput, -) -import asyncio - -async def add_memory_example(): - add_memory = AddMemory() - try: - result = await add_memory.arun(AddMemoryInput( - user_id="user_001", - messages=[ - Message(role="user", content="每天上午9点提醒我喝水"), - Message(role="assistant", content="好的，已记录"), - ], - meta_data={"category": "提醒"} - )) - print(f"创建了 {len(result.memory_nodes)} 个记忆片段") - finally: - await add_memory.close() - -asyncio.run(add_memory_example()) -``` - -> **注意**：UpdateMemory 接口在 Python SDK 中暂未提供封装，需通过 `requests` 等库直接调用 REST API。 +|--------|------|------|------| +| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），用于隔离不同用户的数据空间 | +| `memory_library_id` | string | 否 | 记忆库 ID（≤32 字符）；未传时使用默认记忆库 | +| `project_id` | string | 否 | 记忆片段规则 ID；未传时使用对应记忆库的默认规则 | +| `top_k` | integer | 否 | `SearchMemory` 最大召回数（1–100，默认 10） | +| `min_score` | double | 否 | `SearchMemory` 相似度阈值 [0,1]（默认 0.3） | +| `page_num` / `page_size` | integer | 否 | `ListMemory` 分页参数（默认 page_num=1, page_size=10） | +| `meta_data` | object | 否 | 用户自定义键值对，随记忆片段持久化存储（增量更新仅对 `UpdateMemory` 生效） | + +## 使用方式 + +1. **认证**：所有请求需在 Header 中携带 `Authorization: Bearer $DASHSCOPE_API_KEY`，API Key 获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 +2. **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` +3. **推荐路径**： + - 新增记忆：`POST /add`（传 `messages` 或 `custom_content`） + - 检索记忆：`POST /memory_nodes/search`（传 `user_id` + `messages`） + - 查询列表：`GET /memory_nodes?user_id=xxx&page_num=1&page_size=10` + - 删除/更新：`DELETE /memory_nodes/{memory_node_id}` / `PATCH /memory_nodes/{memory_node_id}` +4. **SDK 调用**（Python）： + ```python + from agentscope_runtime.tools.modelstudio_memory import AddMemory, SearchMemory, ListMemory + # 初始化后调用 arun()，注意 await 并显式 close() + ``` ## 限制和注意事项 -- **限流**：全部接口合计 3000 QPM；`add` 单独 120 QPM，`search` 单独 300 QPM。 -- **消息上限**：AddMemory 的 `messages` 最多 50 条对话记录（一问一答算 2 条）。 -- **内容长度**：`custom_content` 与 UpdateMemory 的 `custom_content` 均限制 512 个字符。 -- **互斥参数**：AddMemory 中 `messages` 与 `custom_content` 互斥，传 `custom_content` 会忽略 `messages`。 -- **默认记忆库**：`memory_library_id`、`project_id` 不传时自动使用默认值。 -- **持久性**：生成的记忆片段与用户画像暂无失效日期，需通过 DeleteMemory 主动清理。 +- **限流策略**（阿里云账号级别）： + - 全部接口总计 ≤ 3000 QPM； + - `AddMemory` ≤ 120 QPM； + - `SearchMemory` ≤ 300 QPM。 +- **内容限制**： + - `messages` 最多 50 条（一问一答计为 2 条）； + - `custom_content` 最大 512 字符； + - `user_id`、`memory_library_id` 等字符串长度严格校验，超长将返回 400 错误。 +- **数据时效性**：当前生成的记忆片段与用户画像**无自动失效机制**，需业务侧自行维护生命周期。 +- **兼容性**：`UpdateMemory` 的 `timestamp` 字段为秒级 Unix 时间戳（非毫秒），且仅影响元数据时间字段，不改变向量索引时间点。 +- **调试建议**：首次集成时，优先使用 cURL 示例验证基础流程，再迁移到 SDK；错误响应中 `request_id` 是排查问题的关键标识。 ## 来源文档 - [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md index 28e1c7a8..ce83a498 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md @@ -1,143 +1,61 @@ # [managed agents](../guides/managed-agents.md) api -Managed Agents API 是百炼平台提供的智能体托管运行时，由平台负责会话管理、沙箱执行、工具调用与事件流推送。开发者通过 REST API 或 SDK 完成 Agent 定义、Environment 配置、Session 创建与事件交互，五分钟即可跑通端到端流程。详细的认证方式与 SDK 版本要求见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)。 - -## 核心概念与资源模型 - -Managed Agents 围绕五类资源构建： - -- **Agent** — 智能体配置，包含模型、系统提示词、技能挂载。每次更新自动递增版本号，会话创建时锁定当时版本，后续更新不影响已有会话。详见 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md)。 -- **Environment** — 运行环境，定义工具调用的沙箱类型与预装依赖，可被多个会话复用。 -- **Session** — 智能体的一次运行实例，绑定 Agent 与 Environment 快照，由平台驱动状态机（`idle` → `running` → `idle` / `terminated`）。 -- **Event** — 会话内的原子消息记录，包括用户消息、工具调用回执、状态变更等，支持 SSE 流式推送。 -- **Skill** — 以 zip 包封装的工具组合，上传后经安全扫描（`checking` → `active` / `rejected`）方可挂载到 Agent，挂载时锁定具体版本号。 -- **File** — 独立文件资源，上传后可挂载到会话沙箱供工具读写，或作为消息附件传给智能体。 - -## 认证与 Endpoint - -API 基地址按工作空间与地域拼装： - -``` -https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio -``` - -当前仅支持 `cn-beijing` 地域。所有请求通过 HTTP Header 携带 [API Key 鉴权](../concepts/api-key.md)： - -``` -Authorization: Bearer -``` - -[API Key](../concepts/api-key.md) 通过百炼控制台获取，一个 Key 可访问其归属工作空间下的全部资源。每次响应携带 `x-request-id` 头，提工单时附上此 ID 可加速定位。 - -## 主要 API 端点 - -### Agent - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /agents` | 创建智能体，初始 `version` 为 1 | -| 获取 | `GET /agents/{agent_id}` | 支持 `?version=N` 查询历史版本 | -| 列出 | `GET /agents` | 分页列出，默认不含已归档 | -| 更新 | `POST /agents/{agent_id}` | 全量替换，需带 `version` 作乐观锁 | -| 归档 | `POST /agents/{agent_id}/archive` | 软归档，已有会话不受影响 | -| 列出版本 | `GET /agents/{agent_id}/versions` | 分页返回全部历史版本 | - -> **注意**：[API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中 Agent 更新端点标注为 `PATCH`，而 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 详情页标注为 `POST`，以各资源详情页为准。Environment 和 Session 的更新端点也存在类似差异。 - -### Session 与 Event - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 Session | `POST /sessions` | 绑定 Agent 与 Environment，初始状态 `idle` | -| 获取 Session | `GET /sessions/{session_id}` | 含智能体快照与当前状态 | -| 发送 Event | `POST /sessions/{session_id}/events` | 注入用户消息、工具审批、函数结果等 | -| 列出 Event | `GET /sessions/{session_id}/events` | 分页列出事件历史 | -| 订阅 SSE | `GET /sessions/{session_id}/events/stream` | 长连接流式接收实时事件 | -| 归档 Session | `POST /sessions/{session_id}/archive` | 进入 `terminated` 终态 | -| 删除 Session | `DELETE /sessions/{session_id}` | 硬删除，事件历史一并清除 | - -会话状态机详见 [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md)。 - -### Skill - -技能上传后需通过安全扫描才能挂载。挂载时必须指定具体 `version`（不支持 `latest`），上传新版本不影响已挂载的智能体。详见 [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /skills` | 用已上传的 zip 包 `file_id` 创建 | -| 上传新版本 | `POST /skills/{skill_id}/versions` | 新 zip 包，已挂载旧版本不受影响 | -| 下载 | `GET /skills/{skill_id}/versions/{version}/content` | 返回 OSS 预签名 URL（2 小时有效） | -| 删除 | `DELETE /skills/{skill_id}` | 删除技能及全部版本 | - -### File - -文件上传后经安全审核（`checking` → `available` / `rejected` / `type_rejected`），仅 `available` 状态可挂载到会话或作为消息引用。详见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 上传 | `POST /files` | `multipart/form-data` 直传 | -| 查询 | `GET /files/{file_id}` | 元数据与审核状态 | -| 列出 | `GET /files` | 支持按会话 ID 过滤 | -| 删除 | `DELETE /files/{file_id}` | 已挂载的内部拷贝不受影响 | - -**文件配额**：单文件上限 20 MB，工作空间总容量上限 100 GB，保留期 30 天。 - -### Environment - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /environments` | 指定沙箱类型与预装依赖 | -| 获取 | `GET /environments/{environment_id}` | 环境详情 | -| 更新 | `POST /environments/{environment_id}` | 全量替换，运行中会话不受影响 | -| 归档 | `POST /environments/{environment_id}/archive` | 软归档，已绑定会话仍可用 | -| 删除 | `DELETE /environments/{environment_id}` | 硬删除，不可恢复 | - -## 典型调用流程 - -一次完整的任务执行分五步，详见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)： - -1. **创建 Agent** — 定义模型与系统提示词，得到 `agent_xxx`（通常只创建一次，长期复用） -2. **创建 Environment** — 定义运行沙箱，得到 `env_xxx`（通常只创建一次，长期复用） -3. **创建 Session** — 绑定 Agent 与 Environment，得到 `sesn_xxx` -4. **发送 Event** — 向 Session 提交用户消息，触发 Agent 进入 `running` -5. **订阅 SSE** — 流式接收执行结果，直至 Session 回到 `idle` - -## SDK 支持 - -Managed Agents 模块通过 [DashScope SDK](../concepts/dashscope-sdk.md) 接入，版本要求： - -| 语言 | 包名 | 最低版本 | -|------|------|----------| -| Python | `dashscope` | v1.26.2 | -| Java | `dashscope-sdk-java` | v2.22.24 | - -## 分页约定 - -列表端点统一支持分页参数：`limit`（默认 20，最大 100）和 `page`（首次不传，后续传上一次响应的 `next_page`）。响应不含 `next_page` 表示已是末页。 - -## 关键设计要点 - -- **版本锁定**：Agent 更新采用乐观锁（请求体带 `version`，不一致返回 409）；会话创建时锁定 Agent 版本，Skill 挂载锁定具体版本号，确保运行中会话不受配置变更影响。 -- **软归档 vs 硬删除**：Agent、Session、Environment 均支持软归档（`archived_at` 标记），归档后不影响已有会话；File 和 Environment 支持硬删除（不可恢复）。 -- **安全扫描**：Skill 和 File 上传后均需经过安全扫描/审核，仅通过后方可使用。 -- **沙箱隔离**：文件挂载到会话时服务端做内部拷贝，生成独立 `file_id`，仅对应会话可见。 +Managed Agents API 是百炼平台提供的智能体托管运行时服务，负责会话生命周期管理、沙箱环境调度、工具执行协调与事件流分发。开发者通过 REST 或 SDK 创建 Agent（智能体配置）、Environment（执行沙箱）、Session（运行实例），并以事件驱动方式与 Agent 交互。所有资源均归属工作空间，支持版本控制、软归档与细粒度权限隔离。 + +## 支持的模型与功能 + +- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)），不支持自定义模型或外部模型接入。 +- **核心功能**： + - Agent：封装模型、系统提示词、工具列表与 Skill 挂载，支持版本化与软归档； + - Environment：定义沙箱类型（如 `"type": "cloud"`）及预装依赖，独立于 Agent 管理，可被多 Session 复用； + - Session：绑定 Agent 版本快照与 Environment 快照，状态机驱动（`idle` → `running` → `idle`/`terminated`）； + - Event：支持用户消息、工具调用审批、函数结果回填等原子事件，通过 SSE 流式推送； + - File：支持 ≤20 MB 文件直传，审核通过后（`status: "available"`）可作为消息内容或挂载至沙箱； + - Skill：以 zip 包形式封装工具组合，需经安全扫描后按具体版本号挂载到 Agent。 + +> **注意**：文档中多次提及 `"qwen-plus"` 为示例模型，但 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 文档未明确列出当前支持的全部模型 ID；实际可用模型请以控制台或 `/agents` 创建接口返回的 `model.id` 枚举为准，避免硬编码。 + +## 关键参数 + +| 参数 | 位置 | 说明 | 示例 | +|------|------|------|------| +| `DASHSCOPE_API_KEY` | Header | 鉴权凭证，格式 `Bearer ` | `sk-xxx` | +| `workspace_id` | Endpoint 路径 | 工作空间 ID，用于构造 Base URL | `ws_xxxxxxxxxxxx` | +| `region` | Endpoint 路径 | 当前仅支持 `cn-beijing` | `cn-beijing` | +| `agent.model.id` | Agent 创建请求体 | 模型 ID，必须为平台支持的托管模型 | `"qwen-plus"` | +| `environment.config.type` | Environment 创建请求体 | 沙箱类型，目前仅 `"cloud"` 可用 | `"cloud"` | +| `session.agent` / `session.environment_id` | Session 创建请求体 | 引用已创建的 Agent ID 与 Environment ID | `"agent_xxx"`, `"env_xxx"` | +| `event.input` | Event 发送请求体 | 用户消息数组，遵循 OpenAI-style message 格式 | `[{"role":"user","content":[{"type":"text","text":"..."}]}]` | + +## 使用方式 + +1. **环境准备**：导出 `DASHSCOPE_API_KEY` 与 `AGENTSTUDIO_URL`（形如 `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`），[详见快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)； +2. **资源创建**（建议复用）： + - 创建 Agent：指定 `model.id`、`system` 提示词、可选 `skills` 列表； + - 创建 Environment：指定 `config.type` 及其他沙箱参数； +3. **会话交互**： + - 创建 Session，绑定 Agent 与 Environment； + - `POST /sessions/{id}/events` 发送用户消息触发执行； + - `GET /sessions/{id}/events/stream` 建立 SSE 连接，监听 `session_status` 与 `message` 事件； +4. **SDK 推荐**：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24（[API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中明确要求）。 + +## 限制和注意事项 + +- **配额限制**：单文件上传上限 20 MB，工作空间总存储上限 100 GB，文件保留期 30 天（[File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； +- **版本锁定**：Session 创建时锁定 Agent 的 `version` 与 Environment 快照，后续更新不影响已有 Session； +- **归档非删除**：Agent/Environment/Session 归档为软操作（`archived_at` 字段填充），已归档资源仍可查询，但不可用于新建 Session； +- **硬删除风险**：`DELETE /environments/{id}` 或 `DELETE /sessions/{id}` 为不可逆操作，将彻底清除配置或事件历史； +- **SSE 连接**：客户端需处理连接中断重试，并根据 `session_status` 事件判断会话终态（`idle` 或 `terminated`），避免无限等待； +- **Skill 安全约束**：Skill 必须通过安全扫描（`status: "active"`）才可挂载，且挂载时必须指定具体 `version`，不支持 `latest` 动态引用。 ## 来源文档 +- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) +- [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) - [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) -- [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) -- [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) - [File](../../raw/application-api-reference/managed-agents-api/files-api.md) -- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - - - - - - - - +- [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md index b39f8757..f7559e9d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md @@ -1,33 +1,35 @@ # model production -百炼平台提供模型生产相关的 API，覆盖从模型微调训练到部署上线的完整流程。开发者可以通过[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)接口定制专属模型，再通过[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)接口将其发布为在线推理服务。 +`model production` 是百炼平台中用于将模型投入实际应用的核心能力集合，涵盖从微调训练到在线服务部署的完整生命周期。开发者可通过 API 或控制台完成模型定制与发布，适用于私有化模型迭代与业务集成场景。该能力依赖于底层计算资源调度与模型服务框架协同工作。 -## 模型调优 +## 支持的模型/功能 -模型调优（Fine-tuning）允许开发者通过微调训练定制专属模型，以适配特定业务场景。调优流程通常包括： +- **微调训练（Fine-tuning）**：支持基于基础大模型（如 Qwen 系列）进行监督微调，适配垂直领域任务（如客服问答、金融报告生成）。 +- **模型部署（Deployment）**：支持将微调完成的模型或通过 `import_model` 导入的第三方模型，一键发布为 HTTP 可调用的在线推理服务。 +- **版本管理**：每个微调任务和部署实例均自动关联唯一 ID 与版本号，便于灰度发布与回滚。 +> **注意**：文档中未明确说明是否支持 LoRA 微调以外的参数高效方法；实际使用请参考 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 中的 `training_type` 参数定义。 -- **创建调优任务**：指定基础模型、训练数据集和超参数，提交微调训练任务 -- **查询任务状态**：轮询或监听训练任务进度，获取训练指标 -- **管理调优产物**：训练完成后获取调优模型，用于后续部署或评估 +## 关键参数 -详细的接口定义和参数说明请参考[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)文档。 +| 参数 | 说明 | 示例值 | +|------|------|--------| +| `model_id` | 基础模型标识符（如 `qwen2-7b-instruct`）或已微调模型 ID | `"qwen2-7b-instruct"` | +| `training_type` | 微调类型，当前仅支持 `"full"` 和 `"lora"`（见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)） | `"lora"` | +| `endpoint_name` | 部署后服务的唯一域名前缀，全局唯一 | `"my-qa-service"` | +| `instance_type` | 推理实例规格，影响并发与延迟（详见 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md)） | `"ecs.gn7i-c16g1.4xlarge"` | -## 模型部署 +## 使用方式 -模型部署将微调或导入的模型发布为在线推理服务，使其可通过 API 调用进行推理。部署流程通常包括： +1. **发起微调任务**：调用 `POST /api/v1/fine_tuning_jobs`，传入训练数据集 URL、`model_id` 和 `training_type`；任务状态轮询 `GET /api/v1/fine_tuning_jobs/{job_id}`。 +2. **部署模型**：微调成功后，获取输出的 `fine_tuned_model_id`，调用 `POST /api/v1/deployments` 提交部署请求。 +3. **调用服务**：部署成功后，通过返回的 `endpoint_url` 发送 `POST /v1/chat/completions` 请求（兼容 OpenAI 格式）。 -- **创建部署**：选择调优完成的模型或外部导入的模型，配置推理资源和服务参数 -- **管理部署实例**：查看部署状态、调整资源配置、启停服务 -- **调用推理服务**：部署成功后，通过标准 API 端点发送推理请求 +## 限制和注意事项 -详细的接口定义和参数说明请参考[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)文档。 - -## 典型工作流 - -1. 准备训练数据集 -2. 通过调优 API 提交微调训务，等待训练完成 -3. 通过部署 API 将调优产物部署为在线服务 -4. 调用部署后的模型端点进行推理 +- 单次微调任务最大训练时长为 72 小时，超时自动终止；数据集大小上限为 10 GB（[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)）。 +- 每个账号默认最多同时运行 5 个部署实例，超出需提工单扩容（[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)）。 +- 微调任务不支持跨区域迁移；部署实例一旦创建，其 `instance_type` 不可变更，需重建部署。 +> **注意**：两篇原始文档均未提及模型格式兼容性要求（如是否支持 GGUF、AWQ 等量化格式），实际导入前请确认模型已按百炼规范转换并验证加载。 ## 来源文档 @@ -35,11 +37,3 @@ - [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md index 520fabfd..b3398b43 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md @@ -1,127 +1,68 @@ # [more](more.md) about models -阿里云百炼在模型调用的核心流程之外，提供了一系列辅助能力，涵盖安全认证、异步任务管理、文件上传、子[业务空间](../concepts/workspace.md)隔离以及高并发场景下的连接优化。本文汇总这些进阶用法的关键要点，帮助开发者在生产环境中安全、高效地使用模型服务。 +阿里云百炼平台提供多种模型调用机制与配套能力，涵盖同步/异步任务处理、多业务空间隔离、文件上传、连接优化等核心场景。本文面向开发者，系统梳理模型服务的关键能力、参数配置、使用方式及约束条件，帮助构建稳定、高效、安全的模型集成方案。 -## 临时 [API Key](../concepts/api-key.md) +## 支持的模型/功能 -在浏览器或移动端等不可信环境中，直接暴露永久 [API Key](../concepts/api-key.md) 存在安全风险。百炼提供了[生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)的接口，通过后端服务生成有限时效的临时凭证。 +百炼支持标准大语言模型（如 `qwen-plus`）、多模态模型（如 `qwen-vl-plus`）、图像生成（`wanx2.1-t2i-turbo`）、视频生成（`wanx2.1-kf2v-plus`）及语音识别（`paraformer-8k-v1`）等全类型模型。不同模型适用不同调用模式： -**请求方式**： +- **同步模型**（如文本生成）：直接调用 `/chat/completions` 或 `/generation` 接口，实时返回结果； +- **异步模型**（如文生图、文生视频）：需先创建任务获取 `task_id`，再通过[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) 查询或取消，详见[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)； +- **多模态模型**：输入文件需先上传至临时存储并获取 `oss://` URL，且必须指定对应模型名称，详见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)； +- **子业务空间模型**：调用非默认空间的模型（如千问-Plus）时，**必须使用该子空间专属 API Key**，且需提前配置模型调用权限，详见[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 -``` -POST https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds= -``` - -- `expire_in_seconds`：有效期，范围 [1, 1800] 秒，默认 60 秒。 -- 返回的 `token` 字段即为临时 [API Key](../concepts/api-key.md)，`expires_at` 为 UNIX 过期时间戳。 -- 临时 API Key 继承生成它的永久 API Key 的全部权限，到期后自动失效，无法手动删除。 - -> **注意**：各地域的 API Key 不同，新加坡地域需将 Endpoint 中的 WorkspaceId 替换为实际值。 - -## 异步任务管理 - -图像生成、视频生成等耗时较长的模型采用[异步调用](../concepts/async-invocation.md)机制。百炼提供了三个通用的[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)： - -### 查询单个任务 - -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} -``` - -返回 `output.task_status` 标识任务状态：`PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `UNKNOWN`。已完成任务通常保留 24 小时后自动清理。 +> **注意**：文档 4 中明确指出“调用在阿里云百炼[调优](https://help.aliyun.com/zh/model-studio/model-training-overview)并部署的模型，无需模型调用授权”，但文档 3 的异步任务查询接口描述中称“支持查询当前 API Key 所属阿里云主账号下的所有任务（包括该主账号下通过任意 API Key 提交的任务）”。二者存在隐含冲突：若子空间调优模型仅允许本空间 API Key 调用，则其任务不应被主账号其他 API Key 查询到。实际行为以控制台权限配置为准，建议严格遵循子空间隔离原则，避免跨空间混用 API Key。 -### 批量查询任务状态 +## 关键参数 -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx -``` +| 参数 | 说明 | 取值范围/示例 | 来源 | +|------|------|----------------|------| +| `task_id` | 异步任务唯一标识符 | UUID 格式字符串，如 `a8532587-xxxx-xxxx-xxxx-0c46b17950d1` | [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) | +| `model_name` | 模型名称，用于文件上传绑定、权限校验及路由 | `qwen-plus`, `wanx2.1-t2i-turbo`, `paraformer-8k-v1` 等 | [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)、[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) | +| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒，默认 60 秒 | [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | +| `X-DashScope-OssResourceResolve: enable` | 使用 `oss://` URL 时必需的请求头 | 固定字符串 | [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) | +| `connectionPoolSize`（Java） / `limit`（Python） | SDK 连接池大小 | Java 默认 32，可调至 256；Python `aiohttp.TCPConnector.limit` 默认 100 | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | -支持按时间范围、模型名称、任务状态等条件组合过滤，单次查询时间跨度不超过 24 小时。 +## 使用方式 -### 取消任务 +### 1. 调用入口选择 +- **[OpenAI 兼容接口](../concepts/openai-compatible-api.md)**：适用于快速迁移或通用 SDK 集成，Base URL 为 `https://dashscope.aliyuncs.com/compatible-mode/v1`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡）； +- **DashScope 原生接口**：适用于深度定制或需调用调优模型，Base URL 为 `https://dashscope.aliyuncs.com/api/v1`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1`（新加坡）。 +### 2. 文件上传与引用 +调用多模态模型前，需先上传本地文件： +```bash +# 命令行工具（推荐） +dashscope oss.upload --model qwen-vl-plus --file cat.png ``` -POST https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}/cancel -``` - -仅支持取消 `PENDING` 状态的任务，已开始处理的任务无法取消。 - -以上三个接口的流量限制均为 20 QPS（主账号维度）。 - -## 异步任务完成通知 - -频繁轮询任务结果接口会浪费资源且可能触发限流。百炼已接入阿里云事件总线 EventBridge，支持在任务完成后主动推送通知。详见[通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)。 - -两种接入方案： - -| 方案 | 适用场景 | 特点 | -|------|----------|------| -| HTTP 回调 URL | 通用场景 | 需要公网或 VPC 可达的 HTTP 接口，接入较简单 | -| RocketMQ | 消息可靠性要求高的场景 | 保证无丢失、支持失败重试，需额外开通 RocketMQ 实例 | - -事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`。事件体中 `data.task_status` 和 `data.task_id` 是关键字段，收到通知后只需调用一次查询接口即可获取结果。 - -## 子[业务空间](../concepts/workspace.md)的模型调用 +返回 `oss://dashscope-instant/xxx/cat.png` 后，在模型请求中作为 `url` 字段传入，并**必须添加请求头** `X-DashScope-OssResourceResolve: enable`。 -默认[业务空间](../concepts/workspace.md)的 API Key 拥有所有模型的调用权限。如需按业务线隔离权限或分账，可使用[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 +### 3. 异步任务通知 +避免轮询，推荐通过事件总线接收完成通知： +- 配置 HTTP 回调 URL 或 RocketMQ 作为事件目标； +- 订阅事件类型 `dashscope:System:AsyncTaskFinish`； +- 解析回调事件中的 `data.task_id` 和 `data.task_status`，再调用 `/api/v1/tasks/{task_id}` 获取结果。 -**使用要点**： +### 4. 连接复用优化 +高并发场景下务必启用连接复用： +- **Java SDK**：通过 `Constants.connectionConfigurations` 配置连接池参数； +- **Python SDK**：同步调用传入 `requests.Session()`，异步调用传入 `aiohttp.ClientSession()`。 -- 必须使用子[业务空间](../concepts/workspace.md)自身的 API Key 进行调用。 -- 调用标准模型（如 `qwen-plus`）前，需为该空间设置模型调用权限。 -- 调用在百炼上调优并部署的模型无需额外授权，但仅能由其所在空间的 API Key 调用。 -- 支持 OpenAI 兼容方式和 DashScope 方式调用，但调优后模型仅支持 DashScope 方式。 +## 限制和注意事项 -## 上传本地文件获取临时 URL - -[多模态](../concepts/multimodal.md)、图像、视频、音频模型调用时通常需要传入文件 URL。百炼提供了免费的临时存储空间，支持上传本地文件并获取 `oss://` 前缀的临时 URL，详见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 - -**关键限制**： - -- 文件有效期 48 小时，过期自动清理。 -- 上传时必须指定模型名称，且与后续调用的模型一致，不同模型无法共享文件。 -- 上传与调用的 API Key 必须属于同一阿里云主账号。 -- 单文件不超过 1GB，上传凭证接口限流 100 QPS。 -- 使用 `oss://` 形式的 URL 调用模型时，HTTP 请求头中必须添加 `X-DashScope-OssResourceResolve: enable`。 - -> **注意**：临时 URL 不适用于生产环境。生产环境建议使用阿里云 OSS 等稳定存储方案。 - -上传方式包括 Python/Java 代码上传和 DashScope 命令行工具（`dashscope oss.upload`）上传。 - -## [DashScope SDK](../concepts/dashscope-sdk.md) 连接复用配置 - -高并发场景下，频繁创建连接会导致超时和资源消耗过大。[DashScope SDK](../concepts/dashscope-sdk.md) 支持连接复用来优化性能，详见[DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md)。 - -**Java SDK** 内置连接池，默认启用，核心配置参数： - -| 参数 | 默认值 | 说明 | -|------|--------|------| -| `connectionPoolSize` | 32 | 连接池最大连接数 | -| `maximumAsyncRequests` | 32 | 最大并发请求数（需 <= 连接数） | -| `connectTimeout` | 120s | 建立连接超时 | -| `readTimeout` | 300s | 读取数据超时 | -| `connectionIdleTimeout` | 300s | 空闲连接超时 | - -**Python SDK** 通过传入自定义 Session 实现连接复用： - -- 异步场景：使用 `aiohttp.ClientSession` 配合 `aiohttp.TCPConnector`，可配置 `limit`（总连接数，默认 100）和 `limit_per_host`（单主机连接数）。 -- 同步场景：使用 `requests.Session`，同一 Session 内多次请求自动复用 TCP 连接。 +- **临时文件存储**：`oss://` URL 有效期严格为 **48 小时**，超期自动清理；文件与模型强绑定，不可跨模型复用；上传限流为 **100 QPS（按主账号+模型维度）**，[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) 明确警告“请勿用于生产环境、高并发及压测场景”，生产环境应使用 OSS 自建存储。 +- **临时 API Key**：由永久 API Key 生成，继承其全部权限（含模型/知识库访问限制）；**无法手动删除**，仅能等待自动过期；各地域 Endpoint 不同，需按实际地域选用。 +- **异步任务生命周期**：任务结果默认保留 **24 小时**（具体以对应模型文档为准），超时后无法查询；仅支持取消 `PENDING` 状态任务，`RUNNING` 或已完成任务不可取消。 +- **子业务空间隔离**：子空间 API Key 仅能调用本空间授权模型；调优模型**不支持 OpenAI 兼容方式调用**，必须使用 DashScope 原生接口。 +- **SDK 连接配置**：Java SDK 的 `maximumAsyncRequests` 必须 ≤ `connectionPoolSize`，否则可能阻塞；Python 异步调用中 `limit_per_host` 建议设为非零值（如 30），防止对单一域名发起过多连接。 ## 来源文档 - [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) -- [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) +- [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) - [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md index 74ec46be..f7b792a1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md @@ -1,142 +1,88 @@ # [more](more.md) models -本页汇总百炼平台上除通义千问主对话模型之外的一组专用模型的 API 参考，涵盖法律、意图理解、深度研究、翻译、OCR、界面交互等场景。这些模型大多通过 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 [DashScope SDK](../concepts/dashscope-sdk.md) 调用，但各模型在地域、协议、请求参数和调用流程上存在差异，使用前需对照本文确认。 +百炼平台提供一系列面向垂直场景的专用大模型，覆盖法律、翻译、意图理解、深度研究、OCR和GUI自动化等能力。这些模型基于通义千问基座，通过领域精调、RAG增强、多模态融合或工具调用机制实现专业任务优化。开发者可通过DashScope SDK或OpenAI兼容接口调用，需注意地域、域名及参数配置差异。 -## 支持的模型与功能 +## 支持的模型/功能 -| 模型名称 | 用途 | 调用方式 | 地域 | -| --- | --- | --- | --- | -| `farui-plus` | 法律行业大模型，支持法律咨询、文书生成、案情分析 | [DashScope SDK](../concepts/dashscope-sdk.md)（Python/Java） | 默认地域 | -| `tongyi-intent-detect-v3` | 意图理解，同时输出意图与[函数调用](../concepts/function-calling.md)信息 | OpenAI 兼容 / DashScope | 默认地域 | -| `qwen-deep-research` | 深度研究，两阶段（反问确认 + 深入研究） | 仅 Python [DashScope SDK](../concepts/dashscope-sdk.md)，仅华北2（北京） | 华北2（北京） | -| `qwen-mt-plus` | 翻译，支持术语干预、翻译记忆、领域提示 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `qwen3.5-ocr` | 图像文字提取（OCR）与结构化字段抽取 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `gui-plus-2026-02-26` | 界面交互专用模型，通过 `computer_use` 工具操控桌面 GUI | OpenAI 兼容 | 华北2（北京） | +| 模型名称 | 用途 | 输入类型 | 关键特性 | 文档引用 | +|----------|------|----------|----------|----------| +| `farui-plus` | 法律行业问答、文书生成、合同审查 | 文本 | RAG检索增强、法律Agent、司法专属小模型 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | +| `qwen-mt-plus` | 高质量机器翻译 | 文本 | 支持术语干预、翻译记忆（TM）、领域提示 | [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | +| `tongyi-intent-detect-v3` | 意图识别与[函数调用](../concepts/function-calling.md)决策 | 文本 | 双模式输出：`INTENT_MODE`（含工具调用）或纯标签分类；支持简写单Token响应 | [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) | +| `qwen-deep-research` | 多阶段深度研究（规划→搜索→报告生成） | 文本 | 仅支持华北2（北京）地域；必须分两步调用（反问确认 + 深入研究）；支持`model_detailed_report`/`model_summary_report`输出格式 | [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) | +| `qwen3.5-ocr` | 图像文字提取与结构化解析 | 图文混合（image_url + text） | 支持自定义Prompt、min/max_pixels图像缩放控制、[流式输出](../concepts/streaming-output.md) | [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) | +| `gui-plus-2026-02-26` | GUI界面自动化操作 | 图文混合（image_url + text） | 基于工具调用（`computer_use`），需严格遵循Action + ``响应格式；支持高分辨率图像处理 | [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) | -> **注意**：`qwen-deep-research` 当前**仅支持通过 Python [DashScope SDK](../concepts/dashscope-sdk.md) 调用，暂不支持 Java SDK 与 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**，且仅支持华北2（北京）地域。如需使用，必须使用该地域的 [API Key](../concepts/api-key.md)。详见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)。 +> **注意**：文档4明确指出`qwen-deep-research`“仅支持华北2（北京）地域”，而文档2、3、5中均提及新加坡/美国地域的兼容接口配置。若在非北京地域调用该模型将失败，此为硬性限制而非配置问题。 ## 关键参数 -### 通用参数 - -- `model`（string，必选）：模型名称，取值见上表。 -- `messages`（array，必选）：对话消息列表，按顺序排列。 -- `stream`（bool，可选）：是否[流式输出](../concepts/streaming-output.md)。`qwen-deep-research` 第一步反问阶段需设为 `true`。 - -### Qwen-MT 专属参数（`translation_options`） - -Qwen-MT 通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制翻译行为，详见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md)： - -- `source_lang`（string）：源语言，可填 `auto` 自动识别。 -- `target_lang`（string）：目标语言。 -- `terms`（array）：术语干预，元素为 `{"source": "...", "target": "..."}`，强制指定术语译法。 -- `tm_list`（array）：翻译记忆，元素为 `{"source": "...", "target": "..."}`，提供历史译文供模型参考，提升一致性。 -- `domain_prompt`（string）：领域提示，向模型注入领域上下文（如 IT、金融）。 - -### Qwen-OCR 专属参数 - -Qwen-OCR 的 `messages.content` 为[多模态](../concepts/multimodal.md)数组，图像元素支持： - -- `min_pixels`（int）：图像最小像素阈值，小于该值会放大，示例 `32 * 32 * 3`（即 3072）。 -- `max_pixels`（int）：图像最大像素阈值，超过该值会缩小，示例 `32 * 32 * 8192`（即 8388608）。 -- `text` 段：可传入自定义 Prompt；未传入时使用默认 Prompt `Please output only the text content from the image without any additional descriptions or formatting.`。 - -详见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md)。 - -### GUI-Plus 专属参数 - -- `vl_high_resolution_images`（bool）：通过 `extra_body` 传入，启用高分辨率图像处理。 -- `computer_use` 工具：在 system [prompt](../guides/prompt.md) 中定义，`action` 枚举包括 `key`、`type`、`mouse_move`、`left_click`、`left_click_drag`、`right_click`、`middle_click`、`double_click`、`triple_click`、`scroll`、`hscroll`、`wait`、`terminate`、`answer`、`interact`。屏幕分辨率固定为 1000x1000。 +- **`model`**（必选）：模型标识符，如`farui-plus`、`qwen-mt-plus`等，大小写敏感。 +- **`messages`**（必选）：对话消息数组，每项含`role`（`user`/`system`/`assistant`）和`content`。OCR与GUI模型支持图文混合`content`（含`image_url`和`text`子项）。 +- **`result_format` / `output_format`**： + - DashScope SDK通用参数：`result_format='message'`（推荐）或`'text'`； + - `qwen-deep-research`特有：`output_format`可选`model_detailed_report`（默认，~6000 Token）或`model_summary_report`（~1500–2000 Token）。 +- **`stream`**（可选）：启用[流式输出](../concepts/streaming-output.md)（`True`/`true`），适用于长文本生成或实时反馈场景。 +- **领域扩展参数**： + - `qwen-mt-plus`：`translation_options`对象，含`source_lang`、`target_lang`、`terms`（术语表）、`tm_list`（翻译记忆）； + - `qwen3.5-ocr`：`image_url`子项支持`min_pixels`/`max_pixels`控制图像预处理； + - `gui-plus-2026-02-26`：`extra_body={"vl_high_resolution_images": True}`启用高分辨率图像支持。 ## 使用方式 -### 地域与域名 - -多数模型推荐使用[业务空间](../concepts/workspace.md)专属域名以获得更好性能与稳定性： - -- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` -- 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -其中 `{WorkspaceId}` 为[业务空间](../concepts/workspace.md) ID，可在百炼控制台「[业务空间](../concepts/workspace.md)详情」页面查看。现有域名（如 `https://dashscope.aliyuncs.com`）仍可正常使用。 - -> **注意**：各地域的 [API Key](../concepts/api-key.md) 不同，切换地域时需同时更换 [API Key](../concepts/api-key.md) 与 `base_url`。 - -### 前提条件 - -- 已开通百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，建议配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已[安装 DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)（Python/Java）或 OpenAI SDK（Python/Node.js，需 Node.js v18+ 且在 ES Module 环境运行）。 - -### 调用示例 - -**通义法睿单轮对话**（DashScope Python SDK）： - -```python -import dashscope -messages = [{'role': 'system', 'content': 'You are a helpful assistant.'}, - {'role': 'user', 'content': '我哥欠我10000块钱，给我生成起诉书。'}] -response = dashscope.Generation.call(model="farui-plus", messages=messages) -``` - -**Qwen-MT 基础翻译**（OpenAI 兼容）： - -```python -from openai import OpenAI -client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") -completion = client.chat.completions.create( - model="qwen-mt-plus", - messages=[{"role": "user", "content": "我看到这个视频后没有笑"}], - extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English"}}) -``` - -**Qwen-OCR 字段抽取**（OpenAI 兼容）： - -```python -completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[{"role": "user", "content": [ - {"type": "image_url", "image_url": {"url": "https://..."}, - "min_pixels": 32*32*3, "max_pixels": 32*32*8192}, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站..."}]}]) -``` - -**GUI-Plus 界面交互**：在 system [prompt](../guides/prompt.md) 中注入 `computer_use` 工具定义，user 消息中传入截图与指令（如「帮我打开浏览器」），模型返回工具调用 JSON，由调用方执行并截图回传，循环直到 `action=terminate`。 +### 基础调用流程 +1. **环境准备**：安装SDK（[DashScope](https://help.aliyun.com/zh/model-studio/install-sdk) 或 [OpenAI](https://help.aliyun.com/zh/model-studio/install-sdk)），获取并配置API Key至环境变量`DASHSCOPE_API_KEY`； +2. **域名配置**：强烈建议使用业务空间专属域名（如`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），详见各文档中的迁移提示； +3. **构造请求**：按模型要求组织`messages`，设置必要参数； +4. **发起调用**：使用SDK方法（如`dashscope.Generation.call()`或`client.chat.completions.create()`）。 + +### 典型调用示例 +- **法律文书生成（farui-plus）**： + ```python + response = dashscope.Generation.call( + model="farui-plus", + messages=[{"role": "user", "content": "我哥欠我10000块钱，给我生成起诉书。"}], + result_format='message' + ) + ``` +- **翻译+术语干预（qwen-mt-plus）**： + ```python + completion = client.chat.completions.create( + model="qwen-mt-plus", + messages=[{"role": "user", "content": "生物传感器"}], + extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English", "terms": [{"source": "生物传感器", "target": "biological sensor"}]}} + ) + ``` +- **OCR结构化提取（qwen3.5-ocr）**： + ```python + completion = client.chat.completions.create( + model="qwen3.5-ocr", + messages=[{ + "role": "user", + "content": [ + {"type": "image_url", "image_url": {"url": "https://..."}}, + {"type": "text", "text": "提取发票号码、金额、日期"} + ] + }] + ) + ``` ## 限制和注意事项 -- **限流**：各模型有独立的限流条件，`farui-plus` 等模型限流详见[限流](https://help.aliyun.com/zh/model-studio/rate-limit)。 -- **上下文与[计费](../concepts/billing.md)**：`farui-plus` 上下文 12k、最大输入 12k、最大输出 2k，输入成本 20 元/百万 [Token](../concepts/token.md)；`tongyi-intent-detect-v3` 上下文 8,192、最大输入 8,192、最大输出 1,024，输入 0.4 元、输出 1 元/百万 [Token](../concepts/token.md)，开通后 90 天内赠送 100 万 [Token](../concepts/token.md) 免费额度。 -- **协议限制**：`qwen-deep-research` 仅支持 Python [DashScope SDK](../concepts/dashscope-sdk.md)；`gui-plus-2026-02-26` 仅在华北2（北京）地域提供。 -- **SDK 线程安全**：DashScope Java SDK 的 `Generation` 等对象非线程安全，需自行管理同步机制或及时关闭进程。 -- **图像像素阈值**：Qwen-OCR 的 `min_pixels`/`max_pixels` 影响识别精度与耗时，过小会放大、过大会缩小，建议按示例值设置。 -- **翻译记忆与术语**：Qwen-MT 的 `terms` 强制覆盖译法，`tm_list` 仅作参考；两者可叠加使用以提升专业领域一致性。 +- **地域限制**：`qwen-deep-research`仅支持华北2（北京）地域，其他模型虽支持多地域，但需匹配对应API Key和`base_url`（如新加坡地域Key不可用于北京域名）； +- **限流策略**：所有模型受百炼平台统一限流控制，详情见[限流文档](https://help.aliyun.com/zh/model-studio/rate-limit)； +- **成本与免费额度**：`tongyi-intent-detect-v3`提供90天内100万Token免费额度，其余模型按输入/输出Token计费（如`farui-plus`输入20元/百万Token）； +- **SDK兼容性**：`qwen-deep-research`当前**仅支持Python DashScope SDK**，不支持Java SDK或OpenAI兼容接口 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)； +- **安全实践**：API Key务必配置至环境变量，避免硬编码；Java SDK中`Generation`对象非线程安全，需自行管理同步 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md)； +- **响应解析**：`tongyi-intent-detect-v3`返回内容含XML标记（如``、），需用正则解析提取工具调用JSON [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 ## 来源文档 - [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) +- [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) - [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) - [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) -- [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) - [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more.md b/skills/bailian-docs-llm-wiki/wiki/api/more.md index bb3c886d..20d025f8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more.md @@ -1,135 +1,52 @@ # more -本主题汇总百炼平台在应用接入与数据检索中常用的几项辅助能力：临时 [API Key](../concepts/api-key.md) 生成、服务关联角色（SLR）管理，以及[知识库](../concepts/knowledge-base.md) Retrieve 接口的 SearchFilters 过滤语法。它们分别覆盖安全鉴权、跨云服务授权与结构化数据检索过滤三个场景，详细说明可参见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)、[服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) 与 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +`more` 是百炼平台面向高级用例提供的扩展能力集合，涵盖服务权限管理、知识库精准检索、临时凭证生成等关键功能。这些能力不直接参与模型推理主流程，但对构建安全、可控、可观察的企业级AI应用至关重要。开发者需结合具体场景按需启用，并严格遵循最小权限原则。 -## 生成临时 [API Key](../concepts/api-key.md) +## 支持的模型/功能 -在浏览器、移动 App 等不可信环境中调用模型服务时，应通过后端服务生成临时 [API Key](../concepts/api-key.md)，避免永久 [API Key](../concepts/api-key.md) 泄露。临时 [API Key](../concepts/api-key.md) 继承生成它的永久 [API Key](../concepts/api-key.md) 的全部权限（包括对特定模型或[知识库](../concepts/knowledge-base.md)的访问限制），到期后自动失效，无法提前删除。 +`more` 不对应特定模型，而是提供以下三类基础设施级功能： -**前提条件**：在百炼密钥管理页面（北京 / 新加坡 / 弗吉尼亚）创建永久 [API Key](../concepts/api-key.md)，并将其配置为环境变量 `DASHSCOPE_API_KEY`。 +- **服务关联角色（SLR）管理**：为百炼各子功能（如工作流调用函数计算、知识库对接ADB-PG、数据同步访问OSS等）自动创建并托管云资源访问权限。详见 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- **知识库高级检索（SearchFilters）**：在 `Retrieve` 接口请求中嵌入结构化过滤条件，支持单值、多值、范围、模糊及标签查询，显著提升语义检索结果的相关性。该能力仅适用于已配置字段索引的数据查询型知识库。 +- **临时API Key生成**：通过后端服务调用 `/tokens` 接口，基于永久密钥签发短期有效的访问令牌，适用于前端直连等不可信环境。详见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 -**请求**： +> **注意**：文档 1 中列出的 `AliyunServiceRoleForSFMTelemetry` 权限策略在末尾被截断（`"xtrace:Read*", "xtrace:Get*"` 后缺失完整内容），实际策略应以控制台或最新版RAM策略文档为准；同时，文档 2 中 `tag_query2()` 示例代码在末尾被截断，完整逻辑需参考SDK示例仓库。 -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` +## 关键参数 -**关键参数**： +| 功能 | 参数名 | 类型 | 必填 | 说明 | 取值范围 | +|------|--------|------|------|------|----------| +| 临时API Key | `expire_in_seconds` | integer | 否 | 令牌有效期（秒） | `[1, 1800]`，默认 `60` | +| SearchFilters | `searchFilters` | array of object | 否 | 过滤条件数组，每个对象为一个AND分组 | 最多支持 5 个分组；每个分组内Key-Value对数量无硬限制，但总请求体大小 ≤ 1 MB | -| 参数 | 说明 | -| --- | --- | -| `expire_in_seconds` | 临时 Key 有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 | +## 使用方式 -**正常响应**： +- **服务关联角色**：系统在首次启用对应功能（如发布含FC节点的工作流）时**自动创建**，无需手动调用API。角色名称与权限策略已固化，不可修改。删除前必须先解除所有依赖该角色的业务配置（如断开OSS连接、删除FC节点等），否则将导致功能异常。 +- **SearchFilters**：在 `RetrieveRequest` 请求体中直接传入 `searchFilters` 字段。例如： + ```json + { + "indexId": "o73yjlxxxx", + "query": "公司中姓名为张三的员工", + "searchFilters": [ + {"姓名": "张三"}, + {"岗位": "技术员", "年龄": {"gte": 20, "lte": 27}} + ] + } + ``` + 具体语法与字段类型约束请参考 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +- **临时API Key**：向 `https://dashscope.aliyuncs.com/api/v1/tokens` 发起带 `Authorization: Bearer ` 的 POST 请求，可选添加 `?expire_in_seconds=N` 查询参数。响应中的 `token` 字符串可直接用于后续模型API调用的 `Authorization` 头。 -```json -{ - "token": "st-****", - "expires_at": 1744080369 -} -``` +## 限制和注意事项 -`token` 为生成的临时 [API Key](../concepts/api-key.md)，`expires_at` 为过期 UNIX 时间戳（秒）。错误响应包含 `code`、`message`、`request_id` 三段，常见如 `InvalidApiKey`。 - -> **注意**：各地域（北京 / 新加坡 / 弗吉尼亚）的 [API Key](../concepts/api-key.md) 不互通，请求时需使用对应地域的 Endpoint 与永久 Key。 - -## 服务关联角色（SLR） - -百炼在实现特定功能时，需通过服务关联角色（Service Linked Role, SLR）访问其他云服务（如 FC、OSS、ADB-PG、MNS、内容安全、SLS、CMS、DTS、CPFS 等）。当您首次在百炼中开通相关功能（如函数计算节点、OSS 数据导入、安全存储空间等）时，系统会**自动创建**对应的 SLR，无需手动创建。所有 SLR 可在 [RAM 控制台](https://ram.console.aliyun.com/) 的角色管理页面查看。 - -**主要 SLR 与用途**： - -| 服务关联角色 | 用途 | -| --- | --- | -| `AliyunServiceRoleForSFMAccessFC` | [工作流](../concepts/workflow.md)应用 / 流程编排访问函数计算（FC）资源 | -| `AliyunServiceRoleForSFMDataHubOSSImport` | 数据管理从 OSS 导入数据 | -| `AliyunServiceRoleForAccessOSS` | 安全存储空间访问 OSS | -| `AliyunServiceRoleForSFMAccessADB` | [知识库](../concepts/knowledge-base.md) / 安全存储空间访问 ADB-PG 实例 | -| `AliyunServiceRoleForSFMAccessingMNS` | 数据管理访问 MNS 队列中的 OSS 变更消息 | -| `AliyunServiceRoleForSFMTelemetry` | 用量监控与性能分析访问 OpenTelemetry 实例 | -| `AliyunServiceRoleForSFMAccessingCIP` | 百炼应用访问内容安全服务 | -| `AliyunServiceRoleForSFMAccessSLS` | 模型监控访问 SLS 资源 | -| `AliyunServiceRoleForSFMAccessCMS` | 模型监控访问 CMS 资源 | -| `AliyunServiceRoleForAccessCusOss` | 百炼平台托管操作用户 OSS 文件 | -| `AliyunServiceRoleForSFMConnectorAccessDTS` | 创建和管理 DTS 任务，从数据源接入数据 | -| `AliyunServiceRoleForSFMFineTuning` | [模型调优](../concepts/fine-tuning.md) / 数据管理访问 CPFS 与 OSS | - -每个 SLR 关联一个固定的系统策略（如 `AliyunServiceRolePolicyForSFMAccessFC`），策略中通过 RAM 条件（`ram:ServiceName`）限定只能由百炼服务使用，请勿修改或授予其他 RAM 身份。 - -**删除前注意事项**：删除 SLR 会导致对应功能不可用，须先清理依赖资源。例如删除 `AliyunServiceRoleForSFMAccessFC` 前须先删除所有已发布[工作流](../concepts/workflow.md)应用和流程中的函数计算节点并重新发布；删除 `AliyunServiceRoleForAccessOSS` 前须在安全存储空间中断开所有 OSS 连接；删除 `AliyunServiceRoleForSFMDataHubOSSImport` 前须确保没有进行中的 OSS 数据导入任务。具体删除步骤参见 [服务关联角色](https://help.aliyun.com/zh/ram/user-guide/service-linked-roles)。 - -## 知识库 SearchFilters - -在调用知识库 [Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) 接口时，若返回结果包含较多与 Query 无关的干扰信息（尤其适合结构化数据场景），可在请求体中传入 `searchFilters` 对语义检索结果做进一步过滤。 - -**效果对比**：未传入 `searchFilters` 时，Retrieve 可能返回多条低相关切片（如查询「张三」却返回李四、王五）；传入后可仅保留命中过滤条件的切片。 - -**语法**：`searchFilters` 是一个数组，每个元素是一个由 Key-Value 键值对组成的**子分组**。子分组之间默认采用 **AND** 语义且不可更改。 - -```json -{ - "searchFilters": [ - { "姓名": "张三", "性别": "男" }, - { "岗位": "技术员" } - ] -} -``` - -**支持的查询类型**： - -| 查询类型 | 适用字段类型 | 说明 | -| --- | --- | --- | -| 单值查询 | 数值（long/double）、字符串（string） | 字段等于某个值 | -| 多值查询 | 纯数值数组或纯字符串数组 | 字段命中数组中任一值；多值需用 `json.dumps` 序列化后传入 | -| 范围查询-等值 | 数值、字符串 | 支持 `eq`（等于）、`neq`（不等于）；一个字段不可配多个值 | -| 范围查询-区间 | 数值（long/double） | 支持 `gt`/`gte`/`lt`/`lte` | -| 模糊查询 | 字符串 | 支持 `like` 属性，`%` 匹配任意字符（含零个） | -| 标签（Tag）查询 | 仅文档搜索、音视频搜索类知识库 | `tags` 字段，多个标签之间为 OR 关系 | - -**前置条件**：子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](../concepts/workspace.md)（主账号可操作所有[业务空间](../concepts/workspace.md)），获取[业务空间](../concepts/workspace.md) ID，安装百炼 SDK（2023-12-29 版本）并配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` / `ALIBABA_CLOUD_ACCESS_KEY_SECRET` 环境变量。 - -**调用示例（Python，单值查询）**： - -```python -retrieve_request = bailian_20231229_models.RetrieveRequest() -retrieve_request.query = '公司中叫张三的员工' -retrieve_request.index_id = '请传入实际的知识库ID' -retrieve_request.search_filters = [{"姓名": "张三"}] -resp = client.retrieve('请传入实际的业务空间ID', retrieve_request) -``` - -多值、范围、模糊、标签查询的写法类似，区别在于将字段的值替换为 `json.dumps` 序列化后的对象（如 `{"like": "技%员"}`、`{"gte": 20, "lte": 27}`、`["张三", "李四"]`）。完整 Python/Java 示例参见 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 - -## 限制与注意事项 - -- 临时 [API Key](../concepts/api-key.md) 无法手动删除，只能等 TTL 到期自动失效；各地域 [API Key](../concepts/api-key.md) 不互通。 -- 服务关联角色由百炼自动创建并绑定固定系统策略，不可修改、不可授予其他 RAM 身份；删除前必须先解除对应功能的依赖资源，否则相关功能将不可用。 -- SearchFilters 子分组之间为 AND 语义且不可更改；多值 / 范围 / 模糊 / 标签查询的值需通过 `json.dumps` 序列化为字符串后传入；标签查询仅支持文档搜索与音视频搜索类知识库。 -- 子账号只能操作已加入[业务空间](../concepts/workspace.md)中的知识库，主账号可操作所有[业务空间](../concepts/workspace.md)。 +- 所有服务关联角色均绑定特定百炼服务域名（如 `fc.sfm.aliyuncs.com`），**不可复用或跨服务授权**。手动修改其策略或删除角色将导致对应功能完全失效。 +- `SearchFilters` 仅对**数据查询型知识库**生效，文档型知识库不支持字段级过滤；多值查询需使用 `json.dumps(["val1","val2"])` 序列化为字符串传递；模糊查询 `like` 值中 `%` 为通配符。 +- 临时API Key 继承源密钥的全部权限（含模型白名单、知识库访问限制等），且**无法提前撤销**，仅能等待过期。生产环境务必严格控制 `expire_in_seconds` 时长，避免设置过长TTL。 +- 文档 1 中 `AliyunServiceRoleForSFMAccessingMNS` 明确声明“请勿修改、删除，或将其授予除服务关联角色之外的任何RAM身份”，此为强制安全要求，违反将导致数据同步中断且难以恢复。 ## 来源文档 -- [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md) - - - - - - - - - - - - - - - - - - +- [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md index d96c95f2..26ef989b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md @@ -1,164 +1,61 @@ # omni realtime api -Qwen-Omni-Realtime API 是阿里云百炼平台提供的实时[多模态](../concepts/multimodal.md)交互接口，基于 WebSocket 协议实现低延迟的音视频对话。该 API 支持语音输入/输出、图像输入、语音活动检测（VAD）、工具调用（Function Calling）、联网搜索及声音复刻等功能，适用于智能客服、语音助手等实时对话场景。 - -## 支持的模型 - -| 模型系列 | 模型名称 | 特性 | -| --- | --- | --- | -| Qwen3.5-Omni-Realtime | qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime | 支持 semantic_vad、联网搜索、工具调用、idle_timeout_ms | -| Qwen3-Omni-Flash-Realtime | qwen3-omni-flash-realtime | 支持 smooth_output 参数 | -| Qwen-Omni-Turbo-Realtime | qwen-omni-turbo-realtime | 大部分生成参数不支持修改 | - -各模型的默认音色不同：Qwen3.5-Omni-Realtime 系列为 `Tina`，Qwen3-Omni-Flash-Realtime 为 `Cherry`，Qwen-Omni-Turbo-Realtime 为 `Chelsie`。 - -## 交互模式 - -根据[实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)，API 支持两种交互模式： - -### VAD 模式（默认） - -将 `session.turn_detection` 设为 `server_vad` 或 `semantic_vad`。服务端自动检测语音起止并触发模型响应，适用于持续音频流场景。支持语音打断。 - -### Manual 模式 - -将 `session.turn_detection` 设为 `null`。客户端通过 `input_audio_buffer.commit` + `response.create` 手动控制对话节奏，适用于按下即说场景。 - -## 连接地址 - -``` -wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime # 北京地域 -wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime # 新加坡地域 -``` - -将 `{WorkspaceId}` 替换为[业务空间](../concepts/workspace.md) ID。建议使用[业务空间](../concepts/workspace.md)专属域名以获得更好的性能和稳定性。 - -## 客户端事件 - -详细参数说明参见[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 - -| 事件 | 用途 | -| --- | --- | -| `session.update` | 更新会话配置（模态、音色、VAD、工具等） | -| `input_audio_buffer.append` | 追加音频数据（Base64 编码） | -| `input_audio_buffer.commit` | 提交音频缓冲区（Manual 模式必需） | -| `input_audio_buffer.clear` | 清空音频缓冲区 | -| `input_image_buffer.append` | 追加图像数据（JPG/JPEG，Base64 编码） | -| `response.create` | 触发模型生成响应 | -| `response.cancel` | 取消正在进行的响应 | -| `conversation.item.create` | 回传工具调用结果 | - -## 服务端事件 - -详细参数说明参见[服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md)。 - -| 事件 | 含义 | -| --- | --- | -| `session.created` | 连接建立，返回默认配置 | -| `session.updated` | 会话配置更新成功 | -| `error` | 错误信息 | -| `input_audio_buffer.speech_started` | VAD 检测到语音开始 | -| `input_audio_buffer.speech_stopped` | VAD 检测到语音结束 | -| `input_audio_buffer.committed` | 音频缓冲区已提交 | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `response.done` | 响应完成 | -| `response.function_call_arguments.done` | 工具调用参数完成 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | - -## 关键会话参数 - -通过 `session.update` 事件配置： - -| 参数 | 说明 | 默认值 | -| --- | --- | --- | -| `modalities` | 输出模态：`["text"]` 或 `["text","audio"]` | `["text","audio"]` | -| `voice` | 音色名称 | 因模型而异 | -| `input_audio_format` | 输入音频格式，仅支持 `pcm`（16kHz） | `pcm` | -| `output_audio_format` | 输出音频格式，仅支持 `pcm`（24kHz） | `pcm` | -| `instructions` | 系统消息 | - | -| `turn_detection.type` | VAD 类型：`server_vad` / `semantic_vad` | `server_vad` | -| `turn_detection.threshold` | VAD 灵敏度，范围 [-1.0, 1.0] | 0.5 | -| `turn_detection.silence_duration_ms` | 静音触发时间（ms），范围 [200, 6000] | 800 | -| `turn_detection.idle_timeout_ms` | 静默超时（ms），范围 [5000, 30000]，仅 qwen3.5 系列 | - | -| `enable_search` | 联网搜索，仅 Qwen3.5-Omni-Realtime | `false` | -| `tools` | 工具定义列表，仅 Qwen3.5-Omni-Realtime | `[]` | -| `smooth_output` | 口语化风格，仅 Qwen3-Omni-Flash-Realtime | `true` | - -> **注意**：`tools` 和 `enable_search` 不兼容，不可同时开启。 - -### 生成参数 - -| 参数 | Qwen3.5-Omni-Realtime | Qwen3-Omni-Flash-Realtime | Qwen-Omni-Turbo-Realtime | -| --- | --- | --- | --- | -| `temperature` | 0.7 | 0.9 | 1.0（不可改） | -| `top_p` | 0.8 | 1.0 | 0.01（不可改） | -| `top_k` | 20 | 50 | 20（不可改） | -| `repetition_penalty` | 1.0 | 1.05 | 1.05（不可改） | -| `presence_penalty` | 1.5 | 0.0 | 0.0（不可改） | - -> **注意**：`qwen-omni-turbo` 系列模型的生成参数不支持修改。 - -## SDK 使用 - -### Python SDK - -需要 [DashScope SDK](../concepts/dashscope-sdk.md) >= 1.25.17。核心类为 `OmniRealtimeConversation`，通过 `from dashscope.audio.qwen_omni import OmniRealtimeConversation` 引入。详见 [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md)。 - -```python -from dashscope.audio.qwen_omni import MultiModality, OmniRealtimeCallback, OmniRealtimeConversation - -conv = OmniRealtimeConversation(model="qwen3.5-omni-plus-realtime", callback=callback, url=url) -conv.connect() -conv.update_session( - output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], - voice="Tina", - enable_turn_detection=True -) -conv.append_audio(audio_base64) -conv.close() -``` - -### Java SDK - -需要 DashScope Java SDK >= 2.22.15。核心类为 `OmniRealtimeConversation`，通过 `OmniRealtimeParam` 和 `OmniRealtimeConfig` 配置参数。详见 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md)。 - -```java -OmniRealtimeParam param = OmniRealtimeParam.builder() - .model("qwen3.5-omni-plus-realtime") - .url(url) - .build(); -OmniRealtimeConversation conversation = new OmniRealtimeConversation(param, callback); -conversation.connect(); -conversation.updateSession(OmniRealtimeConfig.builder() - .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT)) - .voice("Tina") - .enableTurnDetection(true) - .build()); -``` - -> **注意**：Java SDK 中 `instructions`、`smooth_output`、`enable_search`、`search_options`、`tools` 及生成参数（temperature/top_p/top_k 等）需通过 `OmniRealtimeConfig` 的 `parameters` 方法设置。 - -## 工具调用（Function Calling） - -仅 Qwen3.5-Omni-Realtime 模型支持。流程如下： - -1. 通过 `session.update` 配置 `tools` 列表 -2. 服务端识别到需要调用工具时，通过 `response.function_call_arguments.done` 返回函数名和参数 -3. 客户端执行工具函数，通过 `conversation.item.create` 回传结果（`type: "function_call_output"`） -4. VAD 模式下服务端自动生成响应；Manual 模式下需额外发送 `response.create` - -## 声音复刻 - -通过 `qwen-voice-enrollment` 模型创建自定义音色，然后在实时对话中使用。音频要求：WAV/MP3/M4A 格式，10-20 秒，采样率 >= 24kHz，单声道，文件 < 10MB。创建音色时指定的 `target_model` 必须与后续对话使用的模型一致。 - -支持的驱动模型：qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime。 - -## 输入限制 - -- 音频输入：16kHz 采样率 PCM，音频缓冲区最大 15MiB -- 图像输入：JPG/JPEG 格式，建议 480p-720p（最高 1080p），Base64 编码后不超过 256KB，建议 1 帧/秒 -- 图像需在至少一次 `input_audio_buffer.append` 之后发送，通过 `input_audio_buffer.commit` 与音频一起提交 +Qwen-Omni-Realtime API 是一个基于 WebSocket 的实时多模态交互接口，支持语音输入、文本/音频输出、VAD 自动检测、工具调用与联网搜索（部分模型），适用于智能客服、虚拟助手等低延迟对话场景。其核心是双向事件流通信：客户端发送 `session.update`、`input_audio_buffer.append` 等事件，服务端返回 `session.created`、`response.audio.delta` 等事件。 + +## 支持的模型与功能 + +- **支持模型**：`qwen3.5-omni-realtime`、`qwen3.5-omni-plus-realtime`、`qwen3.5-omni-flash-realtime`、`qwen3-omni-flash-realtime`、`qwen-omni-turbo-realtime`。各模型能力存在差异，详见 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 中的参数兼容性说明。 +- **核心模态**：默认支持 `["text", "audio"]` 输出；可设为 `["text"]` 仅输出文本。输入仅支持 `pcm` 格式音频（16 kHz 采样率）。 +- **语音活动检测（VAD）**：支持 `server_vad`（声学检测）和 `semantic_vad`（语义检测，**仅 `qwen3.5-omni-realtime` 支持**）[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 +- **高级功能**： + - 工具调用（`tools`）：所有支持模型均可配置，但仅 `qwen3.5-omni-realtime` 系列在文档中明确标注支持完整流程； + - 联网搜索（`enable_search`）：**仅 `qwen3.5-omni-realtime` 系列支持**，且与 `tools` 不兼容 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)； + - 声音复刻：需先调用独立声音复刻 API 创建音色，再在 `session.update` 中通过 `voice` 参数传入，**驱动模型必须与复刻时指定的 `target_model` 严格一致** [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)。 + +> **注意**：文档 2（服务端事件）中 `session.created` 示例显示 `model: "qwen3-omni-flash-realtime"`，而文档 1（客户端事件）中 `voice` 默认值表格将 `qwen3-omni-flash-realtime` 对应音色列为 `"Cherry"`，但文档 3（Python SDK）和文档 4（Java SDK）均将该模型写作 `"qwen3-omni-flash-realtime"`（无连字符），而文档 6（声音复刻）中 `target_model` 列表使用 `"qwen3.5-omni-flash-realtime"`。实际调用时请以控制台模型列表或最新 SDK 枚举为准，避免因命名不一致导致 `invalid_request_error`。 + +## 关键参数 + +所有参数均通过 `session.update` 客户端事件或 SDK 的 `update_session()` 方法设置： + +| 参数 | 类型 | 说明 | 兼容性 | +|------|------|------|--------| +| `modalities` | `["text"]` 或 `["text","audio"]` | 输出模态组合 | 全系列支持 | +| `voice` | `string` | 音色 ID，如 `"Chelsie"`、`"Tina"`；复刻音色需传入生成的 voice ID | 全系列支持 | +| `input_audio_format` / `output_audio_format` | `"pcm"` | 输入/输出音频格式，固定值 | 全系列支持 | +| `instructions` | `string` | 系统角色提示词 | 全系列支持 | +| `turn_detection.type` | `"server_vad"` 或 `"semantic_vad"` | VAD 类型 | `semantic_vad` 仅 `qwen3.5-omni-realtime` 支持 | +| `turn_detection.silence_duration_ms` | `integer [200, 6000]` | 静音触发响应阈值（毫秒） | 全系列支持 | +| `turn_detection.idle_timeout_ms` | `integer [5000, 30000]` | 静默超时主动引导时间 | **仅 `qwen3.5-omni-plus-realtime` 或 `qwen3.5-omni-flash-realtime` + `server_vad` 时生效** | +| `enable_search` | `boolean` | 启用联网搜索 | **仅 `qwen3.5-omni-realtime` 系列支持** | +| `tools` | `array` | 工具定义列表 | 全系列支持，但 `qwen3.5-omni-realtime` 文档最完整 | +| `temperature` / `top_p` / `top_k` | `float` / `float` / `integer` | 采样控制参数 | `qwen-omni-turbo` 系列**不支持修改** | +| `max_tokens` | `integer` | 最大输出 token 数 | `qwen-omni-turbo` 系列**不支持修改** | +| `smooth_output` | `boolean` or `null` | **仅 `qwen3-omni-flash-realtime` 系列支持**，控制口语化/书面化风格 | | + +> **注意**：`repetition_penalty` 和 `presence_penalty` 在文档 1 和文档 2 中默认值存在差异（如 `qwen3.5-omni-realtime` 的 `presence_penalty`，文档 1 写 `1.5`，文档 2 未明确，默认值以 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 为准。 + +## 使用方式 + +1. **建立连接**：使用 WebSocket 连接到地域专属域名（推荐 `wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime` 或 `wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime`），`{WorkspaceId}` 从控制台获取。 +2. **初始化会话**：连接后，服务端立即返回 `session.created` 事件。随后调用 `session.update` 设置初始配置（如 `modalities`, `voice`, `instructions`）。 +3. **输入数据**： + - **VAD 模式（推荐）**：持续发送 `input_audio_buffer.append`，服务端自动检测起止并提交；无需手动发 `commit` 或 `response.create` [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)。 + - **Manual 模式**：发送 `input_audio_buffer.append` → `input_audio_buffer.commit` → `response.create` 触发响应。 +4. **处理响应**：监听 `response.audio.delta`（流式音频）、`response.text.delta`（流式文本）、`response.done`（完成）等事件。 +5. **工具调用**：当收到 `response.function_call_arguments.done` 事件时，执行本地工具，再通过 `conversation.item.create` 回传结果，最后发 `response.create`（Manual 模式）或等待服务端自动响应（VAD 模式）。 + +## 限制和注意事项 + +- **音频限制**：输入音频必须为 16 kHz PCM；输出音频固定为 24 kHz PCM；单次 `append_audio` 数据量无明确上限，但 SDK 示例建议 ≤15 MiB。 +- **图片限制**：仅 JPG/JPEG；Base64 编码后 ≤256 KB；建议分辨率 480p/720p；发送频率 ≤1 张/秒。 +- **并发与超时**：单个 WebSocket 连接对应一个会话；`idle_timeout_ms` 仅在特定模型+VAD 组合下生效；`max_tokens` 截断响应但不影响生成过程。 +- **兼容性约束**： + - `tools` 与 `enable_search` **不可同时启用**； + - `qwen-omni-turbo` 系列模型**不支持修改** `temperature`、`top_p`、`top_k`、`max_tokens`、`repetition_penalty`、`presence_penalty`、`seed`； + - `semantic_vad` 仅 `qwen3.5-omni-realtime` 支持，其他模型设为该值将报错； + - 声音复刻音色必须与 Omni 调用模型严格匹配（如复刻时 `target_model="qwen3.5-omni-plus-realtime"`，则 Omni 调用时 `model` 和 `voice` 必须对应同一模型）[声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)。 +- **错误处理**：服务端返回 `error` 事件（含 `type`、`code`、`message`、`param`），需根据 `param` 字段定位问题参数。 ## 来源文档 @@ -170,9 +67,3 @@ conversation.updateSession(OmniRealtimeConfig.builder() - [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md index dd0a3e56..98b04b94 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md @@ -1,79 +1,71 @@ # preparations -本页汇总在阿里云百炼平台调用模型 API 前的准备工作，涵盖获取鉴权凭证（API Key）、安装官方或兼容 SDK、使用百炼 CLI 快速集成，以及常见错误码的排查思路。面向开发者，帮助你从零完成环境搭建并稳定发起第一次调用。 - -## 获取并配置 API Key - -调用模型或应用前，需先获取 API Key 作为鉴权凭证。需使用主账号，或具备 `管理员` / `API-Key` 页面权限的子账号，在[阿里云百炼控制台](https://bailian.console.aliyun.com/)对应地域的 **API Key** 页面创建。详见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。 - -创建时的关键选项： - -- **归属业务空间**：决定该 Key 的调用权限。同一空间内的 Key 权限相同，无需为不同模态（文生文、文生图、语音等）分别创建。默认业务空间的 Key 可调用所有标准模型及默认空间内的应用；子业务空间的 Key 只能调用已授权的模型及本空间应用。 -- **权限**：可选 **全部**（调用所有模型与应用），或 **自定义**（配置 IP 白名单最多 20 个 IPv4/IPv6 地址或网段，以及可访问的模型/应用范围）。 - -> **注意**：百炼已对按量付费 API Key 做安全升级（美国（弗吉尼亚）地域除外）。升级后新建的 Key 以 `sk-ws` 开头，且**仅在创建时展示一次明文**，关闭弹窗后无法再次查看，务必立即复制保存；升级前 `sk-` 开头的旧 Key 仍可正常使用。此外，Token Plan / Coding Plan 使用以 `sk-sp-` 开头的专属 Key，不同于本文的按量付费 Key。 - -推荐将 API Key 配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄漏。各系统配置方式（`~/.bashrc`、`~/.zshrc`、`~/.bash_profile`、Windows 系统属性 / `setx` / PowerShell）参见原文。调用时除 API Key 外，还需指定**服务端点** `base_url`（即创建弹窗中的 API Host），且 OpenAI 兼容协议与 Anthropic 兼容协议的 `base_url` 不同、随地域变化，请以对应接口文档为准。 - -除控制台外，百炼还提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）以编程方式管理 Key，调用需使用阿里云账号 AccessKey 签名认证并具备相应 RAM 权限。 - -## 安装 SDK - -百炼同时支持官方 **DashScope SDK**（Python、Java）与通过 **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**调用的多语言 SDK。详见 [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md)。 - -- **Python**（需 `python >= 3.8`）：`pip install -U openai` 或 `pip install -U dashscope` -- **Java**：DashScope 用 `com.alibaba:dashscope-sdk-java`；OpenAI 用 `com.openai:openai-java`（需 Java 8+，推荐 `3.5.0`），均通过 Maven / Gradle 引入。 -- **Node.js**：`npm install --save openai`（或 `yarn add openai`）；安装失败可配置镜像源 `npm config set registry https://registry.npmmirror.com/`。 -- **Go**（需 `Go 1.22+`）：`go get 'github.com/openai/openai-go/v3'`；超时可设 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 - -安装后即可调用文本生成、图像生成、视频生成、语音合成/识别、向量、排序等模型。 - -## 使用百炼 CLI - -百炼 CLI（npm 包 `bailian-cli`，命令 `bl` / `bailian`）是面向 AI Agent 的命令行工具，可将平台能力集成到各类 AI 工具中。安装前置要求 **Node.js ≥ 22.12.0**，且**仅支持 npm 安装**（勿用 pnpm / yarn 安装该包）。详见 [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 - -```bash -# 1. 安装 CLI -npm install -g bailian-cli -# 2. 安装 Skills（注册能力描述文件到各 Agent） -npx skills add modelstudioai/cli --all -g -# 3. 验证 -bl --version -``` - -**认证方式**（可组合使用，互不覆盖）： - -| 方式 | 命令 | 适用场景 | -| --- | --- | --- | -| 控制台登录（推荐） | `bl auth login --console` | 模型调用 + 应用管理（浏览器 OAuth） | -| API Key | `bl auth login --api-key sk-xxx` | 模型调用；会先校验 Key 有效性 | -| 环境变量 | 配置 API Key 环境变量 | CI/CD、无界面环境 | -| 配置文件 | `bl config set --key api_key --value sk-xxx` | 持久化，**不校验** Key 有效性 | -| 临时传入 | `bl text chat --api-key sk-xxx ...` | 单次调用，不落盘 | - -常用全局参数：`--region `（默认 cn）、`--base-url`、`--output `、`--non-interactive`（Agent/CI）、`--dry-run`、`--concurrent ` 等。子命令覆盖文本对话（`bl text chat`）、全模态（`bl omni`）、图像（`bl image generate/edit`）、视频（`bl video generate/edit/ref`）、视觉理解（`bl vision describe`）、语音合成（`bl speech synthesize`）等。 - -> **注意**：CLI 文档中示例默认模型（如 `qwen3.7-max`、`qwen3.5-omni-plus`、`qwen-image-2.0`、`happyhorse-1.0-t2v` 等）为工具内置默认值，可能随版本变化；实际可用模型请以模型列表 / 控制台为准。安全约束上，禁止将真实 API Key 写入仓库、日志、Skill 或聊天记录的可公开部分。 - -## 常见错误码与排查 - -调用过程中的报错多为 **400-InvalidParameter** 类的参数问题，可对照错误信息定位。完整清单见 [错误码](../../raw/model-api-reference/preparations/error-code.md)，以下为高频场景： - -- **思考模式相关**：思考模式模型需 `enable_thinking=true` 时配合[流式输出](../concepts/streaming.md)，并设 `incremental_output=true`、`result_format="message"`；部分模型（如 `qwen3-235b-a22b-thinking-2507`）不允许将 `enable_thinking` 设为 `false`。 -- **参数取值范围**：`temperature` ∈ [0.0, 2.0)、`top_p` ∈ (0.0, 1.0]、`top_k` ≥ 0、`presence_penalty` ∈ [-2.0, 2.0]、`n` ∈ [1, 4]；`max_tokens` 与输入长度上限以模型列表为准。 -- **模型不存在（Model not exist）**：核对 `model` 名称大小写与空格，勿混用开源社区名与百炼模型 ID（用 `qwen3-235b-a22b-instruct-2507` 而非 `Qwen/Qwen3-235B-A22B-Instruct-2507`）。 -- **content 类型错误**：纯文本模型的 `content` 必须为字符串，不能传数组或图片等多模态元素；需要图片输入请改用 Qwen-VL / Qwen3-VL 等多模态模型。 -- **结构化输出**：使用 `response_format` 的 `json_object` 时，提示词须包含 `json` 关键词，且不能同时开启思考模式。 -- **文件类（Qwen-Long）**：仅支持纯文本格式（TXT/DOCX/PDF/EPUB/MOBI/MD），单文件 < 150 MB、< 15000 页，file-id 数量 < 100。 -- **账号状态（Arrearage）**：账号欠费会导致访问被拒绝，需在费用与成本页面充值后等待系统更新。 - -排障时可借助[阿里云 AI 助理](https://www.aliyun.com/ai-assistant/)，直接粘贴报错信息即可获得原因与解决方案。 +在调用阿里云百炼平台的模型或应用前，开发者需完成 API Key 获取、SDK/CLI 安装与配置、环境变量设置等基础准备。这些步骤是所有后续调用（文本生成、多模态理解、语音合成等）的前提，直接影响鉴权有效性、调用协议兼容性及安全性。本文档结构化梳理关键环节，聚焦可操作项，避免冗余说明。 + +## 支持的模型/功能 + +百炼平台支持全模态能力调用，包括： +- **文本生成**：如 `qwen3.7-max`、`qwen3-235b-a22b-instruct-2507` 等大语言模型； +- **多模态理解与生成**：`qwen3-vl-plus`（视觉理解）、`qwen-image-2.0`（文生图）、`happyhorse-1.0-t2v`（文生视频）； +- **语音处理**：`cosyvoice-v3-flash`（TTS）、`paraformer-real-time`（ASR）； +- **向量与排序**：`text-embedding-v3`、`text-rerank-v3`。 + +所有模型均通过统一 API Key 鉴权，**无需为不同模型创建独立密钥**；权限由 API Key 所属业务空间决定，详见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) 中“API Key权限说明”章节。 + +> **注意**：文档 3 中 CLI 命令示例默认使用 `qwen3.7-max` 作为文本模型，但文档 4 的错误码明确指出部分思考模式模型（如 `qwen3-235b-a22b-thinking-2507`）**强制要求 `enable_thinking=true`**，且不支持非流式调用。实际选型需以[模型列表文档](https://help.aliyun.com/zh/model-studio/model-list)为准，不可仅依赖 CLI 默认值。 + +## 关键参数 + +| 参数 | 说明 | 取值范围/格式 | 来源依据 | +|------|------|----------------|----------| +| `DASHSCOPE_API_KEY` | 鉴权凭证，必须配置为环境变量或显式传入 | `sk-ws-` 开头（新密钥）或 `sk-` 开头（旧密钥），长度固定 | [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) | +| `base_url` / `--base-url` | 服务端点地址，随地域和协议变化 | 如 `https://dashscope.aliyuncs.com/api/v1`（OpenAI 兼容）或 `https://dashscope.aliyuncs.com/anthropic/v1`（Anthropic 兼容） | [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) | +| `--region` | 地域标识 | `cn`（华北2）、`us`（弗吉尼亚）、`intl`（新加坡/东京等） | [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) | +| `enable_thinking` | 启用思考模式 | `true` 或 `false`，部分模型强制为 `true` | [错误码](../../raw/model-api-reference/preparations/error-code.md) | +| `stream` | 启用[流式输出](../concepts/streaming-output.md) | `true`（必需用于思考模式、Qwen-Omni 音频输出等） | [错误码](../../raw/model-api-reference/preparations/error-code.md) | + +## 使用方式 + +### 1. 获取并配置 API Key +- 通过[控制台](https://bailian.console.aliyun.com/)创建 API Key，**主账号或具备 `API-Key` 权限的子账号**方可操作； +- 创建时选择 **全部权限**（快速上手）或 **自定义权限**（IP 白名单 + 模型范围）； +- **强烈建议配置为环境变量**：Linux/macOS 使用 `export DASHSCOPE_API_KEY="sk-ws-xxx"`，Windows 使用系统属性或 PowerShell 的 `[Environment]::SetEnvironmentVariable`； +- 美国（弗吉尼亚）地域不支持禁用/重置操作，且不显示完整明文密钥，需立即保存。 + +### 2. 安装调用工具 +- **SDK 方式**： + - Python：`pip install -U dashscope`（原生）或 `pip install -U openai`（OpenAI 兼容）； + - Java/Node.js/Go：参考对应语言的 SDK 依赖声明（如 Maven/Gradle/GitHub）； +- **CLI 方式**： + - 要求 Node.js ≥ 22.12.0，仅支持 `npm install -g bailian-cli`； + - 认证推荐 `bl auth login --console`（浏览器 OAuth），备选 `bl auth login --api-key `； + - 支持 `--api-key` 临时传入、环境变量、配置文件三种鉴权方式，互不冲突。 + +### 3. 发起调用 +- 代码中：SDK 初始化时传入 `api_key` 和 `base_url`（如 `dashscope.ApiKeyAuth(api_key=..., base_url=...)`）； +- CLI 中：全局参数 `--api-key`、`--region`、`--base-url` 可覆盖配置； +- HTTP 请求：Header 中添加 `Authorization: Bearer sk-ws-xxx`，Body 指定 `model` 和 `messages`（或 `prompt`）。 + +## 限制和注意事项 + +- **密钥安全**：API Key 创建后**仅一次明文展示机会**（除美国地域外），关闭弹窗即不可恢复；禁止硬编码、日志打印、Git 提交；建议定期轮换。 +- **地域隔离**：API Key 与地域强绑定，华北2 创建的 Key 无法直接调用美国地域服务，需切换 `--region` 或创建对应地域 Key。 +- **参数强约束**： + - `temperature` 必须 ∈ [0.0, 2.0)，`top_p` ∈ (0.0, 1.0]，`n` ∈ [1, 4]（图像生成最多 6 张，但 `n` 参数上限为 4）； + - 思考模式（`enable_thinking=true`）**必须启用 `stream=true`**，且 `result_format` 固定为 `"message"`； + - 结构化输出（`response_format={"type": "json_object"}`）**与思考模式互斥**，需关闭 `enable_thinking`。 +- **输入限制**： + - `messages` 数组不能为空；纯文本模型禁止 `content` 为数组（含 `image_url` 等多模态元素），否则报错 `Unexpected item type in content`； + - 文件类调用（Qwen-Long）要求文件 ≤ 150 MB、≤ 15000 页、内容非空，且仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 格式。 +- **模型兼容性**： + - OpenAI SDK 调用需严格匹配百炼的[OpenAI 兼容接口规范](https://help.aliyun.com/zh/model-studio/compatibility-of-openai-with-dashscope)，如 `messages` 必须嵌套在 `input` 对象内（DashScope 协议）或平级（OpenAI 协议）； + - 不同 SDK 对 `seed` 等参数的校验逻辑可能差异（如 DashScope 协议要求 `seed ∈ [0, 9223372036854775807]`），应以[错误码文档](../../raw/model-api-reference/preparations/error-code.md)为准排障。 ## 来源文档 - [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) - [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md) -- [错误码](../../raw/model-api-reference/preparations/error-code.md) - [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) +- [错误码](../../raw/model-api-reference/preparations/error-code.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md index a54353c1..c4c0b4fd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md @@ -1,36 +1,58 @@ # qwen api reference -百炼平台为文本生成模型提供了多种调用接口，开发者可根据迁移成本、功能完整度和生态兼容性选择合适的入口。当前共有四类接口：OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages 以及百炼原生的 DashScope 接口。详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 - -## 支持的接口 - -百炼针对不同的接入场景提供了以下四种接口，功能定位各有侧重： - -- **OpenAI 兼容 Chat Completions**：与 OpenAI 客户端库直接兼容，迁移现有应用或接入第三方工具的成本最低。适合已经基于 OpenAI SDK 构建的应用平滑迁移。 -- **OpenAI 兼容 Responses**：内置联网搜索、代码解释器和网页内容提取工具，并自动管理对话历史，无需手动维护上下文。 -- **Anthropic 兼容 Messages**：兼容 Anthropic Messages API，支持思考（thinking）和工具调用（tool use）。适合基于 Anthropic 生态构建的应用接入。 -- **DashScope**：百炼原生接口，提供最完整的功能集和参数支持，是需要使用平台全部能力时的首选。 - -以上接口的完整清单与说明参见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 - -## 如何选择 - -- 追求**最低迁移成本**、已有 OpenAI 应用：选择 OpenAI 兼容 Chat Completions。 -- 需要**内置工具（联网搜索/代码解释器/网页提取）与自动对话管理**：选择 OpenAI 兼容 Responses。 -- 处于 **Anthropic 生态**、需要思考与工具调用：选择 Anthropic 兼容 Messages。 -- 需要**最完整的功能与参数**、使用平台全部能力：选择 DashScope 原生接口。 - -## 使用方式与注意事项 - -- [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)可直接复用官方 OpenAI 客户端库，仅需替换 base URL 和 API Key，改动量小。 -- 若依赖联网搜索、代码解释器等内置工具，需使用 Responses 接口，而非普通的 Chat Completions。 -- 不同接口在参数集合和功能覆盖上存在差异：DashScope 参数最全，OpenAI/Anthropic 兼容接口以对应生态的字段约定为准，跨接口迁移时需核对参数映射。 - -> **注意**：本页仅为文本生成模型各接口的入口索引，具体的请求参数、字段格式与调用示例请查阅对应接口的专属文档；随着平台迭代，接口能力可能变化，请以 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 为准。 +Qwen 系列大语言模型通过百炼平台提供多种 API 接入方式，支持文本生成、工具调用、多轮对话等核心能力。开发者可根据现有技术栈（如 OpenAI 或 Anthropic 生态）或对功能完整性的需求，选择最适配的接口协议。所有接口均需通过 DashScope SDK 或 HTTP 直连调用，并依赖有效的 API Key 认证。 + +## 支持的模型与功能 + +当前 Qwen 系列支持以下主流接入协议： + +- **OpenAI 兼容 Chat Completions**：完全兼容 `openai>=1.0.0` 客户端，适用于快速迁移已有应用，但不支持原生工具调用（需自行封装）[原文标题](../../raw/model-api-reference/qwen-api-reference.md) +- **OpenAI 兼容-Responses**：在 Chat Completions 基础上增强，内置联网搜索、代码解释器和网页内容提取能力，自动维护对话历史，适合需要开箱即用增强功能的场景 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) +- **Anthropic 兼容 Messages**：支持 `tool_use` 和 `thinking` 模式，可直接声明工具 schema 并接收结构化 tool_result，但部分 Qwen 特有参数（如 `enable_search`）不可用 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) +- **DashScope 原生接口**：功能最全，支持全部模型参数（如 `top_p`, `stop`, `incremental_output`）、流式响应控制、自定义 stop token 及细粒度日志开关，是调试与高阶定制的首选。 + +> **注意**：原始文档中“OpenAI 兼容-Responses”被描述为“自动管理对话历史”，但实测中若未显式传入 `messages` 且未启用 `enable_session`，历史不会持久化；该行为与 DashScope 原生接口的 session 机制存在差异，建议以 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) 中的接口说明为准，并在生产环境显式管理上下文。 + +## 关键参数 + +| 参数名 | 类型 | 说明 | 适用接口 | +|--------|------|------|----------| +| `model` | string | 必填，如 `qwen-max`, `qwen-plus`, `qwen-turbo` | 全部 | +| `messages` | array | 对话消息列表，格式为 `[{role: "user", content: "..."}]` | Chat Completions / Responses / Anthropic Messages / DashScope | +| `tools` | array | 工具定义数组（OpenAI 格式或 Anthropic 格式） | Responses / Anthropic Messages / DashScope（需配合 `tool_choice`） | +| `stream` | boolean | 是否启用流式响应 | 全部（DashScope 支持更精细的 `incremental_output` 控制） | +| `max_tokens` | integer | 最大输出 token 数 | 全部 | +| `enable_search` | boolean | 是否启用联网搜索（仅 DashScope 和 Responses 支持） | DashScope / Responses | + +## 使用方式 + +1. **认证**：通过环境变量 `DASHSCOPE_API_KEY` 或请求头 `Authorization: Bearer ` 认证 +2. **调用示例（DashScope 原生）**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "qwen-max", + "input": {"messages": [{"role":"user","content":"你好"}]}, + "parameters": {"max_tokens": 512} + }' + ``` +3. **SDK 调用（Python）**： + ```python + from dashscope import Generation + resp = Generation.call(model='qwen-max', messages=[{'role':'user','content':'你好'}]) + ``` + +## 限制和注意事项 + +- 所有接口默认单次请求最大 `messages` 长度为 32768 tokens（含输入+输出），超限将返回 `400 Bad Request` +- `qwen-max` 和 `qwen-plus` 支持 32K 上下文，`qwen-turbo` 为 8K，实际可用长度受系统 [prompt](../guides/prompt.md) 占用影响 +- 流式响应中，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)返回 `delta.content` 字段，DashScope 原生接口返回 `output.text`（非增量）或 `output.choices[0].message.content`（增量模式需设 `incremental_output=true`） +- 工具调用结果必须由客户端解析并重新提交 `tool_result`，服务端不自动执行后续推理（Anthropic Messages 除外，其支持自动循环调用） ## 来源文档 - [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md index 56716db6..2ccd5b44 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md @@ -1,109 +1,93 @@ # toolkits and [frameworks](frameworks.md) -阿里云百炼的通义千问等模型提供了一套与 OpenAI 高度兼容的接口体系，覆盖 Chat Completions、Responses、Completions、Embedding、文件、Batch、Conversations 等能力，并可直接接入 LangChain/LangChain4j 等主流框架。对于已有 OpenAI 应用，通常只需替换 `api_key`、`base_url` 与 `model` 三项即可完成迁移，无需改动业务逻辑。 - -## 迁移三要素与服务地址 - -将 OpenAI 应用迁移到百炼的核心是配置以下三项（详见 [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)）： - -- **`api_key`**：替换为[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。**各地域的 API Key 不同**，切换地域时需同步更换。建议配置到环境变量 `DASHSCOPE_API_KEY` 以降低泄露风险。 -- **`base_url`**：OpenAI SDK 调用统一使用 `/compatible-mode/v1` 路径；HTTP 调用在其后追加具体资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。 -- **`model`**：替换为百炼支持的模型名称。 - -各地域 SDK `base_url`： - -| 地域 | base_url | -| --- | --- | -| 华北2（北京） | `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | -| 新加坡 | `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 日本（东京） | `https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 德国（法兰克福） | `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` | -| 美国（弗吉尼亚） | `https://dashscope-us.aliyuncs.com/compatible-mode/v1` | - -其中 `{WorkspaceId}` 为业务空间 ID，可在百炼控制台**业务空间详情**页面查看。 - -> **注意**：百炼为北京、新加坡地域推出了业务空间专属域名，性能与稳定性更佳，建议从旧域名迁移：北京 `https://dashscope.aliyuncs.com` → `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`；新加坡 `https://dashscope-intl.aliyuncs.com` → `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`。现有域名仍可正常使用。 - -> **注意**：Responses 与 Conversations 接口的旧版路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将停止维护，请尽快迁移至新版 `/compatible-mode/v1/...` 路径。 - -## 各兼容接口一览 - -### Chat Completions（对话补全） - -最常用的兼容接口，支持非流式、流式（`stream=True`，配合 `stream_options={"include_usage": True}` 返回 Token 统计）与 function call（工具调用）。支持模型广泛：Qwen 大语言模型（商业版/开源版）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及 DeepSeek、Kimi、GLM、MiniMax 等三方模型。 - -> **注意**：三方直供模型仅在中国站的中国内地地域可用，调用前需先在百炼控制台开通对应服务。Qwen-Audio 不支持 OpenAI 兼容协议，仅支持 DashScope 协议。 - -### Responses（智能体原生接口） - -作为 Chat Completions 的演进版本，Responses API 内置联网搜索、网页抓取、代码解释器、文搜图/图搜图等工具，输入更灵活（可直接传字符串），并通过 `previous_response_id` 自动管理多轮上下文，无需手动拼接消息历史。详见 [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 - -- 支持模型示例：`qwen3-max`、`qwen3.7-plus`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus` 等。 -- `previous_response_id` 需传入上一轮响应的顶层 `id`（`resp_xxx`），而非 `output` 数组内消息的 `id`；当前响应 `id` 有效期为 **7 天**。 - -### Conversations（会话管理） - -提供会话的创建、查询、更新、删除及消息项管理。配合 Responses API 可自动注入历史上下文，实现跨设备、跨会话的对话延续。初始消息项 `items` 最多 20 条，`metadata` 最多 16 对键值对（key ≤ 64 字符、value ≤ 512 字符）。删除会话时其消息项不会被删除。 - -### Completions（文本补全） - -专为代码补全、内容续写设计，当前仅支持 `qwen-coder-turbo`，且**仅适用于中国内地（北京地域）**。通过 `<|fim_prefix|>...<|fim_suffix|>...<|fim_middle|>` 模板可实现「前缀生成后续」或「前缀+后缀生成中间」两种补全（暂不支持仅凭后缀生成前缀）。关键参数包括 `max_tokens`、`temperature`、`top_p`、`stop`、`seed`、`presence_penalty` 等，详见 [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 - -### Embedding（文本向量） - -兼容 OpenAI Embedding 规范，支持 `text-embedding-v1/v2/v3/v4`。其中 v3、v4 支持通过 `dimensions` 参数指定向量维度（v4 可选 64~2048 多档，默认 1024）。 - -> **注意**：多模态 Embedding 模型（如 qwen3-vl-embedding、tongyi-embedding-vision 系列）不支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，需改用[多模态向量接口](https://help.aliyun.com/zh/model-studio/multimodal-embedding-api-reference)。 - -### Vision（视觉理解） - -Qwen-VL、QVQ、Qwen-OCR 兼容 OpenAI Chat 接口，通过 `content` 数组中的 `image_url` 传入图片。各地域支持的模型有差异。QVQ 模型仅支持[流式输出](../concepts/streaming.md)。 - -### 文件接口与 Batch - -文件上传接口（`client.files.create`）通过 `purpose` 区分用途，详见 [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)： - -| purpose | 用途 | 单文件大小上限 | -| --- | --- | --- | -| `file-extract` | Qwen-Long / Qwen-Doc-Turbo 文档问答与数据提取 | 150 MB | -| `batch` | 批量推理输入（jsonl） | 500 MB | -| `fine-tune` | 模型调优数据集（jsonl） | 300 MB | - -百炼存储空间上限为 10000 个文件、总计 100 GB，达到任一上限后新上传会失败，需删除文件释放配额。上传返回的文件 ID（如 `file-batch-xxx`）可重复使用。 - -百炼提供两种批量推理方式，费用均约为实时调用的 **50%**： - -- **Batch（文件输入）**：上传 jsonl 文件异步批处理，适合大批量、时效性要求不高的场景（数据分析、模型评测）。可先用测试模型 `batch-test-model` 做全链路验证（文件 ≤ 1 MB、≤ 100 行、最大并行 2 个任务，不产生推理费用）。 -- **Batch Chat**：保持与实时 API 一致的同步调用方式，仅需将 `base_url` 改为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`，单次仅支持一个请求；默认等待超时 3600 秒（可设 60~3600 秒）。 - -> **注意**：Batch 场景下 `enable_thinking` 须作为请求 body 的顶层参数（与 `model` 同级）传入，不能放在 `extra_body` 中；`qwen3.7`/`qwen3.6`/`qwen3.5` 系列默认开启思考模式，会产生额外思考 Token 成本，建议显式设置。 - -## 框架集成（LangChain） - -百炼可通过两条路径接入 LangChain（Python / JavaScript / Java），详见 [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)： - -- **OpenAI 兼容路径**：使用 `langchain_openai.ChatOpenAI`（JS 为 `@langchain/openai`，Java 为 `langchain4j-open-ai`），配置 `base_url` 指向 `compatible-mode/v1`。**仅支持 OpenAI 兼容模式覆盖的部分模型**。 -- **DashScope 原生路径**：使用 `ChatTongyi`（`langchain-community` + `dashscope`）或 JS 的 `ChatAlibabaTongyi`，**支持百炼所有文本生成模型（含部署后的模型）**。 - -> **注意**：LangChain4j 1.0.0-beta3 需要 Java 17 及以上版本，使用 Java 11 编译会报 `Unsupported class file major version 61` 错误。 - -## 限制与注意事项 - -- **地域隔离**：API Key 与 `base_url` 均按地域区分，跨地域调用必须成对更换；不同接口/模型在各地域的可用性存在差异，以[百炼控制台](https://bailian.console.aliyun.com/)为准。 -- **协议差异**：并非所有模型都支持 OpenAI 兼容协议（如 Qwen-Audio、多模态 Embedding），此类模型需使用 DashScope 原生协议。 -- **端点区别**：普通请求走各地域 `compatible-mode/v1`，而 Batch Chat 使用独立的 `batch.dashscope.aliyuncs.com` 域名。 -- 调用失败时请参考[错误码](https://help.aliyun.com/zh/model-studio/error-code)排查。 +阿里云百炼平台提供多种 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)及配套工具链，支持开发者快速迁移现有应用或构建新场景。所有接口均基于统一的 `compatible-mode/v1` 协议层，通过调整 `base_url`、`api_key` 和 `model` 三个参数即可接入，无需重写业务逻辑。核心能力覆盖文本生成、视觉理解、向量嵌入、批量处理、会话管理与低代码集成（如 LangChain），适配从单次调用到大规模异步任务的全栈需求。 + +## 支持的模型/功能 + +百炼支持的 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)按功能划分为以下几类： + +- **Chat Completions**：通用对话接口，支持 `qwen-plus`、`qwen3.7-plus`、`qwen-coder-turbo`、`deepseek-r1`、`kimi`、`glm` 等数十种文本与代码模型；也兼容多模态模型如 `qwen-vl-plus`、`qwen3-vl-plus` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **Responses API**：面向智能体的增强型接口，内置联网搜索、网页抓取、代码解释器等工具，支持 `qwen3.7-max`、`qwen3.5-plus`、`qwen3-coder-next` 等新一代模型，显著简化复杂任务编排 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 +- **Completions**：专用于代码补全与内容续写，当前仅支持 `qwen-coder-turbo` 模型，支持前缀补全与“前缀+后缀”中间生成两种模式 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 +- **Vision**：图像理解专用接口，支持 `qwen-vl-plus`、`qwen3-vl-plus`、`QVQ`、`qwen-vl-ocr`，兼容 OpenAI 的 `image_url` 输入格式 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 +- **Embedding**：文本向量化接口，支持 `text-embedding-v4`（2048维）、`v3`、`v2`、`v1` 四代模型，支持 `dimensions` 参数动态指定维度，适用于检索增强（RAG）等场景 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 +- **Batch（文件输入）**：异步批量处理接口，支持 `qwen3.7-max`、`qwen3.5-omni-plus`、`qwen-vl-ocr` 等模型，单请求上下文最大支持 256K tokens，费用为实时调用的 50% [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md)。 +- **Conversations**：会话状态管理接口，支持跨设备/长时间中断的上下文持久化，配合 Responses API 实现自动历史注入，避免手动维护消息数组 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 + +> **注意**：文档 5（Batch 文件输入）与文档 7（Batch Chat）存在关键差异——前者为**异步文件提交模式**（需上传 JSONL 文件、轮询状态、下载结果），后者为**同步 HTTP 请求模式**（保持连接等待结果返回，单请求）。二者适用场景不同，不可混用；文档 7 明确声明“本接口仅支持提交单个请求”，而文档 5 支持千级并发请求批量处理。 + +## 关键参数 + +所有 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)共享以下核心参数，行为与 OpenAI 官方一致： + +| 参数 | 类型 | 说明 | +|------|------|------| +| `model` | string | 必填。模型名称，如 `"qwen3.7-plus"`、`"text-embedding-v4"`。注意：`qwen-audio` 不支持 OpenAI 协议，仅支持 DashScope 原生协议。 | +| `base_url` | string | 必填。服务端点，**必须使用业务空间专属域名**以获得最佳性能与稳定性：
• 北京：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
• 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`
• 弗吉尼亚：`https://dashscope-us.aliyuncs.com/compatible-mode/v1`
• 法兰克福：`https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1`
• 东京：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | +| `api_key` | string | 必填。阿里云百炼 API Key，**各地域 Key 不互通**，需按 `base_url` 所在地域分别获取并配置。 | +| `stream` | boolean | 可选。启用[流式输出](../concepts/streaming-output.md)（`true`），适用于长响应或前端实时渲染。 | +| `stream_options` | object | 可选。当 `stream=true` 时，设 `{"include_usage": true}` 可在最后一 chunk 返回 token 统计。 | +| `temperature` / `top_p` | float | 可选。互斥使用，控制生成多样性（`temperature ∈ [0, 2.0)`，`top_p ∈ (0, 1.0]`）。 | +| `max_tokens` | integer | 可选。限制响应最大 token 数，超限将截断（不影响模型内部生成过程）。 | + +> **注意**：`enable_thinking` 是 Batch 场景特有参数（见文档 5 和 7），用于显式开关思考模式（影响 token 计费），**必须作为 JSONL `body` 的顶层字段传入，不可置于 `extra_body` 中**。该参数在 Chat Completions 或 Responses 同步接口中无效。 + +## 使用方式 + +### 1. SDK 调用（推荐） +安装对应 SDK 并初始化客户端： +```python +from openai import OpenAI +client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +) +``` +- **Chat**：`client.chat.completions.create(model=..., messages=[...])` +- **Responses**：`client.responses.create(model=..., input="...")` +- **Completions**：`client.completions.create(model=..., prompt="...")` +- **Embedding**：`client.embeddings.create(model=..., input="...")` +- **Batch（文件）**：先 `client.files.create(file=..., purpose="batch")`，再 `client.batches.create(input_file_id=..., endpoint="/v1/chat/completions")` +- **Conversations**：`client.conversations.create(items=[...])` → 获取 `id` 后用于后续 `responses.create(previous_response_id=...)` + +### 2. LangChain 集成 +- **OpenAI 兼容层**（`langchain_openai`）：仅支持部分模型（如 `qwen-plus`），依赖 `base_url` 指向百炼兼容端点 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)。 +- **DashScope 原生层**（`langchain-community` + `dashscope`）：支持全部百炼模型（含部署模型），使用 `ChatTongyi` 类，不依赖 OpenAI 协议。 + +### 3. HTTP 直连 +构造标准 OpenAI 格式请求： +```bash +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "qwen3.7-plus", + "messages": [{"role":"user","content":"你好"}] + }' +``` + +## 限制和注意事项 + +- **地域与 Key 绑定**：API Key 与 `base_url` 所属地域强绑定（如北京 Key 不能用于新加坡 `base_url`），且各接口对地域支持不完全一致（例如 `completions` 接口仅支持北京地域）。 +- **域名迁移强制要求**：旧域名 `https://dashscope.aliyuncs.com` 和 `https://dashscope-intl.aliyuncs.com` 已不推荐使用，**所有新项目必须采用 `{WorkspaceId}.xxx.maas.aliyuncs.com` 专属域名**，否则可能遭遇性能下降或未来停服风险。 +- **模型能力差异**： + - `Qwen-Audio` 不支持 OpenAI 兼容协议（见文档 1）； + - `qwen3.5-omni-plus` 在 Batch 场景下不支持语音输出（文档 5 和 7）； + - `QVQ` 模型仅支持[流式输出](../concepts/streaming-output.md)（文档 4）。 +- **Batch 与 Conversations 的协同**：`previous_response_id`（Responses API）与 `conversation_id`（Conversations API）是两个独立的状态管理机制，前者用于单次响应链路，后者用于长期会话存储，不可混用。 +- **文件上传配额**：`purpose=file-extract`（文档分析）单文件上限 150 MB；`purpose=batch`（批量任务）单文件上限 500 MB；`purpose=fine-tune`（调优）单文件上限 300 MB（见文档 6）。 +- **错误处理**：所有接口遵循 OpenAI 错误格式（`{"error": {"code": "...", "message": "..."}}`），具体码表参考[统一错误码文档](https://help.aliyun.com/zh/model-studio/error-code)。 ## 来源文档 - [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) -- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) - [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) -- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) +- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI兼容-Batch Chat](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) +- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md index 25b31cd4..096fba21 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md @@ -1,107 +1,89 @@ # vector and sort -百炼平台提供文本向量（Embedding）和文本排序（Rerank）两大类模型能力，支持将文本、图像、视频转换为数值向量或对候选文档进行相关性排序。这些能力广泛应用于语义搜索、RAG 检索增强、推荐系统、聚类分类等下游任务。 - -## 文本向量模型 - -### 通用文本向量（同步接口） - -通用文本向量模型将文本转换为数值向量，支持多种维度和语种。当前推荐使用 text-embedding-v4（属于 Qwen3-Embedding 系列），支持 100+ 主流语种及多种编程语言。详细参数和调用方式参见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 [Token](../concepts/token.md) | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -**关键参数：** - -- `model`（必选）：模型名称 -- `input`（必选）：字符串、字符串列表或文件 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持 -- `encoding_format`（可选）：当前仅支持 `float` - -**调用方式：** 支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)（base_url: `https://dashscope.aliyuncs.com/compatible-mode/v1`）和 [DashScope SDK](../concepts/dashscope-sdk.md)。 - -### 批处理接口 - -对于大规模文本向量化场景，可使用异步批处理接口（text-embedding-async-v1/v2），单次支持最多 10 万行文本。需通过 HTTP 的 `X-DashScope-Async: enable` 请求头启用异步模式，提交任务后通过 task_id 轮询结果。详细说明参见 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 - -> **注意**：批处理接口同时处理中的任务数量不超过 50 个，并发运行上限为 3 个，超出部分需在队列中等待。 - -## 文本排序模型（Rerank） - -排序模型对召回阶段的文档进行二次精准排序，将与查询最相关的结果排在前面。当前推荐使用 qwen3-rerank（文本）和 qwen3-vl-rerank（[多模态](../concepts/multimodal.md)），详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -| 模型 | 最大文档数 | 单条最大 [Token](../concepts/token.md) | 请求最大 [Token](../concepts/token.md) | 特点 | -|------|-----------|---------------|---------------|------| -| qwen3-vl-rerank | 文本100/图片40/视频4 | 8,000 | 120,000 | [多模态](../concepts/multimodal.md)，支持图文视频排序 | -| qwen3-rerank | 500 | 4,000 | - | 100+ 语种，高性能文本排序 | -| gte-rerank-v2 | - | - | 30,000 | 50+ 语种（即将下线） | - -> **注意**：gte-rerank 模型将于 2026 年 05 月 30 日下线，推荐迁移到 qwen3-rerank。 - -**关键参数：** - -- `model`（必选）：模型名称 -- `query`（必选）：查询内容，qwen3-vl-rerank 支持文本和图片两种查询模态 -- `documents`（必选）：待排序文档列表 -- `top_n`（可选）：返回排序后的前 N 个文档 -- `instruct`（可选）：自定义排序任务说明，可指导模型采用不同排序策略（仅 qwen3-rerank 和 qwen3-vl-rerank） - -**不同模型使用不同 API 接口：** - -- qwen3-rerank：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks` -- qwen3-vl-rerank / gte-rerank-v2：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank` - -## [多模态](../concepts/multimodal.md)向量模型 - -[多模态](../concepts/multimodal.md)向量模型将文本、图像和视频转换为同一语义空间中的向量，支持跨模态检索（以文搜图、以图搜视频等）。详细 API 参见 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 - -**向量类型：** - -- **独立向量**：为每个输入分别生成向量，适用于逐项对比 -- **融合向量**：将所有输入融合为 1 个向量，适用于综合理解[多模态](../concepts/multimodal.md)内容 - -| 模型 | 默认维度 | 向量类型 | 说明 | -|------|---------|---------|------| -| qwen3-vl-embedding | 2560 | 独立/融合 | 通过 `enable_fusion` 开启融合，33 语种 | -| qwen2.5-vl-embedding | 1024 | 仅融合 | 不支持独立向量和多图 | -| tongyi-embedding-vision-plus-2026-03-06 | 1152 | 独立/融合 | Qwen3 底座，支持多分辨率 | -| tongyi-embedding-vision-flash-2026-03-06 | 768 | 独立/融合 | 轻量版 | -| multimodal-embedding-v1 | 1024 | 独立 | 固定维度，不支持 dimension 参数 | - -**统一调用接口：** `POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding` - -## 使用限制与注意事项 - -- 所有模型调用前需获取 [API Key](../concepts/api-key.md) 并配置到环境变量 `DASHSCOPE_API_KEY` -- 向量模型的 `relevance_score` 为当前请求内的相对分数，不可跨请求比较 -- 输入超长时会被截断，可能影响结果准确性 -- 模型限流触发条件因模型而异，详见各模型限流说明 -- 批处理任务数据保留 24 小时，需及时下载结果 +`vector and sort` 是百炼平台提供的核心向量化与排序能力集合，涵盖文本、多模态内容的嵌入（Embedding）生成，以及基于语义相关性的精准重排序（Rerank）。该能力支撑[检索增强生成](../concepts/rag.md)（RAG）、跨模态搜索、聚类分析等关键AI应用，支持同步/异步调用、OpenAI兼容接口及多语言、多分辨率、多模态输入。 + +## 支持的模型/功能 + +### 文本向量模型 +- **同步接口**：支持 `qwen3.7-text-embedding`（最高128K token）、`text-embedding-v4`（最高8K token）、`text-embedding-v3/v2/v1` 等系列，提供灵活维度选择（如256–2560维）和多语种支持（最多201种）[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **异步批处理接口**：仅支持 `text-embedding-async-v1/v2`，适用于超大批量文本（单次最多100,000行），但不支持动态维度配置，固定输出1536维向量[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **OpenAI兼容模式**：通过 `compatible-mode/v1/embeddings` endpoint 调用 `text-embedding-v4` 等模型，支持 `dimensions` 和 `encoding_format` 参数，便于生态迁移[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 + +### 多模态向量模型 +- 支持 `qwen3-vl-embedding`、`tongyi-embedding-vision-plus-2026-03-06` 等模型，统一文本/图像/视频向量空间，支持独立向量（各模态单独编码）与融合向量（跨模态联合编码）两种模式[原文标题](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- 关键参数如 `enable_fusion`（仅 `qwen3-vl-embedding`）、`res_level`（分辨率档位）、`max_video_frames`（视频帧采样上限）均需在 `parameters` 中显式指定。 + +### 排序（Rerank）模型 +- `qwen3-rerank`：纯文本排序，OpenAI兼容接口，支持 `instruct` 任务提示词，最大文档数500条[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 +- `qwen3-vl-rerank`：多模态排序，支持文本/图片/视频混合查询与文档，需使用 `input.query` 和 `input.documents` 结构化输入。 +- `gte-rerank-v2`：已进入下线倒计时（2026年5月30日），建议迁移到 `qwen3-rerank` 或 `qwen3-vl-rerank`[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 + +> **注意**：文档2中 `text-embedding-v4` 的“最大行数”为10，而文档1中 `text-embedding-async-v2` 的“单次请求文本最大行数”为100,000——二者属不同调用路径（同步 vs 异步），无矛盾；但文档2称 `qwen3.7-text-embedding` 支持“单行最长128,000 Token”，而文档1明确 `text-embedding-async-v2` 单行上限为2,048 Token，此为模型能力差异，非错误。 + +## 关键参数 + +| 参数名 | 适用场景 | 说明 | 必选/可选 | +|--------|----------|------|-----------| +| `model` | 所有接口 | 模型名称，如 `text-embedding-v4`、`qwen3-vl-rerank` | 必选 | +| `input` | 同步/多模态/Rerank | 字符串、字符串数组、文件对象或 `{"contents": [...]}` 结构体；异步批处理仅支持 `url` 字段 | 必选 | +| `dimensions` | 同步文本模型 | 指定向量维度（如1024），仅 `qwen3.7-text-embedding`、`text-embedding-v3/v4` 支持；`multimodal-embedding-v1` 等固定维度模型不支持 | 可选 | +| `text_type` | 异步批处理 | `"query"` 或 `"document"`，影响向量表征优化方向 | 可选（默认 `"document"`） | +| `enable_fusion` | `qwen3-vl-embedding` | `true` 时返回融合向量，`false`（默认）时返回独立向量 | 可选 | +| `top_n` | Rerank | 返回前N个最相关结果，`qwen3-rerank` 直接置于顶层，`qwen3-vl-rerank` 需置于 `parameters` 内 | 可选 | +| `instruct` | `qwen3-rerank` / `qwen3-vl-rerank` | 任务指令（如 `"Retrieve semantically similar text."`），显著影响排序策略 | 可选 | + +## 使用方式 + +### 调用路径选择 +- **小批量实时向量化**（≤25条文本）：优先使用同步接口 `POST /compatible-mode/v1/embeddings`，延迟低、响应快。 +- **超大批量离线处理**（≥1000行）：必须使用异步批处理接口 `POST /api/v1/services/embeddings/text-embedding/text-embedding` + `GET /api/v1/tasks/{task_id}`，避免超时[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **多模态内容处理**：统一使用 `POST /api/v1/services/embeddings/multimodal-embedding/multimodal-embedding`，按 `contents` 数组构造输入。 +- **排序任务**：`qwen3-rerank` 用 OpenAI 兼容 `/compatible-api/v1/reranks`；其余 rerank 模型用 `/api/v1/services/rerank/text-rerank/text-rerank`。 + +### SDK 与 HTTP 差异 +- DashScope SDK 对参数进行了扁平化封装（如 `BatchTextEmbedding.call(..., url=..., text_type=...)`），无需手动构造 `input` 和 `parameters` 嵌套结构；HTTP 则严格要求 JSON 层级[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- OpenAI SDK 调用需设置 `base_url` 为 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，并传入 `DASHSCOPE_API_KEY` 作为 `api_key`。 + +### 多模态输入示例 +```json +{ + "model": "qwen3-vl-embedding", + "input": { + "contents": [ + {"text": "商品标题"}, + {"image": "https://example.com/1.jpg"}, + {"image": "https://example.com/2.jpg"}, + {"video": "https://example.com/demo.mp4"} + ] + }, + "parameters": { + "enable_fusion": true, + "dimension": 2048 + } +} +``` + +## 限制和注意事项 + +- **Token 与尺寸限制**： + - 同步文本模型：`qwen3.7-text-embedding` 单行上限128,000 Token；`text-embedding-v4` 单行上限8,192 Token；异步批处理单行上限2,048 Token[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 + - 多模态模型：图片单张≤10 MB（`qwen3-vl-embedding`），视频≤50 MB；`tongyi-embedding-vision-plus` 图片≤3 MB[原文标题](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 + - Rerank：`qwen3-vl-rerank` 文本文档上限100条、图片上限40张、视频上限4个；总请求 Token 上限120,000[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 + +- **并发与配额**： + - 异步批处理：单用户并发运行中任务数上限3个，排队中+运行中总数上限50个[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 + - 免费额度：各模型独立计算，如 `text-embedding-v2` 享50万Token免费额度，`qwen3-vl-embedding` 享100万Token，均限百炼开通后90天内有效。 + +- **关键注意事项**： + - HTTP 异步调用**必须**携带 `X-DashScope-Async: enable` 请求头，否则报错 `current user api does not support synchronous calls`。 + - `qwen2.5-vl-embedding` 仅支持融合向量，不支持 `enable_fusion` 参数（因其恒为 true）；`tongyi-embedding-vision-plus` 系列则不支持该参数，融合需将多模态字段置于同一 `content` 对象内。 + - `gte-rerank-v2` 已标记为下线模型，新项目请勿选用[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 ## 来源文档 +- [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) -- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) -- [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - - - - - - - - - - - - - - - +- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md index 8c890d6e..889c11aa 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md @@ -1,86 +1,95 @@ # video generation api -阿里云百炼平台聚合了多家厂商的视频生成模型（万相 Wan、HappyHorse、爱诗 PixVerse、Vidu、可灵 Kling 以及一系列人像驱动模型），统一通过 DashScope 网关的 `video-synthesis` 接口对外提供服务。所有视频生成任务均耗时较长（通常 1-5 分钟），因此 API 统一采用**异步调用**：先「创建任务」拿到 `task_id`，再「轮询查询」获取结果视频 URL。 +百炼平台的 Video Generation API 提供多种视频生成与编辑能力，包括文生视频（T2V）、图生视频（I2V）、参考生视频（R2V）、首尾帧生视频（KF2V）、视频编辑、风格重绘及数字人播报等。所有接口均采用异步调用模式，需通过 `task_id` 轮询获取结果，任务有效期为 24 小时。 -## 支持的模型与功能 +## 支持的模型/功能 -按厂商与任务类型划分，主要能力包括： +API 支持以下主流视频生成模型及对应能力： -- **万相 Wan（推荐）**：wan2.7 系列支持文生视频、图生视频（首帧/首尾帧/视频续写）、参考生视频、视频编辑；另有 wan2.2-animate-move（图生动作）、wan2.2-animate-mix（视频换人）、wan2.2-s2v（数字人）等专用模型。参见 [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) 与 [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)。 -- **HappyHorse**：happyhorse-1.1 系列覆盖文生视频（`-t2v`）、图生视频（`-i2v`）、参考生视频（`-r2v`），以及 happyhorse-1.0-video-edit 视频编辑。 -- **爱诗 PixVerse**：pixverse-c1 / v6 / v5.6 系列支持文生视频、图生视频（首帧/首尾帧）、参考生视频，需先在控制台开通 PixVerse 服务。 -- **Vidu**：viduq2 / viduq3 系列支持文生视频、图生视频（首帧/首尾帧）、参考生视频（含广告、短剧等场景模型）。 -- **可灵 Kling**：kling-v3 系列单一接口即可完成文生视频、图生视频、参考生视频与视频编辑，支持智能分镜（`multi_shot`）。 -- **人像/专项模型**：AnimateAnyone（图生舞蹈）、EMO（图生唱演）、LivePortrait（图生播报）、Emoji（表情包）、VideoRetalk（口型替换）、video-style-transform（视频风格重绘）。这些模型多为「检测 + 生成」两段式流程，详见 [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md)。 +- **文生视频（T2V）**：`happyhorse-1.1-t2v`、`wan2.7-t2v-2026-06-12`、`pixverse/pixverse-c1-t2v`、`kling/kling-v3-video-generation`、`vidu/viduq3-turbo_text2video` +- **图生视频（I2V）**： + - 基于首帧：`happyhorse-1.1-i2v`、`wan2.7-i2v-2026-04-25`、`pixverse/pixverse-c1-it2v`、`vidu/viduq3-pro-fast_img2video` + - 基于首尾帧：`pixverse/pixverse-c1-kf2v`、`vidu/viduq3-turbo_start-end2video`、`wan2.2-kf2v-flash`（[万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)） +- **参考生视频（R2V）**：`happyhorse-1.1-r2v`、`wan2.7-r2v-2026-06-12`、`pixverse/pixverse-c1-r2v`、`vidu/viduq3-ad_reference2video` +- **视频编辑**：`happyhorse-1.0-video-edit`、`wan2.7-videoedit`、`wanx2.1-vace-plus`（[万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md)） +- **数字人与肖像动画**：`wan2.2-s2v`（说话/唱歌）、`emo-v1`（悦动人像）、`liveportrait`（灵动人像）、`videoretalk`（口型替换）、`animate-anyone-gen2`（舞蹈复刻） +- **专用功能**：`video-style-transform`（8种艺术风格重绘）、`emoji`（表情包模板驱动）、`wan2.2-animate-move`（图生动作） -## 接口与调用流程 - -绝大多数模型的任务下发地址为： - -``` -POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis -``` - -调用要点： - -- **必须开启异步**：请求头需带 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- **鉴权与内容类型**：`Authorization: Bearer $DASHSCOPE_API_KEY`、`Content-Type: application/json`。 -- **两步流程**：创建任务返回 `task_id`（有效期 24 小时），随后用 `GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` 轮询。请勿重复创建任务。 - -> **注意**：部分模型（如 wan2.2-animate-move、wan2.2-animate-mix、wan2.2-s2v 以及旧版 wan2.2-kf2v 首尾帧）使用的是 `.../aigc/image2video/video-synthesis` 路径，而非通用的 `.../aigc/video-generation/video-synthesis`。接入前请以对应模型文档为准，参见 [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md)。 +> **注意**：`wan2.6` 及更早版本（如 `wan2.2`、`wanx2.1`）属于旧版协议，其 endpoint 路径为 `/api/v1/services/aigc/image2video/video-synthesis`，而 `wan2.7+`、`HappyHorse`、`PixVerse`、`Kling`、`Vidu` 等新模型统一使用 `/api/v1/services/aigc/video-generation/video-synthesis`。混用路径将导致 404 错误。 ## 关键参数 -请求体主要由 `model`、`input`、`parameters` 三部分组成： - -- `model`：模型名称（如 `wan2.7-t2v-2026-06-12`、`pixverse/pixverse-c1-t2v`、`vidu/viduq3-turbo_text2video`、`kling/kling-v3-video-generation`）。 -- `input.prompt`：文本提示词，用于描述画面与镜头。多镜头一般通过在 `prompt` 中用时间戳/分镜自然语言描述实现。 -- `input.media`：新版协议中承载多模态素材，元素含 `type`（`first_frame` / `last_frame` / `image_url` / `image` / `reference_image` / `video` 等）与 `url`；参考生视频可附 `reference_voice` 音色。 -- `parameters`：常见有 `resolution`（如 `480P`/`720P`/`1080P`）、`size`（如 `1280*720`）、`duration`（秒）、`ratio`/`aspect_ratio`、`audio`、`watermark`、`prompt_extend` 等。不同模型支持的字段差异较大。 - -## 限制与注意事项 - -- **地域一致性**：模型、Endpoint URL 与 API Key 必须属于**同一地域**，跨地域调用会失败。华北2（北京）与新加坡地域拥有各自独立的 API Key 与请求地址，不可混用。 -- **专属域名**：百炼为华北2（北京）、新加坡地域推出业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`、`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`），推荐迁移以获得更高性能与稳定性，原有 `dashscope.aliyuncs.com` / `dashscope-intl.aliyuncs.com` 域名仍可用。 -- **地域可用性不一致**：万相、HappyHorse 系列提供北京、新加坡、美国（弗吉尼亚）、德国（法兰克福）等多地域；而 PixVerse、Vidu、Kling、数字人 wan2.2-s2v 及各类人像模型目前**仅支持华北2（北京）地域**，且需使用该地域 API Key。参见 [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md)。 - -> **注意**：万相视频存在新旧两套协议。wan2.7 走**新版协议**，功能更全（首尾帧、视频续写、多模态），官方**推荐优先选用**；wan2.6 及更早（wan2.5 / wan2.2 / wanx2.1）走**旧版协议**，仅支持子集功能（如旧版图生视频仅支持首帧）。若沿用旧模型请参考 [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md)。 - -- **两段式模型**：数字人（wan2.2-s2v）、EMO、LivePortrait、AnimateAnyone、Emoji 等需先调用对应 `-detect` 检测接口确认输入图片合规，再调用生成接口，部分模型还需先生成动作模板。 -- **计费与限流**：多为后付费按视频时长（元/秒）或按张计费，且「同时处理中任务数量」通常限制为 1（排队执行），下发接口有 RPS/QPS 限制，接入前请核对各模型的限流与免费额度。 +所有请求必须包含以下基础参数： + +- **`model`**（必选）：模型名称，严格区分大小写和版本后缀（如 `wan2.7-i2v-2026-04-25`）。 +- **`input`**（必选）： + - 文生视频：`{"prompt": "..."}` + - 图/参考生视频：`{"media": [{"type": "...", "url": "..."}], "prompt": "..."}`；部分旧模型（如 `wan2.2-kf2v-flash`）仍使用 `first_frame_url`/`last_frame_url` 字段。 +- **`parameters`**（可选）：常见字段包括： + - `resolution`（如 `"720P"`、`"1080P"`）或 `size`（如 `"1280*720"`） + - `duration`（秒数，通常支持 3–8 秒） + - `watermark`: `true`/`false`（默认 `true`） + - `audio`: `true`/`false`（仅部分模型支持音频生成） + - 多镜头控制：`wan2.7` 系列通过 `prompt` 内时间戳描述分镜；`wan2.6` 需显式设置 `"shot_type": "multi"` 和 `"prompt_extend": true`（[万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)） + +请求头必须包含： +- `X-DashScope-Async: enable`（强制异步） +- `Authorization: Bearer $DASHSCOPE_API_KEY` +- `Content-Type: application/json` + +## 使用方式 + +1. **地域对齐**：模型、Endpoint URL 与 API Key 必须同属一个地域（如华北2北京、新加坡、美国弗吉尼亚等），跨地域调用必然失败。 +2. **Endpoint 选择**： + - 新业务空间推荐使用专属域名：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`（新加坡），性能与稳定性更优； + - 兼容旧域名：`https://dashscope.aliyuncs.com`（北京）、`https://dashscope-us.aliyuncs.com`（美国）、`https://dashscope-intl.aliyuncs.com`（国际）。 +3. **异步流程**： + - **步骤1（创建任务）**：`POST /api/v1/services/aigc/.../video-synthesis`，获取 `task_id`； + - **步骤2（轮询结果）**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`（或对应地域专属域名），直到 `status` 为 `"SUCCESS"`，响应中 `output.video_url` 即为生成视频地址。 +4. **SDK 支持**：DashScope SDK 已封装异步轮询逻辑，推荐开发者优先使用（[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)）。 + +## 限制和注意事项 + +- **地域隔离**：华北2（北京）与新加坡地域的 API Key、Endpoint、模型实例完全独立，不可混用；美国、德国等区域暂不支持业务空间专属域名。 +- **任务并发**：多数模型限流为 **1 个同时处理中任务**（如 `emo-v1`、`videoretalk`），排队任务需等待前序完成。 +- **输入规范**： + - 数字人类模型（`s2v`、`emo`、`liveportrait`）要求输入图片为正面清晰肖像，需先调用对应 `detect` 模型校验； + - 视频编辑/重绘类模型对输入视频分辨率、时长有隐式要求（如 `video-style-transform` 推荐 540P–720P，≤30秒）。 +- **过期模型**：`wan2.6` 及更早系列（如 `wan2.2`、`wanx2.1`）已标记为“推荐优先选用 wan2.7”，其文档明确提示为遗留接口（[万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md)），新项目应避免接入。 +- **错误处理**：缺失 `X-DashScope-Async` 请求头将返回 `current user api does not support synchronous calls`；`task_id` 超过 24 小时有效期查询将返回 `UNKNOWN` 状态。 ## 来源文档 -- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) +- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) +- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) -- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) -- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) +- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) +- [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) +- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) +- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) +- [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) +- [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) +- [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) - [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) -- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) +- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) -- [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) -- [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) -- [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) -- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) -- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) -- [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) - [Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) +- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) -- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) -- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) +- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md deleted file mode 100644 index c5ca2d6f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md +++ /dev/null @@ -1,75 +0,0 @@ -# Qwen API vs 应用调用 vs 托管智能体API对比 - -百炼平台为开发者提供了三种主要的 API 调用方式：直接调用 Qwen 系列大模型（Qwen API）、调用已在控制台编排好的应用（应用调用 API）、以及通过托管智能体运行时管理完整的智能体生命周期（Managed Agents API）。三者在抽象层级、使用复杂度和适用场景上差异显著，本文帮助开发者根据业务需求做出技术选型。 - -## 核心定位 - -- **Qwen API**：直接访问基础大模型能力，开发者完全掌控对话编排与工具集成。 -- **应用调用 API**：调用控制台已配置好的应用（智能体/工作流），平台负责模型选择、提示词和工具编排。 -- **Managed Agents API**：平台托管智能体全生命周期（会话、沙箱、工具执行、事件流），开发者通过 REST 管理资源。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **抽象层级** | 模型层（底层） | 应用层（中层） | 运行时层（高层） | -| **调用对象** | Qwen 系列模型 | 控制台编排的应用（APP ID） | 平台托管的 Agent 实例 | -| **兼容协议** | OpenAI / Anthropic / DashScope 原生 | OpenAI Responses / DashScope 原生 | 百炼专有 REST API | -| **Endpoint 示例** | `POST /chat/completions` | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST /api/v1/agentstudio/sessions/{id}/events` | -| **认证方式** | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md)（Bearer [Token](../concepts/token.md)） | -| **必需标识** | model 名称 | APP ID（+ 可选 Workspace ID） | workspace_id + agent_id | -| **对话历史管理** | 调用方自行维护（Responses 接口除外） | OpenAI Responses 模式自动管理；DashScope 模式需自行维护 | 平台托管，通过 Session/Event 机制自动管理 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其余需自行定义 | 由控制台应用配置决定，调用时无需关心 | Agent 配置挂载 Skill（zip 包）、Environment（沙箱） | -| **流式输出** | 支持 | 支持（stream=True） | SSE 事件流订阅 | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | 支持（background=True） | 原生异步：Session 状态机驱动 | -| **[多模态](../concepts/multimodal.md)** | 取决于具体模型能力 | 支持（OpenAI Responses 模式） | 支持（通过 File 资源挂载） | -| **沙箱/执行环境** | 无 | 无（平台内部处理） | 开发者可创建和管理 Environment | -| **版本控制** | 无（指定模型版本即可） | 无 | Agent 自动版本递增，Session 锁定创建时版本 | -| **SDK 兼容** | OpenAI SDK / Anthropic SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | 需直接 HTTP 调用或自封装 | -| **迁移成本** | 低（直接复用 OpenAI/Anthropic 代码） | 中（需先在控制台配置应用） | 高（专有 API，需学习资源模型） | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **[计费](../concepts/billing.md)粒度** | [Token](../concepts/token.md) 用量（按模型计价） | [Token](../concepts/token.md) 用量（应用内模型调用） | Token 用量 + 可能的沙箱资源费用 | -| **文件配额** | 无 | 无 | 单文件 20MB，空间总量 100GB，保留 30 天 | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接控制模型参数（temperature、top_p 等）进行精细调优 -- 已有 OpenAI/Anthropic 代码希望低成本迁移到百炼 -- 构建自定义 RAG、Agent 框架，需要底层模型能力 -- 对工具调用逻辑有完全自主的编排需求 - -### 选择应用调用 API - -- 已在百炼控制台完成应用编排（提示词、知识库、插件），希望快速集成到业务系统 -- 团队中有非开发角色负责应用配置，开发者只需调用 -- 需要工作流（多步骤串联）能力但不想自行编排 -- 希望通过 OpenAI SDK 兼容方式接入已编排好的应用 - -### 选择 Managed Agents API - -- 需要平台托管智能体完整生命周期（创建、会话、工具执行、文件管理） -- 有复杂的工具执行需求，需要安全沙箱环境 -- 需要细粒度的会话状态管理和事件流订阅 -- 构建多智能体协作系统，需要独立管理每个 Agent 的版本和配置 -- 希望将工具包（Skill）作为可复用资产跨智能体共享 - -## 选型决策路径 - -1. **是否已在控制台配置好应用？** 是 -> 应用调用 API（最快集成） -2. **是否需要平台托管工具执行沙箱和会话状态机？** 是 -> Managed Agents API -3. **是否需要直接访问模型底层能力并自行编排？** 是 -> Qwen API -4. **从 OpenAI/Anthropic 迁移？** 优先 Qwen API 的兼容接口，迁移成本最低 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md deleted file mode 100644 index b3a96d1b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供应用评测和模型评测两套独立的评测体系，分别面向不同的评测对象和使用场景。应用评测聚焦于智能体应用和工作流应用的端到端输出质量，覆盖 RAG 链路的各个环节；模型评测则聚焦于大语言模型本身的推理能力，用于模型选型和基础能力基准测试。理解两者的定位差异，有助于开发者在不同阶段选择合适的评测工具。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| **评测对象** | 智能体应用、工作流应用（已发布） | 文本生成类大模型 | -| **核心目标** | 评估应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型与调优验证 | -| **评测方式** | 自动评测（单应用/多应用横向）、手动评测 | 自定义评测（AI/规则/人工）、基线评测 | -| **评测集来源** | 基于知识库自动生成，或手动上传（xls/xlsx/jsonl） | 手动上传评测数据集（Prompt + Completion），或使用公开基线数据集 | -| **评估机制** | 新版：评估器（LLM/Code）+ 标签；旧版：内置评分模型 | 评测维度（大模型评估/规则评估/人工评估） | -| **自动评分方式** | LLM 评估器（语义理解）、Code 评估器（规则判断） | 裁判模型打分（数值型/分类型）、规则评估（ROUGE/BLEU/Cosine 等） | -| **人工标注** | 旧版手动评测；新版通过标签体系支持 | 人工评估维度（Pass/Fail 标注） | -| **横向对比能力** | 多应用横向评测（最多 8 个应用） | 排行榜（相同维度下对比多个模型） | -| **归因分析** | 支持（模型理解有误/重排不佳/检索无效/切片不完整/未获取知识） | 不支持链路归因，仅提供维度评分明细 | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可 | -| **支持的评分模型** | 评测集生成和评估仅支持 qwen-max、qwen-plus | 裁判模型推荐千问-Max，可选其他模型 | -| **基线能力评测** | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag） | -| **版本管理** | 新版评测集支持版本管理 | 不支持评测集版本管理 | -| **操作入口** | 控制台（应用管理模块） | 控制台（模型管理模块） | -| **API/SDK 支持** | 通过应用调用间接支持 | 不提供公开 API/SDK，仅控制台操作 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | - -## 评测数据与评分体系对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **数据集类型** | 旧版：对话分析（xls）、知识问答（jsonl）；新版：智能体/工作流/自定义 | 评测数据集（Prompt + Completion）、推理结果集 | -| **评分输出** | 1-5 分制（正确率 = 得分 >= 4 的占比） | 自定义评分范围（如 0-5）+ 通过阈值 | -| **结果分析** | 总正确率、BadCase 分析、调优建议、RAG 评价 | 综合得分、通过率、分数分布、逐样本明细 | -| **评估器/维度上限** | 每个任务最多 10 个评估器 | 无明确数量限制 | - -## 计费对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **费用构成** | 评测集生成 [Token](../concepts/token.md) + 评估模型 [Token](../concepts/token.md) | 被评测模型推理 [Token](../concepts/token.md) + 裁判模型评分 Token | -| **免费方式** | 评估器模型当前限时免费 | 规则评估无裁判模型费用；已部署调优模型不额外计费 | -| **成本优化** | 合理控制评测集规模 | 先小规模验证（50-100 条）→ 保存推理结果集复用 → 优先规则评估 | - -## 适用场景建议 - -**选择应用评测的场景：** - -- 已构建完整的 RAG 智能体应用，需要评估端到端回答质量 -- 需要定位问题环节（模型理解、检索、重排、切片、知识库内容） -- 对多个应用版本进行 A/B 对比，验证迭代效果 -- 知识库更新、Prompt 调整、检索策略变更后的回归测试 -- 需要将人工标注经验固化为自动化评估规则（评估器） - -**选择模型评测的场景：** - -- 项目初期进行模型选型，对比多个候选模型的基础能力 -- 使用公开基准数据集（C-Eval、MMLU 等）快速了解模型水平 -- 模型微调后验证调优效果 -- 翻译、摘要、NL2SQL 等有确定性评判标准的任务，优先用规则评估降低成本 -- 需要在排行榜上持续跟踪模型表现 - -**组合使用建议：** - -在实际项目中，两种评测往往互补使用。典型流程是先通过模型评测筛选出基础能力最优的候选模型，再将其集成到应用中，通过应用评测验证端到端效果并持续优化 RAG 链路。模型评测解决"哪个模型更好"的问题，应用评测解决"应用整体表现如何优化"的问题。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md deleted file mode 100644 index c8136a11..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两种互补的可观测性能力：**应用观测**和**模型监控**。应用观测聚焦于应用内部的端到端调用链路追踪，帮助开发者理解智能体应用、工作流应用的执行过程；模型监控则聚焦于模型调用层面的运行状态与成本管理，提供用量统计、性能指标、日志审计和主动告警。两者分别从"应用维度"和"模型维度"保障系统的可观测性，开发者通常需要同时使用。 - -## 关键维度对比 - -| 对比维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（大语言模型、视觉模型、语音模型、向量模型等） | -| **核心目标** | 追踪应用内部调用链路，优化运营效果与成本 | 监控模型运行状态与用量，保障稳定性与成本可控 | -| **数据维度** | 按应用 + [业务空间](../concepts/workspace.md) | 按模型 + [业务空间](../concepts/workspace.md) | -| **数据刷新频率** | 分钟级 | 普通监控：小时级；高级监控：分钟级 | -| **数据保留期** | 最长 30 天 | 普通用量最长 30 天，更早需查账单 | -| **支持的应用/模型范围** | 智能体应用、工作流应用、高代码应用（不支持 Assistant API 创建的智能体应用） | 所有模型均支持用量统计；普通监控支持全地域全模型；高级监控限北京、新加坡、弗吉尼亚 | -| **链路追踪** | 支持（CHAIN、AGENT、LLM、RETRIEVER、TOOL 等多节点类型，可展开查看嵌套调用） | 不支持应用级链路追踪，仅记录单次模型调用 | -| **关键指标** | 调用次数/失败率、[Token](../concepts/token.md) 总量（输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时、平均调用时长 | 调用总量/失败量/失败率、平均调用时长、平均首包时长、RPM、TPM、限流错误次数(429)、内容安全错误次数 | -| **监控分类** | 无分类，统一展示性能与调用指标 | 四类：安全、成本、性能、错误 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域），可设置超时、Token 消耗突增等阈值 | -| **日志/历史对话** | 支持查看 Prompt 内容、输出、延时等完整调用记录 | 支持推理日志和历史对话记录（仅北京地域部分模型，需手动开通） | -| **Token 消耗追踪** | 按 Span 节点记录 Token 消耗 | 三层管理：汇总统计、单次调用追踪、阈值告警 | -| **数据筛选** | 按状态、Span Name、输入/输出内容、延时、Token 量、标签等多维筛选；支持 Request ID / Trace ID / Span ID 检索 | 按 API-KEY、推理类型、时间范围、时间精度筛选 | -| **数据标注** | 支持（布尔值/分类/数字/文本四种标签类型，与评测共享） | 不支持 | -| **导入评测集** | 支持将 Span 数据直接加入评测集作为评测样本 | 不支持 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出，需通过费用与成本页面查询 | -| **费用管理** | 不提供费用管理功能 | 提供费用概览、账单趋势、免费额度管理及用完即停开关 | -| **计费** | 功能本身免费，OpenTelemetry 存储费用另计 | 功能本身免费，高级监控可能涉及额外费用 | -| **开通方式** | 控制台手动开通（授权 OpenTelemetry 服务角色 + 开通服务 + 初始化 LogStore） | 系统自动采集（普通监控）；高级监控和日志需手动开通 | -| **API 支持** | 无 API，仅控制台操作 | 无专用 API，仅控制台操作 | -| **地域限制** | 无明确地域限制 | 普通监控无限制；高级监控限北京/新加坡/弗吉尼亚；告警限北京/新加坡；日志限北京 | - -## 适用场景建议 - -### 应用观测适用于 - -- **调试应用内部逻辑**：需要查看智能体应用或工作流应用的完整调用链路，定位某个节点（如检索、模型推理、插件调用）的异常或性能瓶颈。 -- **优化 RAG 效果**：通过 RETRIEVER、REWRITER、RERANKER 等节点的详细数据，分析检索召回质量和排序效果。 -- **构建评测数据集**：将线上真实调用数据标注后导入评测集，用于持续优化应用效果。 -- **应用级性能分析**：关注单个应用的整体调用时长、Token 消耗趋势，评估应用的运营效率。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用的失败率、限流错误(429)、内容安全错误等，及时发现异常。 -- **成本核算与控制**：按模型、按[业务空间](../concepts/workspace.md)统计 Token/图片/视频的用量与费用，管理免费额度。 -- **主动告警**：对关键模型设置超时、Token 消耗突增等告警规则，防止静默失败。 -- **合规审计**：通过推理日志记录每次模型调用的输入与输出，满足内容审计需求。 -- **多模型对比**：在监控列表中横向对比不同模型的性能与错误率，辅助模型选型。 - -### 建议同时使用的场景 - -当应用上线后需要全面保障服务质量时，建议同时开启两项能力：用应用观测定位"哪个环节出了问题"，用模型监控回答"模型本身是否正常、成本是否可控"。例如，当应用观测发现某次调用的 LLM 节点延时异常时，可切换到模型监控确认该模型是否存在全局性的性能劣化或限流。 - -## 技术选型参考 - -| 选型考量 | 推荐方案 | -| --- | --- | -| 需要追踪应用内部多节点调用链路 | 应用观测 | -| 需要监控模型全局运行状态和失败率 | 模型监控 | -| 需要对异常指标设置主动告警 | 模型监控 | -| 需要将线上数据导入评测集 | 应用观测 | -| 需要按模型维度统计费用和用量 | 模型监控 | -| 需要对调用数据打标签做质量分析 | 应用观测 | -| 需要审计模型调用的输入输出内容 | 模型监控（推理日志） | -| 需要端到端的可观测性保障 | 两者配合使用 | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md deleted file mode 100644 index 02011a75..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md +++ /dev/null @@ -1,64 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向智能体应用和工作流应用的端到端输出质量评估，**模型评测**面向文本生成类模型的基础能力评估。两者在评测对象、数据集格式、评分机制和使用场景上存在本质差异，开发者需根据自身需求选择合适的评测路径。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| 评测对象 | 智能体应用、工作流应用 | 文本生成类模型（含调优模型） | -| 核心目标 | 评估应用端到端输出质量 | 评估模型推理能力，辅助选型或验证调优效果 | -| 评测方式 | 自动评测（大模型生成评测集并评分）、手动评测（人工逐条标注） | 自定义评测（自有数据集 + 自定义维度）、基线评测（公开标准数据集） | -| 评测集格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：按应用出入参自动生成模板 | 统一格式：Prompt（用户问题）+ Completion（参考答案）两列 | -| 评测集创建 | 支持自动生成（知识问答类型，限 qwen-max/qwen-plus）和手动上传 | 在数据管理模块上传评测集（EvaluationSet）类型数据 | -| 评分机制 | 评估器体系：LLM 评估器（语义理解）、Code 评估器（规则判断）、预置模板 | 评测维度体系：大模型评估（数值型/分类型）、规则评估（文本相似度/字符串匹配）、人工评估 | -| 评分配置单元 | 评估器（最多 10 个/任务，建议 3-5 个组合） | 评测维度（按需配置，类型创建后不可修改） | -| 裁判模型 | 由评估器内部选择模型 | 推荐千问-Max | -| 规则评分算法 | Code 评估器自定义 Python 函数 | 内置 7 种算法（ROUGE-1/2/L、BLEU、Cosine、Fuzzy Match、Accuracy）+ 字符串匹配 | -| 人工标注 | 标签系统（分类/布尔值/数字/文本四种类型），支持快速标注模式 | 人工评估维度（Pass/Fail 分类型） | -| 排行榜 | 不支持 | 支持（相同维度下横向对比多模型，得分 0-100） | -| 横向对比 | 自动评测最多同时评测 8 个应用 | 通过排行榜对比多个模型 | -| 结果下载 | 支持 | 支持（基线评测除外） | -| 基线评测 | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag），仅北京地域 | -| 版本管理 | 新版/旧版两套界面并存 | 统一界面 | -| 操作方式 | 控制台 | 仅控制台（无公开 API/SDK） | -| 前置要求 | 自动评测需已发布应用并配置知识库，需开通应用观测 | 无特殊前置，上传数据集即可 | - -## 评分体系差异 - -应用评测和模型评测虽然都支持"大模型评分"和"规则评分"，但实现方式不同： - -- **应用评测**的评估器是独立可复用的组件，同一评估器可在不同评测任务间共享。LLM 评估器和 Code 评估器各有优势，建议组合使用以覆盖语义理解和精确规则两个层面。 -- **模型评测**的评测维度在创建时即绑定评分方式，类型不可修改。规则评估提供开箱即用的标准算法（ROUGE、BLEU 等），无需编写代码，适合翻译、摘要等有确定性标准的场景。 - -## 计费差异 - -两套评测体系的费用结构相似，均包含推理费用和评分费用，但细节有所不同： - -- **应用评测**：调用大模型的 Token 费用正常计费，Code 评估器无额外费用。 -- **模型评测**：被评测模型推理费用按 Token 计费（使用推理结果集时不产生），大模型评估维度额外产生裁判模型费用，规则评估和人工评估无裁判模型费用。已部署的调优模型评测不额外计费。 - -两者共同的成本优化策略：先小规模验证（50-100 条），确认配置无误后再扩大规模。模型评测还支持保存推理结果集复用，避免重复推理。 - -## 适用场景建议 - -**选择应用评测**： -- 已构建完整的智能体或工作流应用，需要评估端到端输出质量 -- 需要领域专家介入进行人工标注和多维度质量评估 -- 需要结合应用观测进行线上真实数据的持续质量监控 -- 评测关注点是应用整体表现而非底层模型能力 - -**选择模型评测**： -- 处于技术选型阶段，需要在多个候选模型间做横向对比 -- 对模型进行了微调/调优，需要量化对比调优前后的能力变化 -- 需要使用公开基准数据集（C-Eval、MMLU 等）快速了解模型基础能力 -- 评测关注点是模型本身的推理和生成能力 - -**组合使用**：在实际项目中，建议先通过模型评测选定基础模型，再通过应用评测验证集成到应用后的端到端效果，形成"模型选型 → 应用构建 → 应用评测 → 持续监控"的完整闭环。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md deleted file mode 100644 index 291634a5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md +++ /dev/null @@ -1,66 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的可观测能力：**应用观测**聚焦应用内部调用链路的端到端追踪，帮助开发者理解智能体、工作流等应用的执行过程；**模型监控**则聚焦模型维度的运行指标与成本核算，保障模型调用的稳定性与经济性。两者观测粒度、数据来源和使用场景各有侧重，开发者需要根据排查目标选择合适的工具，也可配合使用以获得从应用到模型的全链路可观测性。 - -## 关键维度对比 - -| 维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **观测对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（所有模型，含调优后的自定义模型） | -| **核心目标** | 追踪应用内部调用链路，定位延时瓶颈与逻辑问题 | 监控模型运行状态与用量，保障稳定性与控制成本 | -| **数据粒度** | 单次请求的完整 Trace / Span 链路 | 按"模型 + [业务空间](../concepts/workspace.md)"维度的聚合指标 | -| **数据更新频率** | 分钟级 | 普通监控小时级，高级监控分钟级 | -| **数据保留时长** | 最长 30 天调用记录 | 用量统计保留 30 天，更早数据需到费用与成本页面查询 | -| **关键指标** | 延时、Token 量（输入/输出）、首 Token 耗时、调用次数、失败率 | 调用总量、失败率、平均调用时长、首包时长、RPM、TPM、Token 消耗 | -| **指标分类** | 按节点类型（CHAIN、LLM、RETRIEVER 等）查看 | 按安全、成本、性能、错误四类分类查看 | -| **链路追踪** | 支持，可展开查看完整 Span 树及每个节点的输入输出 | 不支持链路追踪，仅提供模型级聚合数据 | -| **日志能力** | Trace 详情中可查看 Prompt 内容与模型输出 | 需额外开通推理日志，仅北京地域部分模型支持 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域） | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出（需通过费用与成本页面获取账单数据） | -| **数据标注** | 支持对 Span 数据添加标签（布尔值/分类/数字/文本） | 不支持 | -| **与评测集联动** | 支持将 Span 数据直接加入评测集 | 不支持 | -| **地域限制** | 无特殊地域限制 | 高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，日志限北京 | -| **费用** | 功能免费，OpenTelemetry 存储另计 | 功能免费，高级监控的底层存储可能产生费用 | -| **开通方式** | 需手动配置：授权服务角色 → 开通 OpenTelemetry → 初始化 LogStore | 普通监控自动采集；高级监控和日志需手动开通 | -| **操作方式** | 仅控制台，无 API | 仅控制台，无 API | -| **用量/计费统计** | 不涉及计费统计 | 提供按模型类型的用量统计（Token/张/秒），可用于成本核算 | - -## 适用场景建议 - -### 应用观测适用于 - -- **应用调试与优化**：需要查看智能体或工作流应用内部各环节（检索、重写、模型推理、插件调用等）的执行顺序与耗时，定位性能瓶颈。 -- **Prompt 审查**：需要查看每次调用的具体 Prompt 内容与模型输出，排查回答质量问题。 -- **评测数据积累**：将线上真实调用数据标注后加入评测集，用于持续改进应用效果。 -- **单次请求排障**：通过 Request ID / Trace ID / Span ID 精确定位某一次调用的异常环节。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用失败率、限流错误（429）、内容安全错误等指标，及时发现服务异常。 -- **成本管理**：按[业务空间](../concepts/workspace.md)和模型维度统计 Token 消耗和调用量，核算各模型的使用成本。 -- **容量规划**：通过 RPM（每分钟请求数）、TPM（每分钟 Token 数）等指标评估负载水平，预判扩容需求。 -- **主动告警**：设置 Token 消耗或错误率阈值，在异常发生时第一时间收到通知，避免静默故障。 -- **合规审计**：开通推理日志后可回溯历史对话的输入输出，满足内容审计需求。 - -### 建议组合使用 - -在生产环境中，建议同时开启应用观测和模型监控：应用观测负责应用层面的链路追踪与调试，模型监控负责基础设施层面的稳定性与成本管理。当模型监控发现某模型失败率上升时，可在应用观测中按时间范围筛选相关 Trace，进一步定位是应用逻辑问题还是模型服务问题。 - -## 技术选型参考 - -| 排查目标 | 推荐工具 | -| --- | --- | -| 某次调用为什么返回了错误答案 | 应用观测（查看 Trace 中各节点的输入输出） | -| 某个模型最近的失败率是否正常 | 模型监控（查看错误类指标） | -| 应用整体响应变慢，瓶颈在哪个环节 | 应用观测（对比各 Span 延时） | -| 本月 Token 消耗是否超出预算 | 模型监控（查看用量统计与费用概览） | -| 需要对线上 bad case 做标注并加入评测集 | 应用观测（数据标注 + 添加到评测集） | -| 模型调用量突增需要告警 | 模型监控（配置告警规则） | -| 回溯某次模型调用的完整输入输出 | 模型监控（推理日志，仅北京地域部分模型） | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md deleted file mode 100644 index b2325ef4..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用调用方式对比 - -阿里云百炼的应用（智能体应用、[工作流](../concepts/workflow.md)应用、新版智能体 Agent 2.0）可通过 API 集成到业务系统中。官方文档中存在两篇高度相关的主题页：一篇偏 **API 参考**（`application call`），重点介绍两套调用 API（OpenAI 兼容 Responses API 与 DashScope 原生 `/completion`）的端点、参数与同步/异步模式；另一篇偏 **使用指南**（`bailian application calling`），重点介绍通过 DashScope SDK / HTTP 调用智能体与[工作流](../concepts/workflow.md)应用的实操步骤、多轮对话与自定义参数透传。本文对两者做维度对比，帮助开发者在技术选型时快速定位所需信息。 - -## 关键维度对比 - -| 维度 | [application call](../api/application-call.md)（API 参考） | bailian [application call](../api/application-call.md)ing（使用指南） | -| --- | --- | --- | -| 文档定位 | API 参考，强调端点、参数、调用模式 | 使用指南，强调 SDK 实操与场景示例 | -| 所属分类 | api | guides | -| 覆盖的 API 模式 | 两套：OpenAI 兼容 Responses API + DashScope 原生 `/completion` | 一套：DashScope 原生 `/completion`（`Application.call` / `POST /apps/{app_id}/completion`） | -| 主 Endpoint | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`（OpenAI 兼容）
`POST /api/v1/apps/{APP_ID}/completion`（DashScope 原生） | `POST /api/v1/apps/{APP_ID}/completion` | -| SDK 推荐 | OpenAI 兼容模式用 OpenAI SDK；DashScope 原生模式用 DashScope SDK | DashScope SDK（Python / Java），Node.js 用 `axios` | -| 支持应用类型 | 智能体、[工作流](../concepts/workflow.md)、新版智能体 Agent 2.0 | 智能体应用、工作流应用（智能体编排应用已被工作流应用替代） | -| 输入格式 | `input` 为 string 或 messages 数组；多模态支持 `input_text` / `input_image` / `input_file` | `input.prompt` 字符串 或 自行管理 `messages` 数组 | -| 多轮对话 | 通过 messages 数组传递完整对话历史；`pre_response_id` / `conversation_id` 上下文后续支持 | 两种方式：`session_id`（云端托管，1 小时 / 50 轮）或自行管理 `messages`（推荐） | -| 同步/异步 | 支持 `stream` 流式、`background` 异步；异步暂不支持流式 | 默认同步；多轮对话通过 `session_id` 或 `messages` 实现 | -| 多模态 | 显式支持图像、文件（`input_image` / `input_file`，文件仅智能体应用支持） | 未专门展开 | -| 自定义参数透传 | 未展开 | 支持 `biz_params.user_defined_params` 透传业务参数到自定义插件 / 工作流插件节点 | -| 业务空间 | 明确说明子业务空间需 Workspace ID，多地域 Base URL 含 Workspace ID | 提及业务空间对插件与应用关联的约束（同一业务空间内） | -| 地域说明 | 明确给出华北2（北京）默认 Endpoint，并列出德国、新加坡、日本等地域需带 Workspace ID | 未专门说明地域差异 | -| 典型代码示例 | Python（OpenAI SDK 同步多轮） | Python / Java / Node.js / curl（基础调用） | -| 凭证准备 | APP ID + Workspace ID + API Key + SDK | API Key + APP_ID + DashScope SDK | - -## 适用场景建议 - -- **选 `application call`（API 参考）**：当你需要复用现有 OpenAI 生态代码库与工具链；需要使用[流式输出](../concepts/streaming-output.md)或异步执行；需要调用新版智能体 Agent 2.0；需要多模态输入（图像、文件）；或部署在非北京地域需要明确 Workspace ID 与 Base URL 拼接规则时。 -- **选 `bailian application calling`（使用指南）**：当你首次接入百炼应用、需要 Python / Java / curl 的最小可运行示例；需要通过 `session_id` 实现云端托管多轮对话；需要向自定义插件或工作流插件节点透传业务参数（`biz_params.user_defined_params`）；或团队已习惯使用 DashScope SDK 时。 - -## 技术选型参考 - -两篇文档并非互斥，而是互补：`application call` 给出"调哪套 API、用什么端点、传哪些字段"的契约层信息，`bailian application calling` 给出"用哪个 SDK、怎么写代码、怎么传业务参数"的实操层信息。建议的选型路径： - -1. 先读 `application call` 确定调用模式（OpenAI 兼容 vs DashScope 原生），明确端点与参数契约； -2. 若选择 DashScope 原生 `/completion`，再读 `bailian application calling` 获取多语言 SDK 示例与多轮、自定义参数等进阶能力； -3. 若选择 OpenAI 兼容 Responses API，则以 `application call` 为主，参考 OpenAI SDK 既有用法，`bailian application calling` 中的 `session_id` / `biz_params` 等能力在该模式下暂不适用。 - -> 注：两篇文档对应用类型的命名略有差异（"新版智能体 Agent 2.0" vs "智能体编排应用已被工作流应用替代"），接入前请以控制台实际应用类型与最新 API 参考为准。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md deleted file mode 100644 index c3c897e5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用调用方式对比：API 直调与百炼应用调用 - -阿里云百炼平台为已编排好的应用（智能体、工作流、新版智能体 Agent 2.0）提供两套对外调用路径：一是面向 OpenAI 生态的 **Responses API（OpenAI 兼容模式）**，二是面向百炼原生的 **DashScope `Application.call` / `/completion` API**。两者底层均指向同一个 `APP_ID`，但在端点形态、SDK 选型、输入结构、多轮与多模态能力、扩展参数等方面存在差异。本文从技术选型视角对比两种方式，帮助开发者根据现有技术栈与功能需求做出取舍。 - -## 关键维度对比 - -| 维度 | OpenAI 兼容 Responses API（API 直调） | DashScope 原生 API（百炼应用调用） | -| --- | --- | --- | -| 调用端点 | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK 选型 | OpenAI SDK（多语言） | DashScope SDK（Python / Java），或直接 HTTP | -| base_url 配置 | `https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1` | 无需 base_url，SDK 内置或直接 POST | -| 输入格式 | `input` 为字符串或消息数组，`role` 取 `system`/`user`/`assistant`，多模态 `content` 为数组（`input_text`/`input_image`/`input_file`） | `input.prompt` 字符串，或 `messages` 数组（自行管理多轮历史） | -| 输出格式 | OpenAI Responses 结构，`response.output` 等 | `{"output": {"finish_reason","session_id","text"}, "usage":{...}, "request_id":"..."}`，业务侧消费 `output.text` | -| 支持模型 | 智能体、工作流、新版智能体 Agent 2.0 | 智能体应用、工作流应用（[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代） | -| 同步/异步 | 支持 `background` 异步执行，同步默认；流式 `stream=true` | 主要为同步调用，`session_id` 由云端管理历史 | -| [流式输出](../concepts/streaming-output.md) | 支持（`stream=true`），异步暂不支持流式 | 通过 SDK / HTTP 支持（详见调用文档） | -| 多轮对话 | 传递完整 `input` 消息数组；基于 `pre_response_id`/`conversation_id` 的上下文能力后续支持 | 两种方式：`session_id`（云端托管，1 小时有效、最多 50 轮）或自行维护 `messages`（推荐，更灵活） | -| 多模态 | 原生支持文本、图像、文件（`input_file` 仅智能体应用支持） | 通过 `messages` 与应用内编排支持 | -| 自定义参数透传 | 通过 `input`/应用编排间接实现 | `biz_params.user_defined_params` 透传至自定义插件与工作流插件节点 | -| 业务空间 | 默认空间仅需 APP ID；子空间或海外地域需在请求中包含 Workspace ID | 同样需要 APP_ID；子空间按地域 Base URL 处理 | -| 典型场景 | 复用现有 OpenAI 代码库与工具链、多模态交互、统一 OpenAI 协议接入 | 全面功能与更高性能、自定义插件参数透传、Java/Node.js 直接 HTTP 集成 | - -## 适用场景建议 - -**OpenAI 兼容 Responses API 适合：** - -- 已有 OpenAI SDK 代码资产、希望以最小改动接入百炼应用的团队。 -- 需要多模态输入（文本 + 图像 + 文件）的智能体交互场景。 -- 希望统一在 OpenAI 协议生态下做模型/应用切换、保持代码中立。 -- 需要异步执行（`background`）与[流式输出](../concepts/streaming-output.md)能力的实时或长任务交互。 - -**DashScope 原生 `/completion` API 适合：** - -- 追求更全面功能与更高性能，使用百炼原生能力（如自定义插件参数透传 `biz_params`）。 -- Java/Node.js 项目希望直接以 HTTP 方式集成，不引入 OpenAI SDK 依赖。 -- 工作流应用需要通过 `session_id` 让云端托管对话历史，简化多轮实现。 -- 需要在工作流大模型节点中配合 `historyList` 变量精细控制提示词与上下文。 - -## 技术选型建议 - -1. **优先看协议生态**：若团队代码栈已围绕 OpenAI SDK 构建（含观测、重试、流式解析），选 Responses API 可降低迁移与维护成本；若以阿里云/DashScope 体系为主，选原生 API 更顺。 -2. **看扩展能力**：自定义插件参数透传（`biz_params.user_defined_params`）目前是原生 API 的明确能力，需要此能力的场景应选原生 API。 -3. **看多轮管理偏好**：希望云端托管历史、降低客户端状态复杂度，用原生 API 的 `session_id`；希望完全自控历史与上下文，两套 API 都支持 `messages` 数组方式。 -4. **看多模态需求**：图像、文件等多模态输入在 Responses API 中有标准化的 `content` 数组结构，接入更直接；原生 API 需结合应用编排实现。 -5. **看地域与业务空间**：两套 API 均支持默认空间仅凭 APP ID 调用；子业务空间或海外地域需携带 Workspace ID，选型不影响该约束，但需在请求中正确拼装。 -6. **凭证一致**：两套方式都使用同一份 `DASHSCOPE_API_KEY`，无需为不同调用方式分别管理密钥，切换成本主要在 SDK 与请求结构层面。 - -综上，两种方式并非互斥：同一 `APP_ID` 可同时被两套 API 调用，团队可按业务模块分别选型——面向外部生态集成用 Responses API，面向内部能力扩展用原生 API。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md deleted file mode 100644 index 044ec3fc..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用评测与应用监控对比 - -阿里云百炼平台同时提供"应用评测"与"应用监控（应用观测）"两套围绕应用质量的能力：前者面向**离线**场景，用评测集 + 评估器系统化衡量应用输出质量；后者面向**在线**场景，端到端追踪已发布应用的真实调用链路并采集延时、Token 量等运行时指标。两者通过共享的"标签"组件打通——线上 Span 可标注后直接沉淀为评测样本，实现"线上观测 → 离线评测 → [模型调优](../concepts/fine-tuning.md)"的闭环。本文从目的、数据来源、输入输出、计费、典型场景等维度对比两者，帮助开发者做技术选型。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 应用监控（应用观测） | -| --- | --- | --- | -| 核心目的 | 离线评估应用输出质量，产出评分、BadCase、归因与调优建议 | 在线追踪应用真实调用链路，采集延时、Token 量等运行时指标 | -| 数据来源 | 评测集（手动上传或大模型自动生成） | 已发布应用在线接收的真实 Prompt 与调用 | -| 触发方式 | 手动发起评测任务，运行评测集 | 应用添加到观测列表后自动追踪，分钟级同步 | -| 应用范围 | 智能体应用（自动评测需配置知识库）、[工作流](../concepts/workflow.md)应用；新版支持自定义类型 | 智能体应用、[工作流](../concepts/workflow.md)应用、高代码应用；暂不支持 Assistant API 创建的智能体应用 | -| 输入格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：`.xls`/`.xlsx`，按应用出入参生成模板，单文件 ≤ 20MB，单次最多 10 个文件 | 真实线上请求，无需上传文件；导出支持 JSONL 与 EXCEL | -| 输出形式 | 评分（1-5 / 0-100 / 0-1 等）、评估器结果、BadCase 分析、归因报告、调优建议 | 调用链路树（CHAIN/AGENT/RETRIEVER/LLM 等节点）、延时、Token 量、监控统计图表 | -| 评估器/评分机制 | 预置模板 + LLM 评估器 + Code 评估器；每任务最多 10 个评估器 | 不做评分，仅记录指标；可对 Span 人工打标签 | -| 标签能力 | 分类/布尔值/数字/文本四种类型，与监控共用 | 同一套标签体系，可对每个 Span 标注，自动保存 | -| API 端点 | 通过控制台操作（评测集、评测任务、评估器、标签管理） | 无 API，仅控制台操作；底层依赖 OpenTelemetry 服务 | -| 数据时效 | 任务执行后产出，按版本留存 | 指标分钟级更新，调用记录最长可查 30 天 | -| 计费方式 | 评测任务调用大模型产生 Token 费用，正常计费 | 应用观测功能本身不收费；观测数据存储费用由 OpenTelemetry 服务收取 | -| 关键限制 | 自动评测仅面向已发布且配置知识库的智能体应用，单次最多 8 个应用横向评测；评测任务发起后配置不可修改 | 高代码应用仅能观测到入口 CHAIN 节点，不支持内部链路追踪；暂不支持长期记忆检索过程观测 | -| 版本差异 | 区分"旧版"与"新版"两套界面，评测集类型、关联应用范围、评估器机制不同 | 无新旧版本区分 | - -## 两者协同关系 - -应用监控与评测并非孤立能力，平台在数据层做了打通： - -- **Span → 评测集**：应用观测支持将 Span 数据直接加入评测集（选择目标评测集、导入方式、字段映射，每评测集最多 50 个字段映射），把真实线上调用沉淀为评测样本。 -- **共享标签**：标签管理为评测与监控共用，线上 Span 标注的标签可在评测任务中复用，反之亦然，便于跨阶段质量分析。 - -## 适用场景建议 - -- **选应用评测** when：需要系统化、可重复地评估应用输出质量；要对响应做相关性、有害性、幻觉等语义判断；有领域专家介入做端到端标注；发布前做回归验证或横向对比多个应用（最多 8 个）。 -- **选应用监控** when：应用已上线，需要排查真实调用的延时、失败率、Token 成本；想查看智能体/[工作流](../concepts/workflow.md)内部检索、重写、重排、LLM 等子节点的执行链路；需要分钟级性能监控与最长 30 天的调用记录检索；想从线上真实数据中沉淀评测样本。 -- **组合使用** when：希望构建"线上观测 → 标注/BadCase 沉淀 → 离线评测 → 调优迭代"的闭环质量运营体系。此时建议先开通应用观测（自动评测的前置条件之一），再用观测到的 Span 数据补充评测集。 - -## 技术选型小结 - -若问题偏"质量好不好"（输出是否准确、完整、合规）→ 用应用评测；若问题偏"跑得稳不稳"（延时、失败、成本、链路追踪）→ 用应用监控；若希望用线上真实数据驱动持续调优，则两者配合使用，通过共享标签与 Span 导入评测集的能力打通闭环。开通自动评测前需先开通应用观测，这也是两者协同的一个隐含前置条件。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md deleted file mode 100644 index 2194f12d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md +++ /dev/null @@ -1,59 +0,0 @@ -# 应用监控与应用评测对比 - -阿里云百炼平台围绕[智能体应用](../concepts/agent-application.md)、工作流应用提供两类数据驱动的运营能力：**应用监控（应用观测）**与**应用评测**。两者都服务于"用数据评估应用质量、辅助调优"这一目标，但定位不同——应用监控聚焦**线上真实流量的运行时可观测性**（延时、[Token](../concepts/token.md)、调用链路），应用评测聚焦**离线/预上线数据集的质量评估**（评分、BadCase、归因）。本页通过关键维度对比帮助开发者在不同阶段做出技术选型。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 应用评测 | -| --- | --- | --- | -| 核心定位 | 线上运行时可观测，追踪调用链路与性能指标 | 离线/预上线数据集质量评估，量化输出好坏 | -| 数据来源 | 自动采集应用实际调用（Prompt、输出、延时、[Token](../concepts/token.md)） | 评测集（自动生成或手动上传）+ 被测应用输出 | -| 触发方式 | 开启观测后自动同步，分钟级更新 | 手动发起评测任务，任务完成后产出报告 | -| 支持应用类型 | [智能体应用](../concepts/agent-application.md)、工作流应用、高代码应用（仅入口 CHAIN 节点） | [智能体应用](../concepts/agent-application.md)、工作流应用（自动评测需配置[知识库](../concepts/knowledge-base.md)且已发布） | -| 不支持场景 | 暂不支持 Assistant API 创建的智能体；高代码内部链路不可追踪；无 API | 自动评测仅面向配置[知识库](../concepts/knowledge-base.md)的已发布智能体，单次最多 8 个应用 | -| 输入格式 | 实际线上 Prompt 及调用数据（无需用户准备数据集） | 评测集：旧版（`.xls`/`.xlsx` 对话分析、`.jsonl` 知识问答）；新版（`.xls`/`.xlsx`，单文件 ≤20MB，单次 ≤10 文件） | -| 输出格式 | 调用记录、Trace 列表、监控统计图表；可导出 **JSONL** / **EXCEL** | 评测报告：总正确率、BadCase、归因分析、调优建议；任务详情含数据明细与指标统计 | -| 评估方式 | 不评分，仅呈现延时/[Token](../concepts/token.md)/失败率等指标，支持人工标注标签 | 自动评测用大模型评 1-5 分；手动评测人工打标（较差/一般/较好或 1-5 分）；新版支持 LLM/Code 评估器 | -| 评估器/评分 | 无评分器，依赖指标与人工标注 | 预置模板（通用质量/智能体/文本匹配/相似度/格式校验）+ 自定义 LLM/Code 评估器，每任务最多 10 个 | -| 关键指标 | 调用次数（含失败率）、Token 总量（输入/输出）、平均首 Token 耗时、平均调用时长、节点级延时 | 总正确率（≥4 分占比）、BadCase Top-5、RAG 智能体按问题类型分档得分、各评估器通过率 | -| 调用链路追踪 | 支持，按节点类型展示（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等，可嵌套） | 不追踪链路，仅按评测集条目组织输入输出 | -| 数据时效与保留 | 分钟级更新，调用记录最长查 30 天，支持按 Request ID/Trace ID/Span ID 检索 | 任务结果持久保存，评测集每次发布生成新版本，可选特定版本评测 | -| 标签与标注 | 支持对 Span 标注标签（布尔/分类/数字/文本），与应用评测共享统一管理 | 标签为核心组件，用于数据明细筛选；支持普通模式与快速标注 | -| 与对方联动 | 可将 Span 数据直接加入评测集（选目标集、导入方式、字段映射，最多 50 字段映射） | 评测集可承接来自应用观测的真实线上样本 | -| 前提条件 | 需开通 OpenTelemetry 服务并初始化 LogStore；子账号需 `AliyunBailianFullAccess` + 应用观测权限 + `ram:CreateServiceLinkedRole` | 自动评测需已开通应用观测；评测集需发布后可用；任务配置后不可修改 | -| 计费方式 | 功能本身不收费，数据存储费用由 OpenTelemetry 服务收取 | 评测任务调用大模型产生的 Token 费用正常计费；Code 评估器无额外费用 | -| API 支持 | 无 API，仅控制台操作 | 无 API，仅控制台操作（评测集/任务/评估器/标签均通过控制台管理） | -| 典型场景 | 线上性能监控、故障排查、Token 成本分析、瓶颈节点定位、把真实调用回流为评测样本 | 上线前质量验收、Prompt/检索/[知识库](../concepts/knowledge-base.md)调优验证、多应用横向对比、BadCase 归因与优化建议 | - -## 适用场景建议 - -**优先选择应用监控（应用观测）的场景**： - -- 应用已上线，需要持续观察**真实流量**下的延时、Token 消耗、失败率等运行时指标。 -- 需要**端到端调用链路追踪**，定位某个检索/重排/模型节点是性能瓶颈或错误来源。 -- 需要把线上真实调用**回流为评测样本**，让评测集贴近实际分布。 -- 关注**成本治理**，希望按应用/节点维度统计 Token 用量。 - -**优先选择应用评测的场景**： - -- 应用上线前或迭代中需要**系统化评估输出质量**，得到可量化的评分与归因。 -- 需要用**领域专家或自动评估器**对一批固定用例打分，产出 BadCase 与调优建议。 -- 需要对比多个应用/版本在同一评测集上的**横向表现**（自动评测单次最多 8 个应用）。 -- 需要**可复用的评分逻辑**（LLM 评估器做语义判断，Code 评估器做格式/数值精确校验）。 - -## 技术选型参考 - -两者并非二选一，而是**互补的闭环**：应用监控提供"线上发生了什么"的运行时事实，应用评测提供"这些输出到底好不好"的质量判定。推荐组合使用—— - -1. **开发/调优阶段**：用手动或自动评测在固定评测集上迭代 Prompt、检索配置、知识库切片，借助评估器与归因建议收敛质量。 -2. **上线/运营阶段**：开启应用观测监控真实流量，识别性能瓶颈与成本异常，并用 Span 标注或"加入评测集"把线上问题转化为下一轮评测输入。 -3. **数据回流**：将应用观测中标记为错误或低质的 Span 直接加入评测集，形成"线上 → 评测 → 调优 → 线上"的持续优化闭环。 - -选型时关键判据是**数据来源**：要用真实流量就看监控，要用受控数据集就看评测；要量化"好坏"就用评测，要看"运行状态"就用监控。两者共享标签体系，便于跨阶段贯通标注与筛选。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [application evaluation](../guides/application-evaluation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md new file mode 100644 index 00000000..e7ec2ed1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md @@ -0,0 +1,81 @@ +# 应用编排能力对比：托管智能体、应用组件与模型上下文协议 + +## 背景与目的 +在百炼平台构建复杂 AI 应用时，开发者需在不同抽象层级间进行技术选型：是直接调度原子能力（如知识检索、文件解析），还是封装为可复用的智能体实例？是通过标准化协议接入外部工具，还是在底层数据与模型之间建立结构化桥梁？本对比聚焦三大核心编排能力——**托管智能体（Managed Agents）**、**应用组件（Application Components）** 和 **模型上下文协议（Model Context Protocol, MCP）**，旨在帮助开发者清晰理解其定位差异、能力边界与协同关系，避免方案错配（例如用 Application Component API 实现会话状态管理，或用 MCP 直接替代知识库构建），从而做出高效、可维护、可扩展的技术决策。 + +--- + +## 关键维度对比 + +| 维度 | 托管智能体（Managed Agents） | 应用组件（Application Components） | 模型上下文协议（MCP） | +|------|------------------------------|-------------------------------------|------------------------| +| **核心定位** | **面向会话的智能体运行时服务**：提供带状态、带沙箱、事件驱动的 Agent 全生命周期托管能力 | **面向数据与知识的基础设施层**：提供知识库构建、文件/类目管理、Prompt 模板等基础能力的 OpenAPI 接口集合 | **面向工具调用的标准化协议层**：定义大模型与外部工具（搜索、地图、图表等）安全交互的通用通信契约，不处理模型推理本身 | +| **输入格式** | OpenAI-style message 数组（`[{"role":"user","content":[{"type":"text","text":"..."}]}]`），支持文本、图片、文件引用（需先上传并审核） | 多样化结构化输入：
• 文件上传：二进制流 + `AddFile` 元数据
• 知识库构建：`CreateIndex` + `SubmitIndexJob` JSON 配置
• Prompt 模板：含 `${variable}` 占位符的字符串 | 工具调用请求（Tool Call）JSON：
• 智能体场景：由模型自动生成，含 `tool.name` 与 `tool.input`
• 工作流场景：人工配置，支持变量引用（如 `上游节点.output`）
• 外部调用：符合 MCP Streamable HTTP 规范的 POST 请求体 | +| **输出格式** | SSE 流式事件（`message`, `tool_call`, `tool_result`, `session_status`），含完整会话状态变迁与中间结果 | 同步 REST 响应：
• 成功：标准 JSON（如 `{"IndexId": "idx_xxx", "Status": "CREATING"}`）
• 列表接口：分页结构（`NextToken`, `Items[]`）
• 检索结果：`Retrieve` 返回 `Chunks[]` 及相关性分数 | 工具执行结果（Tool Result）JSON：
• 格式由工具 `outputSchema` 定义（如天气服务返回 `{"temperature": 25.3, "condition": "sunny"}`）
• 支持流式响应（`streamableHttp` 协议下）或同步返回 | +| **支持模型** | **仅限百炼托管模型**（如 `qwen-plus`, `qwen-max`, `qwen3`），不支持自定义/外部模型接入；模型 ID 必须在创建 Agent 时显式指定且不可变更 | **不直接涉及模型调用**；为模型提供上下文支撑（如 RAG 检索结果、结构化 Prompt），可被任意百炼模型（包括千问系列、第三方模型）在应用层消费 | **不绑定模型**；作为工具通道服务于智能体/工作流中的任意百炼托管模型（推荐 Qwen-Max/Qwen3 提升工具调用准确率），**不可用于原始千问 API 调用** | +| **API 端点** | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`（REST + SSE） | `https://bailian.{region}.aliyuncs.com`（ROA 风格，如 `/bailian/2023-12-29/indexes`） | 无统一端点：
• 官方服务：`https://dashscope.aliyuncs.com/api/v1/mcps/{service}/{tool}`
• 自定义服务：由用户部署地址决定（如 FC 函数 URL） | +| **计费方式** | **按会话（Session）计费**：
• 基础费用：Agent 运行时长（秒） × 单价
• 附加费用：沙箱资源（CPU/GPU）、文件存储（≤20 MB/文件，30 天保留期）、Skill 执行 | **按操作与资源计费**：
• 文件上传/解析：按文件数与大小计费
• 知识库：按索引容量（GB）、构建时长（CU）、检索调用次数（QPS）计费
• Prompt 模板：免费 | **按工具调用计费**：
• 官方云服务（如 WebSearch）：29 元/千次 + QPS 限制
• 自定义服务：按调用时长（秒）计费（0.000156 元/秒），分“基础模式”与“极速模式” | +| **典型场景** | • 多轮对话客服机器人（需记忆用户偏好、调用订单系统）
• 自动化数据分析助手（上传 Excel → 解析 → 生成图表 → 解释结论）
• 内部 IT 支持 Agent（集成 Jira、Confluence 工具链） | • 构建企业专属知识库（PDF/Word/网页入库 + 分片优化）
• 管理客户资料类目体系与附件文件
• 统一维护销售话术、产品 FAQ 等 Prompt 模板 | • 智能体中自动触发天气查询、路径规划、联网搜索
• 工作流中串联“网页爬取 → 文本摘要 → PPT 生成”工具链
• 第三方应用（如 Cherry Studio）接入百炼工具生态 | + +--- + +## 适用场景建议 + +### ✅ 选择 **托管智能体** 当： +- 你需要一个**有状态、可中断、可审计的会话级执行单元**； +- 业务逻辑涉及**多步骤、条件分支、工具审批（如人工确认支付）**； +- 必须保障**沙箱隔离性**（如运行用户上传的 Python 脚本、调用敏感内部 API）； +- 团队希望**复用已验证的 Skill 包**（经安全扫描的 zip 工具组合）； +- 对**事件流实时性要求高**（如实时推送工具执行进度、用户消息确认）。 + +> ⚠️ 注意：若仅需单次问答（无状态）、或模型固定无需版本控制、或工具链简单无沙箱需求，则过度使用托管智能体将增加运维复杂度。 + +### ✅ 选择 **应用组件** 当： +- 你的核心诉求是**构建和管理知识资产**（RAG 知识库、文档中心、FAQ 库）； +- 需要**批量导入/更新/删除文件与类目**，并精细控制解析策略（如 PDF 用 DocMind、图片用 Qwen-VL）； +- 希望**标准化提示词工程**，实现跨应用复用与 A/B 测试（如不同销售话术模板）； +- 数据源来自内部系统（如 CRM 导出 CSV），需通过 API 自动化同步至百炼。 + +> ⚠️ 注意：应用组件不提供模型推理、会话管理或工具调用能力；它本质是“燃料供给系统”，需与智能体或工作流配合使用。 + +### ✅ 选择 **模型上下文协议（MCP）** 当： +- 你希望**解耦模型与工具**，让同一套智能体配置可灵活切换不同地图/搜索服务商； +- 需要**快速接入多个异构工具**（如同时用高德地图 + WebSearch + QuickChart），避免为每个工具单独开发适配器； +- 在**工作流中精确控制工具调用顺序与参数传递**（如将 OCR 结果作为搜索关键词）； +- 计划将百炼工具能力**嵌入第三方 IDE 或低代码平台**（通过 MCP SDK 标准化对接）。 + +> ⚠️ 注意：MCP 不解决知识库构建、文件管理、Prompt 版本控制等问题；它专注“调用什么工具”和“如何传参”，而非“从哪获取数据”或“如何组织会话”。 + +--- + +## 技术选型参考（面向开发者） + +| 你的问题 | 推荐方案 | 关键理由 | +|----------|----------|----------| +| “我需要一个能记住用户上句话、调用数据库查订单、再生成总结的聊天机器人” | ✅ **托管智能体** | 唯一支持会话状态机（`idle`→`running`→`idle`）、SSE 事件流、沙箱环境三者结合的方案 | +| “我要把公司 500 份产品手册 PDF 自动转成向量知识库，并支持按章节检索” | ✅ **应用组件** | `AddFile` + `CreateIndex` + `SubmitIndexJob` 是知识库构建的标准 API 流程，支持分片、监控与权限隔离 | +| “我的智能体有时需要查天气，有时需要搜新闻，能否不改代码就切换服务商？” | ✅ **MCP** | 通过控制台更换 MCP 服务绑定即可，模型调用逻辑（`tool.name`）完全不变，真正实现工具解耦 | +| “我想在 Python 脚本里批量上传 1000 个文件到百炼，并分类打标” | ✅ **应用组件** | `ApplyFileUploadLease` + `AddFile` + `AddCategory` 提供稳定、幂等的批量文件管理能力 | +| “我有一个自研的股票分析 API，想让百炼模型能调用它” | ✅ **MCP（自定义服务）** | 用 `streamableHttp` 类型接入，无需修改模型代码；KMS 加密密钥保障安全性；支持流式返回实时行情 | +| “我只需要调用一次千问 API 回答问题，不需要状态、不调用工具” | ❌ 三者均不适用 | 应直接使用 [DashScope ChatCompletion API](https://help.aliyun.com/zh/dashscope/developer-reference/quick-start) | + +### 🧩 协同使用最佳实践 +- **典型组合**：`应用组件`（构建知识库） → `MCP`（接入搜索工具） → `托管智能体`（封装为可对话的 Agent） +- **示例流程**： + 1. 用 Application Component API 将产品文档入库（`AddFile` → `SubmitIndexJob`）； + 2. 用 MCP 配置 WebSearch 服务，补充实时信息； + 3. 创建 Managed Agent，挂载该知识库（通过 RAG 插件）+ MCP 服务，设定系统提示词； + 4. 用户提问时，Agent 自动融合知识库检索结果与 WebSearch 结果生成回答。 + +> 💡 **关键提醒**：三者非互斥关系，而是分层协作——应用组件提供“数据燃料”，MCP 提供“工具插件”，托管智能体提供“运行引擎”。合理分层可显著提升系统可维护性与迭代效率。 + +--- +*最后更新：2025年4月* + +## 被对比主题页 + +- [managed agents api](../api/managed-agents-api.md) +- [application component api reference](../api/application-component-api-reference.md) +- [model context protocol](../guides/model-context-protocol.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md deleted file mode 100644 index 9c8ed744..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md +++ /dev/null @@ -1,56 +0,0 @@ -# 百炼应用与托管 Agent 对比 - -阿里云百炼平台提供两条构建 AI 应用的路径：**[智能体应用](../concepts/agent-application.md)**（含新版 Agent 2.0、旧版 Agent 1.0、工作流、高代码应用）与 **Managed Agents（托管 Agent）**。二者都能组合模型、提示词、工具与 MCP 服务，但在运行模式、状态管理、执行环境和适用任务上有本质差异。本页帮助开发者理解两者定位，做出正确的技术选型。 - -核心区别在于：**[智能体应用](../concepts/agent-application.md)是无状态的应用构建与发布形态**，由应用侧维护上下文，面向问答、对话等交互式场景；而 **Managed Agents 是服务端托管的运行时**，在独立云端沙箱中持久化会话状态并支持中断续接，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 关键维度对比 - -| 对比维度 | [智能体应用](../concepts/agent-application.md)（LLM Application） | Managed Agents（托管 Agent） | -|---------|------------------------------|------------------------------| -| 定位 | 应用构建与发布形态（Agent / 工作流 / 高代码） | 服务端托管的智能体运行时 | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，百炼托管的云端容器 | -| 状态持久化 | 不持久化，依赖调用方传递历史 | 事件历史在服务端持久化 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流（User/Agent/Tool/Tool_output/Error/Model/System） | -| 核心对象 | 应用（Agent/工作流/高代码应用） | 智能体 / 运行环境 / 会话 / 事件（四对象解耦复用） | -| 工具能力 | 内置沙箱工具（bash/read/write/edit/glob/grep/download_file）、知识库、MCP、插件 | 7 个内置工具（bash/read/write/edit/glob/grep/download_file）、MCP、Skill | -| 依赖安装 | 不支持自定义环境依赖 | 环境可通过 `config.packages` 预装 apt / pip 依赖并设网络策略 | -| 文件挂载 | 单会话最多 10 个文件、单文件 ≤10MB（文件问答） | 资源挂载到 `/mnt/session/uploads`，独立于会话、可多会话复用；单文件 ≤10MB | -| 记忆能力 | 新版 Agent 短期记忆 0-30 轮，长期记忆未支持 | 会话状态由状态机管理，天然承载长程上下文 | -| 主要 API 端点 | 各应用发布后经统一 API 集成调用 | `POST /agents`、`POST /environments`、`POST /sessions`、`POST /sessions/{id}/events`、`GET /sessions/{id}/events/stream` | -| 开发方式 | 零代码（Agent）/ 低代码（工作流）/ Python（高代码） | 控制台向导或 API 编排 | -| 计费方式 | 按模型 Token、知识库召回、MCP/插件、高代码资源等分项计费 | 按模型 Token 用量及托管沙箱资源计费 | -| 典型场景 | 问答、对话、RAG 检索、固定流程自动化、轻量任务 | 多步工具调用、代码执行、文件批处理等长时任务 | - -## 适用场景建议 - -### 选择智能体应用 - -- **快速交付交互式问答/对话**：新版智能体（Agent 2.0）以自然语言配置即可上线，适合业务人员和产品经理。 -- **知识库 / RAG 问答**：需要挂载企业私有知识库并做标签过滤检索时，智能体应用的知识库能力开箱即用。 -- **流程固定、精确可控**：意图分类、多步骤审批等确定性流程，用工作流做可视化节点编排更稳定。 -- **需要完整代码控制或企业级部署**：用高代码应用（Serverless Function / K8s），配套网关、可观测、日志等企业能力。 -- **多渠道发布**：需要发布到钉钉、微信公众号等第三方平台时，智能体应用原生支持。 - -### 选择 Managed Agents - -- **长时运行任务**：单次任务需要多轮工具调用、可能被中断后续接，服务端持久化事件历史更可靠。 -- **代码执行与文件处理**：需要在独立沙箱中执行 shell 命令、读写文件、安装依赖（apt/pip）、下载资源等。 -- **智能体自主规划**：让智能体在云端容器中自主决定命令与工具调用顺序，无需应用侧管理复杂上下文。 -- **资源复用与会话隔离**：同一环境 / 挂载资源被多个会话复用，且会话间修改互不影响时。 - -## 技术选型参考 - -- 若你的应用本质是**一问一答或短流程交互**，优先用智能体应用；无旧版依赖时选新版 Agent 2.0。 -- 若任务是**服务端长跑、含代码执行或多步工具编排**，选 Managed Agents，避免在应用侧自行拼接和维护会话状态。 -- 两者并非互斥：可以用智能体应用承载前端交互体验，用 Managed Agents 承接后端的长时执行任务。 -- 注意文档中 Managed Agents 的模型名存在不一致（向导示例 `qwen3.7-plus`、API 示例 `qwen3-max`），实际以控制台可选模型 ID 为准；智能体应用推荐使用工具调用能力强的千问-Max 系列。 -- 两类方案的文件上传上限均为单文件 10MB，超限需走文件上传 API。 - -## 被对比主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents](../guides/managed-agents.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md deleted file mode 100644 index d23dd1fd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md +++ /dev/null @@ -1,61 +0,0 @@ -# 知识库与数据连接对比 - -在阿里云百炼平台中，**知识库**与**数据连接**是为大模型应用补充外部数据的两条主要路径，二者定位互补但常被混淆：知识库面向 RAG 检索增强生成，强调"先建索引、再语义召回"；数据连接面向外部数据源的统一接入与实时访问，强调"按需查询、按需引用"。本文从输入格式、输出形态、支持模型、调用方式、计费模型、典型场景等维度对两者做技术选型对比，帮助开发者根据数据形态与访问模式选择合适方案。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 数据连接（Data Connection） | -| --- | --- | --- | -| 定位 | 基于向量化检索增强生成，为模型补充私有数据与最新信息 | 外部数据源统一入口，安全访问企业数据库、文档系统、对象存储 | -| 数据形态 | 非结构化文档、表格、图片、音视频（向量化后检索） | 非结构化文件、结构化表格、关系型数据库、对象存储、语雀文档 | -| 支持的数据源类型 | 本地上传、OSS、数据连接器导入 | 文件、表格、MySQL、PostgreSQL、PolarDB-X 2.0、语雀、OSS | -| 输入格式 | pdf/docx/pptx/xlsx/txt/markdown/html/png/jpg/mp4/mkv 等 | 文件类同知识库；表格 xlsx/xls；数据库走 SQL；不支持直接导入 JSON/CSV/YAML | -| 处理方式 | 切片 + 向量化 + 召回排序（建立索引） | 文件类解析入库；数据库走流式实时查询；OSS 走向量检索服务 | -| 访问模式 | 语义检索召回切片（TopK 1–20，相似度阈值过滤） | 文件/表格按类目导入与查询；数据库实时执行 SQL；OSS 工具调用 | -| 支持模型 | 千问 QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、千问 VL 系列、开源版、第三方（DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 等） | 由挂载的应用决定，本身不绑定模型；数据库/SQL 访问依赖应用编排 | -| 向量模型 | text-embedding-v4/v3（512 维）、multimodal-embedding-v1（1024 维） | OSS 连接器需开通向量检索服务；其他类型不强制向量化 | -| API 端点 | 知识库 API（上传租约→文件→类目→索引任务轮询） | 数据连接 API + 应用内调用（searchOSSFile、searchOSSFileByFileName 等工具） | -| 计费方式 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时（1 RCU≈50 QPS） | 文件/表格平台存储限时免费、超额按量；数据库走源实例计费；OSS 走 OSS 计费 | -| 地域限制 | 仅中国站华北2（北京）可开通使用 | 数据库类需私网/公网可达；PolarDB-X 2.0 仅私网；语雀仅公网版本 | -| 数据时效 | 增量导入后需重新解析与建索引，存在更新延迟 | 数据库与语雀类为实时访问；文件/表格/OSS 导入后立即可用 | -| 配额要点 | 业务空间类目 500、文件 10 万、数据表 1 千；标准版 100 GB、旗舰版 9,999 GB | 平台存储 10 万文件、1 TB 免费额度；导入文件仅保留 90 天可查看 | -| 典型场景 | 文档问答、企业知识问答、图文并茂回复、音视频检索问答 | 实时查询业务库、引用语雀文档、检索 OSS 对象、结构化表格问答 | - -## 适用场景建议 - -### 优先选择知识库 - -- 数据以**非结构化文档**为主，需要**语义检索**（即使关键词不匹配也能召回）。 -- 需要图文并茂回复、视觉理解、音视频内容检索等 RAG 场景。 -- 应用形态为智能体应用、工作流应用，或通过 SDK 集成 RAG 检索能力。 -- 对**召回质量**有可调优诉求（相似度阈值、TopK、切片策略、元数据过滤）。 - -### 优先选择数据连接 - -- 数据源是**关系型数据库**（MySQL、PostgreSQL、PolarDB-X 2.0），需要在对话中**实时执行 SQL** 查询最新业务数据。 -- 数据存储在**语雀**或**OSS**，希望按需引用而非预先向量化。 -- 数据以**结构化表格**为主，且字段含义清晰，适合模型直接理解。 -- 团队希望以"连接器"方式统一管理企业数据资产，按权限安全访问。 - -### 二者结合使用 - -- 同一批数据既需语义召回又需结构化查询：可先将结构化部分接入表格连接器，再将非结构化部分导入知识库，在应用编排中按查询意图分流。 -- 文件已在 OSS：既可作为知识库数据来源导入（向量化召回），也可通过 OSS 连接器在对话中按需检索——前者适合"问答"，后者适合"取文件"。 -- 语雀文档：通过语雀连接器实时访问，避免频繁同步；若需历史版本语义检索，仍需导入知识库。 - -## 技术选型参考 - -1. **看数据形态**：非结构化文档/图片/音视频 → 知识库；结构化数据库/表格 → 数据连接；混合 → 两者并行。 -2. **看时效要求**：实时业务数据（订单、库存）→ 数据连接（流处理类）；可容忍延迟的知识沉淀 → 知识库。 -3. **看访问模式**：语义相似度召回 → 知识库；精确 SQL 查询或工具取文件 → 数据连接。 -4. **看成本结构**：知识库按规格/RCU 计费，检索量大时旗舰版更经济；数据连接平台存储限时免费，主要成本在源数据库与 OSS。 -5. **看地域与权限**：知识库仅北京地域可用；数据库类连接器需保证网络可达并完成 DMS/EventBridge/DTS 等角色授权。 - -> 选型结论：**知识库解决"模型不知道"的问题，数据连接解决"模型拿不到最新数据"的问题**。在百炼平台中两者并不互斥，建议依据数据源类型与访问模式组合使用，由应用编排层按查询意图分发。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md deleted file mode 100644 index fd5d160e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md +++ /dev/null @@ -1,52 +0,0 @@ -# 应用[评测](../concepts/evaluation.md)与模型[评测](../concepts/evaluation.md)对比 - -百炼平台提供两套互相独立的[评测](../concepts/evaluation.md)能力：**应用评测**评估的是已编排好的智能体/工作流应用的端到端输出质量，**模型评测**评估的是大语言模型本身在标准或自定义数据集上的基础能力。两者面向的评测对象、数据组织方式、评估器机制、计费路径均不同，开发者需根据选型阶段（选模型 vs 验应用）选择合适工具。本文从评测对象、数据格式、评估机制、结果产出、计费、典型场景等维度进行对比，供技术选型参考。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -| --- | --- | --- | -| 评测对象 | 已发布的智能体应用、工作流应用（端到端整体输出） | 大语言模型本身（千问系列、开源版、第三方文本模型、调优模型） | -| 评测范式 | 自动评测（大模型生成评测集并评分 1–5）、手动评测（人工打标） | 自动评测（系统跑模型推理后评分）、人工评测（人工打标） | -| 评测集类型 | 旧版：对话分析（.xls/.xlsx）、知识问答（.jsonl）；新版：智能体/工作流/自定义三类（.xls/.xlsx，≤20MB，单次≤10 文件） | 自定义评测（评测数据集 / 推理结果集，Excel）、基线评测（系统预置榜单数据集，无需自备） | -| 评测集字段 | Prompt/Completion/SessionId（对话分析）、query/referenceAnswer/fineKeywords/coarseKeywords（知识问答），新版按应用出入参自动生成模板 | `${prompt}`（输入）、`${output}`（模型输出）、`${completion}`（参考答案）；评测数据集含 Prompt+Completion，推理结果集额外含 Output | -| 评估器/评测维度 | 评估器可复用：预置模板（通用质量/智能体/文本匹配/文本相似度/格式校验）+ 自定义（LLM 评估器、Code 评估器、基于评测任务创建）；单任务最多 10 个 | 评测维度 5 种：大模型评估-数值型、规则评估-文本相似度、大模型评估-分类型、规则评估-字符串匹配、人工评估-分类型；可组合多维度 | -| 评分尺度 | LLM 评估器自定义范围（0-100 / 1-5 / 0-1），Code 评估器返回数值；通过阈值决定 Pass/Fail | 数值型 1–5 分，分类型 Pass/Fail，规则型相似度/匹配度；综合得分=维度平均分 | -| 评分器 Prompt 变量 | 映射到评测集字段或应用输出（query、reference_response、context 等） | 固定三变量 `${prompt}` / `${output}` / `${completion}` | -| 预置模板 | 通用质量、智能体、文本匹配、文本相似度、格式校验五类 | 综合评测（5 维度）、语义相似度、自定义评测三种 | -| 关联范围 | 单次自动评测最多 8 个智能体应用横向对比 | 支持多模型同维度排行榜对比 | -| 基线/榜单能力 | 无 | 提供基线评测（C-Eval、MMLU、ARC、GSM8K、HellaSwag、BBH），仅支持调优模型，不支持预置模型 | -| API/触发方式 | 控制台编排评测任务，调用应用产生 Token 计费 | 控制台创建任务，系统自动推理或读取已有 Output | -| 计费方式 | 评测任务调用大模型产生的 Token 费用正常计费；LLM 评估器裁判评分产生 Token 费用；Code 评估器无额外费用 | System Prompt 产生被评测模型推理费用，评分器 Prompt 产生裁判模型评分费用；推理结果集直接读 Output 不推理，可降本 | -| 标签与人工标注 | 分类/布尔值/数字/文本四类标签，支持快速标注模式，单条标注页三栏布局 | 人工评估-分类型由人工打标签 | -| 结果产出 | 数据明细 + 指标统计；自动评测含总正确率、BadCase Top-5、归因分析、调优建议 | 数据明细（Prompt/Completion/Output/评分）+ 指标统计（综合得分、通过率、分数分布）；基线评测含分学科明细、能力雷达图、行业对比 | -| 版本差异 | 区分旧版/新版应用评测（评测集类型、关联范围、评估器机制显著不同） | 无新旧版之分 | - -## 适用场景建议 - -### 选应用评测 - -- 已完成智能体/工作流编排，需验证端到端输出质量（含知识库检索、Prompt 编排、工具调用整体效果）。 -- 需要 BadCase 归因与调优建议，指导应用迭代。 -- 需对线上真实 Span 做标签标注，与应用观测联动做质量监控。 -- 多应用横向对比（最多 8 个），选优上线。 -- 评测数据为多轮对话或带 fineKeywords 的知识问答，需要按 SessionId 或信息点粒度评估。 - -### 选模型评测 - -- 处于模型选型阶段，需对比千问系列、开源版或第三方模型的基础能力。 -- 需用标准榜单（C-Eval、MMLU、GSM8K 等）做基线验证，且被测模型为调优后模型。 -- 业务场景需自定义评分标准（问答质量、内容安全、Function Calling、NL2SQL、翻译摘要等），希望以数据驱动选型。 -- 已有模型推理结果，希望直接评分以降低推理成本（用推理结果集）。 -- 需要参与排行榜对比多模型同维度表现。 - -### 组合使用 - -复杂项目可先做模型评测锁定底座模型，再做应用评测验证编排后的端到端质量；两者评分器/评估器都依赖大模型裁判，需同时关注裁判模型 Token 成本。应用评测的 LLM 评估器与模型评测的大模型评估-数值型在机制上同源（都靠裁判模型按 Prompt 评分），但前者参数映射到应用输出，后者固定三变量，迁移评分逻辑时需调整变量绑定方式。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md deleted file mode 100644 index 673a548f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md +++ /dev/null @@ -1,50 +0,0 @@ -# 应用评测与模型评测对比 - -阿里云百炼平台提供两套面向不同对象的评测能力：**应用评测**关注智能体/工作流应用的端到端输出质量（含 RAG 归因、调优建议），**模型评测**关注单个文本生成模型在给定数据集上的能力表现（含基线评测、排行榜）。二者虽然都围绕"评测集 + 评估规则 + 报告"展开，但评测对象、可用能力、操作方式和计费构成差异明显。本页面为开发者做技术选型时提供横向参考。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| 评测对象 | 已发布的智能体应用 / 工作流应用（可含 RAG 知识库） | 单个文本生成类模型（含调优模型） | -| 评测目标 | 评估应用回答质量、定位 RAG BadCase、给出调优建议 | 评估模型基础能力、对比选型、验证调优效果 | -| 评测模式/方式 | 自动评测、手动评测；单应用 / 多应用横向（最多 8 个） | 自定义评测、基线评测（公开标准数据集） | -| 评测集/数据来源 | 旧版（对话分析 `.xls`/`.xlsx`、知识问答 `.jsonl`）、新版（智能体 / 工作流 / 自定义，含版本管理）；可从应用观测导入 | 评测数据集（Prompt + Completion，产生推理费用）或推理结果集（已含 Output，免推理费用） | -| 评分/评估器类型 | 评估器：LLM 评估器、Code 评估器、预置模板；每任务最多 10 个 | 评测维度：大模型评估（数值/分类）、规则评估（文本相似度/字符串匹配）、人工评估（分类） | -| 支持模型 | 评测集生成与评分当前仅 `qwen-max` / `qwen-plus` | 被评测：文本生成类模型；裁判模型推荐千问-Max | -| 报告能力 | 总正确率、BadCase 分析、RAG 归因分析、调优建议 | 综合得分、通过率、逐条评分、可参与排行榜 | -| 操作入口/API | 控制台操作（新旧两套系统） | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | -| 地域限制 | 无特殊地域限制（需开通应用观测） | 基线评测仅北京地域可用 | -| 计费构成 | 调用大模型（评测集生成 + 评分）产生的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费用） | -| 典型场景 | 智能体上线前质量把关、版本迭代对比、RAG 优化闭环 | 模型选型、调优前后能力对比、基础能力基准测试 | - -## 评分/评估规则对比 - -两者的评估规则思路相通，但组织方式不同：应用评测以"评估器"为组件挂载到评测任务，模型评测以"评测维度"作为可复用模板。 - -| 规则类别 | 应用评测（评估器） | 模型评测（评测维度） | -|----------|--------------------|----------------------| -| 语义理解打分 | LLM 评估器（相关性、幻觉、有害性等，产生 Token 费用） | 大模型评估-数值型 / 分类型（裁判模型打分或 Pass/Fail，有费用） | -| 确定性/规则判断 | Code 评估器（Python 规则，无额外费用） | 规则评估-文本相似度（ROUGE/BLEU/Cosine）、字符串匹配（无费用） | -| 人工标注 | 手动评测 + 标签管理（分类/布尔/数字/文本标签） | 人工评估-分类型（Pass/Fail 标注） | - -## 适用场景建议 - -- **优化一个已上线的智能体/RAG 应用** → 选应用评测。其归因分析能把 BadCase 精确定位到"模型理解有误 / 重排不佳 / 检索无效 / 切片不完整 / 未获取知识"等环节，并直接给出 Prompt、检索配置或知识库切片的优化建议，形成"识别—归因—优化—回归"闭环。 -- **在多个候选模型间选型，或验证微调效果** → 选模型评测。基线评测可用 C-Eval、MMLU、GSM8K、BBH 等公开数据集快速摸底；自定义评测配合排行榜可做定量对比。 -- **需要跨应用/跨版本横向对比** → 应用评测的多应用横向评测（同一评测基准下最多 8 个应用/版本，须关联相同知识库）。 -- **需要编程化、可自动化的评测流水线** → 两者当前都以控制台为主，模型评测无公开 API/SDK；如需 CI 集成，模型评测侧可参考 PAI Judge Model API。 - -## 技术选型参考 - -1. **先看评测对象**：评"应用整体表现（尤其 RAG）"用应用评测；评"模型本身能力"用模型评测。二者不可互相替代。 -2. **控制成本**：有确定性标准的场景优先使用规则评估 / Code 评估器（无裁判/LLM 费用）；模型评测可先用 50-100 条小规模验证，再保存推理结果集复用以免除重复推理费用。 -3. **注意能力边界**：应用评测的评测集生成与评分当前仅支持 `qwen-max` / `qwen-plus`；模型评测当前仅支持文本生成类模型，基线评测仅北京地域可用，任务提交后目标模型、评测维度类型均不可修改，选错需删除重建。 -4. **对待评测噪声**：模型评测中 1-3% 的分差通常为噪声，LLM 评分器存在位置偏差与自我偏好偏差，建议定期人工抽查校准；应用评测同样建议在知识库更新、Prompt 调整、模型升级后触发回归评测。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md deleted file mode 100644 index deadf313..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向已构建的智能体应用和工作流应用，评估端到端的输出质量与 RAG 链路效果；**模型评测**面向底层大模型本身，评估模型的推理能力和指令遵循表现。两者的评测对象、数据流、评分机制和适用场景均有显著差异，开发者需要根据当前所处的开发阶段选择合适的评测方式。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| **评测对象** | 智能体应用、工作流应用（已发布的完整应用） | 文本生成类大模型（基础模型或调优后模型） | -| **核心目标** | 验证应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型或调优验证 | -| **评测方式** | 自动评测（单应用 / 多应用横向）、手动评测 | 自定义评测（AI / 规则 / 人工）、基线评测（公开数据集） | -| **评测集来源** | 基于应用关联知识库自动生成，或手动上传 | 手动上传评测数据集，或使用公开标准数据集（C-Eval、MMLU 等） | -| **评估器 / 评分机制** | 新版评估器（LLM 评估器 + Code 评估器 + 预置模板）；旧版由平台内置评分 | 评测维度模板（大模型评估数值型/分类型、规则评估相似度/匹配、人工评估） | -| **归因分析** | 支持 RAG 链路归因（模型理解有误、重排不佳、检索无效、切片不完整、未获取知识） | 不提供链路归因，仅输出维度得分和通过率 | -| **横向对比能力** | 最多 8 个应用同基准横向对比 | 支持多模型评测结果排行榜对比 | -| **人工标注** | 新版通过标签体系支持四种类型标注（分类 / 布尔值 / 数字 / 文本） | 人工评估维度（Pass/Fail 标注） | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可评测 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | -| **API 支持** | 通过控制台操作 | 仅控制台操作，不提供公开 API/SDK（可参考 PAI Judge Model API） | -| **计费构成** | 评测集生成 + 应用调用 + 评估器模型的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用 | - -## 评分体系差异 - -| 对比项 | 应用评测 | 模型评测 | -|--------|----------|----------| -| **评分范围** | 1-5 分（正确率 = 得分 >= 4 的占比） | 可自定义整数区间（默认 0-5，建议不超过 10） | -| **自动评分方式** | LLM 评估器（语义）+ Code 评估器（规则） | 大模型评估（裁判模型）+ 规则评估（ROUGE/BLEU/Cosine/字符串匹配） | -| **评估器数量** | 每任务最多 10 个，建议组合 3-5 个 | 按评测维度配置，无上限说明 | -| **评分模型** | 评测集生成和评估仅支持 qwen-max 和 qwen-plus | 裁判模型推荐千问-Max，被评测模型不限 | - -## 适用场景建议 - -### 优先选择应用评测的场景 - -- 智能体应用已发布上线，需要持续监控输出质量 -- 需要定位 RAG 链路中的具体瓶颈（检索、重排、切片、模型理解） -- 知识库更新或 Prompt 调整后需要回归验证 -- 多个应用版本之间需要横向对比，选出最优配置 -- 需要将人工标注经验固化为自动评估规则（通过评估器模板化） - -### 优先选择模型评测的场景 - -- 项目初期的模型选型，需要在多个候选模型间对比基础能力 -- 模型微调（SFT）后需要验证调优效果是否达标 -- 使用公开基准（C-Eval、MMLU、GSM8K、BBH）快速了解模型通用能力 -- 需要用规则评估（ROUGE/BLEU）做确定性指标验证（如翻译、摘要场景） -- 关注模型推理能力本身，而非上层应用的端到端效果 - -### 组合使用建议 - -典型的开发流程中，两种评测可以分阶段配合使用：先通过**模型评测**完成基础模型选型和调优验证，确定最优模型后构建应用，再通过**应用评测**验证端到端效果并持续迭代优化。 - -## 成本优化对比 - -| 策略 | 应用评测 | 模型评测 | -|------|----------|----------| -| **减少推理费用** | 缩小评测集规模 | 使用推理结果集（复用已有推理输出） | -| **减少评分费用** | 使用 Code 评估器替代 LLM 评估器 | 使用规则评估或人工评估替代大模型评估 | -| **渐进式评测** | 先小规模自动评测，再针对 BadCase 人工复核 | 先 50-100 条验证，再扩大到 200-500 条正式评测 | - -## 来源文档 - -- [application evaluation](../guides/application-evaluation.md) (guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) (guides/model-evaluation-introduction.md) - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md deleted file mode 100644 index a6f32795..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md +++ /dev/null @@ -1,38 +0,0 @@ -# 框架、工具包与 MCP 对比 - -阿里云百炼为开发者提供了多种接入与扩展大模型能力的方式，常见的选择包括三类：**开源框架集成**（LlamaIndex、Spring AI Alibaba）、**[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与官方 SDK/工具包**（compatible-mode/v1、DashScope SDK、LangChain 适配）、以及**模型上下文协议（MCP）服务**。三者面向的诉求不同——框架侧重在既有编程语言生态中拼装 RAG/[智能体应用](../concepts/agent-application.md)；兼容接口族侧重用最小改动复用 OpenAI 代码与生态；MCP 则侧重让智能体/工作流动态调用外部工具与云资源。本文从接入方式、语言/运行时、能力范围、适用场景、计费与限制等维度做横向对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 开源框架（LlamaIndex / Spring AI Alibaba） | [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包 | MCP 服务（官方 + 自定义） | -| --- | --- | --- | --- | -| 接入方式 | 框架 SDK + 百炼云端[知识库](../concepts/knowledge-base.md) / 应用 ID | 替换 `api_key`、`base_url`、`model` 三参数，复用 OpenAI 路径 | 在智能体/工作流中挂载 MCP 服务，或外部通过 Streamable HTTP 调用 | -| 语言/运行时 | Python 3.9+（LlamaIndex）、Java JDK 17+ / Spring Boot 3.x（Spring AI Alibaba） | 任意支持 OpenAI SDK 的语言；官方同时适配 LangChain、LangChain4j | 语言无关（协议层）；自定义服务可由 npx/uvx/http 部署 | -| 鉴权 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY` 等环境变量）；子[业务空间](../concepts/workspace.md)需[业务空间](../concepts/workspace.md) ID | [API Key](../concepts/api-key.md)（推荐 `DASHSCOPE_API_KEY`）；新加坡与北京地域 Key 不同 | [API Key](../concepts/api-key.md) + MCP 服务自身鉴权（如 `Authorization` 头）；仅主账号及授权 RAM 用户可访问自定义服务 | -| 主要能力 | 云端[知识库](../concepts/knowledge-base.md)构建、RAG 应用、调用百炼智能体/工作流应用、[知识库](../concepts/knowledge-base.md)检索 | Chat、Responses、Completions、Embedding、Vision、File、Batch、Conversations | 官方工具（地图、联网搜索等）、自定义脚本工具、封装 RESTful API、操作阿里云 OpenAPI（OSS、ECS 等） | -| 知识库/RAG | LlamaIndex 用云端智能切分与官方向量模型，不支持自定义切分/嵌入；Spring AI Alibaba 通过 `DashScopeDocumentRetriever` 检索百炼知识库 | Embeddings 接口做向量化；RAG 需自行在应用层编排 | 不直接提供 RAG；可作为工具被智能体调用，间接参与检索/查询 | -| 模型范围 | LlamaIndex 传 `qwen-max` 等；Spring AI Alibaba 调用智能体/工作流应用（应用背后绑定模型） | Qwen 全系（商业/开源/VL/Coder/Omni/Math）、DeepSeek、Kimi、GLM、MiniMax 等；Responses 支持 qwen3-max/plus/flash、qwen3-coder-plus 等 | 由承载 MCP 的智能体/工作流模型决定；调用准确性依赖提示词，必要时换用千问 3 系列等更强推理模型 | -| 调用形态 | 非流式与流式（Spring AI Alibaba `agent.call` / `agent.stream`） | 流式与非流式；Responses 支持 `previous_response_id` 多轮接续；Batch 异步批量（费用 50%） | 智能体自动判断是否调用；工作流中每节点单工具、手动串联；外部调用走 Streamable HTTP | -| 计费方式 | 按所调用模型/知识库的百炼标准计费 | 按模型 token 计费；Batch/Batch Chat 半价；Responses 上下文关联 7 天 | 云部署 MCP 限时免部署费；联网搜索 2000 次免费后 29 元/千次；自定义基础模式 0.000156 元/秒，极速模式另加 0.000036 元/秒部署时长 | -| 典型场景 | 已使用 Python/Java 生态、希望以框架方式构建 RAG 或集成百炼智能体 | 已有 OpenAI/LangChain 代码、希望低成本迁移或复用生态工具 | 让智能体/工作流动态调用第三方工具或阿里云资源，避免逐个写接口 | - -## 适用场景建议 - -- **选开源框架（LlamaIndex / Spring AI Alibaba）**：团队以 Python 或 Java/Spring 为主技术栈，希望以框架抽象快速搭建 RAG 应用或集成百炼智能体/工作流应用，且可接受云端智能切分与官方向量模型（LlamaIndex）或预先在控制台创建应用/知识库（Spring AI Alibaba）。若需要完全自定义文档切分与嵌入模型，LlamaIndex 路线并不适合，应改用本地知识库方案。 -- **选 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包**：已有基于 OpenAI SDK 或 LangChain/LangChain4j 的存量代码，希望以最小改动（`api_key`/`base_url`/`model`）迁移到百炼，或需要使用 Completions（FIM 代码补全）、Batch（半价批量推理）、Responses（智能体原生能力、内置工具、多轮接续）等专项接口。适合追求协议兼容、跨语言复用与生态工具接入的团队。 -- **选 MCP 服务**：核心诉求是让百炼智能体或工作流在运行时动态调用外部工具（地图、联网搜索、自建脚本、RESTful API、阿里云 OSS/ECS 等），而非固定编写接口。适合需要多工具协同、逐步推理、或把已有业务 API 快速封装给模型使用的场景。注意 MCP 只能在智能体/工作流应用中使用，不能在直接调用千问 API 时接入；且会因工具返回内容进上下文而增加 token 消耗。 - -## 技术选型参考 - -1. **先明确诉求边界**：是"迁移/复用现有 OpenAI 代码"（走兼容接口族）、"用框架拼装 RAG/[智能体应用](../concepts/agent-application.md)"（走开源框架），还是"让运行时智能体动态调用外部工具"（走 MCP）。三者并非互斥，常组合使用——例如用兼容接口族做模型调用，同时用 MCP 扩展工具能力。 -2. **地域与鉴权一致性**：兼容接口族需按[业务空间](../concepts/workspace.md)专属域名拼装 `base_url`，弗吉尼亚地域使用固定域名且不带 `{WorkspaceId}`；Spring AI Alibaba 应用集成与知识库检索对 API Key 变量名约定不同（`DASHSCOPE_API_KEY` vs `AI_DASHSCOPE_API_KEY`），关键是 `application.yml` 占位符与实际变量名一致；子业务空间一律需要业务空间 ID。 -3. **能力限制与成本**：LlamaIndex 云端方案不支持自定义切分/嵌入；Completions 仅限北京地域；MCP 单智能体最多 5 个服务、工作流每节点单工具、自定义服务托管在 FC 无固定出口公网 IP（访问云资源需配白名单或打通 VPC）。批量推理与 Batch Chat 可享 50% 费用优惠；MCP 联网搜索有免费额度与 QPS 限制，自定义服务按响应速度分基础/极速两种计费。 -4. **协议演进**：MCP 已从旧版 SSE 升级为 Streamable HTTP，已开通用户需"取消开通"后重新"立即开通"完成升级；Responses API 旧版路径即将停用，应使用 `/compatible-mode/v1/responses`，且 `previous_response_id` 传顶层 `id`（UUID，有效期 7 天）。 - -## 被对比主题页 - -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [model context protocol](../guides/model-context-protocol.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md deleted file mode 100644 index 033ebdfe..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md +++ /dev/null @@ -1,73 +0,0 @@ -# 模型微调 vs 模型压缩 vs 模型部署 - -模型微调、模型压缩与模型部署是百炼平台模型生产链路中的三个关键环节。开发者在将基础模型投入生产时，需要理解这三者的定位、能力边界和协作关系，以便做出最佳的技术选型。三者的典型流转顺序为：模型微调 → 模型压缩（可选）→ 模型部署，但并非所有场景都需要经历完整链路。 - -## 关键维度对比 - -| 维度 | 模型微调 | 模型压缩 | 模型部署 | -|------|----------|----------|----------| -| 核心目标 | 基于自有数据定制模型能力 | 降低模型精度以减小部署规格和成本 | 将模型上线为可调用的推理服务 | -| 在链路中的位置 | 上游（生产模型） | 中游（优化模型，可选） | 下游（上线模型） | -| 输入 | 基础模型 + 训练数据集（JSONL/ZIP） | 微调产出的全精度自定义模型 | 预置模型或微调/压缩后的自定义模型 | -| 输出 | 微调后的自定义模型 | 量化后的低精度模型 | 在线推理服务（API 端点） | -| 操作方式 | 控制台 + DashScope API | 仅控制台 | 控制台 + DashScope API | -| 支持的模型范围 | Qwen3/2.5 系列、千问VL、Wan图像/视频、CosyVoice | 仅百炼平台微调产出的特定模型 | 预置模型 + 微调模型 + 压缩模型 + OSS导入LoRA模型 | -| 计费方式 | 按训练 [Token](../concepts/token.md) 用量 x 循环次数 | 限时免费（截止时间以控制台公告为准） | PTU / 模型单元(MU) / [Token](../concepts/token.md) 用量 三选一 | -| 可逆性 | 可基于同一基础模型多次微调 | 不可逆，压缩后无法继续微调或二次压缩 | 可随时上线/下线，切换计费需重新部署 | -| 地域限制 | 无特殊限制 | 仅华北2（北京） | 无特殊限制 | -| 耗时 | 取决于数据量和训练轮次，通常数小时 | 取决于模型规格和量化模板 | 部署上线通常数分钟 | - -## 方法详解与选择 - -### 模型微调：让模型学会新能力 - -模型微调适用于需要模型掌握特定领域知识或遵循特定行为模式的场景。百炼提供三种递进式方法： - -- **CPT（继续预训练）**：注入领域知识，需 1000 万+ [Token](../concepts/token.md) 无标签文本 -- **SFT（监督微调）**：学会遵循指令，需 1000+ 条高质量问答对 -- **DPO（直接偏好优化）**：对齐人类偏好，需 100+ 组正负样本对 - -每种方法支持全参训练和高效训练（LoRA）两种模式，LoRA 收敛快、成本低，适合快速验证。 - -### 模型压缩：在精度与成本间取舍 - -模型压缩通过量化技术降低参数精度，实现部署成本优化。例如 qwen3.5-flash 微调模型压缩后部署规格从 MU1x2（108 元/小时）降至 MU8x1（47 元/小时），节省约 56%。需要注意： - -- 仅支持百炼平台微调产出的自定义模型 -- 压缩不可逆，建议在免费期尝试多个量化模板后择优上线 -- 校准数据应与目标推理场景语义相近 - -### 模型部署：将模型能力转化为服务 - -模型部署是链路的最终环节，三种计费方式适用于不同业务特征： - -- **PTU（预置吞吐）**：流量稳定的高负载生产环境，TPS 提升约 1.5-2.0 倍 -- **模型单元（MU）**：资源独占，性能可自定义，支持 PD 分离 -- **Token 用量**：按调用量付费，仅支持部分 LoRA 调优模型 - -## 典型场景与推荐路径 - -| 场景 | 推荐路径 | 说明 | -|------|----------|------| -| 快速验证 PoC | 微调(SFT-LoRA) → 部署(Token用量) | 最低成本，按调用付费 | -| 垂直领域生产 | CPT → SFT → 部署(PTU/MU) | 效果最佳，适合稳定流量 | -| 成本敏感的生产环境 | SFT → 压缩 → 部署(MU) | 压缩可节省 50%+ 部署成本 | -| 高并发低延迟 | 微调 → 部署(PTU) | 预留资源保障吞吐，支持溢出按量兜底 | -| 已有本地 LoRA 模型 | OSS 导入 → 部署(Token用量/MU) | 跳过平台微调，直接部署 | -| 多模态定制（图像/视频/语音） | 微调(SFT-LoRA) → 部署(MU) | 压缩暂不支持多模态模型 | - -## 技术选型建议 - -1. **先明确需求再选路径**：如果预置模型已能满足需求，直接部署即可；如果需要定制能力，先微调再部署。 -2. **成本优化考虑压缩**：当部署成本是瓶颈时，在微调和部署之间插入压缩环节，可显著降低 MU 规格。 -3. **计费方式匹配流量模式**：流量稳定选 PTU/MU，流量波动或低频调用选 Token 用量。 -4. **注意压缩的不可逆性**：压缩前应充分验证微调效果，压缩后无法回退。 -5. **善用免费期**：模型压缩限时免费，建议在此期间多方案对比测试，选出精度与成本的最优平衡点。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md deleted file mode 100644 index 025bfd66..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -[模型微调（Fine-tuning）](../concepts/fine-tuning.md)与模型压缩（量化）是百炼平台模型生产链路中两个相邻但目标截然不同的环节。微调解决的是「模型能不能做好某项任务」的**能力问题**，通过训练修改模型参数来提升特定行业/业务表现；模型压缩解决的是「模型部署贵不贵」的**成本问题**，通过量化把全精度微调模型转为低精度版本，从而降低部署所需的 MU 规格。二者在完整链路中的位置为：**模型调优 →（可选）模型压缩 → 模型部署**——压缩的输入正是微调的产出。本文从技术选型角度对比两者的关键差异，帮助开发者判断在什么阶段该用哪个能力。 - -## 关键维度对比 - -| 维度 | [模型微调（Fine-tuning）](../concepts/fine-tuning.md) | 模型压缩（量化） | -| --- | --- | --- | -| 核心目的 | 提升模型在特定任务/领域的能力 | 降低部署规格与推理成本 | -| 处理对象 | 基础模型（千问系列、VL、图像/视频、语音等） | 仅限百炼平台微调产出的自定义模型 | -| 输入 | 训练数据集（ChatML / 纯文本 / ZIP 音视频等） | 上游全精度微调模型 + 可选校准数据集 | -| 输出 | 新的自定义微调模型 | 低精度（量化）版本的自定义模型 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA 高效训练） | 量化（不含结构剪枝、知识蒸馏） | -| 使用方式 | 控制台可视化 或 API/命令行（DashScope HTTP） | 控制台：模型 → 模型训练 → 模型压缩 → 创建压缩任务 | -| API 端点 | `POST /api/v1/files`、`/api/v1/fine-tunes`、`/api/v1/deployments` | 以控制台操作为主（未提供公开 API 流程） | -| 关键配置 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 量化模板（MU 编号）、量化产出后缀、校准数据 | -| 计费方式 | 按训练 Token 计费（CosyVoice 0.2 元/千 Tokens；部署另计） | 压缩任务本身限时免费；压缩后模型按部署 MU 规格计费 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 可逆性/复用 | 微调模型可继续被压缩、部署 | 不可逆；压缩后不支持继续微调或二次压缩 | -| 典型收益 | 指令遵循、领域知识、偏好对齐、风格定制 | 部署成本下降（示例约节省 56%） | -| 典型场景 | 通用模型无法满足行业需求、需注入专业知识或对齐偏好 | 微调模型已达标、需在保持能力前提下压低上线成本 | - -## 适用场景建议 - -### 优先选择模型微调 - -- Prompt 工程、插件调用等方法已用尽，仍无法满足业务对准确率、专业性或风格的要求。 -- 需要向模型注入领域词汇与事实（CPT）、教会模型遵循特定指令或执行任务（SFT）、或对齐人类偏好并抑制幻觉（DPO）。这三种方式可按 `CPT（可选）→ SFT → DPO（可选）` 递进组合。 -- 涉及多模态定制：视觉理解（VL，仅 SFT）、图像/视频生成（SFT-LoRA，需触发词）、语音合成（CosyVoice，仅 API 发起）。 -- 建议：若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）；对成本/时间敏感或数据集较小时选 LoRA 高效训练。 - -### 优先选择模型压缩 - -- 已经拥有一个训练达标的全精度微调模型，主要痛点是**部署/推理成本偏高**。 -- 希望在尽量保持模型能力的前提下降低部署 MU 规格（MU 编号越大规格越小、成本越低，但精度损失可能越大）。 -- 建议：利用压缩任务限时免费的窗口，对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选出成本与精度平衡最优的方案再正式上线；若面向特定场景（如客服问答），校准数据应选语义相近的数据集以提升量化精度。 - -### 二者结合的典型链路 - -对绝大多数生产落地而言，两者不是「二选一」而是「先后使用」：先用**微调**把模型能力打磨到位并验证效果，再用**压缩**在上线前压低部署成本。需特别注意压缩不可逆——若后续还想继续微调或调整，必须回到上游全精度微调模型重新压缩，因此应先冻结微调版本、确认效果达标后再进入压缩环节。 - -## 技术选型速查 - -- 目标是「让模型更会做事」→ 微调。 -- 目标是「让模型更省钱部署」→ 压缩。 -- 需要通过 API/命令行自动化全流程（上传数据 → 训练 → 部署）→ 微调具备完整 HTTP 接口；压缩目前以控制台操作为主。 -- 数据集较小或预算有限 → 微调选 LoRA 高效训练；压缩选较大 MU 编号模板并做好精度验证。 -- 两者均受**华北2（北京）地域**约束，须使用该地域 API Key 与工作空间。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md deleted file mode 100644 index b77b31c6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -在百炼平台的模型生产链路中，模型微调（Fine-tuning）与模型压缩（量化）是两个位置相邻但目标截然不同的环节。完整链路为：**模型调优 → 模型压缩（可选）→ 模型部署**。微调解决的是「模型效果不够好」的问题，通过修改模型参数让模型在特定行业/业务上表现更佳；模型压缩解决的是「模型部署成本太高」的问题，通过降低参数精度（量化）在尽量保持能力的前提下缩小部署所需的 MU 规格。两者并非替代关系，而是先后衔接：通常先微调得到高精度自定义模型，再对其进行压缩以降低推理成本。本文面向开发者，对两者的关键维度做对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | -| --- | --- | --- | -| 核心目标 | 提升模型在特定任务/领域的效果 | 降低部署 MU 规格，减少推理成本 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA） | 量化（不含结构剪枝、知识蒸馏） | -| 链路位置 | 生产链路起点 | 位于微调与部署之间（可选环节） | -| 输入对象 | 基础模型 + 训练数据集 | 百炼平台微调产出的自定义模型 | -| 输入数据格式 | ChatML（SFT/DPO）、纯文本（CPT）、ZIP/OSS（[多模态](../concepts/multimodal.md)）等 | 源模型 +（条件选填）最多 5 个校准数据集 | -| 输出产物 | 微调后的自定义模型（可部署、可再微调） | 低精度自定义模型（不可再微调、不可二次压缩） | -| 支持模型 | 千问文本/VL、Wan 图像/视频、CosyVoice 语音等[多模态](../concepts/multimodal.md) | 以控制台展示为准，如 qwen3.5-flash-2026-02-23 微调模型 | -| 使用方式 | 控制台（可视化）或 API/命令行（DashScope HTTP） | 控制台操作：模型训练 → 模型压缩 → 创建压缩任务 | -| 关键 API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/`、`POST /api/v1/deployments` | 暂无公开 API，通过控制台创建压缩任务 | -| 关键配置项 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 任务名称、源模型、量化模板（MU 编号）、量化后缀、校准数据 | -| 任务状态 | PENDING → RUNNING → SUCCEEDED | PENDING → QUEUING → RUNNING → SUCCEEDED / FAILED / CANCELED | -| 计费方式 | 按训练 Token 计费；CosyVoice 0.2 元/千 Tokens + 部署时长 | 压缩任务限时免费，压缩后模型按部署 MU 规格计费 | -| 可逆性 | 微调模型可继续训练、再压缩 | **不可逆**，不支持继续微调或二次压缩 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 典型场景 | 注入领域知识、学会指令遵循、对齐人类偏好、定制风格 | 高精度模型上线前的成本优化 | - -## 各方案的适用场景建议 - -### 优先使用模型微调 - -- **Prompt 工程/插件调用仍无法满足需求**：需要模型掌握专业词汇、事实性知识（CPT），或稳定地遵循特定指令与任务格式（SFT）。 -- **需要对齐人类偏好、抑制幻觉**：在 SFT 之上叠加 DPO，用「更好-更差」回答对进一步优化。 -- **[多模态](../concepts/multimodal.md)定制**：图像/视频生成的风格定制（Wan LoRA + 触发词）、视觉理解任务、语音合成音色克隆（CosyVoice，仅 API）。 -- **效果为先且能接受较高部署规格**：官方推荐若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）。 - -### 适合使用模型压缩 - -- **已有微调完成的高精度自定义模型，且部署成本偏高**：例如需要长期在线服务、对每小时 MU 费用敏感的场景。以 qwen3.5-flash 微调模型为例，压缩后可从 MU1\*2（108 元/小时）降至 MU8\*1（47 元/小时），成本节省约 56%。 -- **业务对少量精度损失可接受**：MU 编号越大部署规格越小、成本越低，但精度损失可能越大，需按业务权衡。 -- **有语义相近的校准数据**：客服问答等场景应选择贴近实际推理语义的校准数据集，以提升量化精度。 - -## 面向开发者的技术选型参考 - -1. **先明确瓶颈**：效果不达标 → 微调；效果已达标但推理太贵 → 压缩。二者不冲突，常规链路是「先微调，再按需压缩」。 -2. **顺序与不可逆性**：压缩必须以微调产出的自定义模型为输入，且压缩不可逆。若后续还想继续微调或调整训练，务必保留上游全精度微调模型；压缩模型无法二次微调或二次压缩。 -3. **成本估算**：微调按训练 Token 计费（CosyVoice 另计），部署按 MU 规格计费；压缩任务本身限时免费，收益体现在部署阶段更低的 MU 规格。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证效果，选最优方案上线。 -4. **接入方式差异**：微调支持控制台与 API/命令行（可编排到 CI/CD），压缩目前以控制台操作为主。 -5. **共同约束**：两者均仅在华北2（北京）可用，须使用该地域 API Key；压缩仅支持百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md deleted file mode 100644 index 8b952a86..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md +++ /dev/null @@ -1,54 +0,0 @@ -# 图像、视频与3D生成对比 - -百炼平台把图像、视频、3D 三类生成能力统一收口在模型推理网关下，使用相同的 `Authorization: Bearer ` 鉴权与 JSON 承载的输入输出协议，但三者在输入模态、输出形态、调用模式、耗时、计费维度与典型场景上差异显著。本文面向需要做多模态生成技术选型的开发者，横向对比三类能力的关键维度，帮助快速锁定适合业务的接口族。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 输入格式 | 文本 [prompt](../guides/prompt.md)；参考图/蒙版（URL 或 Base64） | 文本；首帧/首尾帧图像；参考多图；视频；音频 | 文本 [prompt](../guides/prompt.md)（≤1024 字符）；单图 URL；多图（前/左/后/右 4 视角，固定数组） | -| 输出格式 | 单张或多张静态图像（临时 URL） | 视频文件（临时 URL） | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| 支持模型族 | 通义千问图像、万相（Wan）、Z-Image、可灵、创意工具系列 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| API 端点 | OpenAI 兼容 `/compatible-mode/v1/images/generations`；或 DashScope 原生 `/services/aigc/text2image/image-synthesis` | `/services/aigc/video-generation/video-synthesis`（万相2.7 等）；或 `/services/aigc/image2video/video-synthesis`（动作/换人/数字人/旧版首尾帧） | `/services/aigc/video-generation/3d-generation` | -| 调用模式 | 同步（兼容模式）或异步轮询（万相/创意工具，返回 task_id） | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | -| 必要请求头 | `Authorization`；异步任务需 `X-DashScope-Async: enable` | `Authorization`、`Content-Type: application/json`、`X-DashScope-Async: enable`（缺异步头报 `does not support synchronous calls`） | `Authorization`、`X-DashScope-Async: enable` | -| 典型耗时 | 秒级（同步）到数十秒（异步） | 1–5 分钟；视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 异步任务 24 小时 | 24 小时 | 24 小时；超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 临时 URL，需及时下载/转存 OSS | 临时 URL | 2 小时 | -| 地域约束 | 各地域通用，按模型开通 | 模型/Endpoint/API Key 须同地域；PixVerse、Vidu 仅华北2（北京） | 仅华北2（北京），且须用北京 API Key | -| 内容安全 | 内置审核，违规返回 `DataInspectionFailed` | 内置审核 | 内置审核，失败返回 `code`/`message` | -| 计费方式 | 按张/按次（视模型） | 按任务/时长 | 按任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）计数，`usage` 含生成数量 | -| 典型场景 | 文生图、图像编辑、电商/营销垂类、人像玩法、艺术文字 | 文生视频、图生视频、视频编辑、数字人、肖像动态、风格重绘 | 文生 3D、单图生 3D、多图生 3D，产出可二次加工的 GLB 资产 | - -## 调用模式差异 - -图像生成是三类中唯一支持**同步调用**的：千问-文生图等可走 OpenAI 兼容 `images/generations` 直接拿结果。万相与创意工具则多用 DashScope 原生异步协议，提交后拿 `task_id` 轮询。 - -视频与 3D 一律异步，且都强制 `X-DashScope-Async: enable` 头。视频接口明确要求"同地域"约束（模型、Endpoint、API Key 必须同地域），3D 则更严格——仅华北2（北京）可用。两者都强调"请勿重复创建任务，直接轮询"。 - -## 输入模态对比 - -- **图像生成**：以文本 [prompt](../guides/prompt.md) 为主，部分编辑接口接受参考图与蒙版，输入图支持公网 URL 或 Base64。 -- **视频生成**：多模态输入最丰富，万相2.7 支持文本/图像/音频/视频混合输入，参考生视频可保持角色与音色一致性；首尾帧、参考多图等模式扩展了可控性。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥。多图模式视角顺序固定为前/左/后/右，数组长度固定 4，不需要的视角传空对象 `{}`，这是 3D 独有的约束。 - -## 输出与产物处理 - -图像与视频输出都是临时 URL，文档建议及时下载或转存 OSS。3D 输出更结构化：根据 `pbr`/`texture` 参数组合返回 `pbr_model_url`（PBR 材质 GLB）或 `base_model_url`（无贴图基础模型），并附 `rendered_image_url` 预览图，链接有效期仅 2 小时，短于图像/视频的临时 URL 生命周期。 - -## 选型建议 - -- **静态视觉物料（海报、商品图、人像）**：选图像生成。中文语义优先千问-文生图或万相-文生图 V2；按指令改图用万相-图像生成与编辑 2.7；电商/营销垂类用虚拟模特、AI 试衣、创意海报；高美感/艺术风格用 Z-Image 或可灵。 -- **动态视频内容（短剧、营销视频、数字人播报）**：选视频生成。通用文生/图生优先万相2.7（`wan2.7-t2v`/`wan2.7-i2v`）；多镜头叙事用万相2.6 `shot_type: multi`；数字人/换人用 `wan2.2-s2v`、`wan2.2-animate-mix`；北京地域可按需选 PixVerse、Vidu。 -- **可复用 3D 资产（游戏、电商 3D 展示、工业建模）**：选 3D 生成，仅限北京地域。高精度需求用 `Tripo/Tripo-H3.1`（最高 200 万面，`geometry_quality: ultra`）；追求速度用 `Tripo/Tripo-P1.0`（最高 2 万面）。需要 PBR 材质保留默认 `pbr: true`，仅需白模则同时关 `texture` 与 `pbr`。 -- **跨模态组合**：可先用图像生成产出关键帧，再喂给视频生成的图生/首尾帧接口生成动态内容；3D 则更适合独立资产管线，与图像/视频管线并行而非串行。 - -新接入一律优先最新版本（万相 2.7、通用图像编辑 2.5、wan2.7、Tripo-H3.1/P1.0），旧版接口保留兼容但不再增强。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md deleted file mode 100644 index f6d09469..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md +++ /dev/null @@ -1,60 +0,0 @@ -# 入门路径对比：开始使用与模型快速上手 - -阿里云百炼同时面向"应用构建者"和"模型调用开发者"两类用户，提供了两条不同的入门路径。`start-using` 侧重于在控制台零代码搭建基于私有知识的问答应用（智能体应用、工作流应用、知识库等）；`get-started-with-models` 则侧重于通过兼容 OpenAI 的 API 直接调用大模型完成第一次推理。本页从目标用户、输入形式、输出形式、支持模型、API 端点、计费方式、典型场景等维度对比两条路径，帮助开发者根据需求做出技术选型。 - -## 关键维度对比 - -| 维度 | [start using](../guides/start-using.md)（应用构建路径） | [get started with models](../guides/get-started-with-models.md)（模型调用路径） | -| --- | --- | --- | -| 主要目标 | 零代码/低代码搭建端到端的私有知识问答应用 | 通过 API 直接调用大模型完成文本/多模态推理 | -| 目标用户 | 业务人员、应用开发者、希望快速上线 RAG 应用的团队 | 开发者、需要将模型推理集成到自有代码或后端服务的工程师 | -| 上手时长 | 约 5 分钟完成第一个智能体应用（含知识库构建） | 几行代码即可完成首次调用，开通账号后即可跑通 | -| 输入形式 | 控制台可视化配置：System Prompt、知识库、技能、MCP 工具、工作流节点 | HTTP 请求 / OpenAI 兼容 SDK / DashScope SDK，传入 `messages`、`model`、参数 | -| 输出形式 | 发布后的应用（Web、微信、钉钉、音视频实时互动）+ Responses API 同步/异步调用 | 模型推理结果（文本、图像、音频、视频、向量、重排序结果） | -| 应用类型 | 智能体应用（Agent 2.0）、工作流应用、高代码应用、MCP 服务 | 直接调用模型 API；可叠加模型调优（SFT/CPT/DPO）、模型部署、模型评测 | -| 支持模型 | 千问系列（推荐 qwen3.7-max 用于问答）、QwQ 推理系列、DeepSeek 系列、视觉模型 qwen-vl-plus/max、嵌入 text-embedding-v3/v4 | 文本生成（qwen3.7-max/plus、qwen3.6-flash、deepseek-v4-pro/flash、kimi-k2.7-code、glm-5.2 等）、图像/视频/3D/音频/全模态/向量/重排序全谱系 | -| 知识库 | 核心能力：文档/数据/图片/音视频知识库、智能切分、检索调优、图文检索、监控 API | 不直接提供知识库；如需 RAG 需自行搭建或与百炼应用/知识库集成 | -| API 端点 | Responses API（同步 + `background=true` 异步），调用时需传入自定义参数；通过应用 ID 调用 | OpenAI 兼容 `/compatible-mode/v1`、Anthropic 兼容地址、DashScope SDK 端点；按计费方案选择 Base URL | -| 接入域名 | 应用调用走百炼统一接入，发布渠道含微信/钉钉/音视频 SDK | 业务空间专属 `{WorkspaceId}.{region}.maas.aliyuncs.com`（生产推荐）、`dashscope.aliyuncs.com`（存量）、`trial.{region}.maas.aliyuncs.com`（验证）、Token/Coding Plan 专属域名 | -| 凭证与鉴权 | 控制台操作为主；API 调用时使用应用相关凭证 | API Key（按业务空间隔离，不可跨地域混用）；建议写入环境变量 `DASHSCOPE_API_KEY` | -| 地域选择 | 应用开发与批量推理、模型调优仅在北京与新加坡支持 | 多地域可选（北京、新加坡、法兰克福、东京、弗吉尼亚），地域决定接入点与数据存储；服务部署范围可限定中国内地/国际/全球 | -| 计费方式 | 大模型调用计费 + 知识库规格费用 + 知识库模型调用费用（2026-01-04 起正式计费）；提供限时免费额度 | 按量付费（Dashscope 域名 / 业务空间专属 / 试用域名）、Token Plan（交互式，不可用于后端）、Coding Plan（AI 编码套餐）；不同方案对应不同 Base URL | -| 调用模式 | 同步调用（实时交互，可复用 OpenAI 代码库）、异步调用（`background=true` 返回 Task ID） | 同步、流式（SSE）、批量推理、异步任务；请求超时最高 3600 秒（业务空间专属域名） | -| 可观测性 | 应用观测（端到端流程）、应用评测（智能体/工作流/自定义评测集）、长期记忆与用户画像 | 模型告警（仅北京/新加坡）、模型评测、模型调优 | -| 典型交付物 | 可对外发布的问答应用（含欢迎语、预设问题、发布渠道、音视频互动） | 一次模型推理调用或集成了模型能力的后端服务 | - -## 适用场景建议 - -### 选择 [start using](../guides/start-using.md)（应用构建路径）当： - -- 需要在 5 分钟内零代码搭建一个能回答私有领域问题的问答应用。 -- 业务方或非工程师角色希望可视化配置 System Prompt、知识库、技能、MCP 工具。 -- 需要完整 RAG 流程（知识库构建、切分策略、检索调优、图文检索）且不想自行实现。 -- 需要发布到微信、钉钉、H5/APP 等渠道，或需要音视频实时互动。 -- 需要端到端应用观测、评测、长期记忆与用户画像管理。 - -### 选择 [get started with models](../guides/get-started-with-models.md)（模型调用路径）当： - -- 开发者需要将大模型推理直接集成到自有代码、后端服务或 AI 工作流中。 -- 需要使用 OpenAI 兼容 SDK 或 DashScope SDK，复用现有 OpenAI 代码库。 -- 需要访问文本、图像、视频、3D、音频、全模态、向量/重排序等完整模型谱系。 -- 对地域、服务部署范围、接入域名、并发上限、SLA 有精细控制需求。 -- 需要使用 Token Plan / Coding Plan 等专项计费方案，或需要进行模型调优（SFT/CPT/DPO）、模型部署、模型评测。 - -### 两条路径结合使用： - -实际项目中两条路径常常互补——先用 `get-started-with-models` 跑通模型调用、选定合适模型与地域，再用 `start-using` 将模型能力封装为带知识库、技能、发布渠道的完整应用；反过来，已发布的应用也可通过 Responses API 被后端服务以 OpenAI 兼容方式调用。开发者可先明确"我要的是模型推理能力还是端到端应用"，再据此选择起点。 - -## 来源主题页 - -- [start using](../guides/start-using.md)（guides/start-using.md） -- [get started with models](../guides/get-started-with-models.md)（guides/get-started-with-models.md） - -## 被对比主题页 - -- [start using](../guides/start-using.md) -- [get started with models](../guides/get-started-with-models.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md deleted file mode 100644 index c57c9b93..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成 - -百炼平台同时提供图像生成、视频生成和 3D 模型生成三大视觉内容创作能力。三者在输入输出格式、模型生态、调用方式、计费模式和适用场景上存在显著差异。本文从开发者技术选型角度，对三类能力进行系统对比，帮助开发者根据业务需求选择最合适的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| **输入格式** | 文本([prompt](../guides/prompt.md))、参考图片、涂鸦草图 | 文本([prompt](../guides/prompt.md))、首帧图片、参考图/视频、音频 | 文本([prompt](../guides/prompt.md))、单图、多图(4视角) | -| **输出格式** | PNG 图片(512x512 至 4K) | MP4 视频(5-10秒，1280x720等) | GLB 模型(带PBR材质贴图或无贴图) | -| **核心模型** | 千问(Qwen-Image)、万相(Wan)、Z-Image、可灵(Kling) | 万相(Wan 2.7)、HappyHorse、Pixverse、Vidu、Kling | Tripo-H3.1、Tripo-P1.0 | -| **调用方式** | 同步/异步均支持 | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | -| **典型响应时间** | 秒级至十秒级 | 1-5 分钟 | 分钟级(需轮询，建议间隔15秒) | -| **API端点** | [DashScope SDK](../concepts/dashscope-sdk.md) / HTTP | `/api/v1/services/aigc/video-generation/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持地域** | 多地域(部分模型仅北京) | 多地域(北京、新加坡等) | 仅华北2(北京) | -| **SDK兼容性** | [DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | OpenAI兼容SDK、[DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | 仅HTTP | -| **产物有效期** | 即时返回，无时效限制 | task_id 24小时有效 | 下载链接2小时有效，task_id 24小时有效 | -| **模型数量** | 20+ 模型(含创意工具) | 10+ 模型系列 | 2 个模型 | -| **编辑能力** | 局部重绘、风格迁移、扩图、超分 | 视频重绘、风格转换、口型替换 | 无编辑能力 | - -## 输入输出能力详细对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 文生内容 | 支持(文生图) | 支持(文生视频) | 支持(文生3D) | -| 图生内容 | 支持(图像编辑、参考生图) | 支持(图生视频、参考生视频) | 支持(单图/多图生3D) | -| 多模态输入 | 多图参考、涂鸦 | 关键帧序列、音频驱动 | 4视角多图(前左后右) | -| 批量生成 | 单次1-9张 | 单次1条视频 | 单次1个模型 | -| 最大输出分辨率 | 4K(wan2.7-image-pro) | 1920x1080 | 最高200万面(H3.1) | - -## 计费与商业化对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 计费单位 | 按张计费 | 按任务/时长计费 | 按任务计费 | -| 免费体验 | 部分模型有免费额度 | 部分模型有免费额度 | 需开通Tripo服务 | -| 商业化程度 | 大部分已商业化 | 主力模型已商业化 | 已商业化 | - -## 适用场景建议 - -**选择图像生成的场景**： - -- 电商商品图、营销海报、社交媒体配图等静态视觉内容制作 -- 需要精确文字渲染(如广告文案嵌入图片) -- 图像局部编辑、风格转换、AI试衣等垂直场景 -- 对响应速度要求高(秒级出图) -- 需要批量生成多张候选图供筛选 - -**选择视频生成的场景**： - -- 短视频内容创作、广告视频制作 -- 数字人驱动(音频/文本驱动说话、唱歌) -- 服装展示、舞蹈动作等动态展示 -- 需要关键帧精确控制镜头运动 -- 已有视频的风格转换或口型替换 - -**选择3D生成的场景**： - -- 游戏资产、AR/VR场景中的3D物体快速原型 -- 电商3D商品展示(可旋转查看) -- 建筑/工业设计的概念验证模型 -- 需要标准PBR材质的可渲染模型 - -## 技术选型决策参考 - -1. **内容维度**：静态画面选图像生成；需要动态表现选视频生成；需要空间立体展示选3D生成。 -2. **时效要求**：图像生成响应最快(秒级)，适合实时交互；视频和3D均为分钟级异步任务，适合离线批处理。 -3. **生态成熟度**：图像生成模型最丰富、功能最全面；视频生成处于快速发展期，模型迭代频繁；3D生成目前模型较少，但输出质量已达可用水平。 -4. **地域限制**：3D生成仅限北京地域，部分图像创意工具同样限北京；视频生成地域覆盖较广。 -5. **集成复杂度**：图像生成支持同步调用，集成最简单；视频和3D需要实现异步轮询逻辑，建议封装任务状态管理层。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md new file mode 100644 index 00000000..8b4f6c1a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md @@ -0,0 +1,70 @@ +# 图像、视频与3D生成能力对比 + +为帮助开发者快速理解百炼平台在多模态生成领域的技术边界与工程适配特性，本文系统对比图像生成（Image Generation）、视频生成（Video Generation）与3D生成（3D Generation）三大核心能力。对比聚焦于实际开发中高频关注的技术维度——包括调用模式、模型生态、输入输出约束、地域与计费策略等，旨在支撑产品规划、架构设计与模型选型决策。所有结论均基于当前（2024年Q3）百炼平台正式发布的API文档与运行时行为。 + +## 关键能力维度对比 + +| 维度 | 图像生成 | 视频生成 | 3D生成 | +|------|----------|----------|--------| +| **输入格式** | 文本（`prompt`）、图像URL（`image_url`）、局部掩码（`mask`）、风格参考图（`style_ref_url`）、涂鸦（`sketch`）等多模态组合；支持批量输入（如海报多文案） | 文本（T2V）、单图/首尾帧/参考图（I2V/KF2V/R2V）、视频片段（编辑类）、数字人肖像图（S2V/Emo）；`input` 结构统一为 `{"media": [...], "prompt": "..."}` 或纯 `{"prompt": ...}` | 文本（文生3D）、单张图像URL（单图生3D）、4张有序视角图数组（多图生3D，空位用 `{}` 占位）；三者互斥，不可混合 | +| **输出格式** | JPEG/PNG 图像（URL），支持水印控制；部分模型返回优化提示词（`prompt_extend`）、分割掩码（`instance_mask`）等辅助数据 | MP4 视频（URL），含可选音频轨道；部分模型返回分镜渲染图、关键帧序列；数字人模型额外输出 `.fbx` 或 `.glb` 动作文件 | GLB 模型文件（`pbr_model_url` 带PBR材质 / `base_model_url` 无贴图），及配套预览图 `rendered_image_url`；所有URL有效期仅2小时 | +| **支持模型（代表）** | `qwen-image-2.0-pro`（文字精度）、`wan2.7-image-pro`（4K高清）、`kling/kling-v3-*`（强风格）、`virtualmodel-v2`（电商虚拟模特）、`facechain-portrait-generation`（人物写真） | `wan2.7-t2v/i2v/r2v`（全链路新协议）、`kling/kling-v3-video-generation`（高动态）、`vidu/viduq3-*`（快节奏）、`emo-v1`（悦动人像）、`liveportrait`（灵动人像） | `Tripo/Tripo-H3.1`（高精度，≤200万面）、`Tripo/Tripo-P1.0`（快速，≤2万面）；仅Tripo官方模型，无第三方接入 | +| **API端点（标准路径）** | `/api/v1/services/aigc/image-generation/generation`（同步）
`/api/v1/services/aigc/image-generation/generation_async`（异步） | `/api/v1/services/aigc/video-generation/video-synthesis`（新模型统一路径）
`/api/v1/services/aigc/image2video/video-synthesis`（`wan2.6`及更早旧路径，已弃用） | `/api/v1/services/aigc/video-generation/3d-generation`（注意：路径含`video-generation`但属3D服务，为历史兼容命名） | +| **调用模式** | **混合模式**：
• 同步：`z-image-turbo`、`qwen-image-*`、`wan2.6-t2i` 等（响应 <15s）
• 异步：虚拟模特、背景生成、局部重绘等（需轮询 `task_id`） | **强制异步**：
所有模型均需 `X-DashScope-Async: enable`，创建任务后轮询 `GET /api/v1/tasks/{id}`；任务有效期24小时 | **强制异步**：
必须携带 `X-DashScope-Async: enable`；创建任务后轮询；`task_id` 有效期24小时，结果URL有效期仅2小时 | +| **计费方式** | 按**成功生成的图片张数**计费（非请求次数）；免费额度500张/90天（主账号+RAM共享）；单价因模型而异（例：`wanx-v1`: 0.16元/张，`wanx-style-repaint-v1`: 0.12元/张） | 按**成功生成的视频条数**计费；无公开免费额度说明，需按用量购买资源包或开通后付费；单价未在文档中统一公示，以控制台实时报价为准 | 按**成功生成的3D模型个数**计费；无免费额度；单价依模型版本区分（`H3.1` > `P1.0`），具体见控制台定价页；失败任务不扣费 | +| **典型场景** | • 电商素材生成（商品图、海报、模特图）
• 设计辅助（背景替换、风格迁移、AI试衣）
• 内容创作（插画、头像、锦书文字艺术）
• 工业应用（缺陷标注补全、图纸增强） | • 营销短视频（文生广告片、产品演示）
• 数字人播报（新闻、客服、培训）
• 影视预演（分镜动画、角色动作测试）
• 社交内容（表情包、GIF动图、AI舞蹈） | • 工业设计（概念建模、零部件快速原型）
• 游戏开发（低多边形资产生成）
• AR/VR内容生产（可交互3D对象）
• 电商3D展示（商品360°视图基础模型） | +| **地域可用性** | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；各区域API Key与Endpoint独立 | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；跨地域调用必然失败 | **仅华北2（北京）地域可用**；其他地域调用返回403或模型不可见错误 | +| **SDK支持** | DashScope SDK（Python/Java）完整封装同步/异步调用、自动重试、凭证管理 | DashScope SDK 封装异步轮询逻辑，推荐使用；避免手动实现长轮询 | DashScope SDK 支持，但需显式指定北京地域Endpoint；无专用3D模块，复用通用异步任务接口 | + +## 各方案适用场景建议 + +### ✅ 图像生成 —— 适合「高频、轻量、多样化」视觉内容生产 +- **首选场景**：需要快速产出大量静态图像的业务，如电商平台每日上新图生成、营销海报A/B测试、设计团队灵感草稿、个性化头像/证件照批量处理。 +- **技术优势**：同步调用降低延迟（<1s响应），支持精细控制（分辨率、水印、风格索引），垂直模型丰富（虚拟模特、鞋靴、海报专用）。 +- **规避风险**：避免用图像API生成含复杂运动/时间逻辑的内容（如“挥手动作”），此类需求应转向视频生成。 + +### ✅ 视频生成 —— 适合「动态表达、人机交互、时间序列」内容构建 +- **首选场景**：数字人驱动（企业IP形象播报）、短视频自动化生产（图文转视频）、影视工业预演（分镜动画）、社交互动内容（AI跳舞、口型同步）。 +- **技术优势**：原生支持多镜头描述（`wan2.7` [prompt](../guides/prompt.md)内时间戳）、首尾帧控制、参考图动作迁移；数字人模型提供人脸/语音/动作联合生成能力。 +- **规避风险**：勿用于生成超长视频（当前最大8秒）或高精度物理仿真（如流体、布料动力学）；3D空间一致性弱于专用3D生成。 + +### ✅ 3D生成 —— 适合「几何结构明确、需下游渲染/交互」的三维资产创建 +- **首选场景**：工业设计快速建模（如家具、小家电概念验证）、游戏美术管线中的基础网格生成、AR应用中轻量化3D商品模型、教育可视化教具制作。 +- **技术优势**：输出标准GLB格式（含PBR材质），可直接导入Unity/Unreal/Three.js；支持多视角输入提升几何准确性；`H3.1`版本达200万面，满足中等复杂度建模。 +- **规避风险**：不适用于生成无明确几何结构的抽象艺术（如“一团流动的光”）；不支持纹理编辑、UV展开等后期操作；仅北京地域可用，需提前规划部署架构。 + +## 面向开发者的选型参考指南 + +1. **评估输入复杂度** + - 若输入仅为文本或单图 → 优先评估图像生成（成本低、速度快）； + - 若需表达时间变化（动作、过渡、节奏）→ 必选视频生成； + - 若目标为可旋转、可光照、可碰撞的三维实体 → 唯一选择3D生成。 + +2. **检查地域与基础设施约束** + - 若业务已部署于新加坡或美国 → **排除3D生成**，并确认视频/图像模型在对应地域的可用性（部分垂直模型仅限北京）； + - 若需混合调用（如先图生图再图生视频）→ 确保所有服务使用**同一地域API Key与Endpoint**，避免跨域认证失败。 + +3. **权衡成本与质量要求** + - 追求极致性价比（千张级/日）→ 图像生成（有免费额度+单价透明）； + - 接受中等成本换取动态表现力 → 视频生成（按条计费，单条成本高于单图）； + - 愿为专业3D资产支付溢价 → 3D生成（`H3.1`单价显著高于`P1.0`，但面数与材质质量跃升）。 + +4. **验证端到端工作流可行性** + - 图像生成：检查是否需后续处理（如抠图→合成→视频），若链路过长，考虑直接使用视频生成的I2V； + - 视频生成：确认输入图是否满足数字人模型的正面肖像要求（需先调用`detect`接口校验）； + - 3D生成：验证输入图是否符合视角顺序（前/左/后/右），多图生3D对拍摄规范性要求高，建议先用单图模式快速验证。 + +5. **上线前必做事项** + - 所有能力均需配置 `X-DashScope-Async: enable` 请求头（视频/3D强制，图像部分模型强制）； + - 生产环境务必使用**业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），避免旧域名限流与延迟问题； + - 对异步任务实现健壮轮询（带指数退避、超时熔断、`task_id`有效期校验），禁止无限循环轮询。 + +> **最后提醒**：模型能力持续迭代，`wan2.7+`、`kling-v3`、`Tripo-H3.1` 等新版本已逐步替代旧模型（如 `wan2.2`、`wanx2.1`）。新项目开发请严格参照各能力文档顶部的「最新版API参考」链接，避免依赖已标记为“遗留接口”的旧路径与参数。 + +## 被对比主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md deleted file mode 100644 index 19131b2a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md +++ /dev/null @@ -1,41 +0,0 @@ -# 图像、视频与 3D 生成对比 - -百炼平台的多媒体生成能力覆盖三条主线：**图像生成**（文生图、图像编辑、翻译、创意工具）、**视频生成**（文生/图生/参考生视频、数字人与人像动画）、**3D 生成**（基于 Tripo 的文生/图生 3D 资产）。三者都构建在 DashScope 兼容接口之上，共享「创建任务 → 轮询结果」的异步范式和统一的鉴权/地域约束，但在输入输出形态、可用模型、调用协议、耗时、地域覆盖和计费产物上差异明显。本文汇总关键维度对比，帮助开发者按业务场景快速选型与集成。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 输入格式 | 文本 `prompt`、图像（`image_url`/`images`/`base_image_url`/`mask_image_url`）、多模态 `messages`；图像翻译需 `source_lang`+`target_lang` | 文本 `prompt` + 多模态 `media`（首帧/尾帧/图像/视频/参考图/音色） | 三选一互斥：文本 `prompt`（≤1024 字符）、单图 `image`、多图 `images`（固定 4 元素，前/左/后/右） | -| 输出格式 | 图片 URL（有效期 24 小时） | 视频 URL | GLB 模型（`pbr_model_url` 或 `base_model_url`）+ 预览渲染图，链接有效期仅 **2 小时** | -| 主要模型系列 | 千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu | 万相 Wan、HappyHorse、爱诗 PixVerse、Vidu、可灵，及数字人/EMO/LivePortrait/AnimateAnyone | Tripo（`Tripo/Tripo-H3.1` 高精度、`Tripo/Tripo-P1.0` 快速专业） | -| 调用协议 | 多数异步（两步式），新版万相 2.6/2.7、Z-Image 支持 HTTP 同步 | 全部异步（两步式） | 全部异步（两步式） | -| 典型 API 端点 | `.../aigc/text2image/image-synthesis`、`.../aigc/image2image/image-synthesis`、`.../aigc/image-generation/generation`；同步走 `.../aigc/multimodal-generation/generation` | `POST .../aigc/video-generation/video-synthesis`（部分数字人/首尾帧走 `.../image2video/video-synthesis`） | `POST .../aigc/video-generation/3d-generation` | -| 轮询查询 | `GET .../api/v1/tasks/{task_id}` | `GET .../api/v1/tasks/{task_id}` | `GET .../api/v1/tasks/{task_id}`（建议间隔约 15 秒，RPS 默认 20） | -| 典型耗时 | 秒级至 1-2 分钟（编辑类偏长） | 1-5 分钟（VACE 约 5-10 分钟） | 较长（面数越高越久） | -| 关键异步头 | `X-DashScope-Async: enable`（同步模型除外） | `X-DashScope-Async: enable`（必选） | `X-DashScope-Async: enable`（必选） | -| 地域覆盖 | 华北2（北京）、新加坡、美国（弗吉尼亚）；多数创意工具/翻译/可灵/Vidu 仅北京 | 华北2（北京）、新加坡为主，部分模型另有美国、德国 Endpoint | **仅华北2（北京）** | -| 计费产物 | 仅对成功生成的输出图片计费，输入图/失败任务不计费 | 按成功生成的视频计费 | 仅对成功结果计数（`text-to-3d`/`image-to-3d`/`multi-image-to-3d`） | -| 典型场景 | 海报/电商主图、创意设计、图文编辑、图像翻译、试衣/虚拟模特 | 短视频/广告、动态营销、数字人播报、人像动画 | 游戏/电商 3D 资产、AR/VR 素材、工业设计原型 | - -## 各方案适用场景建议 - -- **图像生成**：需求量大、迭代快、单次成本低的静态视觉内容首选。文本渲染复杂选千问系列，通用文生图与创意工具选万相，追求低延迟同步返回选 Z-Image / 万相 2.6/2.7。电商场景可直接用试衣、虚拟模特、创意海报等垂直工具，减少自研成本。 -- **视频生成**：需要动态叙事、营销短视频或数字人播报时采用。纯创意用文生视频（`wan2.7-t2v` 等），有首帧素材用图生视频，需精确控制起止画面用首尾帧，多素材融合用参考生视频；口播/唱歌数字人用 `wan2.2-s2v` 系列。注意其耗时最长，需设计好轮询与超时重试。 -- **3D 生成**：面向游戏、电商展示、AR/VR 与设计原型的三维资产生产。有明确外观参考用单图/多图生 3D（多图按前/左/后/右提供更稳定），只有创意描述用文生 3D；对面数与贴图有要求时用 `Tripo-H3.1` + `geometry_quality: ultra`，追求速度用 `Tripo-P1.0`。务必在 2 小时内下载产物。 - -## 面向开发者的技术选型参考 - -1. **地域与鉴权先行**：三者都要求「模型 + Endpoint + API Key 同地域」。3D 生成及大量图像创意工具仅限北京地域，跨国部署需评估地域可用性；北京/新加坡建议迁移到业务空间专属域名以提升性能。 -2. **统一异步框架，复用轮询逻辑**：三条线均以 `X-DashScope-Async: enable` 创建任务并轮询 `tasks/{task_id}`，`task_id` 有效期 24 小时，切勿重复创建。可封装统一的任务提交/轮询/状态机（`PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED` 等）模块复用。 -3. **同步 vs 异步权衡**：仅图像生成部分新版模型支持 HTTP 同步（低延迟、实现简单）；视频与 3D 强制异步，需引入队列与回调（3D 支持异步任务回调）。 -4. **产物时效差异**：图像/视频链接有效期 24 小时，3D 下载链接仅 2 小时，务必在生成成功后尽快转存到自有 OSS。 -5. **成本控制**：三者均只对成功产物计费，失败与输入不计费；主账号与子账号共享额度与限流，注意统一配额规划。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md deleted file mode 100644 index e3b03503..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md +++ /dev/null @@ -1,57 +0,0 @@ -# 知识库与记忆库对比 - -知识库（Knowledge Base）与记忆库（Memory Library）都是百炼平台为大模型补充"外部数据"的能力，但二者解决的问题截然不同：**知识库**面向"领域知识补全"，通过 RAG 检索私有文档切片提升特定领域回答的准确性；**记忆库**面向"跨会话上下文持久化"，自动从对话中提取用户偏好与历史信息并在后续会话中召回注入 Prompt。本页从技术维度做横向对比，帮助开发者根据业务场景做选型。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| 核心定位 | RAG 检索增强，补充私有/领域知识 | 长期记忆，跨会话保持用户上下文 | -| 数据来源 | 预先上传的文档、表格、图片、音视频 | 从对话中自动提取，或 `custom_content` 直写 | -| 存储粒度 | 文档切片（单切片 ≤ 6,000 Token） | 记忆片段 / 用户画像（结构化属性） | -| 输入格式 | pdf/docx/txt/md/html、Excel/CSV、图片、音视频 | 对话 `messages` 数组或自定义内容 | -| 输出/召回 | 相关文档切片（TopK 1–20，混合检索 + Rerank） | 相关记忆片段/画像，注入 Prompt（`top_k` 建议 3–10） | -| 用户隔离 | 按知识库/业务空间 | 按 `user_id` 命名空间完全隔离 | -| API 端点 | `bailian.cn-beijing.aliyuncs.com`（OpenAPI/SDK） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | -| 鉴权 | AccessKey + `WORKSPACE_ID` + 子账号数据权限 | `DASHSCOPE_API_KEY`（`sk-` 开头） | -| 典型调用流程 | ApplyFileUploadLease → 上传 → AddFile → CreateIndex → SubmitIndexJob → 轮询 | AddMemory 写入 → SearchMemory 检索 → 注入 Prompt | -| 零侵入接入 | 智能体/工作流应用挂载知识库 | OpenClaw 记忆插件（自动捕获/自动召回） | -| 地域限制 | 仅中国站华北2（北京） | 通过 DashScope 公共端点，无华北2限制 | -| 有效期 | 数据长期保留，删除即永久清除 | 记忆片段默认 180 天（可配 7/30/180 天或永不过期） | -| 计费方式 | 规格费用（按小时）+ 模型调用费用（向量化/Rerank 按 Token） | 随长期记忆 API 调用计费（依 DashScope 定价） | -| 配额限制 | 存储 100GB–9,999GB；检索并发 1–10,000 QPS；单次召回 ≤ 20 | 按 `user_id` 记忆空间管理，默认记忆库不可删除 | - -## 各方案适用场景 - -### 优先选知识库 - -- 需要基于**大量私有文档**（产品手册、法规、FAQ、财报等）做准确问答，答案必须有据可查、可溯源。 -- 对**召回质量与可控性**要求高：需要 Meta 过滤、多路混合检索、Rerank 精排、引用来源展示。 -- 领域知识**相对稳定、可批量导入**，且以"内容准确性"为核心指标。 -- 需要企业级配额与监控：SLS 日志审计、QPS 扩容、多知识库联合检索。 - -### 优先选记忆库 - -- 需要智能体**跨会话记住用户**（偏好、习惯、历史事件、结构化画像），提升长期交互的连贯性与个性化。 -- 数据是**在对话中动态产生**的，无法预先整理成文档。 -- 希望**零侵入接入**：OpenClaw Agent 通过插件在生命周期钩子中自动捕获/召回。 -- 按用户维度隔离记忆空间，强调"个体记忆"而非"公共知识"。 - -### 两者组合使用 - -知识库与记忆库并不互斥，典型的个性化领域助手可同时使用：用**知识库**保证领域答案的准确性与可溯源，用**记忆库**记住每个用户的偏好与历史，从而在正确的知识之上提供个性化体验。 - -## 技术选型建议 - -- **知识稳定 vs 上下文动态**：内容能预先整理、追求答案准确性选知识库；内容随对话产生、追求个性化连贯选记忆库。 -- **地域约束**：知识库当前仅支持华北2（北京），若部署地域受限需评估；记忆库走 DashScope 公共端点，约束更少。 -- **接入成本**：记忆库提供 HTTP API 与 OpenClaw 插件，接入更轻量；知识库需完成上传-索引-轮询的完整流水线，前置配置更重。 -- **成本模型**：知识库有明确的规格费用（按小时）叠加模型调用费用，Rerank 费用与初步召回切片数相关，需关注 TopK 优化；记忆库随长期记忆 API 调用计费。 -- **鉴权体系差异**：知识库使用 AccessKey + WORKSPACE_ID 的阿里云 RAM 体系，记忆库使用 DashScope API Key，接入时注意区分。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md deleted file mode 100644 index bb4bf118..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 知识库、记忆库与数据接入对比 - -百炼平台提供了三种互补的数据管理机制：**知识库**（Knowledge Base）、**记忆库**（Memory Library）和**数据连接**（Data Connection）。三者分别面向不同的数据形态和业务需求，开发者在构建[智能体应用](../concepts/agent-application.md)时往往需要根据数据特征、实时性要求和集成复杂度做出技术选型。本文从核心定位、数据来源、检索方式、[计费](../concepts/billing.md)模式等关键维度对三者进行系统对比，帮助开发者快速找到最适合自身场景的方案。 - -## 核心定位对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心技术 | RAG（检索增强生成） | 长期记忆 API（语义提取 + 持久化） | 数据源连接器（实时访问） | -| 解决的问题 | 为大模型补充私有文档和最新信息 | 解决跨会话上下文丢失问题 | 统一管理和访问企业外部数据源 | -| 数据形态 | 非结构化文档、结构化表格、图片、音视频 | 对话中的关键事件、用户画像属性 | 数据库、文档系统、对象存储 | -| 数据生命周期 | 持久存储，手动管理 | 可配置有效期（7/30/180 天或永不过期） | 实时连接，数据留在原系统 | -| 数据所有权 | 平台托管（上传后由平台管理） | 平台托管（自动提取并存储） | 数据留在原处，平台仅建立连接 | - -## 数据输入与支持格式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 输入方式 | 本地上传、OSS 导入 | 对话消息自动提取或 custom_content 直写 | 连接器配置（数据库凭证、OSS Bucket、[Token](../concepts/token.md)） | -| 支持格式 | PDF/DOCX/DOC/PPTX/TXT/MD/HTML/XLSX/XLS/PNG/JPG/BMP/GIF/音视频 | 对话 messages（JSON）或自定义文本 | MySQL/PostgreSQL/PolarDB-X 2.0/语雀/OSS/文件/表格 | -| 单文件限制 | 文档最大 150 MB（1000 页）；文本最大 10 MB；图片最大 20 MB；音视频最大 512 MB | 无文件概念，按对话轮次写入 | 文件连接器：平台存储最多 100,000 个文件、1 TB；表格连接器：1 TB 免费额度 | -| 数据解析 | 电子文档/文档智能/大模型/Qwen VL/音视频 五种解析方式 | 系统自动从对话中提炼关键信息 | 文件连接器支持与知识库相同的五种解析方式 | - -## 检索与集成方式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 检索机制 | 语义向量检索 + 关键词混合检索 + Rerank 排序 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 百炼 SDK 知识检索/知识问答接口 | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用内工作流节点或内置工具调用 | -| 集成入口 | [智能体应用](../concepts/agent-application.md)、工作流应用、外部 API | API 直连、OpenClaw 插件（零侵入） | 应用内数据连接器节点 | -| 多源联合 | 支持最多 15 个知识库联合检索 | 按 user_id 自动聚合同一用户记忆 | 支持同时配置多个连接器 | -| 实时性 | 需重新导入和索引后才能检索新内容 | 对话结束后即时写入，下次对话可召回 | 流处理类实时访问源数据库最新数据 | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 规格费用 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时 | 以长期记忆 API 调用[计费](../concepts/billing.md) | 平台存储限时免费；自有 OSS 按 OSS 费用 | -| 模型调用费用 | 向量化 + Rerank 排序（按 [Token](../concepts/token.md) 计费） | 记忆提取和检索的模型调用费用 | 大模型文档解析按模型调用计费 | -| 免费额度 | 新用户 720 小时标准版（开通后 30 天有效） | 参见长期记忆 API 计费说明 | 文件/表格平台存储限时免费（1 TB） | -| 并发限制 | 标准版 1 QPS；旗舰版 50-10,000 QPS | 参见 API 限流说明 | 取决于源数据库和网络配置 | -| 地域限制 | 仅华北2（北京） | 参见 DashScope 服务地域 | 流处理类需网络可达；PolarDB-X 仅支持私网 | - -## 适用场景建议 - -**选择知识库**适用于： -- 企业有大量私有文档（产品手册、技术规范、FAQ 等），需要让大模型基于这些文档进行准确问答 -- 需要图文并茂的回复、复杂 PDF/图表理解、音视频内容检索等[多模态](../concepts/multimodal.md)场景 -- 对检索精度有较高要求，需要向量检索 + Rerank + 多轮对话改写等完整 RAG 流水线 -- 数据更新频率中等（可接受重新导入和索引的延迟） - -**选择记忆库**适用于： -- 智能体需要记住用户的历史偏好、习惯和关键事件，实现个性化持续服务 -- 需要跨会话保持上下文连贯性（如客服机器人记住用户之前反馈的问题） -- 希望零侵入接入（通过 OpenClaw 插件自动捕获和召回，无需改业务代码） -- 数据以对话形式产生，不需要事先准备结构化文档 - -**选择数据连接**适用于： -- 企业数据存储在 MySQL、PostgreSQL 等关系数据库中，需要智能体实时查询最新数据 -- 数据不适合或不允许复制到平台，要求数据留在原系统 -- 需要对接语雀文档、OSS 存储等已有数据系统 -- 业务场景需要执行 SQL 查询获取精确的结构化数据（如订单查询、库存查询） - -## 组合使用建议 - -三种方案并非互斥，在复杂业务场景中推荐组合使用： - -- **知识库 + 记忆库**：知识库提供专业领域知识，记忆库记住用户偏好和历史交互，二者结合实现既专业又个性化的智能服务。 -- **知识库 + 数据连接**：知识库提供静态文档知识，数据连接提供实时业务数据，适合需要同时参考文档和查询数据库的场景（如技术支持 + 工单系统）。 -- **三者组合**：在全场景智能助手中，知识库负责专业知识问答，数据连接负责实时数据查询，记忆库负责用户画像和交互记忆，共同构建完整的智能体数据底座。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md deleted file mode 100644 index 818cb053..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md +++ /dev/null @@ -1,69 +0,0 @@ -# 知识库与长期记忆对比 - -阿里云百炼平台针对「让大模型使用外部信息」提供了两条不同路径：**知识库（RAG）** 负责为模型补充**私有文档与最新事实**，**长期记忆（Memory Library / 长期记忆 API）** 负责跨会话持久化**用户个性化信息**。二者常被混淆，但设计目标、数据形态、检索机制与计费方式都截然不同。本文面向开发者，从技术选型角度对二者做逐维度对比，帮助你在智能体、问答、Agent 等场景中选对能力。 - -其中长期记忆在平台上有两个入口：控制台侧的**记忆库（Memory Library Overview）** 与开放的 **长期记忆（新）RESTful API**，二者底层为同一套服务，下文将它们合并为「长期记忆」一列，必要处再区分入口差异。 - -## 一句话定位 - -- **知识库**：把企业文档/结构化数据/多模态素材建成可检索索引，回答前先检索、再增强生成，解决「模型不知道我的私有资料」。 -- **长期记忆**：从多轮对话中自动提炼关键事件与用户画像并持久化，后续对话语义召回后注入 Prompt，解决「模型记不住这个用户」。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 长期记忆（Memory Library / 长期记忆 API） | -| --- | --- | --- | -| 解决的问题 | 为模型补充私有数据与最新信息，提升回答准确性 | 跨会话保留用户偏好与历史，实现个性化上下文 | -| 数据来源 | 预先上传的文档、Excel/CSV、图片、音视频等素材 | 从对话消息自动提取，或 `custom_content` 直接写入 | -| 内容形态 | 切片（chunk）+ 向量索引 + Meta 信息 | 记忆片段（关键事件）+ 用户画像（结构化属性） | -| 写入时机 | 建库/导入阶段离线建索引（AddFile→CreateIndex→SubmitIndexJob） | 对话过程中/结束后实时写入（AddMemory） | -| 输入格式 | 文件（pdf/docx/ppt ≤150MB、txt/md/html ≤10MB、图片 ≤20MB、音视频 ≤512MB） | `messages`（对话，最多 50 条）或 `custom_content`（≤512 字符）+ `user_id` | -| 输出/召回 | 召回相关切片（单次最多 20），供大模型增强生成 | 召回相关记忆片段/画像，注入 Prompt | -| 检索机制 | Query 改写 → 向量+关键词混合检索 → Rerank 精排 → 加权返回 | 语义检索，可选 rerank / query 重写 / 意图判别 | -| 隔离维度 | 按知识库 ID / 业务空间 | 按 `user_id`（记忆空间），可再按 `memory_library_id` | -| API 端点 | `bailian.cn-beijing.aliyuncs.com`（阿里云 OpenAPI 风格） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | -| 认证方式 | 子账号策略 `AliyunBailianDataFullAccess` | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 核心接口 | AddFile / CreateIndex / SubmitIndexJob / 知识检索 / 知识问答 | AddMemory / SearchMemory / ListMemory / Update / Delete / ProfileSchema 系列 | -| 关键参数 | 相似度阈值、召回片段数 TopK(1–20)、权重、Meta 抽取、智能切分 | `top_k`(1–100)、`min_score`(默认 0.3)、`enable_rerank/rewrite/judge` | -| 地域限制 | 仅中国站**华北2（北京）**可开通使用 | 通过 DashScope 全局域名接入，无北京地域限制 | -| 数据有效期 | 长期保存，删除即永久清除且停止计费 | API 侧「暂无失效日期」；控制台默认 180 天，可配 7/30/180 天或永不过期 | -| 计费方式 | 规格费用（标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时）+ 模型调用 Token 费；2026-01-04 起计费 | 按接口调用限流管理（全部 ≤3000 QPM，add 120 QPM，search 300 QPM） | -| 零侵入接入 | 关联到智能体/工作流应用（知识库节点） | OpenClaw 记忆插件（`before_agent_start` 召回 + `agent_end` 捕获） | -| 典型场景 | 企业文档问答、产品手册、客服、结构化数据查询、多模态搜索 | 个人助理记住用户习惯、智能体长期陪伴、跨会话偏好延续 | - -## 适用场景建议 - -### 选择知识库（RAG）当 - -- 你有**成体量的私有资料**（文档、手册、报表、图片、音视频）需要模型准确引用。 -- 要求回答**可溯源**（展示引用来源）、可做效果评测与持续优化。 -- 数据是**面向所有用户共享**的事实性知识，而非某个用户的个人信息。 -- 可接受仅在**华北2（北京）**地域使用，并规划好规格费用与 Token 成本。 - -### 选择长期记忆当 - -- 你要让智能体**记住单个用户**的偏好、习惯、历史事件，实现个性化。 -- 数据来自**实时对话**、随时间增长且需自动去重/更新。 -- 需要按 `user_id` 做**多用户记忆隔离**，或用画像模板维护结构化属性。 -- 希望**零侵入接入**（OpenClaw 插件）或用轻量 HTTP API 快速集成。 - -### 组合使用（推荐） - -二者并不互斥，常见的高质量智能体会同时使用：**知识库**提供权威事实与私有知识，**长期记忆**提供该用户的个性化上下文。典型编排为——对话前用 `SearchMemory` 召回用户记忆注入 Prompt，同时用知识检索召回相关文档切片；对话后用 `AddMemory` 沉淀新的用户信息。这样既「答得准」又「记得住」。 - -## 技术选型速查 - -- 「模型不知道我的资料」→ 知识库。 -- 「模型记不住这个用户」→ 长期记忆。 -- 需要**引用溯源 / 结构化数据查询 / 多模态素材** → 知识库。 -- 需要**跨会话个性化 / 用户画像 / 对话自动沉淀** → 长期记忆。 -- 受限于**北京地域**或对**规格费用**敏感 → 优先评估知识库成本；长期记忆走 DashScope 全局接入且以调用限流计。 -- 想**最快接入 Agent** → 长期记忆 OpenClaw 插件（配 `apiKey` + `userId` 即可自动捕获/召回）。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md deleted file mode 100644 index 9ff18a9a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 知识库 vs 记忆库 vs 数据接入对比 - -百炼平台提供三种核心数据管理能力：知识库（RAG）、记忆库（长期记忆）和数据连接。三者均用于为大模型补充外部信息，但在数据组织方式、检索机制和适用场景上存在本质差异。本文从开发者技术选型角度，对三者的能力边界和关键特征进行系统对比。 - -## 关键维度对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心定位 | 基于 RAG 的私有文档检索增强 | 跨会话长期记忆持久化与召回 | 外部数据源统一接入与实时访问 | -| 输入格式 | PDF、DOCX、TXT、Markdown、HTML、XLSX、图片、音视频等非结构化文件 | 对话消息（messages）或自定义内容（custom_content） | 数据库连接串、OSS Bucket、语雀 [Token](../concepts/token.md)、本地文件上传 | -| 数据存储方式 | 平台托管（向量化切片存储） | 平台托管（记忆片段 + 用户画像） | 平台托管或流处理（原数据源实时访问） | -| 检索机制 | 向量检索 + 关键词检索 + Rerank 精排 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 知识检索服务 / 知识问答服务（百炼 SDK） | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用工作流节点或工具调用 | -| 数据粒度 | 文档切片（最大 6,000 [Token](../concepts/token.md)/片） | 记忆片段（事件级）或用户画像（属性级） | 原始数据行/文件/文档 | -| 数据更新方式 | 手动上传 / OSS 导入，需重新索引 | 自动从对话提取，支持去重和动态更新 | 实时连接原数据源，数据变更即时生效 | -| 多用户隔离 | 无内置用户隔离（按知识库粒度管理） | 原生 user_id 级别隔离 | 按连接器实例隔离 | -| 并发规格 | 标准版 1 QPS / 旗舰版 50-10,000 QPS | 未公开 QPS 限制 | 取决于底层数据源能力 | -| [计费](../concepts/billing.md)方式 | 标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时 + 检索/排序 [Token](../concepts/token.md) 费 | 包含在百炼平台使用中（按 API 调用） | 文件/表格平台存储限时免费；流处理按底层数据源[计费](../concepts/billing.md) | -| 数据有效期 | 永久（手动删除） | 默认 180 天（可配置 7/30/180 天或永不过期） | 永久（随原数据源生命周期） | -| 集成方式 | 智能体应用 / 工作流节点 / SDK API | API 直连 / OpenClaw 插件零侵入接入 | 工作流节点 / 应用工具调用 | - -## 适用场景建议 - -### 选择知识库 - -- 企业有大量非结构化文档（产品手册、FAQ、技术文档）需要语义检索 -- 需要高精度的文档问答能力，支持[多模态](../concepts/multimodal.md)（图文、音视频） -- 对检索并发有明确要求（旗舰版支持万级 QPS） -- 需要精细控制检索质量（切片策略、Rerank、标签过滤等） - -### 选择记忆库 - -- 智能体需要跨会话记住用户偏好、历史交互和个性化信息 -- 需要按用户维度隔离记忆空间 -- 希望零侵入集成（通过 OpenClaw 插件自动捕获/召回） -- 数据来源是对话本身，而非预置文档 - -### 选择数据连接 - -- 需要实时查询企业数据库中的结构化数据（MySQL、PostgreSQL、PolarDB-X） -- 数据存储在外部系统（语雀、OSS）且需保持同步 -- 应用需要执行 SQL 查询获取精确结果 -- 数据更新频繁，不适合定期导入知识库 - -## 组合使用建议 - -三种能力并非互斥，实际应用中常组合使用： - -- **知识库 + 记忆库**：知识库提供通用文档检索，记忆库补充用户个性化上下文，实现"懂业务 + 懂用户"的智能体 -- **知识库 + 数据连接**：非结构化文档走知识库语义检索，结构化数据走数据连接 SQL 精确查询 -- **三者结合**：在工作流中编排知识库节点、数据连接工具和记忆注入，构建具备完整数据感知能力的复杂应用 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md deleted file mode 100644 index 83e098a7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md +++ /dev/null @@ -1,63 +0,0 @@ -# 知识库与记忆库对比 - -阿里云百炼平台同时提供**知识库（Knowledge Base）**和**记忆库（Memory Library）**两种数据增强能力，二者分别面向不同的信息管理需求。知识库基于 RAG 技术，用于将企业私有文档、结构化数据等外部知识注入大模型，提升特定领域问答的准确性；记忆库则面向跨会话场景，自动从对话中提取和持久化关键信息，使智能体能够在多轮交互中保持对用户偏好与历史上下文的理解。本文从核心定位、数据模型、接入方式、检索机制等维度进行系统对比，帮助开发者根据业务场景做出合理的技术选型。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| **核心定位** | 基于 RAG 的外部知识检索增强，为大模型补充私有数据与最新信息 | 跨会话长期记忆，自动提取并持久化对话中的关键信息 | -| **数据来源** | 用户主动导入的文档（PDF/Word/Markdown/HTML/Excel/CSV）、图片、音视频等 | 从对话历史中自动提取，或通过 `custom_content` 直接写入 | -| **数据类型** | 文档搜索、数据查询（NL2SQL）、图片问答、音视频搜索四类，创建时选定不可更改 | 记忆片段（事件和信息）和用户画像（结构化属性），可独立或组合使用 | -| **存储粒度** | 切片（Chunk）级别，按智能切分或自定义规则对文档分片，单切片上限 6,000 Token | 记忆片段级别，系统自动提炼为精简的事实描述；用户画像按属性字段存储 | -| **数据持久性** | 持久存储，无过期机制 | 记忆片段可配置有效期（7/30/180 天或永不过期），控制台默认 180 天 | -| **检索机制** | 向量 + 关键词混合检索 + Rerank 排序，支持相似度阈值、TopK、权重等精细调控 | 基于语义的记忆检索（SearchMemory），通过 `top_k` 控制召回数量 | -| **向量模型** | `text-embedding-v4`/`text-embedding-v3`（512 维）；视觉理解用 `qwen3-vl-embedding`；图片问答用 `multimodal-embedding-v1`（1024 维） | 由平台内部自动处理向量化，用户无需选择向量模型 | -| **排序模型** | 支持 `qwen3-rerank`、`qwen3-rerank(hybrid)`、`qwen3-vl-rerank` | 无独立排序模型，由记忆检索服务内部排序 | -| **API 端点** | `bailian.cn-beijing.aliyuncs.com`，通过阿里云 SDK（`alibabacloud_bailian20231229`）调用 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*`，通过 DashScope API Key 认证 | -| **认证方式** | 阿里云 AccessKey（`ALIBABA_CLOUD_ACCESS_KEY_ID` / `SECRET`）+ WorkspaceId | DashScope API Key（`DASHSCOPE_API_KEY`，以 `sk-` 开头） | -| **接入方式** | 控制台可视化 + 开放 API（仅文档搜索类） | API 直连 + OpenClaw 记忆插件（零侵入自动捕获/召回） | -| **数据隔离** | 按知识库实例隔离，单次检索可绑定最多 15 个知识库 | 按 `user_id` 隔离，同一 `user_id` 共享记忆空间 | -| **可挂载模型** | 千问全系列（QwQ/Long/Max/Plus/Turbo/Coder 等）及第三方模型（DeepSeek/Llama/Yi 等） | 不直接挂载模型，作为独立记忆服务由应用侧在 Prompt 中注入 | -| **计费模式** | 标准版 0.03 元/小时；旗舰版 0.2 元/RCU/小时（1 RCU = 50 QPS），另计 SLS 日志存储费用 | 随 DashScope API 调用计费，无独立规格费用 | -| **地域限制** | 仅支持中国站华北2（北京）地域 | 通过 DashScope 全局端点访问 | -| **监控能力** | 内置日志投递至 SLS，支持调用审计、问题排查、用量统计与告警 | 通过控制台查看记忆库统计信息，无独立日志服务集成 | - -## 适用场景建议 - -### 适合选择知识库的场景 - -- **企业知识问答**：需要基于产品手册、内部文档、技术规范等大量非结构化文档进行精准问答。 -- **结构化数据查询**：通过自然语言查询 Excel/CSV 中的结构化数据（NL2SQL）。 -- **[多模态](../concepts/multimodal.md)检索**：需要对图片、音视频内容进行搜索和问答。 -- **高并发生产环境**：旗舰版支持最高 10,000 QPS，适合对吞吐量有要求的业务系统。 -- **精细化检索调优**：需要通过多种向量模型、排序策略、切片方式、相似度阈值等参数精确控制召回质量。 - -### 适合选择记忆库的场景 - -- **个性化智能助手**：需要记住用户偏好、习惯和历史交互信息，提供个性化服务。 -- **跨会话上下文保持**：用户多次对话之间需要保持连贯性，避免重复提供相同信息。 -- **用户画像构建**：需要从对话中自动提取和维护用户的结构化属性（年龄、职业、偏好等）。 -- **轻量级集成**：通过 OpenClaw 插件零侵入接入现有 Agent，无需改造应用代码。 -- **对话式应用**：客服机器人、个人助理等需要"记住"用户的长期交互场景。 - -### 组合使用 - -在实际业务中，知识库与记忆库可以组合使用以实现最佳效果。例如，一个智能客服系统可以同时挂载知识库获取产品文档中的专业知识，又通过记忆库记住每位用户的历史问题和偏好，从而在准确回答专业问题的同时提供个性化的服务体验。 - -## 技术选型参考 - -选择知识库还是记忆库，核心取决于数据的来源和用途： - -- 如果数据是**预先准备好的静态文档**，需要检索后辅助回答 -- 选择**知识库**。 -- 如果数据是**从对话中动态产生**的，需要跨会话持久化 -- 选择**记忆库**。 -- 如果两种需求并存，建议**同时接入**，各司其职。 - -从工程复杂度看，记忆库的接入成本更低（尤其是 OpenClaw 插件方式），而知识库提供了更丰富的检索调优手段和更高的生产环境承载能力。开发者应根据数据规模、并发需求、检索精度要求和集成方式综合评估。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md deleted file mode 100644 index a5f55a41..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md +++ /dev/null @@ -1,58 +0,0 @@ -# 托管智能体与 LLM 应用对比 - -百炼平台既提供面向"应用构建"的 LLM 应用体系（智能体 / 工作流 / 高代码），也提供面向"长时自主任务"的 Managed Agents 托管运行时。二者虽然都以大模型为核心、都能接入知识库与 MCP 工具，但在运行模式、状态管理、执行环境和目标场景上定位截然不同。本文从技术选型视角梳理二者差异，帮助开发者判断"我该用哪一个"。 - -## 概念定位 - -- **托管智能体（Managed Agents）**：平台在服务端托管的智能体运行时，为多步工具调用、代码执行、文件处理等长时任务提供独立云端沙箱容器。会话状态、事件历史由服务端持久化，支持中断与续接，智能体在沙箱内自主执行命令、读写文件、安装依赖。 -- **LLM 应用**：面向应用构建的一整套模式，包含智能体（Agent，含 2.0 / 1.0）、工作流（Workflow）、高代码应用三种类型，覆盖从零代码配置到 Python 编码的不同开发门槛，用于快速搭建可发布、可被 API 调用的 AI 应用。 - -## 关键维度对比 - -| 维度 | Managed Agents（托管智能体） | LLM 应用（智能体/工作流/高代码） | -| --- | --- | --- | -| 核心定位 | 服务端托管的长时任务运行时 | 应用构建平台，多种应用形态 | -| 运行模式 | 服务端维护会话状态，支持中断与续接 | 智能体应用多为无状态调用，应用侧维护上下文 | -| 执行环境 | 独立沙箱、云端容器 | 共享运行时（高代码可选 Serverless/K8s 部署） | -| 事件模型 | 会话级 SSE 事件流，事件历史持久化 | 响应级[流式输出](../concepts/streaming.md) | -| 开发方式 | API 编排（配置 Agent/Environment/Session） | 零代码配置 / 可视化编排 / Python 编码 | -| 主要 API 端点 | `/api/v1/agentstudio/agents`、`/environments`、`/sessions`、`/sessions/{id}/events` | 应用发布后经"发布渠道"提供的调用 API | -| 工具能力 | 内置 7 个工具（bash/read/write/edit/glob/grep/download_file）+ MCP + Skill | 内置沙箱工具、知识库、MCP、插件（按应用类型） | -| 文件处理 | 上传挂载到 `/mnt/session/uploads`，单文件 ≤ 10MB | 文件问答（全文引用/RAG/自定义），单会话 ≤ 10 文件、单文件 ≤ 10MB | -| 记忆/上下文 | 服务端持久化会话事件，可挂载资源复用 | 短期记忆 0-30 轮；长期记忆暂未支持 | -| 发布与集成 | 通过 Agent/Environment/Session API 直接编排 | 需先"发布"，再经发布渠道 API/第三方平台调用 | -| 计费方式 | 按模型 Token 用量 + 沙箱运行资源 | 模型 Token + 知识库召回 + MCP/插件 + 高代码函数计算/网关/存储 | -| 典型场景 | 多步工具调用、代码执行、文件批处理等长时自主任务 | 问答对话、固定流程自动化、企业级后端服务 | - -## 适用场景建议 - -**优先选择 Managed Agents 的场景：** - -- 任务需要**多步自主决策 + 代码执行**，如自动化数据处理、脚本编写与运行、文件批量转换。 -- 需要**长时运行**且要在中途中断、审批、续接的任务，依赖服务端持久化的事件历史。 -- 需要**隔离的沙箱环境**安装依赖（apt/pip）、运行 shell 命令，且不希望自行维护会话上下文。 - -**优先选择 LLM 应用的场景：** - -- **智能体应用（Agent 2.0）**：意图开放、需模型自主规划调用知识库/MCP 的问答与对话类应用，业务人员即可零代码配置。 -- **工作流应用**：流程固定、需精确控制执行链路的多步骤自动化，适合 IT 运维与业务分析师用可视化编排。 -- **高代码应用**：对性能、可观测性、企业级运维有要求的生产级 AI 后端，由工程师用 Python 编码并部署到 Serverless/K8s。 -- **文件问答**：面向文档总结、长文检索、图片/视频分析等，直接在智能体应用中上传文件即可。 - -## 技术选型参考 - -- **控制粒度维度**：从"AI 自主"到"人工精确控制"依次为——Managed Agents / Agent 2.0（模型自主规划）> 工作流（预定义流程）> 高代码（完全代码控制）。需要确定性流程选工作流或高代码；需要自主探索选 Managed Agents 或 Agent 2.0。 -- **状态需求维度**：需要跨轮次持久化会话、支持中断续接的长时任务，只有 Managed Agents 原生支持；LLM 智能体应用多为无状态，需应用侧自行维护上下文。 -- **交付形态维度**：要"对外发布 + 多渠道集成（钉钉/公众号）"选 LLM 应用；要"程序化编排自主任务"直接对接 Managed Agents 的 agentstudio API。 -- **开发门槛维度**：业务人员/产品经理→Agent 2.0；运维/分析师→工作流；AI 工程师→高代码或 Managed Agents API。 - -> 提示：二者并非互斥。可在 LLM 应用中通过 MCP/知识库快速搭建对话入口，同时把重型的多步执行任务下沉到 Managed Agents 沙箱，形成"轻交互层 + 重执行层"的组合架构。 - -## 被对比主题页 - -- [managed agents](../guides/managed-agents.md) -- [llm application](../guides/llm-application.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md deleted file mode 100644 index 12f72f00..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md +++ /dev/null @@ -1,48 +0,0 @@ -# 图像、视频与3D生成对比 - -阿里云百炼平台提供了覆盖多模态内容生成的三大能力：**图像生成**、**视频生成**与**3D 资产生成**。三者都通过 DashScope 网关对外提供服务，共享相似的异步调用范式与鉴权体系，但在支持模型、输入输出格式、接口路径、地域可用性、计费方式和典型应用场景上存在显著差异。本文从技术选型角度对三者做横向对比，帮助开发者根据业务需求快速定位合适的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 支持模型 | 千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等多家族 | 万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及多种人像驱动模型 | 仅 Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| 核心功能 | 文生图、图生图、图像编辑、局部重绘、扩图、虚拟模特、AI 试衣、创意海报等 | 文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑、数字人/口型/舞蹈等 | 文生 3D、单图生 3D、多图生 3D | -| 输入格式 | `prompt` / `messages` / `images`（JPG/PNG/JPEG/BMP/WEBP，[512,4096] 像素，≤10MB） | `input.prompt` + `input.media`（`first_frame`/`last_frame`/`image_url`/`video` 等） | `prompt`（≤1024 字符）/ `image` / `images`（4 元素数组，三者互斥；JPEG/PNG，[20,6000] 像素，≤20MB） | -| 输出格式 | 图像 URL（有效期 24 小时） | 视频 URL（异步返回） | GLB 模型（`pbr_model_url` / `base_model_url`）+ 预览渲染图（下载链接有效期 2 小时） | -| 主要 API 端点 | 多路径：`.../text2image/image-synthesis`、`.../aigc/multimodal-generation/generation`、`.../image-generation/generation` 等 | `POST .../aigc/video-generation/video-synthesis`（部分模型走 `.../aigc/image2video/video-synthesis`） | `POST .../aigc/video-generation/3d-generation` | -| 调用方式 | 以异步为主（`X-DashScope-Async: enable` + 轮询）；新模型支持 HTTP 同步 | 全部异步（创建任务 → 轮询查询） | 全部异步（创建任务 → 轮询，建议 15 秒间隔） | -| 任务耗时 | 通常 1-2 分钟 | 通常 1-5 分钟 | 较长（异步） | -| 地域可用性 | 华北2（北京）、新加坡、美国（弗吉尼亚）等多地域，独立 Key 不可混用 | 万相/HappyHorse 多地域；PixVerse/Vidu/Kling/数字人/人像模型仅北京 | 仅华北2（北京），需该地域 API Key | -| 计费方式 | 仅对成功生成的输出图片计费，含 90 天免费额度 | 多为后付费按视频时长（元/秒）或按张计费 | 按任务成功结果计数（`text-to-3d`/`image-to-3d`/`multi-image-to-3d`） | -| 并发限制 | 主/子账号共享 QPS 与处理中任务数 | 同时处理中任务通常限 1（排队执行） | 查询接口默认 RPS 20 | -| 典型场景 | 电商海报、虚拟模特、AI 试衣、创意插画、图文混排 | 短视频、广告、数字人播报、动画、口型替换 | 游戏/AR/VR 资产、3D 建模、数字孪生 | - -## 共性特征 - -- **统一网关与异步范式**：三者均通过 DashScope 网关调用，且核心流程都是「创建任务获取 `task_id` → 轮询 `GET .../api/v1/tasks/{task_id}`」。创建任务时必须携带 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- **`task_id` 有效期均为 24 小时**，切勿重复创建任务，轮询获取即可。 -- **地域隔离**：不同地域拥有独立的 API Key 与请求地址，跨地域调用会导致鉴权失败。百炼推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）。 -- **请求体结构**：普遍由 `model` / `input` / `parameters` 三部分组成。 - -## 各方案适用场景建议 - -- **图像生成**：适合需要静态视觉内容的场景，模型选择最丰富、地域覆盖最广、计费门槛最低（仅计成功输出图并有免费额度），是多模态生成中最成熟、门槛最低的入口。电商与创意工具（虚拟模特、AI 试衣、创意海报）尤为突出，但需注意部分创意工具模型仅提供免费体验、用尽后不可付费。 -- **视频生成**：适合需要动态内容的场景，功能维度最复杂（文生/图生/参考生/编辑/数字人）。选型时优先选用走新版协议、功能最全的万相 wan2.7 系列；若使用 PixVerse、Vidu、Kling 或人像/数字人模型需注意仅限北京地域。计费按时长且并发通常限 1，需评估吞吐与排队成本。 -- **3D 生成**：能力最聚焦，仅基于 Tripo 模型，产出带 PBR 材质的 GLB 模型。仅限北京地域、下载链接仅 2 小时有效期，需及时下载与转存。适合游戏、AR/VR、工业设计等对 3D 资产有需求的场景；`Tripo-H3.1` 追求高精度（最高 200 万面），`Tripo-P1.0` 追求速度。 - -## 技术选型参考 - -1. **地域约束优先评估**：3D 生成及大量第三方视频模型仅支持北京地域，若业务部署在新加坡/海外，应优先确认图像生成或万相视频系列的多地域可用性。 -2. **同步 vs 异步**：仅图像生成的部分新模型（如 `wan2.6-image`、`z-image-turbo`）支持 HTTP 同步一次返回，对低延迟场景友好；视频与 3D 必须异步轮询，需在客户端实现任务状态管理。 -3. **输出有效期差异**：图像/视频结果与 `task_id` 有效期 24 小时，而 3D 产物下载链接仅 2 小时，集成 3D 时务必在回调或轮询成功后立即下载转存。 -4. **计费模型差异**：图像按成功输出图计费且有免费额度、门槛最低；视频按时长计费、并发受限、成本更高；3D 按成功任务计数。批量或高并发场景需据此估算成本与吞吐。 -5. **接口路径不统一**：三大能力乃至同一能力内不同模型的接口路径都可能不同（尤其图像与视频），接入前务必以对应模型的官方文档为准。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md deleted file mode 100644 index aa5c3e48..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md +++ /dev/null @@ -1,72 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成对比 - -百炼平台提供图像生成、视频生成和 3D 模型生成三大多媒体内容创作能力。三者在输入输出形态、调用模式、模型生态和适用场景上存在显著差异。本文从开发者技术选型角度，系统对比三类生成 API 的核心维度，帮助快速定位最适合业务需求的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| **输入格式** | 文本 [prompt](../guides/prompt.md)、参考图（URL/Base64）、蒙版、涂鸦草图 | 文本 [prompt](../guides/prompt.md)、首帧/首尾帧图像、参考图、音频、源视频 | 文本 [prompt](../guides/prompt.md)、单图（URL）、多图（4视角，前左后右） | -| **输出格式** | 图片 URL（JPEG/PNG），需及时下载 | 视频 URL（MP4），需及时下载 | GLB 模型文件 URL（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用模式** | 同步（OpenAI 兼容）或异步（DashScope 原生） | 仅异步（创建任务 + 轮询 task_id） | 仅异步（创建任务 + 轮询 task_id） | -| **典型耗时** | 秒级（同步）~ 十几秒（异步） | 1-10 分钟 | 数分钟（建议 15 秒间隔轮询） | -| **API 端点** | `/compatible-mode/v1/images/generations` 或 `/api/v1/services/aigc/text2image/image-synthesis` | `/api/v1/services/aigc/video-generation/video-synthesis` 或 `/image2video/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持模型** | 通义千问图像、万相 V1/V2/2.6/2.7、Z-Image、可灵、创意工具（10+） | 万相 HappyHorse/wan2.7/2.6/2.5/2.2/2.1、爱诗 PixVerse、Vidu、可灵 | Tripo-H3.1（高精度）、Tripo-P1.0（专业快速） | -| **模型数量** | 20+ 模型/接口 | 30+ 模型变体 | 2 个模型 | -| **地域限制** | 无特殊限制 | 部分模型仅华北2（北京） | 仅华北2（北京） | -| **[计费](../concepts/billing.md)方式** | 按图片张数/分辨率[计费](../concepts/billing.md) | 按视频时长/分辨率[计费](../concepts/billing.md) | 按任务次数计费 | -| **协议支持** | OpenAI 兼容 + DashScope 原生 | DashScope 原生（异步必填 X-DashScope-Async 头） | DashScope 原生（异步必填 X-DashScope-Async 头） | -| **task_id 有效期** | 异步任务 24 小时 | 24 小时 | 24 小时 | -| **内容安全** | 内置审核，违规拒绝 | 内置审核，违规拒绝 | 遵循平台统一内容安全策略 | - -## 功能丰富度对比 - -| 能力类别 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| 文本生成 | 文生图（多模型可选） | 文生视频 | 文生 3D | -| 图像驱动 | 图像编辑、局部重绘、涂鸦作画 | 图生视频（首帧/首尾帧）、参考生视频 | 单图生 3D、多图生 3D | -| 风格控制 | 风格参数、负面词、艺术风格模型 | 视频风格重绘 | 贴图质量、几何精度 | -| 垂类工具 | 虚拟模特、AI试衣、创意海报、背景生成等 10+ 工具 | 视频换人、数字人、肖像动态（唱演/播报/口型替换） | 无 | -| 后处理 | 擦除补全、画面扩展、人物分割 | 视频编辑（指令+参考图） | 无 | - -## 适用场景建议 - -### 图像生成 - -- 电商商品图批量制作（虚拟模特、背景替换、创意海报） -- 营销素材快速迭代（文生图 + 风格控制） -- 内容创作中的插图与配图需求 -- 人像娱乐玩法（风格重绘、写真） -- 需要低延迟同步返回结果的场景 - -### 视频生成 - -- 短视频/广告创意自动化生产 -- 数字人播报与虚拟主播 -- 电商商品动态展示视频 -- IP 角色动画与多镜头叙事 -- 视频内容二次编辑与换人 - -### 3D生成 - -- 游戏/XR 场景中的 3D 资产快速原型 -- 电商商品 3D 展示与 AR 试用 -- 建筑/工业设计概念验证 -- 需要 PBR 材质的高精度渲染场景 - -## 技术选型建议 - -1. **追求响应速度**：图像生成支持同步调用，秒级返回；视频和 3D 均为异步，分钟级等待不可避免。 -2. **模型生态丰富度**：图像生成 > 视频生成 > 3D生成。图像生成模型和垂类工具最多，3D 目前仅有 Tripo 系列。 -3. **地域部署灵活性**：图像生成地域限制最少；视频和 3D 部分模型仅限华北2（北京）。 -4. **业务垂类覆盖**：电商/营销场景，图像生成的垂类工具链最完善；数字人/播报场景选视频生成；3D 资产生产选 3D 生成。 -5. **[多模态](../concepts/multimodal.md)组合**：可先用图像生成产出关键帧，再送入视频生成做动态化；或先用图像/多视角图生成 3D 模型，形成完整的内容生产流水线。 -6. **成本控制**：图像生成单次成本最低，3D 生成单次成本最高但产出资产复用价值大。建议根据产出资产的复用频次评估 ROI。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md deleted file mode 100644 index ce1262d9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md +++ /dev/null @@ -1,69 +0,0 @@ -# 长期记忆、记忆库与[知识库](../concepts/knowledge-base.md)对比 - -百炼平台为开发者提供了三类用于"补充模型上下文"的能力：**长期记忆（新）API**、**记忆库（Memory Library）**、**[知识库](../concepts/knowledge-base.md)（Knowledge Base）**。三者定位不同：长期记忆 API 是底层的 RESTful 接口集合；记忆库是建立在长期记忆 API 之上、面向"跨会话用户记忆"的产品化封装（含控制台管理、OpenClaw 插件零侵入接入）；[知识库](../concepts/knowledge-base.md)则基于 RAG 技术，面向"私有文档/数据的语义检索"。本页通过关键维度对比，帮助开发者根据业务诉求（用户画像 vs 文档问答 vs 零侵入记忆）做出技术选型。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）API | 记忆库（Memory Library） | 知识库（Knowledge Base） | -| --- | --- | --- | --- | -| 定位 | 底层 RESTful API，存储/检索/更新/删除用户记忆片段与画像 | 长期记忆 API 的产品化封装，含控制台与 OpenClaw 插件 | 基于 RAG 的私有数据语义检索，为模型补充领域知识 | -| 输入格式 | `messages`（对话）或 `custom_content`（自定义文本，≤512 字符） | 同长期记忆 API；OpenClaw 插件自动捕获对话 | 文件（pdf/docx/xlsx/图片/音视频等）、数据表、OSS 导入 | -| 输出格式 | 记忆片段（`memory_nodes`）、用户画像（结构化属性） | 记忆片段 + 用户画像，可注入 Prompt | 召回的文本切片（≤20 个/次），拼装后注入 Prompt | -| API 端点 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | 复用长期记忆 API；OpenClaw 插件通过 Gateway 钩子调用 | 阿里云百炼 SDK / 检索 API（需 AliyunBailianDataFullAccess 权限） | -| 支持模型 | 不直接绑定模型；记忆提取与画像由服务端完成 | 同长期记忆 API；OpenClaw 插件不支持 Coding Plan [API Key](../concepts/api-key.md) | 预置千问系列、DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 及自定义模型 | -| 数据存储 | 记忆片段与画像，按 `user_id` 隔离，按 `memory_library_id` 分库 | 同长期记忆 API；默认库预置"默认有效期 180 天"规则 | 向量索引（text-embedding-v3/v4 512 维；multimodal-embedding-v1 1024 维） | -| 计费方式 | 按 API 调用计费（具体见平台说明） | 同长期记忆 API | 规格费用（按小时）+ 模型调用费用（[Token](../concepts/token.md)）；标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| 并发/限流 | 全部接口 ≤3000 QPM；add 120 QPM；search 300 QPM | 同长期记忆 API | 标准版 1 QPS（固定）；旗舰版 50–10,000 QPS（1–200 RCU） | -| 有效期 | API 直写：暂无失效日期 | 控制台规则：7/30/180 天或永不过期（默认 180 天） | 持久存储，无失效概念 | -| 接入方式 | 直接调用 RESTful API；Python 可用 `agentscope-runtime` | API 直连 / 百炼控制台 / OpenClaw 插件（零侵入） | 控制台创建 + SDK 集成；可挂载到[智能体应用](../concepts/agent-application.md)、工作流应用 | -| 典型场景 | 跨会话个性化、用户偏好持久化、自动提取关键事件 | 用户长期记忆、画像维护、Agent 零侵入记忆接入 | 私有文档问答、领域知识检索、图文/音视频内容搜索 | - -## 各方案适用场景建议 - -### 长期记忆（新）API - -适合需要**精细控制记忆生命周期**的开发者：自行管理写入（AddMemory）、语义检索（SearchMemory）、更新与删除，并通过画像模板（Profile Schema）维护结构化用户属性。当业务需要将记忆能力嵌入自有应用、对 `user_id` 与 `memory_library_id` 做多租户隔离、或希望用 `custom_content` 直接写入指定记忆（绕过对话提炼）时，优先选择此 API。 - -### 记忆库（Memory Library） - -适合希望**以最低接入成本获得跨会话记忆**的场景： - -- **OpenClaw 插件方式**：通过 `before_agent_start`（自动召回）与 `agent_end`（自动捕获）两个 Gateway 钩子实现零侵入记忆，所有提炼、向量化、语义检索由百炼服务端完成。适合基于 OpenClaw 构建的 Agent，无需改动业务代码。 -- **控制台方式**：在百炼控制台可视化管理记忆库与规则，支持配置记忆片段有效期（7/30/180 天或永不过期），适合非技术运营人员参与记忆策略管理。 -- **API 直连方式**：与长期记忆 API 一致，适合自定义接入。 - -注意：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置。 - -### 知识库（Knowledge Base） - -适合需要**让模型基于私有数据回答问题**的 RAG 场景： - -- 文档问答（pdf/docx/markdown/图片等，单文件最大 150MB） -- 数据查询（xlsx 数据表，最大 10 万行） -- 图片问答（multimodal-embedding-v1） -- 音视频搜索（最大 512MB） - -知识库仅在**中国站华北2（北京）**地域可用，提供标准版（1 QPS、≤100 GB）与旗舰版（50–10,000 QPS、≤9,999 GB）两档规格。创建后知识库类型、metadata 抽取与切片策略不可更改，需一次性规划。适合企业知识库、产品手册问答、领域知识检索等"知识供给"型应用。 - -## 技术选型参考 - -| 选型问题 | 推荐方案 | -| --- | --- | -| 需要记住用户偏好、历史事件，实现跨会话个性化 | 长期记忆 API 或记忆库 | -| 基于 OpenClaw 构建 Agent，希望零侵入接入记忆 | 记忆库（OpenClaw 插件） | -| 需要运营人员在控制台管理记忆规则与有效期 | 记忆库（控制台） | -| 需要精细控制记忆的增删改查与画像模板 | 长期记忆（新）API | -| 需要基于私有文档/数据做语义检索问答 | 知识库 | -| 需要处理图片/音视频等多模态内容检索 | 知识库（图片问答/音视频搜索） | -| 高并发检索（>1 QPS） | 知识库旗舰版（最多 10,000 QPS） | -| 同时需要"用户记忆"和"文档知识" | 记忆库 + 知识库组合使用，二者互补 | - -**一句话总结**：长期记忆 API 与记忆库解决"记住用户是谁、做过什么"的问题（个性化上下文），知识库解决"模型不知道的领域知识"的问题（RAG 检索）。二者并不互斥，可在同一应用中组合使用——用记忆库维持用户画像与历史，用知识库供给领域文档，共同提升大模型在特定业务中的表现。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [knowledge base](../guides/knowledge-base.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md deleted file mode 100644 index a5963c3b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 记忆库与长期记忆对比 - -百炼平台为解决大模型跨会话上下文丢失的问题，提供了围绕"长期记忆"的完整能力。在文档体系中，这一能力以两种视角呈现： - -- **记忆库（Memory Library）**：偏产品与方案视角，强调接入方式（控制台 / HTTP API / OpenClaw 插件）、记忆片段与用户画像两类内容形态，以及记忆规则与有效期等可运营配置。 -- **长期记忆（新）API**：偏接口与实现视角，给出 RESTful 端点、请求/响应字段、限流策略和画像模板（Profile Schema）的完整 CRUD。 - -二者底层共用同一套 `https://dashscope.aliyuncs.com/api/v2/apps/memory/` 服务，记忆片段与用户画像的数据模型一致；区别在于"封装层"和"可控粒度"。本文从开发者技术选型角度对两者进行对比。 - -## 关键维度对比 - -| 维度 | 记忆库（Memory Library） | 长期记忆（新）API | -| --- | --- | --- | -| 定位 | 上层方案与产品形态：跨会话记忆的整体接入 | 底层 RESTful 接口参考：程序化记忆管理 | -| 文档位置 | 应用使用指南（guides） | 应用 API 参考（api） | -| 接入方式 | 控制台可视化管理 + HTTP API + OpenClaw 插件零侵入 | 直接 HTTPS 调用 `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 记忆内容 | 记忆片段 + 用户画像（两类可独立或组合使用） | 记忆片段 + 用户画像（同一数据模型） | -| 写入输入 | `messages`（自动提取）或 `custom_content`（直写，最大 512 字符），二选一 | 同左，`messages` 最多 50 条对话 | -| 检索能力 | `SearchMemory`，主要参数 `top_k`（建议 3–10） | `SearchMemory`，含 `top_k`（1–100，默认 10）、`min_score`（默认 0.3）、`enable_rerank`/`enable_judge`/`enable_rewrite`、`project_ids` 多规则混合检索 | -| 管理 API | 重点呈现 `AddMemory` / `SearchMemory`（写入与召回） | `AddMemory` / `SearchMemory` / `ListMemory` / `UpdateMemory` / `DeleteMemory` + 画像模板 CRUD + `GetUserProfile` | -| 画像管理 | 通过 `profile_schema` 参数提取画像，模板在记忆库详情页获取 | 提供 `CreateProfileSchema` / `ListProfileSchemas` / `UpdateProfileSchema` / `DeleteProfileSchema` / `GetProfileSchema` 全套接口 | -| 记忆有效期 | 控制台默认规则 180 天，可配 7/30/180 天或永不过期；按规则可编辑 | API 文档标注"生成的记忆片段与用户画像暂无失效日期"（以控制台记忆规则配置为准） | -| 记忆规则 | 每账号自带默认记忆库 + 默认规则（不可删除），可创建新记忆库与规则 | 通过 `memory_library_id`、`project_id` 参数指定记忆库与规则；不传使用默认 | -| 用户隔离 | `user_id` 命名空间隔离，OpenClaw 插件所有 Agent 共享同一记忆 | `user_id` 最大 64 字符，用于标识记忆归属 | -| 限流 | 未单独列出，复用底层 API 限额 | 全部接口合计 ≤ 3000 QPM；`add` 120 QPM；`search` 300 QPM | -| 客户端 SDK | Python `agentscope-runtime`（`AddMemory` / `SearchMemory` 等封装，需 `close()`） | 以 cURL / REST 为主，参数与字段为权威定义 | -| 插件支持 | OpenClaw `modelstudio-memory-for-openclaw`：`before_agent_start` 自动召回 + `agent_end` 自动捕获 | 不直接涉及，插件内部回调此 API | -| 典型场景 | 跨会话个性化、Agent 偏好记忆、控制台运营记忆规则、OpenClaw 零侵入接入 | 程序化记忆 CRUD、画像模板生命周期管理、批量召回与重排序、自建记忆编排 | - -## 适用场景建议 - -### 选择"记忆库（Memory Library）"视角当 - -- 需要在控制台可视化地创建记忆库、配置记忆片段规则与有效期（7/30/180 天或永不过期）。 -- 希望以"产品方案"形式接入，例如通过 OpenClaw 插件实现 `autoCapture` / `autoRecall` 的零侵入跨会话记忆，而不愿手写每轮的写入与检索调用。 -- 业务侧关注的是"用户偏好持续化""Agent 跨会话理解"等整体能力，而非单个接口字段。 - -### 选择"长期记忆（新）API"视角当 - -- 需要在自研应用中精细控制每一步记忆操作：写入、搜索、列表、更新、删除，以及对画像模板做完整的增删改查。 -- 需要使用高级检索参数（`min_score` 阈值、`enable_rerank` 重排序、`enable_judge` 意图判别、`enable_rewrite` query 重写、`project_ids` 多规则混合检索）来调优召回质量。 -- 需要依据明确的限流（3000 QPM 总量、add 120 QPM、search 300 QPM）做容量规划与重试策略。 -- 需要程序化维护用户画像模板的字段定义，或对接已有用户体系做批量画像写入与读取。 - -## 技术选型小结 - -记忆库与长期记忆（新）API 并非二选一的两套系统，而是**同一能力的产品层与接口层**： - -- 做方案设计与运营配置时，以"记忆库"文档为准（接入方式、记忆规则、有效期、OpenClaw 插件配置）。 -- 做接口对接与字段实现时，以"长期记忆（新）API"文档为准（端点、参数、返回结构、限流、画像模板 CRUD）。 - -实践建议：先用记忆库视角确定接入形态（控制台 / API / 插件）与记忆规则，再在长期记忆（新）API 中查证具体端点与字段；两者配合即可覆盖从产品方案到代码实现的完整链路。注意记忆有效期以控制台记忆规则配置为准——API 文档的"暂无失效日期"指 API 直写且不指定 `project_id` 时使用默认规则的情形。 - -## 被对比主题页 - -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md deleted file mode 100644 index e6d71fca..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md +++ /dev/null @@ -1,43 +0,0 @@ -# 记忆能力对比（长期记忆 vs 记忆库） - -百炼平台提供两种面向"跨会话上下文持久化"的记忆能力描述入口：一是 **长期记忆（新）**（API 参考视角，`api/long-term-memory-new.md`），二是 **记忆库（Memory Library）**（用户指南视角，`guides/memory-library-overview.md`）。两者本质指向同一套底层长期记忆 API（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`），但在文档定位、接入方式、管理入口和适用对象上存在差异。本页从技术选型角度对二者进行对比，帮助开发者快速判断应参考哪一份文档、采用哪种接入路径。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）（API 参考） | 记忆库（Memory Library）（用户指南） | -| --- | --- | --- | -| 文档定位 | RESTful API 接口参考，逐接口说明请求/响应字段 | 能力总览与接入指南，含概念、控制台管理与插件接入 | -| 受众 | 直接调用 HTTP API 的后端/服务端开发者 | 需要端到端方案选型、含控制台与零侵入接入的开发者 | -| Base URL | `https://dashscope.aliyuncs.com/api/v2/apps/memory/` | 同上（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`） | -| 认证方式 | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | 环境变量 `DASHSCOPE_API_KEY`，同样以 `Bearer` 方式携带 | -| 接口覆盖 | AddMemory / SearchMemory / ListMemory / DeleteMemory / UpdateMemory / CreateProfileSchema / ListProfileSchemas / DeleteProfileSchema / UpdateProfileSchema / GetProfileSchema / GetUserProfile 共 11 个 | 重点讲 AddMemory / SearchMemory，并补充 ListMemory、CreateProfileSchema、GetUserProfile 等封装类 | -| SDK 支持 | 以 cURL 示例为主 | 额外提供 Python `agentscope-runtime` 封装类，需在 `finally` 调 `close()` | -| 零侵入接入 | 未涉及 | 提供 OpenClaw 记忆插件，`before_agent_start`/`agent_end` 钩子自动召回/捕获 | -| 控制台管理 | 未涉及 | 支持控制台可视化管理记忆库、记忆规则、默认有效期 | -| 记忆有效期 | 明确"生成的记忆片段与用户画像暂无失效日期" | 控制台默认规则 180 天，可配置 7/30/180 天或永不过期；以控制台规则为准 | -| 限流说明 | 给出账号级 QPM：全局 3000、add 120、search 300 | 未在总览中给出 QPM 数字 | -| 检索增强参数 | 列出 `top_k`/`min_score`/`enable_rerank`/`enable_judge`/`enable_rewrite` | 仅点出 `top_k`、`minScore`（插件）等关键参数 | -| 画像能力 | 完整的 Profile Schema CRUD 与 GetUserProfile 接口 | 介绍画像模板概念与 `profileSchema` 配置项，接口细节指向 API 参考 | -| [计费](../concepts/billing.md)方式 | 未在本页说明 | 未在本页说明（统一走 DashScope [计费](../concepts/billing.md)） | -| 典型场景 | 需要精细控制记忆 CRUD、画像模板、检索召回参数的服务端集成 | 需要快速接入、可视化运维或让 OpenClaw Agent 自动具备记忆 | - -## 适用场景建议 - -- **参考"长期记忆（新）"文档的情况**：你需要直接对接 HTTP API，关心每个接口的请求体、响应字段、`event` 事件类型（ADD/UPDATE/DELETE）、画像模板的完整 CRUD，或需要按 `enable_rerank`/`enable_judge`/`enable_rewrite` 等参数精细调优检索召回；适合自研 Agent 后端、需要严格接口契约的服务端开发者。 -- **参考"记忆库"文档的情况**：你希望先从业务视角理解"记忆片段 vs 用户画像"两类持久化内容的差异与组合用法，或希望通过控制台创建/编辑记忆库与记忆规则、配置有效期，又或者你的 Agent 运行在 OpenClaw 之上，希望以插件方式零侵入获得"自动捕获/自动召回"能力；适合做整体方案选型与低代码运维的团队。 -- **二者结合使用**：多数生产落地建议先读"记忆库"理解概念与管理入口，再用"长期记忆（新）"对照接口字段落地代码；OpenClaw 用户可仅依赖插件配置项即可跑通，深定制时再回查 API 参考。 - -## 技术选型参考 - -1. 接入路径：纯后端 HTTP 调用 → 选 API 直连（两份文档均适用，接口细节以"长期记忆（新）"为准）；OpenClaw Agent → 选记忆插件（仅"记忆库"文档覆盖）。 -2. 有效期策略：若需记忆按天失效，务必以控制台记忆规则配置为准（默认 180 天）；API 直写且不指定 `project_id` 时走默认规则，"暂无失效日期"的说法仅适用于不经过规则提炼的直写场景。 -3. 命名空间隔离：两份文档均强调 `user_id` 为记忆空间隔离维度，不同 `user_id` 完全隔离；OpenClaw 插件目前所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置，也不支持百炼 Coding Plan 的 API Key。 -4. 检索质量：需要重排序、意图判别、query 重写等高级召回能力时，参考"长期记忆（新）"的 SearchMemory 参数；插件场景受 `topK`/`minScore` 配置项约束。 -5. 画像存储：需要固定结构化属性（年龄、职业、偏好等）持久化时使用用户画像，字段命名应清晰具体、避免同义并存；接口细节走"长期记忆（新）"的 Profile Schema 系列。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md new file mode 100644 index 00000000..87267d43 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md @@ -0,0 +1,61 @@ +# [长期记忆](../concepts/long-term-memory.md)与知识库方案对比 + +为帮助开发者在智能体（Agent）与 RAG 应用开发中做出精准技术选型，本文系统对比百炼平台两大核心上下文增强能力：**[长期记忆](../concepts/long-term-memory.md)（Long-Term Memory, 新版）** 与 **知识库（Knowledge Base）**。二者虽均基于向量检索与语义理解，但设计目标、数据来源、生命周期管理及集成范式存在本质差异。本对比聚焦实际工程落地维度，涵盖接口行为、模型依赖、计费逻辑与典型适用场景，旨在提供可操作的选型决策依据。 + +## 关键维度对比 + +| 维度 | [长期记忆](../concepts/long-term-memory.md)（新） | 知识库 | +|------|----------------|---------| +| **核心定位** | 用户级、会话级**个性化上下文持久化**：捕获并结构化用户偏好、意图、习惯、关系等动态语义信息 | **领域/业务级静态知识注入**：为大模型提供私有、结构化或非结构化的外部事实性知识（文档、表格、音视频等） | +| **输入格式** | • `messages`：多轮对话数组（最多50条），自动提取记忆片段
• `custom_content`：纯文本（≤512字符），绕过提取直接写入
• 支持 `meta_data` 自定义元数据 | • 多源文件：PDF/DOCX/TXT/CSV/JSON/MP3/MP4 等（单文件 ≤150MB）
• 支持 API 批量上传或控制台导入
• 索引时自动切片（≤6000 Token/片）并抽取 `filename`/`date`/`author` 等元数据 | +| **输出格式** | • `SearchMemory` 返回结构化记忆片段列表，含 `id`、`content`、`score`、`meta_data`、`created_at`
• `GetUserProfile` 返回 JSON Schema 定义的结构化画像对象 | • 检索服务：返回带 `score`、`source`（文件名/页码）、`content`、`metadata` 的文本切片数组
• 问答服务：返回生成答案 + 引用溯源（高亮原文位置 + 文件链接） | +| **支持模型** | • **记忆提取**：由平台内置专用模型驱动（不暴露给用户选择）
• **检索排序**：默认向量模型 + 可选 `enable_rerank`（使用平台统一 rerank 模型）
• **不支持自定义模型替换** | • **检索模型**：支持指定向量模型（如 `qwen3-embedding`）
• **重排模型（Rerank）**：纯文本知识库仅支持 `qwen3-rerank`；多模态知识库支持 `qwen-vl-rerank` 等视觉专用模型
• **生成模型**：问答服务可自由绑定任意百炼平台支持的 LLM（Qwen 系列、DeepSeek、Llama3.1 等） | +| **API 端点** | • Base URL：`https://dashscope.aliyuncs.com/api/v2/apps/memory/`
• `POST /add`（新增）
• `POST /memory_nodes/search`（检索）
• `GET /memory_nodes?user_id=xxx`（分页查询）
• `DELETE /memory_nodes/{id}` / `PATCH /memory_nodes/{id}`（删/改） | • Base URL：`https://dashscope.aliyuncs.com/api/v2/knowledgebase/`（华北2地域）
• `POST /retrieval`（独立检索）
• `POST /qa`（问答服务）
• `POST /knowledgebases/{kb_id}/documents`（上传）
• 控制台创建后生成专属服务端点（含鉴权Token） | +| **计费方式** | • **按调用量计费**：
 – `AddMemory` / `SearchMemory` / `ListMemory` 等 API 调用按次计费（具体单价见控制台定价页）
 – 无知识库规格费、无向量模型 Token 费
• **无存储容量费**（记忆片段按账号配额管理） | • **双重计费**：
 – **规格费**：按知识库运行时长（标准版 0.03 元/小时）或 RCU（旗舰版 0.2 元/RCU/小时）
 – **模型费**：向量嵌入（Embedding）与 Rerank 模型按实际 Token 消耗计费（独立于规格费）
 – 问答服务中的 LLM 调用另计费 | +| **典型场景** | • 智能客服：记住用户历史投诉、设备型号、服务偏好
• 个人助手：持续跟踪日程提醒、饮食禁忌、旅行计划
• 教育 Agent：记录学生错题类型、薄弱知识点、学习节奏
• 游戏 NPC：维护玩家角色关系、阵营立场、任务进度 | • 企业知识问答：HR 政策、IT SOP、产品手册即时查询
• 法律/医疗辅助：基于法规条文、临床指南生成专业建议
• 客服工单处理：关联历史工单、解决方案库、产品变更日志
• 投研分析：从财报、研报 PDF 中提取关键财务指标与风险提示 | +| **数据生命周期** | • 默认无自动过期（需业务侧通过 `expire_time` 参数或定时任务清理）
• 控制台支持配置全局有效期（7/30/180天或永不过期）
• `UpdateMemory` 仅更新内容与 `meta_data`，不改变向量索引时间戳 | • 文档上传后即构建索引，无显式过期机制
• 更新知识需重新上传文件或调用 `update_document` API 触发增量索引
• 删除文档后，对应切片从向量库中移除（约1-5分钟生效） | +| **地域与权限** | • 全地域可用（与 DashScope API 一致）
• 仅需 `DASHSCOPE_API_KEY`（Bearer Token 认证） | • **仅限中国站华北2（北京）地域**
• 需子账号具备 `AliyunBailianDataFullAccess` 权限
• SDK 调用需配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` 等 AK/SK 环境变量 | +| **SDK 支持** | • Python：`agentscope-runtime>=1.1.5` 提供 `AddMemory`/`SearchMemory`/`ListMemory`/`DeleteMemory` 异步封装
• `UpdateMemory` 需直调 REST API
• OpenClaw 插件开箱即用（`autoCapture`/`autoRecall`） | • Python/Java SDK 提供完整生命周期管理（创建、上传、索引、检索、问答）
• 控制台生成的问答服务支持一键导出 SDK 调用示例
• 不提供 OpenClaw 原生插件 | + +## 各方案的适用场景建议 + +### ✅ 优先选用 **长期记忆（新）** 当： +- 你需要**跨会话记住单个用户的行为特征与主观状态**（如“张三讨厌电话推销”、“李四每周三健身”）； +- 应用逻辑依赖**动态更新的用户画像**（年龄、职业、兴趣标签），且需与对话流深度耦合； +- 场景对**低延迟、高并发写入**有要求（如每轮对话结束自动存记忆），且无法接受知识库的文档上传/索引延迟； +- 数据敏感度高，**拒绝将用户对话原始内容上传至共享知识库**，要求严格按 `user_id` 隔离； +- 工程团队倾向轻量级集成，仅需几行 SDK 代码即可启用记忆能力，无需管理知识库规格与配额。 + +### ✅ 优先选用 **知识库** 当： +- 你的核心需求是**让大模型准确回答基于私有文档的问题**（如“最新版《员工手册》第5章关于年假的规定是什么？”）； +- 知识源为**批量、静态、结构化程度不一的业务文档**（合同模板、产品说明书、会议纪要），且需支持 PDF 表格识别、音视频转文字等多模态解析； +- 要求**答案可溯源、可审计**，必须明确标注引用来源（文件名+页码+段落）； +- 需要**混合检索能力**（向量 + 关键词）、**多知识库联合混排**（如同时查 HR 政策 + IT 流程 + 财务制度），并精细调控权重与标签过滤； +- 团队已具备文档治理流程，能接受知识库创建后**配置不可逆**（需重建），并愿意承担规格费与模型 Token 成本以换取更高精度与稳定性。 + +### ⚠️ 注意边界与组合策略 +- **不要混淆用途**：长期记忆 ≠ 用户文档存储空间；知识库 ≠ 用户偏好数据库。将用户聊天记录直接丢进知识库，既浪费成本又降低检索精度。 +- **推荐组合使用**：典型智能体架构中，**长期记忆负责“用户是谁、想要什么”**（个性化上下文），**知识库负责“世界是什么、规则是什么”**（领域知识）。二者通过不同 `user_id` 和 `kb_id` 隔离，再由 Agent 编排协同调用。 +- **性能兜底建议**：对高 QPS 场景（如百万级用户助手），长期记忆的 `SearchMemory`（300 QPM 限额）可能成为瓶颈，此时可结合本地缓存（如 Redis）暂存高频用户记忆；知识库则需根据并发量选择旗舰版 RCU 规格。 + +## 面向开发者的选型参考 + +| 选型问题 | 长期记忆（新） | 知识库 | 决策建议 | +|----------|----------------|---------|-----------| +| **我的数据是用户对话产生的个性化信息吗？** | 是 | 否 | ✔️ 选长期记忆 | +| **我的数据是公司内部的 PDF/Excel/音视频等业务资料吗？** | 否 | 是 | ✔️ 选知识库 | +| **我需要为每个用户单独隔离数据，且不能共享？** | 是（`user_id` 强隔离） | 否（知识库全局共享，靠权限控制访问） | ✔️ 选长期记忆 | +| **我需要答案附带原文出处，满足合规审计要求？** | 否（仅返回记忆内容） | 是（返回 `source` 字段与文件链接） | ✔️ 选知识库 | +| **我的应用部署在新加坡/法兰克福地域？** | 支持 | ❌ 不支持（仅华北2） | ✔️ 必须选长期记忆 | +| **我追求最低接入成本，希望 SDK 一行代码启用？** | `AddMemory().arun(...)` 即可 | 需先创建 KB、上传文档、触发索引，再调用检索 | ✔️ 选长期记忆 | +| **我需要支持视觉文档（扫描件/PPT）的 OCR 与理解？** | 不支持 | 支持（需创建“多模态知识库”，绑定 VL 模型） | ✔️ 选知识库 | + +> **最后建议**：首次集成时，请务必使用 cURL 或 Postman 验证基础 API 流程（而非直接依赖 SDK 封装），重点关注 `request_id` 日志排查；生产环境务必设置合理的 `min_score`（长期记忆 ≥0.5，知识库 ≥0.3）与 `top_k`（3–10），避免噪声干扰或召回不足。 + +## 被对比主题页 + +- [long term memory new](../api/long-term-memory-new.md) +- [knowledge base](../guides/knowledge-base.md) +- [memory library overview](../guides/memory-library-overview.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md deleted file mode 100644 index 61ec6040..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md +++ /dev/null @@ -1,43 +0,0 @@ -# 模型直调与应用调用对比 - -百炼平台对外提供两类 API 调用路径：**模型直调**（直接调用 Qwen 等文本生成模型）与**应用调用**（调用在控制台预先编排好的智能体、工作流、Agent 2.0 应用）。两者底层都走 DashScope 网关并复用同一套 [API Key](../concepts/api-key.md)，但在调用对象、输入格式、能力范围与典型场景上有显著差异。本页面向做技术选型的开发者，按关键维度对比两种方案，帮助快速判断应走哪条路径。 - -## 关键维度对比 - -| 维度 | 模型直调（Qwen API） | 应用调用（Application API） | -| --- | --- | --- | -| 调用对象 | 单个文本生成模型（`model` 字段指定 Qwen 系列模型名） | 控制台已编排好的应用（智能体 / 工作流 / Agent 2.0），由 `APP_ID` 标识 | -| 核心入口 | OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages、DashScope 原生 | OpenAI 兼容 Responses API（`/apps/agent/{APP_ID}/compatible-mode/v1/responses`）、DashScope 原生 API（`/apps/{APP_ID}/completion`） | -| 前置准备 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY`） | [API Key](../concepts/api-key.md) + 应用 ID（`APP_ID`，控制台手动获取；子[业务空间](../concepts/workspace.md)还需 `Workspace ID`） | -| 输入格式 | `messages` 数组（OpenAI/Anthropic 兼容）或 Responses 的 `input`；需自行维护对话历史（Responses 接口除外） | `input.prompt` 字符串 / `input` 消息数组（OpenAI 兼容）；`input.prompt` + `parameters` + `biz_params`（DashScope 原生） | -| 对话历史 | 仅 OpenAI 兼容 Responses 由平台自动管理；其余接口调用方自行拼接 `messages` | 支持 `session_id`（云端托管，1 小时有效、最多 50 轮）或自行管理 `messages`；同时传两者时以 `messages` 为准 | -| 内置工具能力 | 联网搜索、代码解释器、网页内容提取仅 Responses 接口内置；其他接口需自行定义工具 | 应用编排阶段在控制台绑定插件 / 节点，调用时通过 `biz_params.user_defined_params` 透传业务参数，工具由应用内部装配 | -| 支持模型 | Qwen 全系列（含 VL 多模态），按 `model` 字段切换 | 由应用编排时所选模型决定（如智能体做图像输入需选 Qwen-VL 并设「自定义处理」） | -| [流式输出](../concepts/streaming-output.md) | 各接口均支持 `stream` 参数 | 同步调用支持 `stream=True`；[异步调用](../concepts/async-invocation.md)暂不支持流式；工作流需在输出节点启用「[流式输出](../concepts/streaming-output.md)」并重新发布 | -| 异步执行 | 由调用方自行实现 | OpenAI 兼容 Responses 支持 `background=True`，返回任务 ID 后轮询 `retrieve` 至终态 | -| 多模态 | 各接口按协议支持文本 / 图像 / 文件（`input_image` / `input_file`） | `input` 消息数组支持 `input_text` / `input_image` / `input_file`（`input_file` 仅[智能体应用](../concepts/agent-application.md)支持） | -| 计费方式 | 按 token 用量计费（输入 + 输出） | 按 token 用量计费；应用编排内部多步调用累计计费 | -| 功能完整度 | DashScope 原生接口参数最全；兼容接口为保证协议一致可能不暴露全部原生参数 | DashScope 原生 `/completion` 功能更全、性能更高；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)便于复用 OpenAI 生态 | -| 典型场景 | 单模型推理、文本生成、对话补全、从 OpenAI/Anthropic 平迁、需要精细采样参数 | 复用已编排的多步骤智能体 / 工作流、内置 RAG 与插件、长耗时异步任务、业务系统集成 | - -## 选型建议 - -- **选模型直调**：当你只需要一个模型做单轮或多轮文本生成、补全、对话，且希望直接复用 OpenAI / Anthropic SDK 与既有代码库，或需要最全的采样参数与原生能力时。从外部平台迁入时优先评估兼容接口，需要极致功能时再切到 DashScope 原生接口。 -- **选应用调用**：当业务逻辑已被编排成智能体或工作流（含 RAG、插件、多节点流程），希望把整套能力一次性集成进业务系统，而不是在调用方重新实现编排逻辑时。长耗时任务（报告生成、多步工具调用）选异步 `background=True`；需要复用 OpenAI 工具链选 Responses API，需要更全功能与更高性能选 DashScope 原生 `/completion`。 -- **混合使用**：两者共用同一 `DASHSCOPE_API_KEY`，可在同一业务系统中并存——轻量推理走模型直调，复杂编排走应用调用，按场景而非按模型选路径。 - -## 注意事项 - -- **凭证管理**：API Key 推荐写入 `DASHSCOPE_API_KEY` 环境变量，不要在生产环境硬编码。 -- **应用 ID 获取**：`APP_ID` 与 `Workspace ID` 目前只能在控制台手动复制，不支持 API / CLI 查询；RAM 子账号默认只能查看其已加入的[业务空间](../concepts/workspace.md)。 -- **地域差异**：上述 Endpoint 默认适用于华北2（北京）；德国（法兰克福）、新加坡、日本（东京）等地域或调用子[业务空间](../concepts/workspace.md)下应用时，请求须包含 `Workspace ID`，且该 ID 是对应地域 Base URL 的组成部分。 -- **兼容接口的功能取舍**：OpenAI / Anthropic 兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全参数、插件或业务字段，应改用 DashScope 原生接口。 -- **迁移评估**：从 OpenAI / Anthropic 迁移到模型直调时，先确认目标 Qwen 模型在对应兼容接口下是否支持所需参数（`temperature`、`tools`、`stream` 等）；迁移到应用调用时，注意异步不支持流式、多轮历史需自行管理（除非用 `session_id`）。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md deleted file mode 100644 index af15d86e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md +++ /dev/null @@ -1,64 +0,0 @@ -# 模型微调、压缩与部署对比 - -百炼平台围绕“把通用大模型变成业务专属模型”提供了一条递进式链路：**模型微调（Fine Tuning）→ 模型压缩（Model Compression）→ [模型部署](../concepts/model-deployment.md)（Model Deployment）**。三者并非并列的替代方案，而是同一交付链上前后衔接的环节——微调负责把业务/场景知识写入模型参数，压缩负责把全精度微调模型量化为低精度版本以降低部署门槛，部署负责把模型上线为可调用的推理服务。本文从输入格式、输出产物、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者明确各环节的边界与衔接关系，避免在选型时混淆“训练 / 压缩 / 部署”三件事。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | [模型部署](../concepts/model-deployment.md)（Model Deployment） | -| --- | --- | --- | --- | -| 核心目标 | 将业务/场景知识、风格、偏好写入模型参数 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 将预置或调优后的模型上线为独立、资源专享的推理服务 | -| 在链路中的位置 | 链路起点，产出可被压缩/部署的微调模型 | 微调之后、部署之前的可选优化环节 | 链路终点，对外提供推理 API | -| 输入格式 | CPT：1000 万+ Token 无标签领域文本；SFT：1000+ 条问-答对；DPO：100+ 组偏好对；万相图像/视频：`.zip` 内含 `data.jsonl` 与素材；CosyVoice：多条/数小时录音 | 当前工作空间内已有的自定义微调模型；量化模板要求时还需校准数据（内部上传且已发布的数据集） | 模型 ID（预置模型或调优/压缩产出模型）；API/命令行调用还需 API Key 与业务空间 | -| 输出产物 | 自定义微调模型（全参或 LoRA），可用于压缩或直接部署 | 量化后的低精度模型，落盘到「我的模型」，可直接部署 | 在线推理服务（`deployed_model` 唯一 ID，`status=RUNNING`） | -| 支持模型 | 文本：Qwen3.x / Qwen2.5 系列等；视觉：Qwen3-VL / Qwen2.5-VL；图像：wan2.7；视频：wan2.2/wan2.5；语音：cosyvoice-v3-flash | 仅「支持压缩的模型」列表中的自定义微调模型（如 qwen3.5-flash-2026-02-23 对应的微调版）；已量化模型不可二次压缩 | 预置模型、调优产出模型、压缩产出模型、OSS 导入的 LoRA 模型（仅 LoRA，不支持全参微调模型导入） | -| 操作入口 | 控制台「模型调优」页面；或 DashScope API `POST /api/v1/fine-tunes` | 控制台「模型训练 > 模型压缩」页面；或 OpenAPI | 控制台「[模型部署](../concepts/model-deployment.md)」页面（北京）；或 `https://dashscope.aliyuncs.com/api/v1/deployments` | -| 计费方式 | 按 Token 计费（API 创建的任务仅支持按 Token，不支持训练单元）；控制台任务可用训练单元（预付费/后付费）；计费与 `n_epochs`/`batch_size`/`max_length` 等超参相关 | 压缩功能限时免费；产出模型部署后按所选部署单元规格计费 | 预置吞吐（PTU）、模型单元（MU）、按 Token 用量三种；PTU 支持长输入阶梯系数与前缀缓存折扣；服务创建后无法切换计费方式 | -| 可逆性 | 可重新训练、迭代 | 不可逆：产出模型不支持继续训练，也不支持二次压缩；如需迭代须回到上游全精度模型重新训练 | 可下线重新部署以切换计费方式；按 Token 用量模式一个月不使用自动释放 | -| 任务状态流转 | PENDING → SUCCEEDED（轮询 `GET /api/v1/fine-tunes/{job_id}`） | 待开始 / 排队中 / 运行中 / 停止中 / 压缩成功 / 压缩失败 / 已取消；仅排队中/运行中可取消 | PENDING → RUNNING（轮询 `GET /api/v1/deployments/{deployed_model}`） | -| 关键约束 | `-Base` 后缀模型不可直接用于调优对话；CosyVoice 调优仅支持 API 发起、无法扩展基础模型不支持的语种 | 校准数据控制台 UI 暂不支持 OSS 挂载类型；模板名称决定压缩后可部署规格（如 MU5、MU8） | 仅华北二（北京）地域；导入 LoRA 的 rank 须为 8/16/32/64 且各层一致；VL 模型须冻结 VIT | - -## 各方案适用场景建议 - -### 模型微调 - -适合“Prompt 工程已到上限、需要把知识或风格沉淀进模型参数”的场景： - -- **CPT（继续预训练）**：业务有大量无标注领域语料（专业词汇、行业事实），需要先做领域适应。 -- **SFT（监督微调）**：需要模型遵循特定对话格式或任务执行规范，有 1000+ 条高质量问-答对。 -- **DPO（直接偏好优化）**：SFT 之后想进一步对齐人类偏好、抑制坏答案。 -- **LoRA 高效训练**：数据集较小、需快速验证、追求低成本快迭代；全参训练则用于追求全局效果最优。 -- 多模态场景：文生图/图生图训练人物 IP 或风格 LoRA（万相 wan2.7）、图生视频训练首帧/首尾帧 LoRA（wan2.2/wan2.5）、语音合成训练专属音色（CosyVoice）。 - -### 模型压缩 - -适合“已有全精度微调模型，但希望部署到更小规格单元、降低成本并提升吞吐”的场景： - -- 微调产物部署成本偏高，想通过量化降到 MU5/MU8 等更小规格。 -- 对推理显存和吞吐有明确优化诉求，且不打算继续迭代模型参数（压缩不可逆）。 -- 不适合：尚未完成微调的模型、已量化过的模型、需要继续训练迭代的模型。 - -### 模型部署 - -适合“模型已就绪（无论预置、微调还是压缩产出），需要对外提供高并发、低延迟推理服务”的场景： - -- **PTU（预置吞吐）**：高负载生产环境、稳定吞吐、流量可预估，追求 TPS 提升（约 1.5~2.0 倍于按 Token）。 -- **模型单元（MU）**：资源独占、需自定义性能指标、长时任务；支持 PD 分离以降低首 Token 延迟。 -- **按 Token 用量**：调优后效果验证、性价比优先、对并发延迟要求不高；仅支持部分 LoRA 调优模型，一个月不使用自动释放。 -- **自定义 LoRA 导入**：本地训练的 LoRA 模型（千问3/千问2.5 系列及 VL 版本）需从 OSS 导入后再部署；注意推理参数建议参照 vLLM 默认值调整。 - -## 技术选型参考 - -1. **先判断是否需要动参数**：如果 Prompt 工程能解决，就不必微调；只有当知识/风格/偏好需要沉淀进模型时才进入微调环节。 -2. **微调模式选择**：快速验证用 LoRA，全局最优用全参；多模态按模型类型选 SFT-LoRA；语音合成走 CosyVoice SFT。 -3. **是否插入压缩环节**：微调产物部署成本敏感时，在部署前加一步压缩；但需注意压缩不可逆、不支持二次压缩，迭代须回到全精度模型重训。 -4. **部署计费方式选择**：生产高负载选 PTU（兼顾长输入与前缀缓存优惠）；资源独占与 PD 分离选模型单元；仅做效果验证选按 Token 用量。计费方式创建后不可切换，需下线重部署。 -5. **链路衔接**：微调 →（可选）压缩 → 部署 是单向链路；压缩产出的模型不可继续训练，所以“想压缩又想继续迭代”的需求要把全精度模型作为迭代基线，每次迭代后重新压缩。 -6. **入口选择**：零代码/合规场景走控制台；需要自动化编排（如 CI/CD）走 API/命令行；CosyVoice 调优目前仅支持 API，控制台不可用。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md deleted file mode 100644 index cca2e06a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md +++ /dev/null @@ -1,43 +0,0 @@ -# 微调、压缩与部署方案对比 - -百炼平台围绕「让模型贴合业务」提供了三条相互衔接的路径：**模型调优（Fine Tuning）**把业务/场景知识写进模型参数；**模型压缩（Model Compression）**用量化算法降低微调模型的精度与显存占用；**模型部署（Model Deployment）**把平台预置模型或调优后的模型部署为在线推理服务。三者构成 `调优 → （可选）压缩 → 部署 → 调用` 的链路，本文从输入产出、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者根据「是否需要写知识」「是否要降本」「并发与延迟要求」做技术选型。 - -## 关键维度对比 - -| 维度 | 模型调优（Fine Tuning） | 模型压缩（Model Compression） | 模型部署（Model Deployment） | -| --- | --- | --- | --- | -| 一句话定位 | 将业务/场景知识写入模型参数，压低延迟、抑制幻觉、对齐偏好 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 把预置或调优后的模型部署为资源专享的在线推理服务 | -| 输入 | 训练数据集：CPT 需 1000 万+ [Token](../concepts/token.md) 无标签文本；SFT 需 1000+ 条问答对；DPO 需 100+ 组偏好对；万相/CosyVoice 需 `.zip`（含 `data.jsonl` + 媒体素材） | 当前工作空间内符合条件的**自定义微调模型**（全精度）+ 可选校准数据集 | 平台预置模型、调优产出模型、OSS 导入的 LoRA 模型；部署参数（`plan`、`capacity`、`ptu_capacity` 等） | -| 输出 | 微调后的自定义模型（全参或 LoRA 产物），可继续部署或压缩 | 量化后的低精度模型，可直接用于部署 | 在线推理服务（`deployed_model`），返回 `status` 为 `RUNNING` 后即可调用 | -| 主要调优/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA）；万相/CosyVoice 仅 `efficient_sft` | 量化模板（如 MU5、MU8），模板名决定压缩后可部署规格 | 预置吞吐（PTU）、模型单元（MU）、[Token](../concepts/token.md) 用量（lora）三种计费方式 | -| 支持模型 | 千问3.x/2.5 文本与 VL 系列、万相 wan2.7/wan2.2/wan2.5、CosyVoice-v3-flash 等 | 千问系列对应自定义微调模型（如 `qwen3.5-flash-2026-02-23`），以控制台列表为准 | 平台预置模型 + 部分调优模型；OSS 导入仅支持 LoRA，rank 须为 8/16/32/64 | -| 操作入口 | 控制台「模型调优」页面（零代码）或 DashScope API（`/api/v1/fine-tunes`） | 控制台「模型训练 > 模型压缩」页面或 OpenAPI | 控制台（北京）或 API/命令行（`/api/v1/deployments`） | -| 关键 API/参数 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`（`training_type`、`hyper_parameters`）、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | 任务名称、源模型、量化产出后缀、量化模板、`custom_calibration_file_ids` | `POST /api/v1/deployments`（`plan`=`ptu`/`mu`/`lora`）、`GET /api/v1/deployments/{deployed_model}`、`DELETE /api/v1/deployments/{deployed_model}` | -| 计费方式 | 控制台任务支持模型训练单元（预付费/后付费）；API 任务仅按 [Token](../concepts/token.md) 计费，不支持训练单元 | 当前**限时免费**；产出模型部署后按所选部署单元标准规格计费 | PTU（时长×预置吞吐）、模型单元（时长×数量×单价，可包月）、Token 用量（输入/输出 Token×单价，不使用不计费） | -| 扩缩容 | 不适用（训练任务级） | 不适用（压缩任务级） | PTU/模型单元自助增减；Token 用量需控制台提交申请人工审核 | -| 是否可逆 | 可基于基础模型重新训练迭代 | **不可逆**：产出模型不支持继续训练或二次压缩，需回到上游全精度模型重训 | 可随时 `DELETE` 下线服务，但删除后不可恢复 | -| 典型场景 | 注入领域知识、复刻特定风格/IP、对齐人类偏好、专属音色训练 | 降低微调模型部署成本，部署到更小规格单元、提升吞吐 | 高并发低延迟生产环境、稳定吞吐、调优后效果验证、长时任务 | - -## 各方案适用场景建议 - -- **模型调优（Fine Tuning）**：当 Prompt 工程已无法满足延迟、幻觉或风格复刻要求，需要把业务/场景知识直接写进模型参数时选用。数据量充足（CPT 千万级 Token、SFT 千条级问答、DPO 百组级偏好对）、且希望模型在特定任务上达到全局最优时，优先考虑全参训练；数据量小、需快速验证或仅复刻风格（万相 LoRA、CosyVoice 专属音色）时，优先用 LoRA 高效训练。注意 API 任务仅按 Token 计费，需用训练单元请走控制台。 -- **模型压缩（Model Compression）**：当已有全精度微调模型、希望降低显存占用并部署到更小规格单元时选用。压缩功能当前限时免费，是把调优成果低成本上线的有效手段。务必注意产出模型**不可继续训练、不可二次压缩**，迭代需回到上游全精度模型；模板名（如 MU5、MU8）直接决定后续可部署规格，应在压缩阶段就规划好部署目标。 -- **模型部署（Model Deployment）**：当需要资源专享、高并发、低延迟或稳定吞吐的生产级推理服务时选用。计费方式一旦创建不可更改，需按业务特征提前选定：高负载、流量可预估、追求低延迟选 PTU（TPS 约为按 Token 的 1.5~2.0 倍）；资源独占、长时任务、需 PD 分离选模型单元；调优后效果验证、对并发延迟要求不高、追求高性价比选 Token 用量（仅部分 LoRA 模型支持，一个月不用自动释放）。 - -## 选型链路建议 - -三者并非互斥，而是上下游关系：先用**调优**把知识写进模型；若部署成本敏感，再用**压缩**把全精度产物量化到更小规格；最后用**部署**把模型（无论是否压缩）上线为推理服务。典型组合： - -1. **轻量风格定制**：SFT-LoRA（万相/CosyVoice）→ 部署（Token 用量，快速验证）。 -2. **生产级领域模型**：CPT → SFT（全参）→ 压缩（量化到 MU8）→ 部署（PTU 或模型单元，高并发低延迟）。 -3. **偏好对齐优化**：SFT → DPO → 压缩 → 部署（模型单元，PD 分离降首 Token 延迟）。 - -选型核心判断：**要写知识选调优，要降成本选压缩，要上线选部署**；若同时追求极致效果与极致成本，按 `调优（全参）→ 压缩 → 部署（PTU/MU）` 串联即可。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md new file mode 100644 index 00000000..4390971f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md @@ -0,0 +1,51 @@ +# 模型部署方案对比：高并发推理、生产部署与压缩优化 + +为帮助开发者在不同业务阶段（如流量洪峰应对、长期服务上线、成本敏感型部署）科学选型，本文系统对比百炼平台三大核心模型部署能力：**高并发推理（TPM 预留 & 快速模式）**、**生产级模型部署（[model production](../api/model-production.md)）** 和 **模型压缩（[model compression](../guides/model-compression.md)）**。三者定位互补——高并发推理聚焦 *运行时性能保障*，生产部署解决 *定制化模型落地闭环*，压缩优化则面向 *推理成本与资源效率平衡*。理解其差异是构建稳定、高效、可演进 AI 服务的关键前提。 + +## 关键维度对比 + +| 维度 | 高并发推理（TPM 预留 + 快速模式） | 生产部署（[model production](../api/model-production.md)） | 压缩优化（[model compression](../guides/model-compression.md)） | +|------|----------------------------------|------------------------------|------------------------------| +| **核心目标** | 保障高吞吐稳定性（TPM 预留）或极致首 token/流式延迟（快速模式） | 实现私有化微调模型的端到端上线与服务化 | 降低已训练模型的推理资源消耗与部署成本 | +| **输入格式** | 标准 OpenAI 兼容请求体（`messages`, `stream`, `temperature` 等）；快速模式需额外适配 `reasoning_content` 字段解析 | 微调：JSONL 训练数据集 URL；部署：`model_id` / `fine_tuned_model_id` + 实例规格等配置参数 | 微调成功且状态为 `SUCCEEDED` 的自定义模型 ID；可选校准数据集（最多 5 个已发布数据集） | +| **输出格式** | 标准 OpenAI 流式/非流式响应；快速模式返回含 `delta.reasoning_content` 和 `delta.content` 的双通道结构 | 微调任务输出 `fine_tuned_model_id`；部署后返回 `endpoint_url`（兼容 `/v1/chat/completions`） | 生成新模型 ID（源模型名 + 后缀），如 `my-qwen-ft-w8a8`，存于模型中心，可直接用于部署 | +| **支持模型** | **TPM 预留**：Qwen、GLM、DeepSeek、Kimi 等十余个主流基础模型（如 `qwen3.7-max-2026-05-20`, `deepseek-v4-pro`）；
**快速模式**：仅 `glm-5.2-fast-preview`（Preview 阶段，严格限定） | 支持基于 Qwen 系列等基础模型的全参微调（`full`）与 LoRA 微调（`lora`）；部署对象为微调产出模型或 `import_model` 导入的第三方模型 | **仅限百炼平台内微调产出的自定义模型**（如 `qwen3.5-flash-2026-02-23-finetuned-xxx`）；不支持基础模型、OSS 模型、第三方模型 | +| **API 端点** | **TPM 预留**：复用标准 MaaS 域名（如 `https://{workspace_id}.maas.aliyuncs.com/v1`），但 `model` 参数需替换为专属 TPM code（如 `tpm-qwen37max-xxx`）；
**快速模式**：必须使用专属地域域名（如 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），`model="glm-5.2-fast-preview"` | 微调：`POST /api/v1/fine_tuning_jobs`；部署：`POST /api/v1/deployments`；服务调用：`POST {endpoint_url}/v1/chat/completions` | 控制台操作为主（路径：模型 > 模型训练 > 模型压缩）；无公开 REST API，任务通过控制台提交与管理 | +| **计费方式** | **TPM 预留**：按预留容量（kTPM）预付费，超额部分自动降级为按量计费；缩容按 1.5 倍系数结算；
**快速模式**：Preview 阶段暂未明确独立计费规则，实际按底层资源消耗计费（建议监控） | 微调：按 GPU 小时计费；部署：按所选实例规格（如 `ecs.gn7i-c16g1.4xlarge`）的 MU 小时计费；版本管理不额外收费 | 压缩任务本身限时免费；压缩后模型的部署费用按 MU 规格单独计费（因规格降低而节省成本） | +| **典型场景** | - 大促期间客服机器人流量峰值保障（TPM 预留）
- 编程助手要求 <200ms 首 token 延迟（快速模式）
- Agent 多步推理中对 token 流速敏感的链路 | - 金融领域定制化报告生成模型上线
- 电商客服知识库问答模型迭代与灰度发布
- 将开源模型微调后封装为内部 SaaS 服务 | - 已验证效果的微调模型需降低 30%+ 推理成本
- 边缘侧或轻量级容器环境部署受限于显存/内存
- 快速验证不同量化精度（W4A4/W8A8）对业务指标的影响 | + +## 适用场景建议 + +- **选择高并发推理（TPM 预留）当**： + 你的模型已在生产环境稳定运行，但面临周期性流量高峰（如每日晚 8 点用户咨询激增），且 SLA 要求“99.9% 请求在 1s 内完成”，无法容忍公共资源池的随机限流。此时，TPM 预留是保障容量确定性的最优解，尤其适用于成熟业务线的稳态扩容。 + +- **选择快速模式（Fast mode）当**： + 你的应用对交互实时性极度敏感（如 IDE 内嵌代码补全、语音转文字后的即时意图分析），且能接受 Preview 阶段的技术不确定性。注意：仅 `glm-5.2-fast-preview` 可用，客户端需改造解析逻辑，**严禁用于支付、风控等强 SLA 场景**。 + +- **选择生产部署（[model production](../api/model-production.md)）当**： + 你需要将自有业务数据训练出的专属模型（如医疗问诊微调模型）长期、可靠、可回滚地上线。它提供完整的生命周期管理（训练→部署→版本→监控），是构建企业级 AI 应用的基石能力，适合从 PoC 迈向规模化落地的团队。 + +- **选择压缩优化（[model compression](../guides/model-compression.md)）当**： + 你的微调模型已通过业务验证，但部署成本过高（如需 2×A10 GPU），或目标环境资源受限（如单卡 24GB 显存）。通过 PTQ 量化可显著降低 MU 规格（如从 `gn7i-c16g1.4xlarge` 降至 `gn7i-c8g1.2xlarge`），在精度损失可控前提下实现成本优化，**必须在部署前执行，且不可逆**。 + +## 技术选型参考（面向开发者） + +| 你的关键诉求 | 推荐方案 | 关键动作提醒 | +|--------------|----------|--------------| +| “我的模型流量忽高忽低，怕高峰期被限流崩掉” | ✅ TPM 预留 | 计算真实 kTPM 需求（考虑长文本阶梯系数），使用专属 model code，监控“超额降级统计”避免隐性成本 | +| “用户抱怨补全太慢，首 token 要 1.2 秒，体验差” | ⚠️ 快速模式（仅限 GLM-5.2） | 确认业务能接受 Preview 风险；切换域名；解析 `reasoning_content`；压测排队延迟容忍度 | +| “我要用自己标注的 5000 条合同数据训练一个法律问答模型并上线” | ✅ 生产部署 | 优先选用 LoRA 微调（成本低、速度快）；规划 `endpoint_name`；部署后立即做 A/B 测试验证效果 | +| “这个微调好的模型效果不错，但部署要两台 A10，太贵了，能压小点吗？” | ✅ 压缩优化 | 在华北2地域操作；选 W8A8 模板作为起点；用历史测试集校准；部署后对比 accuracy & latency | +| “我想把 HuggingFace 上下载的 Llama3-8B-GGUF 模型直接部署” | ❌ 三者均不支持 | 百炼当前不支持直接导入 GGUF/AWQ 等外部量化格式；需先转换为百炼兼容格式或通过 `import_model` 流程验证 | +| “我需要同时跑 10 个不同版本的客服模型做灰度” | ✅ 生产部署（+ 版本管理） | 提工单申请提升部署实例配额（默认 5 个）；利用 `version_id` 精确路由流量 | +| “模型压缩后还能不能继续微调？” | ❌ 不可以 | 压缩不可逆！务必保留原始 `SUCCEEDED` 微调模型，所有后续迭代均从此开始 | + +> **重要提醒**：三类能力并非互斥，而是可组合使用——例如：对生产部署的 `qwen3.5-flash-finetuned-xxx` 模型执行压缩得到 `qwen3.5-flash-finetuned-xxx-w8a8`，再为其预留 TPM 并启用快速模式（若该模型未来支持）。但请注意：**快速模式当前仅对 `glm-5.2-fast-preview` 开放，不支持其他模型（含压缩后模型）**。技术演进请持续关注官方文档更新。 + +## 被对比主题页 + +- [model high speed inference](../guides/model-high-speed-inference.md) +- [model production](../api/model-production.md) +- [model compression](../guides/model-compression.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md deleted file mode 100644 index 048238f7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md +++ /dev/null @@ -1,45 +0,0 @@ -# 模型部署与高速推理对比 - -在百炼平台上落地生产级推理服务时，开发者常面临两条技术路线：一是通过**模型部署**为预置或调优模型建立独立、资源专享的推理实例（PTU、模型单元、按 Token 三种计费）；二是通过**高速推理**能力（TPM 预留、快速模式 Fast mode）在标准调用之上叠加容量保障或输出提速。二者定位不同——前者关注「拥有一套专属推理服务」，后者关注「让现有模型调用更稳、更快」。本页梳理两者关键差异，帮助开发者按业务诉求做选型。 - -## 关键维度对比 - -| 对比维度 | 模型部署（PTU / MU / 按 Token LoRA） | 高速推理（TPM 预留 / 快速模式 Fast mode） | -| --- | --- | --- | -| 核心目标 | 建立独立、资源专享的推理服务，满足高并发、低延迟、私有模型落地 | 为已有模型调用锁定专属吞吐（TPM 预留）或提升输出速度（快速模式） | -| 支持模型 | 平台预置模型 + 调优/自训练模型（LoRA 导入，覆盖千问3/千问3-VL/千问2.5/千问2.5-VL） | TPM 预留：以控制台为准（千问3.x、GLM-5.x、DeepSeek-v4、Kimi-K2.6 等）；快速模式：有限模型如 `glm-5.2-fast-preview` | -| API 端点 / 接入 | 部署接口 `POST https://dashscope.aliyuncs.com/api/v1/deployments`，用 `plan` 区分计费；推理走 [DashScope SDK](../concepts/dashscope-sdk.md) 对专属服务调用 | TPM 预留：将 `model` 替换为专属模型 code，请求结构不变；快速模式：base_url 改为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，`model` 指定 fast 版 | -| 计费方式 | PTU：时长×TPM 单价（后付费按小时/预付费按天）；MU：时长×单元数（后付费按分钟/预付费按月）；LoRA：按 Token 使用量 | TPM 预留：按 kTPM 预付费（部署成功即计费）；快速模式：按 token 计费，逻辑与标准 API 一致 | -| 计费切换限制 | 计费方式创建后不可更改，须下线重新部署 | TPM 预留可扩缩容/续费/退订（退费按已用 1.5 倍系数结算） | -| 超额处理 | PTU 超吞吐或超输入上限自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`） | TPM 预留：自动降级公共池按量、不中断；快速模式：进入排队队列（不立即限流） | -| 输出速度 | PTU 相比按 Token TPS 提升约 1.5~2.0 倍；MU 支持 PD 分离降低首 Token 延迟 | 快速模式 TPS 提升至标准 API 的 1.5~2 倍（达 80~100 TPS） | -| 资源专享 | 是（资源独占，MU 性能可自定义） | TPM 预留：是（专属容量刚性兑付）；快速模式：否（按 token 共享，仅提速） | -| 代码改动 | 部署后用专属 `deployed_model` ID 调用 | TPM 预留/快速模式：替换 `model`（快速模式还需改 base_url） | -| 成熟度 | 正式能力（仅华北2·北京地域） | TPM 预留：正式；快速模式：preview 阶段，规格可能调整 | -| 典型场景 | 智能客服、实时内容审核、私有微调模型落地、长时任务 | TPM 预留：流量可预估、不能接受公共限流；快速模式：AI 编程助手、Agent 多步推理、实时对话 | - -## 适用场景建议 - -- **选模型部署（PTU）**：流量稳定、需要并发/延迟确定性的高负载生产环境；希望获得高于按量的 TPS，并利用长输入阶梯系数与前缀缓存折扣优化额度。建议部署前用控制台**容量计算器**估算所需 TPM。 -- **选模型部署（MU）**：需要部署私有微调（LoRA）模型、性能指标可自定义、或有长时任务，并希望用 PD 分离降低首 Token 延迟。 -- **选模型部署（按 Token / LoRA）**：仅用于验证 SFT 高效训练后的自定义模型效果，不使用不计费。 -- **选 TPM 预留**：不想新建独立部署实例，但流量可预估、无法接受被公共资源限流；需要「专属容量刚性兑付、超额自动降级不中断」的确定性保障。 -- **选快速模式（Fast mode）**：对输出速度（TPS）敏感、计费仍希望按 token 走标准逻辑的场景（编程助手、Agent、实时对话），可接受 preview 阶段的规格变动与独立接入域名。 - -## 技术选型参考 - -1. **先看诉求本质**：需要「一套专属服务 + 自定义/私有模型 + 计费确定性」→ 走**模型部署**；只想给现有模型「加容量保障或加速」→ 走**高速推理**。 -2. **计费柔性**：模型部署计费方式创建后不可改（须下线重建），选型前务必确认；TPM 预留支持不中断扩缩容，柔性更高。 -3. **超额行为要区分**：PTU 与 TPM 预留超额均转/降级为按量、服务不中断；而**快速模式超额是排队**，对延迟敏感业务需评估队列影响并做重试机制。 -4. **迁移成本**：模型部署与 TPM 预留仅需替换 `model`/`deployed_model` 参数；快速模式还需改 `base_url`（MaaS 域名），迁移时注意区分。 -5. **成熟度与地域**：模型部署当前限华北2·北京地域；快速模式处于 preview，能力可能随版本调整——正式生产建议优先选正式能力，并以百炼控制台展示的价格、容量起步值、支持模型为准。 -6. **可组合使用**：两条路线并非互斥。对于既要私有模型落地又要输出提速的场景，可分别评估 MU 部署与快速模式，按模型支持情况组合。 - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md deleted file mode 100644 index 488f6c5d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md +++ /dev/null @@ -1,51 +0,0 @@ -# 模型部署与高速推理对比 - -百炼平台为推理调用提供两类专属容量方案：**模型部署**（含 PTU、模型单元、按 Token 用量三种计费形态）与**高速推理**（TPM 预留）。两者都通过预付费锁定专属吞吐、避免公共限流，但在隔离强度、接入方式、适用模型与计费粒度上存在差异。本文并排列出关键维度，帮助开发者根据业务负载特征做出选型。 - -## 关键维度对比 - -| 维度 | 模型部署（PTU） | 模型部署（模型单元） | 模型部署（按 Token 用量） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | --- | -| 计费单位 | 按使用时长 + 输入/输出 TPM | 按使用时长 × 模型单元数量 | 按输入/输出 Token 数 | 按 kTPM（输入/输出分别计价） | -| 容量保障 | 专属部署实例，强隔离 | 专属部署实例，支持 PD 分离 | 不保障容量，仅验证效果 | 专属容量刚性兑付，与公共池隔离 | -| 超额处理 | 自动转按量计费，响应头 `x-dashscope-ptu-overflow:true` | — | 不适用 | 自动降级公共池按量计费，服务不中断 | -| 扩缩容 | 自助增减吞吐量 | 自助增减模型单元数量 | 控制台提交申请，人工审核 | 自助增减 kTPM | -| 接入方式 | `model` 替换为 `deployed_model`（专属服务 ID） | 同 PTU | 同 PTU | `model` 替换为专属模型 code | -| 创建入口 | 控制台或 `POST /api/v1/deployments`（华北2-北京） | 同 PTU | 同 PTU | 控制台「创建 TPM 预留」 | -| 支持模型 | 预置模型 + 导入的 LoRA 模型（千问3/2.5 系列） | 私有模型，性能指标自定义 | 部分 LoRA 调优模型 | 千问3.7-Max/Plus、千问3.6-Flash、GLM-5.2/5.1、DeepSeek-v4-Flash/Pro、Kimi-K2.6（分区域） | -| 长输入/缓存 | 支持阶梯系数与前缀缓存折扣 | 取决于模型 | 取决于模型 | 支持阶梯系数与前缀缓存折扣（容量计算器自动应用） | -| 退订规则 | 按天计费，无法提前退费 | 预付费首月内提前退订，日单价按 1.2 倍计费 | 一个月不使用自动释放 | 缩容/退订按 1.5 倍系数结算已用部分 | -| 到期保留 | 欠费保留 24 小时后停止计费、底层资源删除 | 同 PTU | — | 到期 2 小时内可续费，2~14 小时停止不可调用，14 小时后删除 | -| 地域 | API 部署仅华北2（北京） | 同 PTU | 同 PTU | 华北2（北京）、新加坡 | - -## 适用场景建议 - -- **高负载生产环境、需稳定吞吐与低延迟**：优先选择模型部署的 **PTU** 模式。它提供专属部署实例，支持长输入阶梯系数和前缀缓存折扣，适合对隔离性、性能、可观测性要求高的核心业务。 -- **私有模型部署、PD 分离、自定义性能指标**：选择模型部署的**模型单元**模式。通过 `deploy_spec`（如 `MU1`）与副本数精细控制算力，适合需要 Prefill/Decode 分离以降低首 Token 延迟的场景。 -- **调优后模型效果验证、不使用不计费**：选择模型部署的**按 Token 用量**模式。仅支持部分 LoRA 模型，扩缩容需控制台人工审核，适合实验性验证而非长期承载生产流量。 -- **流量可预估、不能接受限流、希望最小接入改动**：选择**TPM 预留**。预付费按 kTPM 锁定专属容量，超额自动降级公共池按量计费不中断，仅需将 `model` 替换为专属模型 code，适合中高流量但对极致隔离要求不高的在线服务。 -- **费用优化、无需专属容量**：考虑按量付费或资源包/节省计划，不在本对比页范围。 - -## 技术选型参考 - -1. **隔离强度**：PTU/模型单元为专属部署实例，隔离性最强；TPM 预留为专属容量兑付，与公共池共享底层但仍保障吞吐。对隔离性敏感（如合规、性能稳定性）选前者，仅对限流敏感选后者。 -2. **接入成本**：TPM 预留只需替换 `model` 参数为专属模型 code，无 API 端点变更；模型部署需通过 `POST /api/v1/deployments` 创建并等待 `status=RUNNING`，调用时使用 `deployed_model` 作为 `model`，并确保 API Key 与部署在同一业务空间。 -3. **模型范围**：模型部署支持导入自训练 LoRA 模型（千问3/2.5 系列，rank 须为 8/16/32/64，且 vocab、chat_template、VIT 须与基础模型一致）；TPM 预留仅支持平台预置模型，无法承载私有调优模型。 -4. **地域**：API 部署目前仅华北2（北京）；TPM 预留在华北2（北京）与新加坡均有开放，海外业务选 TPM 预留。 -5. **计费灵活性**：PTU 按天计费无法提前退费，模型单元首月内退订有 1.2 倍惩罚，TPM 预留缩容/退订按 1.5 倍系数结算。短期试算建议先用容量计算器评估 RPM、平均输入/输出长度、缓存命中率，避免额度过低频繁降级。 -6. **长输入与缓存**：两者均支持阶梯系数与前缀缓存折扣（如 glm-5.1 输入系数 1.33、缓存折扣 0.2；deepseek-v4-pro 缓存折扣 0.08）。PTU 通过响应头 `service_tier`、`provisioned_tokens`、`cached_tokens` 标识计费方式；TPM 预留在详情页「超额降级统计」中查看降级次数。 - -## 来源文档 - -- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) -- [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) -- [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) -- [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) -- [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md new file mode 100644 index 00000000..3cd824a8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md @@ -0,0 +1,57 @@ +# 模型评估与监控体系对比 + +为帮助开发者在模型研发、上线与运维全生命周期中科学选型，本文系统对比百炼平台三大核心质量保障能力：**模型评测（Model Evaluation）**、**模型监控（Model Monitoring）** 和 **应用评测（Application Evaluation）**。三者定位互补：模型评测聚焦「结果质量」的离线、结构化打分；模型监控侧重「运行状态」的实时、可观测追踪；应用评测则面向「端到端智能体/工作流」的业务逻辑与交互效果评估。本对比旨在厘清能力边界、明确适用阶段与技术约束，避免功能误用或能力缺失。 + +## 关键维度对比 + +| 维度 | 模型评测（Model Evaluation） | 模型监控（Model Monitoring） | 应用评测（Application Evaluation） | +|------|------------------------------|-------------------------------|-------------------------------------| +| **核心目标** | 量化模型推理输出质量（语义正确性、事实一致性、指令遵循度等） | 实时观测模型服务稳定性、性能、成本与异常行为 | 评估智能体/工作流级应用的端到端业务效果（如问答准确率、任务完成率、RAG链路健壮性） | +| **输入格式** | 必须提供三元组：`Prompt` + `Output`（模型生成）+ `Completion`（参考答案）；支持 `.xls`/`.xlsx`/`.jsonl` 格式评测数据集 | 无需显式输入——自动采集所有调用请求的原始 `input` 与 `output`（北京地域可审计），指标基于 API 调用日志聚合 | 支持多种结构化输入：
• 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）
• 新版：按「智能体」「工作流」「自定义」类型自动生成参数模板（支持 `query`/`response`/`history` 等字段映射） | +| **输出格式** | 结构化评分结果：
• 综合得分（各维度平均）
• 维度级通过率/分布图
• 逐样本明细（含 AI 评分、规则分数、人工标签） | 多粒度时序指标：
• 控制台卡片（小时级延迟）
• Prometheus 指标（分钟级，需高级监控）
• 审计日志（北京地域，含原始 I/O 与 Token 用量） | 分层报告输出：
• 总体正确率 / Pass率
• BadCase 归因（检索失败/切片错误/重排偏差等）
• 多评估器并行结果（LLM/Code/历史模型）
• 标签筛选后的细分统计 | +| **支持模型** | **仅文本生成类模型**（预置及调优后模型），不支持多模态、语音、结构化输出模型 | **全量支持**：覆盖控制台所有可选模型（含千问系列、开源快照、三方模型及全部调优模型） | **受限支持**：
• 自动评测：仅 `qwen-max` 和 `qwen-plus` 可作为裁判模型
• 应用关联：支持任意已发布智能体/工作流（不限底层模型） | +| **API 端点** | 无独立公开 API；通过控制台创建评测任务触发，结果通过 `/api/v1/evaluation/tasks/{id}` 查询（需权限） | 提供标准 Prometheus HTTP API：
`GET {endpoint}/api/v1/query_range`
支持 `model_call_count`、`model_usage` 等 20+ 指标查询 | 无标准化 REST API；评测任务通过控制台发起，结果数据可通过 `/api/v1/application-evaluation/tasks/{id}` 获取（内部接口，文档未开放） | +| **计费方式** | • 被评测模型推理费用（使用评测数据集时）
• 裁判模型费用（仅大模型评估维度）
• 规则/人工评估零额外费用 | • 无单独监控费用
• 所有监控数据采集免费
• **但调用本身产生常规模型费用**（Token 计费） | • LLM 评估器调用：按 Token 计费（`qwen-max`/`qwen-plus`）
• 评测集生成：按 Token 计费
• Code 评估器：执行脚本免费（不调用模型） | +| **典型场景** | • A/B 测试新 Prompt 效果
• 验证模型微调前后性能提升
• 选型决策：对比多个候选模型在特定任务上的综合得分
• 合规审计：人工复核高风险输出 | • 生产环境告警（失败率突增、P99 延时超标）
• 成本治理：识别高 Token 消耗应用/用户
• 性能优化：分析首 Token 延时瓶颈
• 安全审计：追踪内容安全拦截事件 | • RAG 应用迭代：定位检索/切片/重排环节短板
• 智能体上线前验收测试
• 多版本横向对比（同一评测集验证不同 workflow 配置）
• 客户反馈归因：将 BadCase 关联至具体评估器标签 | + +## 各方案适用场景建议 + +- **选择模型评测（Model Evaluation）当**： + ✅ 需要对**单次模型输出质量进行深度语义评判**（如“回答是否完整覆盖问题要点”“是否存在幻觉”）； + ✅ 有明确参考答案（Completion），且任务具备可定义的评分维度（如“事实准确性”“指令遵循度”）； + ✅ 处于模型开发/调优阶段，需高频、小批量验证效果； + ❌ 不适用于无参考答案的开放域任务，或需实时响应的线上服务保障。 + +- **选择模型监控（Model Monitoring）当**： + ✅ 需要**7×24 小时守护生产服务稳定性**（如设置“失败率 >5%”告警）； + ✅ 关注**性能基线变化**（如 TPM 下降 30%）、**成本异常**（某 API Key 单日 Token 消耗翻倍）； + ✅ 运维团队需对接 Grafana/Prometheus 构建统一可观测平台； + ❌ 不适用于评估“回答好不好”，仅能回答“调用快不快、成不成、花多少钱”。 + +- **选择应用评测（Application Evaluation）当**： + ✅ 评估对象是**封装了 Prompt、工具、知识库的智能体或工作流**，而非裸模型； + ✅ 需要**归因分析 BadCase 根源**（例如：“80% 错误源于知识切片过短”，而非“模型答错了”）； + ✅ 业务逻辑复杂，需混合 LLM 语义判断 + Code 规则校验（如“返回 JSON 必须含 `status:success` 字段”）； + ❌ 不适用于纯文本生成模型的基础能力 benchmark（此时应选模型评测）；旧版评测集已逐步淘汰，新建项目务必使用新版「智能体/工作流」评测集。 + +## 技术选型参考（面向开发者） + +| 你的需求 | 推荐方案 | 关键理由 | 注意事项 | +|----------|----------|----------|----------| +| **快速验证一个新 Prompt 在 GSM8K 上的效果** | 模型评测（基线评测） | 直接调用预置 C-Eval/GSM8K 数据集，5 分钟获取基准分，无需准备数据 | 仅北京地域可用；结果不可下载，仅作快速参考 | +| **上线后发现客服机器人响应变慢，需定位是模型还是网络问题** | 模型监控（高级监控 + 首Token延时指标） | 分钟级查看 `model_first_token_duration`，结合 `model_call_duration_p99` 对比，若首Token延时高而总延时低，说明网络或前置服务瓶颈 | 需开通高级监控，且模型部署在北京/新加坡/弗吉尼亚 | +| **知识库问答应用上线后客户投诉“答案不相关”，需找出是检索不准还是模型理解错** | 应用评测（新版 + 多评估器） | 创建「检索相关性」Code 评估器 + 「答案事实性」LLM 评估器，用同一评测集并行跑分，BadCase 自动归因到具体环节 | 必须使用新版评测集；需为知识库配置切片策略并发布应用 | +| **为合规要求，每月人工抽检 100 条金融咨询回答是否含违规表述** | 模型评测（人工评估维度） | 创建 Pass/Fail 标签，上传待检数据集，分配标注人员，结果自动统计通过率并留痕 | 人工评估无裁判模型费用；标注完成后才标记“评测完成” | +| **构建企业级 AI 运维看板，集成模型调用量、成本、错误率与业务指标** | 模型监控（Prometheus API + 自建 Grafana） | 所有指标通过标准 HTTP API 拉取，可与现有监控栈无缝融合，支持 `workspace_id`/`apikey_id` 等多维下钻 | 高级监控需手动开启；认证使用 AccessKey，注意密钥安全 | + +> **重要提醒**：三者非互斥关系，而是**协同闭环**—— +> 🔹 模型评测发现“事实性得分低” → 触发应用评测深入归因 → 定位到知识切片问题 → 优化切片策略 → 用模型监控确认线上 P99 延时未恶化 → 再用模型评测验证修复效果。 +> 开发者应根据所处阶段（开发/测试/上线/运维）和关注焦点（质量/性能/成本/归因），组合使用这三类能力，构建完整的 AI 质量保障体系。 + +## 被对比主题页 + +- [model evaluation introduction](../guides/model-evaluation-introduction.md) +- [model monitoring](../guides/model-monitoring.md) +- [application evaluation](../guides/application-evaluation.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md deleted file mode 100644 index a61b79ec..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md +++ /dev/null @@ -1,62 +0,0 @@ -# 模型评估与应用评估对比 - -百炼平台提供了两套定位不同的评测能力：**模型评估**面向文本生成类模型本身的能力打分与横向对比，**应用评估**面向智能体/工作流应用的端到端输出质量。二者虽然都用到"评测集 + 评估器 + 报告"的组合，但评估对象、可用接口、归因粒度差异明显。本文从技术选型角度梳理关键维度，帮助开发者在正确的层面选用正确的工具。 - -## 核心差异一览 - -| 对比维度 | 模型评估 | 应用评估 | -|---------|---------|---------| -| 评估对象 | 单个文本生成模型（基座/调优模型）的推理结果 | 已发布的智能体应用、工作流应用的端到端输出 | -| 评测方式 | 自定义评测 + 基线评测 | 自动评测 + 手动评测（并存新旧两套系统） | -| 打分手段 | 大模型评估、规则评估（ROUGE/BLEU/Cosine 等）、人工评估 | 评估器（LLM 评估器 / Code 评估器）+ 人工标签 | -| 数据基础 | 评测数据集（Prompt+Completion）或推理结果集 | 评测集（旧版：对话分析/知识问答；新版：智能体/工作流/自定义） | -| 参考答案 | 规则/相似度评估需要；大模型/人工评估不需要 | 自动评测可基于知识库自动生成评测集，无需人工准备 | -| 基线能力评测 | 支持公开标准集（C-Eval、MMLU、GSM8K、BBH，仅北京地域） | 不涉及，聚焦应用业务表现 | -| 归因分析 | 侧重维度得分与通过率，无 RAG 环节归因 | 提供 BadCase 到 RAG 环节的归因（理解/重排/检索/切片/知识缺失）+ 调优建议 | -| 横向对比 | 通过排行榜对多模型排名对比 | 多应用横向评测，最多对比 8 个应用或版本 | -| 操作接口 | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | 控制台操作，依赖应用观测能力 | -| 前置条件 | 准备评测集/结果集、创建评测维度 | 应用须已发布并配置知识库、开通应用观测、获取相应权限 | -| 计费方式 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费） | 评测产生的 Token 费用正常计费；评估器模型当前限时免费 | -| 典型场景 | 选型基座模型、验证微调/调优效果、模型能力回归 | 验证智能体回答质量、RAG 效果优化、应用版本迭代对比 | - -## 评分/评估器体系对比 - -| 能力 | 模型评估 | 应用评估 | -|------|---------|---------| -| 语义打分 | 大模型评估（数值型/分类型），推荐千问-Max 作裁判 | LLM 评估器，用于相关性、有害性、幻觉检测 | -| 确定性打分 | 规则评估：文本相似度、字符串匹配 | Code 评估器：Python 规则，格式校验、数值计算、精确匹配 | -| 人工打分 | 人工评估分类型（Pass/Fail） | 标签体系（分类/布尔/数字/文本）+ 快速标注 | -| 复用与固化 | 评测维度可创建为模板被多任务复用 | 评估器支持预置模板、自定义，或从历史评测任务标注结果抽象生成 | -| 组合建议 | 按标准答案有无与语义需求选单一评分方式 | 建议组合 3-5 个评估器（如 LLM 相关性 + Code 格式校验），单任务最多 10 个 | - -## 适用场景建议 - -- **选用模型评估：** - - 需要在多个基座模型之间做选型决策，或验证微调/调优后模型能力是否提升。 - - 有确定性标准答案（翻译、摘要、Function Calling），可用规则评估低成本快速打分。 - - 希望用公开标准集（C-Eval/MMLU/GSM8K/BBH）快速摸底模型基础能力（注意仅北京地域）。 - - 关注纯模型层面的语义质量，愿意用裁判模型打分。 - -- **选用应用评估：** - - 评估的是完整的智能体/工作流应用（含 RAG 检索、Prompt、知识库），而非单一模型。 - - 需要将 BadCase 归因定位到检索、重排、切片、知识缺失等具体环节并获得调优建议。 - - 需要在同一基准下横向对比多个应用或同一应用的不同版本（最多 8 个）。 - - 需要建立持续评测闭环：知识库更新、Prompt 调整、模型升级、检索策略变更后触发回归。 - -## 技术选型参考 - -1. **先看评估对象的层级**：只关心"模型答得好不好"用模型评估；关心"这个 Agent/工作流整体表现"用应用评估。 -2. **接口约束**：模型评估当前仅控制台操作，无公开 API/SDK，需编程化流水线时应提前评估（可考虑 PAI Judge Model API）；应用评估同样以控制台 + 应用观测为主。 -3. **成本控制**：两者都可用规则/Code 评估器规避裁判模型费用；模型评估可先用 50-100 条小规模验证并复用推理结果集，应用评估的评估器模型当前限时免费。 -4. **前置成本**：应用评估要求应用已发布、配置知识库并开通应用观测，接入成本更高；模型评估只需准备数据集与评测维度。 -5. **组合使用**：完整落地一个 RAG 智能体时，可先用模型评估锁定基座模型，再用应用评估在业务层做端到端质量把关与归因优化，形成"选模型 → 调应用"的两级评测链路。 -6. **注意评测噪声**：模型评估中 1-3% 的分差通常为噪声，LLM 评分器存在位置/自我偏好偏差，建议定期人工抽查校准；应用评估同样建议结合人工标签交叉验证。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [application evaluation](../guides/application-evaluation.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md deleted file mode 100644 index adf83f6b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型评测与模型监控对比 - -模型评测（Model Evaluation）与模型监控（Model Monitoring）是百炼平台面向模型全生命周期的两类能力，但解决的问题截然不同：**模型评测聚焦"上线前的质量决策"**——用打分和对比帮助开发者在众多模型或调优版本中选出最优；**模型监控聚焦"上线后的运行观测"**——追踪调用性能、Token 消耗和费用趋势，并在异常时主动告警。二者常被混淆，本文从技术维度做横向对比，为选型与阶段落地提供参考。 - -## 关键维度对比 - -| 对比维度 | 模型评测（Model Evaluation） | 模型监控（Model Monitoring） | -|---------|------------------------------|------------------------------| -| 核心目的 | 评估模型输出质量，做模型/调优版本选型 | 观测线上调用的性能、成本、稳定性 | -| 生命周期阶段 | 上线前 / 选型验证阶段 | 上线后 / 生产运行阶段 | -| 输入 | 评测数据集（Prompt+Completion）或推理结果集 | 真实线上调用流量（无需额外准备数据） | -| 输出 | 综合得分、通过率、逐条评分、排行榜 | 调用时长、失败率、RPM/TPM、Token/费用趋势、告警 | -| 支持模型 | 仅文本生成类模型 | 所有模型（含自定义调优模型） | -| 评估/度量方式 | 大模型评估、规则评估（ROUGE/BLEU/Cosine）、人工评估 | 安全 / 成本 / 性能 / 错误 四类监控指标 | -| 操作方式 | 仅控制台操作，无公开 API/SDK | 控制台面板 + 标准 Prometheus HTTP API | -| 地域限制 | 基线评测仅北京地域；自定义评测更广 | 高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡；日志仅北京部分模型 | -| 数据实时性 | 离线批量评测，任务级产出 | 普通监控小时级、高级监控分钟级；用量约 1 小时延迟 | -| 计费方式 | 被评测模型推理费 + 裁判模型评分费（规则/人工无裁判费） | 监控本身以用量统计为主，关注 Token/张/秒等计量与免费额度 | -| 主动能力 | 无告警，靠人工查看结果 | 支持告警规则（短信/邮件/电话/钉钉/企业微信/Webhook） | -| 外部集成 | 无（可参考 PAI Judge Model API 编程化评测） | 私有 Prometheus + Grafana / 自建应用（Basic Auth） | -| 典型场景 | 模型选型、调优效果验证、能力基线对比 | 生产成本管控、性能退化发现、故障排查与审计 | - -## 各方案适用场景建议 - -### 优先使用模型评测 - -- **模型选型**：在多个候选模型或不同调优版本之间做横向对比，用统一维度打分选出最优。 -- **调优效果验证**：微调 / RAG 改造前后对比，量化改动是否真正带来质量提升。 -- **能力基线摸底**：借助 C-Eval、MMLU、GSM8K、BBH 等公开数据集快速评测模型的通用基础能力。 -- **质量门禁**：上线前用问答质量、内容安全等维度设定通过阈值，作为发布准入标准。 - -选型内部再细分评分方式：确定性、有标准答案的场景（翻译、摘要、Function Calling）用规则评估以省去裁判费用；语义理解类（问答、安全）用大模型评估；创意写作、专业判断用人工评估。 - -### 优先使用模型监控 - -- **生产成本管控**：按业务空间维度统计 Token/张/秒消耗，结合"免费额度用完即停"避免超支。 -- **性能与稳定性观测**：跟踪调用时长、首 Token 延时、RPM/TPM、失败率与限流，及时发现性能退化。 -- **异常主动发现**：为关键模型配置告警规则，紧急问题走电话/短信、一般问题走邮件。 -- **故障排查与内容审计**：开通推理/审计日志，回溯每次调用的输入、输出与 Token 消耗。 -- **可视化与二次分析**：通过 Prometheus HTTP API 接入 Grafana 或自建看板做深度分析。 - -## 面向开发者的技术选型参考 - -1. **二者不是替代关系，而是阶段互补**：评测负责"选对模型"，监控负责"用好模型"。一个完整的落地闭环通常是——先用模型评测做选型与调优验证，再用模型监控守护线上运行。 -2. **按接入方式选**：需要编程化、纳入 CI 的质量流水线时，模型评测目前仅控制台可用（可考虑 PAI Judge Model API）；而监控天然提供 Prometheus API，更适合自动化集成。 -3. **注意地域约束**：基线评测、告警、高级监控、日志均有地域限制，跨地域部署时需提前确认所选地域是否支持相应能力。 -4. **成本策略**：评测阶段用 50-100 条小规模先验证、保存推理结果集复用、确定性场景优先规则评估；运行阶段用监控约束 `max_tokens`、按任务选轻量模型、大批量任务走批量推理。 -5. **决策严谨性**：评测中 1-3% 的分差通常属于噪声，且 LLM 评分器存在位置偏差与自我偏好偏差，建议定期人工抽查；监控数据存在分钟级到小时级延迟，做实时决策时需考虑延迟窗口。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md deleted file mode 100644 index 2a4e108f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型体验、生产与部署对比 - -在百炼平台上，从"选模型"到"训模型"再到"上线专属推理服务"是一条完整的链路，分别对应三个主题：**模型体验**（选型与全模态能力概览）、**模型生产**（微调训练 + 部署 API）、**模型部署**（专属资源推理服务的计费与落地）。三者面向不同阶段的开发者需求，容易混淆但各有侧重。本页对关键维度做横向对比，帮助开发者判断当前所处阶段应该查阅哪一部分文档、采用哪种能力。 - -## 定位对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 核心目标 | 按场景选型、了解模型能力 | 微调训练定制模型 + 发布为服务 | 提供独立、资源专享的推理服务 | -| 所处阶段 | 需求调研 / 选型 | 模型定制与上线（训练→部署） | 生产落地与容量规划 | -| 面向对象 | 应用开发者、方案设计者 | 需要专属模型能力的开发者 | 需要高并发/低延迟私有推理的团队 | -| 文档类别 | guides（指南） | api（接口参考） | guides（指南） | -| 是否需要训练数据 | 否 | 是（微调数据集） | 视情况（可导入 LoRA 或用平台模型） | - -## 能力与技术维度对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 主要内容 | 文本/视觉/视频/图片/3D/语音/音乐/向量全模态选型 | 模型调优（Fine-tuning）+ 模型部署 API | 三种计费方式、PTU 缓存、LoRA 导入、部署流程 | -| 核心接口 | 各品类模型调用 API（DashScope 等） | 调优 API + 部署 API（`POST /deployments`） | `POST/GET/DELETE /deployments`（同生产的部署接口） | -| 输入 | 场景需求（文本/图像/音频等多模态） | 基础模型 + 训练数据集 + 超参数 | 模型 ID / LoRA 产物 + 计费与资源配置 | -| 输出 | 推荐模型清单与参数指引 | 调优模型产物、在线推理服务 | 专属推理服务（`deployed_model` 唯一 ID） | -| 计费方式 | 按模型广场标注的计量（Token / 张 / 时长等） | 训练按量 + 部署按所选 plan | PTU / 模型单元（MU）/ 按 Token（LoRA），创建后不可改 | -| 地域限制 | 因模型而异（如音乐仅华北2·北京） | 依接口而定 | 仅华北2·北京地域 | -| 典型场景 | 聊天/内容生成/OCR/视频生成/TTS/ASR 等 | 定制专属模型、微调后上线 | 智能客服、实时审核等稳定高负载生产 | - -## 部署计费方式细分（属"模型部署"主题） - -| 计费方式 | plan 值 | 资源特性 | 计费粒度 | 适用场景 | -|----------|---------|----------|----------|----------| -| 预置吞吐 PTU | `ptu` | 预留资源保障 TPM，超额自动转按量 | 时长 × TPM 单价（后付费按小时 / 预付费按天） | 流量稳定、需延迟确定性的高负载 | -| 模型单元 MU | `mu` | 资源独占，支持 PD 分离 | 时长 × 模型单元数（后付费按分钟 / 预付费按月） | 私有微调模型、长时任务 | -| 按 Token（LoRA） | `lora` | 仅对 SFT/LoRA 自定义模型开放，不用不计费 | 按 Token 使用量 | 调优效果验证 | - -## 适用场景建议 - -- **还在挑模型 / 评估能力** → 查阅**模型体验**。先按模态（文本、视觉、视频、图片、3D、语音、音乐、向量）定位推荐模型，再到模型广场核对参数与计费，无需自己训练即可直接调用共享 API。 -- **需要专属能力、要微调训练** → 走**模型生产**流程：准备数据集 → 调优 API 提交微调 → 部署 API 发布为在线服务 → 调用端点推理。这是"训练 + 上线"的编排视角，关注接口定义与工作流顺序。 -- **要把模型跑成生产级私有服务** → 深入**模型部署**：根据流量特征选 PTU / MU / 按 Token，利用容量计算器规划 TPM，关注前缀缓存折扣、长输入阶梯系数、LoRA 导入要求（rank 8/16/32/64、必需文件、VL 需冻结 VIT）等落地细节。 - -## 技术选型参考 - -1. **共享推理 vs 专属推理**：仅需通用能力时直接用模型体验推荐的公共模型（按量共享）；对并发、延迟、数据隔离有确定性要求时，才走部署形成专属服务。 -2. **生产与部署的关系**：模型生产是"训练→部署"的**全流程视角**（含调优 API），模型部署是其中"部署"环节的**深度展开**（计费、缓存、导入、排障），两者共用 `POST /deployments` 接口。需要微调则从生产入手，只需部署平台模型可直接看部署文档。 -3. **计费不可逆**：部署计费方式一经创建无法更改，切换须先下线再重新部署；预付费提前退订首月按 1.2 倍计费——上线前务必确认。 -4. **自训练模型限制**：仅支持 LoRA 导入（不支持全参微调），且不能改动 vocab / chat_template，VL 模型须冻结 VIT，OSS Bucket 需打 `bailian-datahub-access` 标签。 -5. **计费自动兜底**：PTU 超出额度或输入超模型上限会自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`），无需改代码，但需通过 `service_tier`、`provisioned_tokens`、`cached_tokens` 字段监控成本。 - -## 被对比主题页 - -- [model experience](../guides/model-experience.md) -- [model production](../api/model-production.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md deleted file mode 100644 index d76323b6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型评测 vs 模型监控 vs 模型生产对比 - -百炼平台围绕模型全生命周期提供三大核心能力：**模型评测**用于上线前的质量验证，**模型监控**用于上线后的运行保障，**模型生产**用于模型的定制与部署。三者分别对应"评估选型 — 生产部署 — 运维保障"三个阶段，开发者需根据当前所处阶段选择合适的工具。本文从功能定位、输入输出、适用模型、费用模型等维度进行系统对比，帮助开发者快速定位所需能力。 - -## 关键维度对比 - -| 维度 | 模型评测 | 模型监控 | 模型生产 | -| --- | --- | --- | --- | -| **功能定位** | 模型质量量化验证 | 模型运行状态观测与告警 | 模型调优、压缩与部署 | -| **生命周期阶段** | 上线前 / 迭代验证 | 上线后持续运维 | 模型定制与交付 | -| **输入** | 评测数据集（Excel）或预置榜单 | 自动采集调用数据（无需手动输入） | 训练数据集 + 基础模型 | -| **输出** | 评测报告（综合得分、通过率、分数分布） | 监控看板、调用日志、告警通知 | 定制模型 + 在线推理服务端点 | -| **交互方式** | 控制台创建评测任务 | 控制台看板 + Prometheus API + Grafana | REST API（异步任务模式） | -| **支持模型范围** | 千问系列、开源模型、部分三方模型；基线评测仅支持调优后模型 | 所有模型（普通监控）；高级监控限北京/新加坡/弗吉尼亚 | 平台支持的基础模型（微调）+ 自定义导入模型（部署） | -| **[计费](../concepts/billing.md)方式** | 被评测模型推理费 + 裁判模型评分费 | 免费（高级监控依赖云监控 Prometheus 实例） | 训练算力费 + 部署推理资源费 | -| **数据延迟** | 任务式，分钟至小时级完成 | 普通监控约 1 小时；高级监控分钟级 | 异步任务，训练耗时视数据量而定 | -| **地域限制** | 无特殊限制 | 日志仅华北2（北京）；告警仅北京/新加坡 | 以控制台和 API 返回为准 | -| **典型操作** | 创建评测维度 → 上传数据 → 运行任务 → 查看报告 | 查看看板 → 配置告警 → 排查日志 | 创建调优任务 → 压缩 → 部署 | - -## 适用场景建议 - -### 模型评测适用场景 - -- **模型选型**：在多个候选模型之间进行横向对比，选择最适合业务的模型。 -- **调优验证**：微调后需要量化验证是否提升了目标能力。 -- **版本回归**：模型升级后确认核心能力未退化。 -- **能力基线**：使用预置榜单（C-Eval、MMLU、GSM8K 等）建立通用能力基准线。 - -### 模型监控适用场景 - -- **线上稳定性保障**：实时观测调用时长、首包延迟、失败率等性能指标。 -- **成本管控**：按[业务空间](../concepts/workspace.md)追踪 [Token](../concepts/token.md) 消耗，发现异常用量。 -- **故障排查**：通过历史对话日志定位推理异常的具体请求。 -- **主动告警**：[Token](../concepts/token.md) 突增、超时率升高时自动通知运维人员。 - -### 模型生产适用场景 - -- **领域定制**：通用模型无法满足垂直场景时，通过微调注入专业知识。 -- **成本优化**：对定制模型进行量化压缩，降低推理资源占用。 -- **服务化交付**：将训练好的模型部署为在线推理端点，供业务应用调用。 - -## 技术选型参考 - -| 开发者需求 | 推荐能力 | 说明 | -| --- | --- | --- | -| 需要知道哪个模型效果好 | 模型评测 | 创建自定义评测任务或基线评测进行量化对比 | -| 需要定制一个专属模型 | 模型生产 | 调优 → 可选压缩 → 部署为推理服务 | -| 模型已上线，需要保障稳定运行 | 模型监控 | 开启高级监控 + 告警规则 | -| 调优后想验证效果 | 模型评测 + 模型生产 | 先用模型生产完成调优，再用模型评测量化对比 | -| 线上模型出现异常，需要定位原因 | 模型监控 | 查看调用日志和失败详情排查 | -| 想在 Grafana 中统一看板 | 模型监控 | 接入高级监控的 Prometheus HTTP API | - -## 三者协作关系 - -在实际项目中，三大能力通常按以下流程串联使用： - -1. **模型生产**：基于基础模型微调出领域定制模型，压缩后部署上线。 -2. **模型评测**：部署前用评测任务验证定制模型的质量是否达标。 -3. **模型监控**：部署后持续观测运行状态，设置告警确保服务可靠性。 - -当监控发现质量劣化时，可回到模型评测进行定量验证，确认后再通过模型生产重新调优迭代，形成闭环。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) -- [model production](../api/model-production.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md deleted file mode 100644 index 1b246309..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md +++ /dev/null @@ -1,54 +0,0 @@ -# 模型监控 vs 应用监控 - -百炼平台提供两种互补的观测能力：**模型监控**关注单个模型 API 调用层面的性能与成本，**应用观测**则从应用维度端到端追踪内部调用链路。理解二者的定位差异，有助于开发者在不同阶段选择合适的监控手段，实现从模型选型到应用上线的全链路可观测。 - -## 关键维度对比 - -| 维度 | 模型监控 | 应用监控（应用观测） | -|------|----------|----------------------| -| 监控对象 | 单个模型的 API 调用 | 整个应用（智能体/工作流/高代码应用）的完整调用链路 | -| 核心关注点 | 模型性能、[Token](../concepts/token.md) 消耗、费用趋势 | 应用内部节点间的执行流程、端到端延时 | -| 支持范围 | 所有模型（含自定义调优模型） | 智能体应用、工作流应用、高代码应用 | -| 地域限制 | 高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | 无特殊地域限制（需开通 OpenTelemetry 服务） | -| 典型指标 | 调用时长、首 [Token](../concepts/token.md) 延时、RPM/TPM、失败率、[Token](../concepts/token.md) 消耗 | 调用次数、失败率、Token 总量、平均首 Token 耗时、平均调用时长 | -| 数据粒度 | 普通监控小时级；高级监控分钟级 | 分钟级更新 | -| 数据保留 | 用量统计 30 天 | 调用记录最长 30 天 | -| 告警能力 | 支持（短信/邮件/电话/钉钉/企微/Webhook） | 不支持（需结合模型监控告警） | -| 外部集成 | Prometheus HTTP API + Grafana（Basic Auth） | OpenTelemetry 服务存储 | -| 日志能力 | 推理日志（输入/输出/Token 逐条记录） | Trace/Span 链路日志（含节点层级关系） | -| 筛选维度 | API-KEY、推理类型、时间范围、时间精度 | Request ID/Trace ID/Span ID、状态、Span Name、延时、Token 量、标签 | -| 数据标注 | 不支持 | 支持对 Span 添加标签（布尔/分类/数字/文本） | -| 评测集联动 | 不支持 | 支持将 Span 数据导入评测集 | -| 费用 | 免费（高级监控需开启） | 功能免费，存储费用由 OpenTelemetry 服务收取 | -| API 支持 | Prometheus HTTP API 可编程查询 | 仅控制台操作，无独立 API | - -## 适用场景 - -### 模型监控适合 - -- **模型选型与调优阶段**：对比不同模型的调用时长、首 Token 延时、成本效益比，为技术选型提供数据依据。 -- **成本管控**：按[业务空间](../concepts/workspace.md)维度统计 Token 消耗，配置免费额度用尽即停策略，避免超支。 -- **稳定性保障**：通过告警规则监控失败率、限流错误，在异常发生时第一时间通知相关人员。 -- **运维自动化**：利用 Prometheus API 接入自建监控系统或 Grafana 大盘，实现统一运维。 - -### 应用监控适合 - -- **应用调试与优化**：通过 Trace 链路追踪定位应用内部的性能瓶颈节点（如检索耗时、模型推理耗时）。 -- **智能体/工作流行为分析**：查看 AGENT、RETRIEVER、LLM、TOOL 等节点的执行顺序和中间结果，理解应用的实际行为。 -- **数据驱动迭代**：将线上真实调用数据标注后导入评测集，形成"观测-标注-评测-优化"闭环。 -- **问题排查**：按 Request ID 精确定位单次调用的完整执行链路，快速定位报错节点。 - -## 技术选型建议 - -1. **两者并非互斥**：生产环境建议同时开启。模型监控提供宏观的性能告警与成本管控，应用观测提供微观的链路诊断能力。 -2. **开发调试阶段**优先使用应用观测，快速定位逻辑问题和性能瓶颈。 -3. **上线运营阶段**优先配置模型监控告警，保障服务稳定性和成本可控。 -4. **高代码应用**目前应用观测仅支持入口节点，内部链路需依赖代码中集成 Tracing 模块并启用 `--telemetry enable` 参数。 -5. 如需将监控数据接入第三方平台，模型监控通过 Prometheus API 对接，应用观测数据则存储在 OpenTelemetry 服务中。 - -## 被对比主题页 - -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md deleted file mode 100644 index 31515cc0..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md +++ /dev/null @@ -1,42 +0,0 @@ -# 模型微调、压缩与高速推理对比 - -在百炼平台上，把一个模型从"能用"打磨到"好用、省钱、够快"，通常会经历三类相互独立又可串联的能力：**模型微调（Fine-tuning）** 把领域知识、任务能力与人类偏好写入参数；**模型压缩** 通过量化在保持能力的前提下降低部署规格与成本；**高速推理** 则在调用侧解决容量兑付与输出速度问题。三者定位不同——微调改变"模型本身"，压缩改变"部署精度/成本"，高速推理改变"调用时的容量与吞吐"，因此常见的完整链路为：模型调优 → 模型压缩（可选）→ 模型部署 → 高速推理（可选）。本文面向开发者，从关键维度对比三者，帮助做技术选型。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | 高速推理（TPM 预留 / 快速模式） | -| --- | --- | --- | --- | -| 解决的问题 | 深度定制：注入领域知识、指令遵循、偏好对齐、特定音色/风格 | 降低部署所需 MU 规格，减少推理成本 | 容量刚性兑付（预留）/ 提升单请求输出速度（快速模式） | -| 处理对象 | 基础模型 + 训练数据集 | 平台微调产出的全精度自定义模型 | 线上可调用的模型（含专属 code / fast 模型） | -| 输入格式 | 数据集：SFT 用 ChatML `{"messages":[...]}`；DPO 追加 `chosen`/`rejected`；CPT 用 `{"text":"..."}`；视觉 ZIP+`data.jsonl`；CosyVoice `{"wav_fn","text"}` | 源模型 + 量化模板（+ 可选校准数据集，最多 5 个） | 标准对话请求（`messages`）；预留替换 `model` 为专属 code；快速模式指定 fast 模型 ID | -| 输出产物 | 微调后模型（`finetuned_output` / `deployed_model`） | 低精度量化模型（新后缀，更小部署规格） | 无新模型产物，仅提升调用容量/速度 | -| 支持模型 | 千问文本/VL、万相图像/视频、CosyVoice 等多模态 | 仅平台微调产出的自定义模型（如 qwen3.5-flash 系列），不支持基础/第三方模型 | 预留：已开放预留的模型；快速模式：`glm-5.2-fast-preview` | -| 训练/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA） | 量化（不含剪枝、蒸馏），MU 编号越大规格越小 | 预留：锁定 kTPM 吞吐；快速模式：提速至 1.5~2 倍 TPS | -| API 端点 / 接入 | `POST /api/v1/files`、`/fine-tunes`、`/deployments`（华北2 API Key） | 主要通过控制台创建压缩任务（模型训练 → 模型压缩） | 预留：标准 `dashscope.aliyuncs.com` + 专属 code；快速模式：`{workspace_id}.cn-beijing.maas.aliyuncs.com` 专属域名 | -| 计费方式 | 按 Token（API 创建仅支持按 Token）或训练单元（仅控制台）；成本与耗时较高 | 压缩任务限时免费；成本体现在部署阶段按 MU 规格计费（示例节省约 56%） | 预留：按 kTPM 预付费，超额降级按量；快速模式：按 token（同标准 API），超额排队 | -| 地域限制 | 仅华北2（北京） | 仅华北2（北京） | 预留：以控制台为准；快速模式：华北2（北京）、新加坡 | -| 可逆性 / 约束 | 可继续叠加训练（CPT→SFT→DPO） | 不可逆；压缩后不支持继续微调或二次压缩 | 预留退订/快速模式为 preview，规格可能变动 | -| 代码改动量 | 大：准备数据集、走训练→部署全流程 | 小：控制台配置任务，调用无需改代码 | 小：预留替换 `model`；快速模式换域名 + fast 模型 ID | -| 典型场景 | 领域问答、安全合规对齐、专属音色/图像风格 | 微调模型上线前的成本优化 | 高峰期专属容量保障；对输出速度敏感的编程助手/Agent/实时对话 | - -## 各方案适用场景建议 - -- **模型微调**：当 Prompt 工程、插件调用、RAG 等手段仍无法达到效果，需要把领域知识或人类偏好"写进参数"时选用。文本生成推荐按 `CPT（可选）→ SFT → DPO（可选）` 递进组合；对成本/时间敏感或数据集较小时用 LoRA 高效训练，追求效果且模型支持时优先全参。多模态（VL、万相、CosyVoice）按各自支持的方式与超参集处理。注意训练成本与耗时较高，且能力仅限华北2（北京）。 -- **模型压缩**：已有微调产出的全精度自定义模型、且部署成本偏高时选用。因压缩任务当前限时免费，建议在免费期内对同一模型尝试多个量化模板，分别部署后用业务测试集验证精度与成本的平衡，再选最优方案上线。切记压缩不可逆，需保留上游全精度模型以便重新压缩。 -- **高速推理 - TPM 预留**：面向高峰期需要"专属容量、不被公共限流"的生产业务，按 kTPM 预付费锁定吞吐，超额自动降级按量、不中断，适合对稳定性和容量兑付要求高的场景。 -- **高速推理 - 快速模式**：面向 AI 编程助手、Agent 多步推理、实时对话等对**输出速度**敏感的场景，TPS 可达标准 API 的 1.5~2 倍。当前仍为 preview，规格可能调整，生产接入前需评估稳定性；接入需使用专属域名，不能与预留的标准域名接入方式混用。 - -## 技术选型参考 - -1. **先定位需求属于哪一层**：要"改变模型能力/风格"→微调；要"降部署成本"→压缩；要"保容量或提速度"→高速推理。三者可组合：微调 → 压缩 → 部署 → 叠加 TPM 预留或快速模式。 -2. **成本与投入权衡**：微调投入最大（数据、训练时间、Token/训练单元费用）；压缩几乎零调用改造且当前限时免费，收益在部署阶段；高速推理中预留是预付费换刚性容量，快速模式按量计费换速度。 -3. **可逆性与前置条件**：压缩不可逆且要求上游为平台微调模型；微调与压缩均仅限华北2（北京）；快速模式为 preview 且模型/地域受限。 -4. **接入方式差异**：预留沿用标准域名 + 专属 model code；快速模式改用 workspace 专属域名 + fast 模型 ID，两者不可混用，选型时需评估现有调用代码的改造范围。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md deleted file mode 100644 index 47044f8e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md +++ /dev/null @@ -1,85 +0,0 @@ -# 模型微调、压缩与部署对比 - -在百炼平台上，将一个基础模型转化为可在生产环境中使用的定制化推理服务，通常涉及三个阶段：**模型微调（Fine-tuning）**、**模型压缩（Quantization）** 和 **模型部署（Deployment）**。三者在模型生产链路中依次衔接（调优 → 压缩（可选）→ 部署），各自解决不同层面的问题。本文从功能定位、适用范围、操作方式和成本等维度进行系统对比，帮助开发者在技术选型时做出合理决策。 - -## 功能定位与链路关系 - -模型微调、压缩与部署构成一条完整的模型生产流水线： - -- **模型微调**处于链路上游，目标是基于自有数据定制模型能力，使模型在特定领域或任务上表现更优。 -- **模型压缩**处于链路中间，是可选环节，通过量化技术降低模型参数精度，从而减小部署所需的算力规格和推理成本。 -- **模型部署**处于链路下游，将训练好（或压缩后）的模型发布为在线推理服务，供应用调用。 - -三者缺一不可地覆盖了"训练 → 优化 → 上线"的完整生命周期，但各自的输入输出、关注点和约束条件差异显著。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（Quantization） | 模型部署（Deployment） | -|------|------------------------|------------------------|----------------------| -| **核心目标** | 基于自有数据定制模型能力，提升特定场景表现 | 降低模型参数精度，减小部署规格与推理成本 | 将模型发布为在线推理服务，供应用调用 | -| **在链路中的位置** | 上游（第一步） | 中间（可选） | 下游（最后一步） | -| **输入** | 基础模型 + 训练数据集（JSONL / ZIP 等） | 微调产出的自定义模型 | 预置模型 / 微调模型 / 压缩后模型 | -| **输出** | 微调后的自定义模型 | 低精度量化模型 | 可调用的在线推理服务（API 端点） | -| **支持的模型范围** | 广泛：Qwen3.6/3.5/3/2.5 系列文本模型、Qwen-VL 视觉模型、Wan 图像/视频模型、CosyVoice 语音模型 | 较窄：仅支持百炼平台微调产出的特定自定义模型（如 qwen3.5-flash） | 最广泛：预置模型（Qwen、DeepSeek、GLM 等）+ 微调模型 + 压缩模型 + OSS 导入的 LoRA 模型 | -| **支持的模态** | 文本生成、视觉理解、图像生成、视频生成、语音合成 | 仅文本生成 | 文本生成、[多模态](../concepts/multimodal.md)、语音合成等 | -| **操作方式** | 控制台 + API | 仅控制台 | 控制台 + API | -| **是否可逆** | 可重复训练和迭代 | 不可逆：压缩后不支持继续微调或二次压缩 | 可随时上线/下线，支持重新部署 | -| **典型耗时** | 数小时至数天（取决于数据量和训练轮次） | 平台自动完成，排队后运行 | 分钟级（状态变为"运行中"即可调用） | -| **计费方式** | 按 Token 用量计费：训练数据 Token 数 x 训练轮次 x 训练单价 | 压缩任务本身限时免费 | 三种模式：预置吞吐（PTU）、模型单元（MU，按时长）、Token 用量（按调用量） | -| **地域限制** | 无特殊限制 | 仅华北2（北京） | 无特殊限制 | - -## 训练方法与压缩模板选择 - -### 微调方法 - -百炼平台提供三种递进式微调方法，推荐按 CPT → SFT → DPO 的顺序使用： - -| 方法 | 目标 | 数据要求 | 典型场景 | -|------|------|----------|----------| -| CPT（继续预训练） | 注入领域知识 | 1000万+ Token 无标签文本 | 金融/医疗/法律等垂直领域适配 | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条高质量问答对 | 客服、代码助手、Agent 工具调用 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组正负样本对 | 安全合规强化、降低幻觉 | - -每种方法支持全参训练和高效训练（LoRA）两种模式。全参训练效果更好但耗时更长、成本更高；LoRA 训练收敛快、成本低，适合快速验证。 - -### 压缩量化模板 - -量化模板决定压缩后的部署规格，MU 编号越大表示部署规格越小、成本越低，但精度损失可能越大。可选配校准数据以提升量化精度，建议选择与推理场景语义相近的数据集。 - -## 部署计费方式对比 - -| 计费方式 | 计费公式 | 适用场景 | 特点 | -|---------|---------|---------|------| -| 预置吞吐（PTU） | 使用时长 x (输入 TPM 单价 x 输入 TPM + 输出 TPM 单价 x 输出 TPM) | 流量稳定的高负载生产环境 | 保障吞吐额度内不限速，超额自动切换按量计费；TPS 约为按量的 1.5~2.0 倍 | -| 模型单元（MU） | 使用时长（小时）x 模型单元数量 x 模型单元单价 | 需要资源独占和自定义性能指标 | 支持 PD 分离计算模式，可降低首 Token 延迟 | -| Token 用量 | 输入 Token 数 x 输入单价 + 输出 Token 数 x 输出单价 | 调用量不稳定、用量较少的场景 | 仅支持部分 LoRA 调优模型，不使用不计费 | - -## 适用场景建议 - -**场景一：快速验证模型定制效果** -推荐路径：SFT 高效训练（LoRA）→ 直接部署（Token 用量计费）。跳过压缩环节，以最低成本快速上线验证。 - -**场景二：生产环境成本敏感** -推荐路径：SFT 全参训练 → 模型压缩 → 部署（MU 计费）。通过压缩降低部署规格（如从 MU1x2 降至 MU8x1，成本节省约 56%），适合长期运行的在线服务。 - -**场景三：高并发低延迟的核心业务** -推荐路径：微调（按需）→ 部署（PTU 计费）。PTU 模式提供预留吞吐保障，TPS 提升约 1.5~2.0 倍，适合流量稳定且对延迟敏感的场景。 - -**场景四：[多模态](../concepts/multimodal.md)模型定制** -推荐路径：视觉/图像/视频/语音 SFT → 直接部署（MU 计费）。当前压缩功能仅支持文本模型，[多模态](../concepts/multimodal.md)微调模型需直接部署。 - -## 技术选型要点 - -1. **是否需要微调**：如果预置模型已满足需求，可直接部署，无需微调。当模型在特定领域表现不佳、需要定制输出格式或降低幻觉时，再考虑微调。 -2. **是否需要压缩**：压缩可显著降低部署成本，但会带来一定精度损失且不可逆。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案。 -3. **如何选择部署计费方式**：流量稳定选 PTU，需要资源独占和灵活配置选 MU，用量少且不稳定选 Token 用量。注意 Token 用量模式仅支持部分 LoRA 模型。 -4. **地域约束**：模型压缩当前仅在华北2（北京）地域可用，规划链路时需考虑地域一致性。 -5. **不可逆操作提醒**：压缩后的模型不支持继续微调或二次压缩，务必保留上游全精度微调模型以备后续迭代。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md deleted file mode 100644 index bb5c6450..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型优化方式对比（微调/压缩/高速推理） - -百炼平台围绕"让模型更贴合业务、跑得更省、调得更稳"提供了三类模型优化能力：**模型微调（Fine Tuning）**、**模型压缩（Model Compression）**、**高速推理（TPM 预留）**。三者作用阶段不同、目标不同，但在生产落地中常被串联使用：先用微调注入业务能力，再用压缩降低部署规格，最后用高速推理保障线上吞吐。本页从输入输出、支持模型、API/控制台入口、[计费](../concepts/billing.md)方式、典型场景等维度横向对比，供开发者在技术选型时参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | -| 一句话定位 | 把业务/场景知识写入模型参数 | 把全精度微调模型量化为低精度版本 | 为指定模型锁定专属吞吐量 | -| 优化阶段 | 训练阶段（改变模型权重） | 部署前阶段（不改权重，只做量化） | 服务运行阶段（不动模型，只保容量） | -| 是否改变模型 | 是，产出新的微调模型 | 是，产出量化后的低精度模型 | 否，仅锁定原模型的推理容量 | -| 前置条件 | 已开通百炼模型服务，准备好训练数据 | 当前工作空间内已有支持压缩的自定义微调模型 | 已开通百炼模型服务，已创建业务空间 | -| 输入格式 | 文本：JSONL 问-答对；图像/视频：`.zip`（含 `data.jsonl` + 训练样本）；语音：录音文件 | 全精度微调模型 + 校准数据集（部分模板要求） | 业务流量预估（RPM、平均输入/输出长度、缓存命中率） | -| 输出格式 | 新的自定义微调模型（`finetuned_output`） | 量化后的低精度模型（部署到更小规格单元） | 专属 model code（替换 API 的 `model` 参数即可） | -| 支持模型 | 千问系列（文本/视觉）、万相 wan2.7/wan2.2/wan2.5（图像/视频）、CosyVoice 语音 | 当前支持 `qwen3.5-flash-2026-02-23` 等基础模型对应的自定义微调模型（以控制台为准） | 按区域开放，如 qwen3.7-max、glm-5.2、kimi-k2.6、deepseek-v4-pro 等（以控制台为准） | -| 控制台入口 | 「模型调优」页面 | 「模型 > 模型训练 > 模型压缩」页面 | 控制台「创建 TPM 预留」 | -| API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | OpenAPI（`custom_calibration_file_ids` 等字段） | 替换 `model` 参数为专属 model code，调用 OpenAI 兼容模式 `/compatible-mode/v1/chat/completions` | -| [计费](../concepts/billing.md)方式 | 按 Token [计费](../concepts/billing.md)（API 创建的任务）；控制台创建可使用模型训练单元（预付费/后付费） | 当前限时免费；产出模型部署上线后按部署单元规格计费 | 按 kTPM 预付费，输入/输出 TPM 分别计价，按天计费；预留内调用不额外收费，超额自动降级按量 | -| 任务状态 | `PENDING` → `SUCCEEDED`；部署 `RUNNING` | 待开始/排队中/运行中/停止中/压缩成功/压缩失败/已取消 | 运行中/待生效/变配中/已停止/已过期/已取消 | -| 是否可逆 | 可迭代训练（CPT→SFT→DPO 递进） | 不可逆：不支持继续训练，不支持二次压缩 | 可扩缩容、续费、退订；退订不可恢复 | -| 典型场景 | 注入领域知识、对齐偏好、复刻风格、压低延迟、抑制幻觉 | 降低显存占用、部署到更小规格、降低部署成本、提升吞吐 | 流量可预估且不能接受限流、业务高峰期保障容量 | - -## 各方案适用场景建议 - -### 模型微调（Fine Tuning） - -适合"通用模型答不好、Prompt 工程已到瓶颈"的场景。当你需要把领域术语/事实、特定对话格式、人类偏好、特定风格写入模型本身时，应选微调： - -- **补知识**：CPT 注入领域文本（1000 万+ Token）。 -- **学做事**：SFT 教会指令遵循（1000+ 问-答对），文本/视觉/图像/视频/语音[多模态](../concepts/multimodal.md)均支持。 -- **做更好**：DPO 对齐人类偏好（100+ 组偏好对）。 -- **快速验证或数据量小**：优先用 LoRA 高效训练，速度快、成本低。 -- CosyVoice 调优仅支持 API 发起，且仅支持 SFT 高效微调。 - -### 模型压缩（Model Compression） - -适合"已经微调出好模型，但部署太贵/规格太大"的场景。它不改变模型能力定位，只做量化降本： - -- 前提是上游已产出全精度微调模型（压缩是微调的下游）。 -- 通过量化模板决定压缩后可部署规格（如 MU5、MU8），把 `MU1 * 2（¥108/小时）` 降为 `MU8 * 1（¥47/小时）`。 -- 适合追求更低部署成本、更高推理吞吐、更小部署单元的线上服务。 -- 注意：压缩产物不可继续训练、不可二次压缩，迭代须回到全精度模型重训。 - -### 高速推理（TPM 预留） - -适合"流量可预估、不能接受公共限流、要保障高峰期容量"的场景。它不动模型本身，只锁定推理吞吐： - -- 业务高峰期不能因公共池限流而抖动。 -- 需要 SLA 刚性兑付、专属容量不与他人共享。 -- 超额自动降级按量计费且不中断服务，适合可预估但偶发突刺的流量。 -- 接入极简：仅替换 `model` 参数为专属 model code，无需改代码逻辑。 - -## 技术选型建议 - -三者并非互斥，建议按"能力注入 → 规格压缩 → 容量保障"的顺序串联使用： - -1. **先微调**：用业务数据训练出符合场景的自定义模型（解决"答得好不好"）。 -2. **再压缩**：对微调产物做量化，部署到更小规格单元（解决"跑得省不省"）。 -3. **最后保容量**：用 TPM 预留锁定线上吞吐，保障高峰期稳定（解决"调得稳不稳"）。 - -如果只是临时验证或流量很小，可跳过压缩与预留，直接按量调用微调后的模型；如果是对外承诺 SLA 的生产服务，建议三步全走。若仅需保障容量而模型本身已满足需求，可单独使用 TPM 预留；若仅需降本而流量平稳，可单独使用模型压缩。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md deleted file mode 100644 index 7cd40cf3..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md +++ /dev/null @@ -1,51 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控能力：**应用监控（应用观测）**聚焦于应用层端到端调用链路，**模型监控**聚焦于模型层用量与运行状态。两者观测对象、指标口径、数据时效、地域支持各不相同，开发者需根据排查目标（应用行为 vs 模型行为）选择合适的能力，或组合使用以实现从业务入口到模型底层的全链路可观测。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 业务空间内的应用（智能体应用、工作流应用、高代码应用） | 主账号下所有业务空间内的模型调用（按"模型 + 业务空间"维度） | -| 观测粒度 | 应用内部调用链路（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等节点，支持嵌套） | 模型维度的聚合指标（调用次数、失败率、延时、Token、RPM/TPM 等） | -| 关键指标 | 延时、Token 量、调用记录（Prompt/输出）、CHAIN 节点状态 | 安全/成本/性能/错误四类：失败率、429 限流、首 Token 延时、Token 消耗等 | -| 数据时效 | 分钟级同步，调用记录最长可查 30 天 | 普通监控小时级、高级监控分钟级；用量数据延迟约 1 小时 | -| 模型范围 | 随应用调用自动覆盖应用内使用的模型 | 普通监控支持所有地域所有模型；高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | -| 日志/历史对话 | 在 Trace 详情中查看 Prompt 内容、输出、Span 原始数据 | 仅华北2（北京）地域部分模型支持请求和响应日志（需开通推理日志） | -| 数据筛选 | Span 模式（Root/All/Model Span）+ 状态/Span Name/延时/Token/标签等多条件 | 按 API-KEY、推理类型、时间范围、时间精度筛选；失败详情可点击查看 | -| 数据导出 | 支持 JSONL / EXCEL 导出 | 高级监控指标存储于私有 Prometheus，支持标准 Prometheus HTTP API，可接入 Grafana | -| 主动告警 | 不支持告警，仅控制台查看 | 支持告警规则（短信/邮件/电话/钉钉/企微/Webhook），按 CRITICAL/ERROR/WARNING/INFO 分级 | -| 计费方式 | 应用观测功能本身不收费；数据存储费由 OpenTelemetry 服务收取 | 监控功能本身不收费；用量按 Token/张/秒计费，存储于云监控 Prometheus | -| 开通方式 | 控制台"应用观测配置"：授权 OTel 角色 → 开通 OTel 服务 → 初始化 LogStore | 控制台"模型监控配置"：开通审计/推理日志、高级监控、告警规则 | -| API 支持 | 无 API，仅控制台操作 | 高级监控提供 Prometheus HTTP API，可接入自建应用 | -| 标注与[评测](../concepts/evaluation.md) | 支持对 Span 加标签（布尔/分类/数字/文本），可加入[评测](../concepts/evaluation.md)集 | 不直接支持标注，但 Token 追踪与告警可辅助[评测](../concepts/evaluation.md)样本筛选 | - -## 适用场景建议 - -### 选用应用监控（应用观测） - -- 需要追踪智能体应用、工作流应用内部端到端调用链路（如检索 → 改写 → LLM → 工具调用）。 -- 排查应用层延时异常、Token 消耗突增的根因节点（定位到具体 Span）。 -- 希望将真实线上调用作为评测样本导入评测集。 -- 应用部署在任意地域，且不依赖主动告警（仅需控制台查询）。 -- 限制：不支持 Assistant API 创建的智能体应用；高代码应用仅能观测入口 CHAIN 节点。 - -### 选用模型监控 - -- 需要按业务空间/模型维度汇总调用量、失败率、延时、Token 消耗，做成本核算与稳定性保障。 -- 需要 429 限流、内容安全错误等模型层错误的明细与趋势分析。 -- 需要主动告警（超时、Token 突增等静默失败），或接入 Grafana/自建看板。 -- 需要 Token 消耗的汇总、追踪、阈值告警三层成本管理。 -- 需要查看请求和响应原文（仅北京地域部分模型，需开通推理日志）。 -- 限制：用量统计不支持按阿里云账号维度汇总；日志/告警有地域限制。 - -### 组合使用 - -当问题既涉及应用行为又涉及模型底层时，建议先用应用监控定位异常 Span（如某个 LLM 节点延时高或失败），再通过 Request ID / Trace ID 关联到模型监控中的对应调用记录，结合模型层失败率、限流、Token 消耗做联合诊断。应用监控提供"是什么调用出了问题"，模型监控提供"模型侧为什么会出问题"，二者互补构成完整的可观测体系。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md deleted file mode 100644 index d9d02b20..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md +++ /dev/null @@ -1,60 +0,0 @@ -# 应用监控与模型监控对比 - -阿里云百炼平台提供了两套互补的可观测能力：**应用监控（应用观测）**面向由智能体、工作流、高代码搭建的完整应用，追踪端到端的调用链路；**模型监控**面向底层模型调用本身，聚焦性能指标、Token 消耗与费用趋势。二者关注的抽象层次不同——前者回答"这个应用内部发生了什么"，后者回答"这个模型跑得怎么样、花了多少钱"。本文对比两者的关键差异，帮助开发者在做可观测性方案选型时快速定位合适的工具。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 智能体应用、工作流应用、高代码应用的端到端调用链路 | 模型调用本身（含调优后的自定义模型） | -| 核心价值 | 追踪应用内部调用链路、节点耗时与思考过程 | 追踪模型性能、Token 消耗与费用趋势 | -| 支持范围 | 三类应用；不支持 Assistant API 创建的智能体；高代码仅观测入口 CHAIN 节点 | 用量统计支持全部模型；高级监控仅限北京/新加坡/弗吉尼亚；告警仅限北京/新加坡 | -| 数据延迟 | 分钟级 | 普通监控小时级、高级监控分钟级；用量统计约 1 小时 | -| 数据留存 | 调用记录最长 30 天 | 用量统计不支持查看 30 天以前数据 | -| 指标维度 | 调用次数/失败率、Token 量、平均首 Token 耗时、平均调用时长 | 安全、成本、性能（RPM/TPM 等）、错误四大类 | -| 筛选维度 | Request ID / Trace ID / Span ID、状态、Span、输入输出、延时、Token、标签 | API-KEY、推理类型（实时/批量）、时间范围、时间精度 | -| 链路/节点粒度 | 提供 CHAIN/AGENT/RETRIEVER/LLM/TOOL 等丰富节点类型与嵌套关系 | 无调用链路，按模型/API-KEY 维度聚合指标 | -| 告警能力 | 无内置告警 | 支持告警规则（短信/邮件/电话/钉钉/企业微信/Webhook），分 CRITICAL/ERROR/WARNING/INFO 等级 | -| 数据标注/评测 | 支持 Span 打标签、直接加入评测集 | 不涉及 | -| 日志能力 | 可查看 Prompt、输出、延时、Token 等调用记录 | 推理日志/审计日志，仅华北2（北京）部分模型，分钟级延迟 | -| 外部系统接入 | 数据导出为 JSONL / EXCEL | 高级监控数据存于私有 Prometheus，支持标准 HTTP API，可接 Grafana（Basic Auth） | -| 使用方式 | 仅控制台操作，无 API | 控制台面板 + Prometheus HTTP API | -| 计费方式 | 功能本身免费，观测数据存储由 OpenTelemetry 服务收费 | 监控功能查看无额外说明；聚焦模型调用本身的 Token/费用管理 | -| 前置开通 | 需授权 OpenTelemetry 角色、开通服务、初始化 LogStore | 高级监控/告警需在模型监控配置中开启；日志需开通审计与推理日志 | -| 典型场景 | 调试应用逻辑、定位慢节点、优化检索/Prompt、构建评测样本 | 成本核算、性能容量规划、异常主动告警、故障排查 | - -## 适用场景建议 - -### 优先选择应用监控（应用观测） - -- 应用由智能体、工作流或高代码搭建，需要**看清内部调用链路**（哪个节点慢、检索命中如何、模型思考过程）。 -- 需要按 Trace ID / Span ID **下钻单次请求**，排查复杂多节点应用的逻辑问题。 -- 希望把**真实线上调用直接沉淀为评测样本**，或对 Span 数据做标注管理。 -- 只在控制台侧使用、可接受分钟级更新、调用记录 30 天内可回溯。 - -### 优先选择模型监控 - -- 关注的是**模型层面的性能与成本**：RPM/TPM、失败率、限流、Token 消耗与费用趋势。 -- 需要**主动告警**（限流、失败率飙升、成本超标），并通过短信/电话/钉钉等渠道通知。 -- 需要**按业务空间精细化管理模型成本**，结合免费额度"用完即停"控制预算。 -- 希望把监控数据**接入 Grafana 或自建系统**做统一可视化（依赖高级监控的 Prometheus 接口）。 -- 需要通过推理日志做**内容审计或故障排查**（注意仅限北京地域部分模型）。 - -## 技术选型参考 - -两套能力并非互斥，成熟的生产系统通常**同时启用**： - -1. **分层定位问题**：先用模型监控发现"某模型失败率/延时异常"，再用应用监控下钻到具体应用与节点，确认是检索、插件还是模型环节导致。 -2. **注意地域限制**：模型高级监控与告警对地域敏感（高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，推理日志限北京）。若业务不在这些地域，应用监控的分钟级链路观测可作为主要抓手。 -3. **权衡数据延迟**：应用监控与模型高级监控均为分钟级；模型普通监控与用量统计为小时级，不适合实时排障，更适合趋势分析与成本复盘。 -4. **成本可观测归口**：需要账单/额度维度治理时以模型监控（用量统计 + 免费额度）为准；需要单次请求成本归因时用应用监控的 Token 量指标。 -5. **数据出口差异**：需要把数据导入外部 BI/表格用应用监控的 JSONL/EXCEL 导出；需要接入监控告警平台（Grafana/Prometheus）走模型高级监控的 HTTP API。 - -简言之：**排查"应用怎么跑的"用应用监控，管理"模型跑得好不好、花多少钱"用模型监控**，两者组合可覆盖从链路调试到成本告警的完整可观测闭环。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md deleted file mode 100644 index 3e3d1202..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控体系：**应用观测**和**模型监控**。应用观测侧重于端到端追踪应用内部的调用链路与节点级性能，帮助开发者定位应用层面的延时与逻辑问题；模型监控则聚焦于模型维度的调用性能、Token 消耗、费用趋势与告警，帮助开发者管控模型使用成本与稳定性。理解两者的定位差异，有助于开发者建立完整的可观测性方案。 - -## 核心维度对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（含自定义模型和调优模型） | -| **观测粒度** | 节点级（CHAIN、LLM、RETRIEVER、TOOL 等多种节点类型） | 模型级（按模型名称、API-KEY、推理类型筛选） | -| **核心指标** | 调用延时、Token 用量（输入/输出）、首 Token 耗时、调用次数与失败率 | 调用时长、首 Token 延时、RPM、TPM、Token 消耗、失败率、限流错误、内容安全错误 | -| **数据更新频率** | 分钟级 | 普通监控为小时级；高级监控为分钟级 | -| **数据保留时长** | 最长 30 天 | 用量统计不支持查看 30 天以前的数据 | -| **告警能力** | 无 | 支持告警规则配置，通知方式含短信、邮件、电话、钉钉机器人、企业微信机器人、Webhook | -| **用量统计** | 无独立用量统计，通过监控统计图表查看 Token 总量 | 按[业务空间](../concepts/workspace.md)维度统计，支持免费额度管理与用完即停 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 高级监控数据可通过 Prometheus HTTP API 接入 Grafana 等外部系统 | -| **数据标注** | 支持对 Span 添加标签（布尔值/分类/数字/文本），可关联评测集 | 不支持 | -| **日志查看** | 通过 Trace 详情查看 Prompt 输入输出与原始数据 | 开通推理日志后可查看每次调用的输入、输出及 Token 消耗（仅华北2北京地域部分模型） | -| **地域限制** | 无特殊地域限制 | 高级监控仅支持北京、新加坡、弗吉尼亚；告警仅支持北京、新加坡 | -| **计费** | 功能免费，观测数据存储由 OpenTelemetry 服务收费 | 功能免费，高级监控数据存储在私有 Prometheus 实例中 | -| **操作方式** | 仅控制台操作，无 API | 控制台操作 + Prometheus API 接入 | - -## 开通与前置条件对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **开通步骤** | 授权 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore | 普通监控默认可用；高级监控需在模型监控配置中手动开启 | -| **子账号权限** | 需要 AliyunBailianFullAccess + 应用观测页面权限 + ram:CreateServiceLinkedRole 策略 | 标准百炼控制台权限即可 | -| **推荐操作账号** | 主账号 | 无特殊要求 | - -## 适用场景建议 - -### 应用观测适合以下场景 - -- **调用链路排查**：应用响应慢或出错时，需要逐节点定位瓶颈，例如区分是检索环节还是模型推理环节导致延时过高。 -- **Prompt 调试**：查看每次调用的完整输入输出，对比不同 Prompt 的效果。 -- **数据质量管理**：通过 Span 筛选与标注功能，对线上真实调用数据进行质量打分，并将优质样本导入评测集。 -- **工作流应用调试**：工作流包含多种节点类型（意图分类、脚本转换、条件判断等），需要观察每个节点的执行情况。 - -### 模型监控适合以下场景 - -- **成本管控**：按[业务空间](../concepts/workspace.md)统计模型用量与费用，配合免费额度管理控制预算。 -- **稳定性保障**：配置告警规则，在失败率上升或限流异常时及时收到通知。 -- **性能基线建立**：通过 RPM、TPM、首 Token 延时等指标建立性能基线，持续跟踪模型表现。 -- **多模型对比**：对比不同模型在相同业务场景下的调用时长、Token 消耗等指标，辅助模型选型。 -- **外部可视化集成**：将监控数据接入 Grafana 等系统，构建统一的运维大盘。 - -### 建议组合使用 - -在生产环境中，推荐同时启用两套监控：用模型监控建立全局的成本与稳定性视图并配置告警，用应用观测在出现异常时深入排查具体调用链路。两者从不同维度覆盖可观测性需求，互为补充而非替代。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md deleted file mode 100644 index a7e7444d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md +++ /dev/null @@ -1,76 +0,0 @@ -# [多模态](../concepts/multimodal.md)生成 API 对比（图像/视频/3D） - -百炼平台提供图像、视频、3D 三类[多模态](../concepts/multimodal.md)生成 API，分别面向不同的内容产出形态。三者都通过 DashScope HTTP 接口调用，统一使用 API Key 鉴权，并遵循"创建任务 → 轮询结果"的异步任务模式（部分图像模型支持同步调用）。本页从输入格式、输出格式、支持模型、API 端点、调用模式、[计费](../concepts/billing.md)与典型场景等维度做横向对比，帮助开发者根据产出目标与技术约束做选型。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 产出形态 | 静态图片（PNG） | 视频文件 | GLB 模型 + 预览渲染图 | -| 输入格式 | 文本、图像（图生图/编辑）、参考图 | 文本、图像（首帧/首尾帧）、参考图、视频、音频 | 文本、单图、多图（前/左/后/右 4 视角，固定数组长度 4） | -| 输出格式 | PNG，1–6 张或多图组图 | 视频 URL | PBR 材质 GLB（`pbr_model_url`）或无贴图基础模型（`base_model_url`），含 1 张预览渲染图 | -| 调用模式 | 同步（千问/万相2.6+/Z-Image 等新版）或异步（V1 及部分编辑/创意类） | 仅异步 | 仅异步 | -| API 端点 | 同步：`POST /api/v1/services/aigc/multimodal-generation/generation`；异步轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/video-synthesis`（部分走 `image2video/video-synthesis`）；轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/3d-generation`；轮询：`GET /api/v1/tasks/{task_id}` | -| 必需请求头 | `Authorization`；异步需 `X-DashScope-Async: enable` | `Content-Type`、`Authorization`、`X-DashScope-Async: enable` | `X-DashScope-Async: enable`（缺少报 `current user api does not support synchronous calls`） | -| 典型耗时 | 同步秒级返回；异步 1–2 分钟 | 1–5 分钟，万相2.1 视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 24 小时 | 24 小时 | 24 小时，超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 随接口返回 | 随接口返回 | 2 小时，需及时下载 | -| 支持模型系列 | 千问图像、万相（Wan/wanx）、Z-Image、可灵 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1` 高精度、`Tripo/Tripo-P1.0` 专业快速） | -| 地域可用性 | 北京/新加坡/弗吉尼亚等多地域，地域独立鉴权不可混用；千问-图像翻译仅北京 | 同地域约束，模型/Endpoint/API Key 必须同地域；PixVerse、Vidu 仅北京 | 仅华北2（北京） | -| 业务空间专属域名 | 支持（`{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` 等） | 支持（北京 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 等） | 走默认 dashscope 域名 | -| SDK 支持 | 部分模型支持 DashScope SDK（Python/Java） | HTTP 为主 | HTTP | -| [计费](../concepts/billing.md)方式 | 按张数/模型[计费](../concepts/billing.md) | 按任务/时长计费 | 按任务计费（`usage` 记录任务类型与生成数量） | -| 典型场景 | 文生图、图生图、图像编辑、虚拟模特、试衣、海报、背景生成、擦除补全、画面扩展、人物写真 | 文生视频、图生视频、参考生视频、视频编辑、视频换人、数字人、肖像动态视频 | 文生 3D、单图生 3D、多图生 3D，游戏/电商/工业设计资产 | - -## 调用模式差异 - -三类 API 在调用流程上高度一致，均采用"创建任务 → 轮询查询"模式，但图像 API 额外提供**同步调用**能力： - -- **图像生成**：千问图像系列（qwen-image-2.0-pro/max/plus）、万相 2.6/2.7 文生图与编辑、Z-Image 等新版模型支持一次请求即返回结果的同步调用，走 `multimodal-generation/generation` 端点；V1 版及部分编辑/创意类模型仍需异步。同步模式流程更简单，适合交互式场景。 -- **视频生成 / 3D 生成**：因耗时较长（视频 1–10 分钟，3D 资产更久），统一仅支持异步。请求必须携带 `X-DashScope-Async: enable`，缺少该头会报错 `current user api does not support synchronous calls`。 - -三者都强调"请勿重复创建任务"，`task_id` 有效期 24 小时，直接轮询即可。 - -## 输入能力对比 - -| 输入方式 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 纯文本 | 支持，复杂文字渲染能力强（千问系列） | 支持（文生视频） | 支持，中英文等多语言，最大 1024 字符 | -| 单图输入 | 支持（图生图、图像编辑） | 支持（首帧生视频） | 支持，JPEG/PNG，宽高 [20,6000]，≤20MB | -| 多图输入 | 部分编辑模型支持多图输入/输出 | 支持（参考生、首尾帧） | 支持，固定 4 视角（前/左/后/右），有效 2–4 张 | -| 视频输入 | 不适用 | 支持（视频编辑、参考生视频） | 不适用 | -| 音频输入 | 不适用 | 万相2.7 支持[多模态](../concepts/multimodal.md)输入含音频 | 不适用 | - -3D 生成的多图输入有严格的视角顺序约束（前/左/后/右），不需要的视角传空对象 `{}`，这与图像/视频的"多图作为参考"语义不同。 - -## 产物与质量参数 - -| 项 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 输出规格 | 总像素 512×512~2048×2048，宽高比 1:4~4:1，1–6 张；万相2.7 支持 4K | 视频文件 URL | 面数：H3.1 最高 200 万面，P1.0 最高 2 万面 | -| 质量参数 | 分辨率、张数、宽高比 | 分辨率、时长、镜头叙事（`shot_type: multi`） | `texture_quality`（标清/高清）、`geometry_quality`（standard/ultra）、`pbr`、`texture` | -| 一致性能力 | 千问编辑支持角色一致性 | 万相2.7 参考生支持角色形象与音色一致性 | 多图视角约束保证几何一致性 | -| 预览能力 | 直接返回图片 | 直接返回视频 | 额外返回 `rendered_image_url` 预览渲染图 | - -## 适用场景建议 - -- **选图像生成 API**：需要静态视觉产出，强调文字渲染、风格化、精确编辑（增删移动物体、改动作）、虚拟模特/试衣/海报等电商与营销场景。优先用同步调用模型（千问图像、万相2.6+/2.7、Z-Image）以简化流程；批量或创意类任务再用异步。 -- **选视频生成 API**：需要动态叙事、数字人、肖像动态视频、视频编辑/换人。文生视频、图生视频（首帧/首尾帧）、参考生视频均可，万相2.7 是推荐的新版协议，支持多模态输入与角色/音色一致性。注意 PixVerse、Vidu 仅北京地域可用且需单独开通。 -- **选 3D 生成 API**：需要可直接导入引擎/3D 软件的 GLB 资产，适用于游戏、电商商品 3D 展示、工业设计。仅北京地域可用，需开通 Tripo。高精度选 `Tripo/Tripo-H3.1`（最高 200 万面），追求速度选 `Tripo/Tripo-P1.0`。 - -## 技术选型参考 - -1. **产出形态决定大类**：图片→图像 API；视频→视频 API；3D 模型→3D API。三者端点不同，不可混用。 -2. **延迟敏感优先同步**：仅图像 API 提供同步调用，适合交互式产品；视频与 3D 必须异步，需在业务侧实现轮询或配置异步任务回调（3D 查询接口默认 RPS 20）。 -3. **地域与鉴权**：三类均要求模型、Endpoint、API Key 同地域。3D 仅北京可用；千问-图像翻译、PixVerse、Vidu 也仅北京。建议迁移到业务空间专属域名以获得更好性能与稳定性。 -4. **任务复用**：`task_id` 24 小时有效，三类都要求轮询而非重复创建任务；3D 产物下载链接仅 2 小时，需及时落盘。 -5. **输入约束**：3D 多图必须按前/左/后/右 4 视角顺序；图像图文混排需开启 `enable_interleave=true` 并配合 SSE 流式；视频首尾帧、参考生有专属模型变体。 -6. **模型开通**：可灵、PixVerse、Vidu、Tripo 均需先在控制台搜索并开通授权，再调用 API。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md deleted file mode 100644 index 9ac91dd6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 图像生成、视频生成与3D生成对比 - -百炼平台提供图像生成、视频生成和3D生成三大[多模态](../concepts/multimodal.md)内容创作能力。三者在输入输出格式、模型生态、调用方式和适用场景上各有侧重。本文从开发者技术选型角度，对这三类生成能力进行系统对比，帮助快速定位最适合业务需求的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|---------|---------|--------| -| **输入格式** | 文本提示词、参考图像（单张/多张）、涂鸦草图 | 文本提示词、首帧/首尾帧图像、参考图像/视频/音频 | 文本提示词、单张图像、多图（4视角：前/左/后/右） | -| **输出格式** | PNG 图像（512x512 至 4K） | MP4 视频 | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用方式** | 同步调用为主，部分模型支持异步 | 全部异步（创建任务 → 轮询结果） | 全部异步（创建任务 → 轮询结果） | -| **主要模型系列** | 千问-图像、万相(Wan/Wanx)、Z-Image、可灵(Kling) | 万相(Wan)、HappyHorse、爱诗(PixVerse)、Vidu、可灵(Kling) | Tripo（H3.1 / P1.0） | -| **模型数量** | 20+ 款模型覆盖各类场景 | 6 大模型家族，任务类型丰富 | 2 款模型（高精度 / 专业快速） | -| **可用地域** | 部分全地域，部分仅华北2（北京） | 多数仅华北2（北京），HappyHorse 支持海外地域 | 仅华北2（北京） | -| **批量输出** | 单次可生成 1-9 张图像 | 单次生成 1 条视频 | 单次生成 1 个3D模型 | -| **产物有效期** | 即时返回，URL 有时效 | task_id 有效期 24 小时 | task_id 有效期 24 小时，下载链接有效期 2 小时 | -| **典型生成耗时** | 秒级至十秒级 | 分钟级 | 分钟级（耗时较长） | - -## 能力覆盖对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|:-------:|:-------:|:-----:| -| 文本生成 | 支持 | 支持 | 支持 | -| 图像/图片参考生成 | 支持 | 支持（首帧/首尾帧） | 支持（单图/多图） | -| 内容编辑 | 支持（局部重绘、风格迁移、扩图等） | 支持（指令编辑、视频迁移） | 不支持 | -| [多模态](../concepts/multimodal.md)混合输入 | 支持（文+图） | 支持（文+图+视频+音频） | 不支持 | -| 中文文字渲染 | 支持（千问、Z-Image 等） | 不适用 | 不适用 | -| 人像/人物专项 | 支持（人像风格重绘、AI试衣） | 支持（数字人、舞动人像、悦动人像等） | 不适用 | -| PBR 材质输出 | 不适用 | 不适用 | 支持 | - -## 计费方式差异 - -- **图像生成**：按张计费，不同模型单价不同（如扩图 0.18 元/张）。部分创意工具仅提供免费体验额度，用完不可付费续用。 -- **视频生成**：按任务计费，费用与视频时长、分辨率、模型版本相关。 -- **3D生成**：按任务计费，费用与贴图质量（standard/detailed）和几何精度（standard/ultra）相关。 - -## 适用场景建议 - -**选择图像生成的场景：** -- 电商商品图、营销海报、社交媒体配图等静态视觉内容 -- 需要精细文字渲染或图文混排的场景（如带中文的宣传图） -- 图像编辑与风格迁移（如局部重绘、背景替换、AI试衣） -- 对生成速度要求高、需要批量出图的场景 - -**选择视频生成的场景：** -- 短视频创作、广告片制作、动态内容营销 -- 人像动画（数字人播报、舞蹈视频、唱演视频） -- 需要多镜头叙事或多角色互动的复杂视频 -- 视频风格转换和口型替换等后期编辑 - -**选择3D生成的场景：** -- 游戏资产、AR/VR 场景中的3D模型快速原型 -- 电商3D商品展示 -- 需要 PBR 材质的高精度3D资产生产（最高 200 万面） -- 从多视角图片重建3D物体 - -## 技术选型要点 - -1. **生成速度**：图像生成最快（秒级），视频和3D生成均需分钟级等待，且必须使用[异步调用](../concepts/async-invocation.md)模式。 -2. **模型生态丰富度**：图像生成和视频生成均拥有多个模型家族可选，3D生成目前仅有 Tripo 系列。 -3. **地域限制**：3D生成仅限北京地域；视频和图像生成的部分模型也有地域限制，选型前需确认目标地域的模型可用性。 -4. **输出后处理**：视频和3D的产物下载链接有时效限制（3D仅 2 小时），需在业务流程中及时下载存储。 -5. **开通流程**：3D生成和部分视频/图像模型需在百炼控制台额外搜索并开通服务，不是默认可用。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md deleted file mode 100644 index 45397164..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md +++ /dev/null @@ -1,74 +0,0 @@ -# Qwen API、应用调用与托管智能体 API 对比 - -百炼平台提供了多种 API 接入方式，开发者在集成大模型能力时常面临选型困惑：是直接调用 Qwen 模型 API，还是通过应用调用 API 使用已编排好的智能体/工作流，亦或是采用 Managed Agents API 获得平台全托管的智能体运行时？本文从接口定位、协议兼容性、会话管理、工具能力、计费模式等维度进行系统对比，帮助开发者根据实际场景做出技术选型。 - -## 定位差异 - -- **Qwen API**：直接调用 Qwen 系列大语言模型，获取文本生成能力。开发者自行管理 [prompt](../guides/prompt.md)、上下文和工具调用逻辑，灵活度最高。 -- **应用调用 API**：调用在百炼控制台中已创建并发布的智能体或工作流应用。应用内部已封装模型选择、知识库检索、插件调用等编排逻辑，开发者只需传入用户输入即可获取最终结果。 -- **Managed Agents API**：平台全托管的智能体运行时，提供 Agent、Session、Environment、Skill、File 等资源抽象。由平台负责会话状态机、沙箱执行、工具调用与事件流推送，适合需要长期运行、多步工具调用的复杂场景。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -|------|----------|-------------|-------------------| -| **定位** | 模型级调用，直接访问 Qwen 系列模型 | 应用级调用，调用已编排好的智能体/工作流 | 平台托管智能体运行时，全生命周期管理 | -| **兼容协议** | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | OpenAI Responses（兼容模式）、DashScope | 百炼原生 REST API | -| **API 端点** | `POST /compatible-mode/v1/chat/completions` 等 | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` 或 `POST /api/v1/apps/{APP_ID}/completion` | `POST /api/v1/agentstudio/sessions/{session_id}/events` 等 | -| **认证方式** | [API Key](../concepts/api-key.md)（`Authorization: Bearer`） | [API Key](../concepts/api-key.md) + APP ID（+ 可选 Workspace ID） | [API Key](../concepts/api-key.md) + Workspace ID | -| **模型选择** | 请求体 `model` 字段指定任意 Qwen 模型 | 控制台配置，调用时无需指定模型 | Agent 创建时配置模型 | -| **会话管理** | 调用方自行维护（Responses 接口除外） | DashScope API 通过 `session_id` 自动维护；Responses API 需传完整历史 | 平台全托管，Session 状态机自动驱动 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其他接口需自定义 | 控制台可视化编排插件、知识库、工具 | Skill（zip 包上传）+ Environment 沙箱执行 | -| **[多模态](../concepts/multimodal.md)支持** | 需选用 VL 系列模型 | Responses API 支持图像和文件输入 | 通过 File 资源挂载到 Session | -| **[流式输出](../concepts/streaming.md)** | 支持（`stream=true`） | 支持（`stream=true`） | SSE 事件流（`GET .../events/stream`） | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | Responses API 支持（`background=true`），DashScope 暂不支持 | 原生异步，Session 状态机驱动 | -| **支持地域** | 多地域 | 仅华北2（北京） | 仅 cn-beijing | -| **SDK 支持** | OpenAI SDK、Anthropic SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | 百炼原生 SDK / HTTP | -| **配置方式** | 纯代码，请求参数控制 | 控制台可视化编排 + API 调用 | API 全程管理（Agent/Environment/Session/Skill） | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接、细粒度地控制模型推理参数（temperature、top_p 等）。 -- 已有基于 OpenAI 或 Anthropic SDK 的应用，希望低成本迁移到百炼平台。 -- 构建自定义的 RAG、Agent 框架，模型调用只是其中一环。 -- 对 [prompt](../guides/prompt.md) 工程有深度需求，需要完整掌控输入输出。 - -### 选择应用调用 API - -- 已在百炼控制台完成智能体或工作流的可视化编排，希望通过 API 将其集成到业务系统。 -- 需要使用控制台配置的知识库检索、插件、工作流节点等平台能力，不想在代码中重新实现。 -- 团队中非开发人员负责应用逻辑编排，开发人员只负责 API 集成。 -- 需要快速上线，应用逻辑变更通过控制台完成而非修改代码。 - -### 选择 Managed Agents API - -- 需要平台全托管的智能体运行时，不想自行管理会话状态和工具执行环境。 -- 智能体任务涉及多步工具调用、代码执行、文件读写，需要沙箱环境保障安全。 -- 希望通过 API 动态创建和管理多个智能体，实现多 Agent 协作。 -- 需要细粒度的事件流（SSE）来追踪智能体执行过程中的每一步操作。 -- 有自定义工具（Skill）需要安全审核后挂载，要求版本锁定和隔离。 - -## 选型决策参考 - -1. **"我只需要一个模型回答问题"** — 选 Qwen API。最简单直接，兼容主流 SDK。 -2. **"我已在控制台搭好应用，想 API 接入"** — 选应用调用 API。零编排代码，改逻辑只需改控制台配置。 -3. **"我需要平台帮我管理 Agent 的执行环境和工具调用"** — 选 Managed Agents API。平台托管状态机、沙箱和事件流，适合复杂任务。 -4. **迁移成本优先** — Qwen API 的 OpenAI/Anthropic 兼容接口迁移成本最低；应用调用 API 也提供 OpenAI 兼容模式。 -5. **功能完整度优先** — Qwen API 的 DashScope 原生接口参数最丰富；Managed Agents API 的资源模型最完整。 - -## 注意事项 - -- Qwen API 的兼容接口可能不暴露 DashScope 原生的全部参数，如需最全功能建议使用 DashScope 接口。 -- 应用调用 API 要求先在控制台创建并发布应用，APP ID 只能通过控制台手动获取。 -- Managed Agents API 当前仅支持 cn-beijing 地域，Skill 上传后需通过安全扫描才能挂载。 -- 三种 API 的计费方式均基于 token 消耗，但应用调用和 Managed Agents 可能涉及额外的平台资源费用（如沙箱、存储），请参考官方定价文档。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md deleted file mode 100644 index a3da8448..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md +++ /dev/null @@ -1,57 +0,0 @@ -# Qwen API vs 全双工实时API vs 托管智能体API - -百炼平台提供多种 API 接口满足不同开发场景。Qwen API 面向文本生成任务，提供多协议兼容的 HTTP 接口；全双工实时 API（Omni Realtime）基于 WebSocket 实现低延迟的音视频实时对话；托管智能体 API（Managed Agents）则提供完整的智能体托管运行时，由平台负责会话编排与沙箱执行。本文从协议、能力、适用场景等维度帮助开发者做出技术选型。 - -## 关键维度对比 - -| 维度 | Qwen API | 全双工实时 API | 托管智能体 API | -| --- | --- | --- | --- | -| 通信协议 | HTTP（REST） | WebSocket（长连接） | HTTP（REST）+ SSE 事件流 | -| 输入格式 | 文本（messages JSON） | 音频流（PCM 16kHz）、图像（Base64） | 文本消息、文件附件 | -| 输出格式 | 文本（支持流式 SSE） | 音频流（PCM 24kHz）+ 文本转录 | 事件流（SSE），含文本、工具调用回执等 | -| 支持模型 | Qwen 系列文本生成模型 | Qwen3.5-Omni-Realtime、Qwen3-Omni-Flash-Realtime、Qwen-Omni-Turbo-Realtime | 可配置任意百炼平台模型 | -| API 端点 | `https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions` 等 | `wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime` | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio` | -| 兼容协议 | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | 无（百炼专有 WebSocket 协议） | 无（百炼专有 REST 协议） | -| 工具调用 | Responses 接口内置联网搜索/代码解释器/网页提取；其他接口需自定义 | 支持 Function Calling 和联网搜索（仅 Qwen3.5 系列） | 平台托管技能（Skill zip 包），沙箱内执行 | -| 会话管理 | 仅 Responses 接口自动管理历史；其他需客户端维护 | 平台维护 WebSocket 会话上下文 | 平台全托管（Session 状态机：idle → running → idle/terminated） | -| 延迟特性 | 标准 HTTP 请求-响应，流式可逐 token 返回 | 超低延迟（毫秒级音频帧推送） | 异步任务式，SSE 实时推送中间事件 | -| 多模态支持 | 纯文本（部分模型支持图像输入） | 音频 + 视频 + 图像 + 文本 | 文本 + 文件（通过 File 资源挂载） | -| 计费方式 | 按 token 计费（输入/输出分计） | 按音频时长 + token 计费 | 按底层模型 token + 沙箱资源用量计费 | - -## 适用场景建议 - -### Qwen API - -- **文本问答与对话**：聊天机器人、客服对话、内容生成等标准 NLP 任务 -- **快速迁移**：已有 OpenAI 或 Anthropic 代码的项目，可通过兼容接口低成本切换到 Qwen 模型 -- **轻量工具调用**：使用 Responses 接口可直接获得联网搜索、代码解释器能力，无需额外开发 -- **批量处理**：适合离线或准实时的文本处理流水线 - -### 全双工实时 API - -- **语音助手**：需要实时语音输入并即时语音回复的场景 -- **智能客服**：基于 VAD 自动检测用户说话意图，实现自然的对话轮转 -- **多模态交互**：需要同时处理音视频输入的实时应用（如视频通话中的 AI 助手） -- **声音定制**：利用声音复刻能力打造品牌专属语音形象 - -### 托管智能体 API - -- **复杂任务编排**：智能体需要多轮推理、工具调用、代码执行的场景 -- **平台托管运行时**：不想自行管理会话状态、沙箱环境和工具执行的团队 -- **企业级智能体**：需要版本管理、权限控制、文件交互等完整生命周期管理 -- **多技能组合**：通过 Skill 机制灵活组装工具链，且由平台保证安全审核 - -## 技术选型指引 - -1. **只需文本生成** → 选择 Qwen API。如已有 OpenAI/Anthropic SDK 代码，使用对应兼容接口可零改动迁移。 -2. **需要实时语音交互** → 选择全双工实时 API。它是唯一支持音频流式双向通信的接口，延迟最低。 -3. **需要平台托管的智能体运行时** → 选择托管智能体 API。适合需要沙箱执行、多工具编排、会话生命周期管理的复杂 Agent 应用。 -4. **组合使用**：三者并非互斥。例如可用托管智能体 API 编排复杂工作流，其底层模型调用仍走 Qwen API；或在语音助手前端使用全双工实时 API，后端通过托管智能体执行复杂任务。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md deleted file mode 100644 index 5b2f9be9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md +++ /dev/null @@ -1,81 +0,0 @@ -# 智能体应用 - -智能体应用（Agent）是阿里云百炼平台的核心应用构建模式之一，通过自然语言零代码配置，让大模型基于角色设定自主决策、动态规划并调用知识库、MCP、Skill 等工具来完成任务。相较于流程固定的工作流应用，智能体强调 AI 的自主性，适合意图开放、需要动态编排的对话与轻量任务场景。 - -## 版本演进：Agent 1.0 与 Agent 2.0 - -百炼提供两代技术架构不同的智能体，**不支持直接升级或版本切换**，迁移需重新创建： - -- **新版智能体（Agent 2.0）**：2025 年 12 月 26 日上线。将知识库、MCP 等能力统一抽象为「工具」，由智能体自主规划调用时机与顺序，并完整展示「规划-执行-反思」链路。无旧版依赖时推荐使用。**仅支持 API 调用，不支持任何分享渠道**（魔笔/UI、钉钉、微信、组件、音视频互动）。 -- **旧版智能体（Agent 1.0）**：通过知识库（RAG）+ 插件扩展能力，先检索知识再决策是否调用工具，适合意图单一、流程固定的简单任务。自定义插件有 **5 秒超时限制**。分享渠道均为 1.0 功能。 - -## 核心能力配置（Agent 2.0） - -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）；可配置最长回复长度、`temperature`、`enable_thinking`（思考模式）等参数。 -- **提示词（System Prompt）**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境下的 `bash`、`write`、`read`、`edit`、`glob`、`grep`、`download_file`，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **Skill**：可扩展能力包，智能体在对话中自动识别匹配任务并调用，无需额外编码。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话内工具调用最大次数。 - -## 文件问答 - -智能体应用支持上传文件进行问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等 | 功能灵活，依赖配置的工具 | - -限制：单会话最多 10 个文件，单文件不超过 10 MB（超出需用文件上传 API）。支持文档、图片、视频、音频等格式。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成，核心步骤：在应用配置页点击「发布」→ 在「发布渠道」查看调用方式。RAM 账号发布前需拥有 `ram:CreateServiceLinkedRole` 权限。 - -API 调用与工作流应用完全一致，通过 `Application.call` / `POST /apps/{app_id}/completion` 触发： - -```python -import os -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -响应结构为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}`，业务侧主要消费 `output.text`。 - -## 分享与组件化（仅 Agent 1.0） - -- **分享渠道**：UI 应用/魔笔、钉钉、微信公众号、音视频实时互动。UI 体验链接与音视频临时二维码有效期均为 **24 小时**。分享产生的费用由应用创建者 UID 账号承担。 -- **组件化**：智能体或工作流可发布为模块化组件供其他应用复用。接入智能体时组件作为工具，大模型据「组件描述」自动判断调用；预设系统参数 `query`、`imageList` 无法删除。注意避免嵌套调用（A↔B）和多级调用（A→B→C 易超时）。 - -## 与 Managed Agents 的区别 - -Managed Agents 是服务端托管运行时，与无状态的智能体应用不同：它在独立云端沙箱容器中维护会话状态，支持中断与续接、事件历史持久化，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 计费说明 - -- **模型调用**：按模型类型和 Token 用量计费。 -- **知识库**：按量付费，召回的文本切片会增加输入 Token；自 2026 年 1 月 4 日起正式计费。 -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 由第三方收取。 - -百炼提供限时免费额度，可在模型广场查看。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [managed agents](../guides/managed-agents.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) -- [start using](../guides/start-using.md) -- [skill](../guides/skill.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md deleted file mode 100644 index 72e01839..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md +++ /dev/null @@ -1,90 +0,0 @@ -# 智能体编排 - -智能体编排是指在阿里云百炼平台上，通过配置或代码组织智能体（Agent）、工具、知识库、子智能体等资源之间的调用关系与执行流程，让大模型在受控边界内自主规划并完成业务任务的能力。 - -## 在百炼中的使用方式 - -百炼支持三种互补充的编排形态，开发者可单独使用或组合集成： - -| 形态 | 编排方式 | 控制力 | 适用场景 | -| --- | --- | --- | --- | -| 新版智能体（Agent 2.0） | 自然语言配置，零代码 | 由大模型自主规划 | 智能客服、知识问答、任务助理 | -| 工作流应用 | 可视化节点编排，低代码 | 节点固定、流程确定可复现 | 报告生成、订单处理、审批流 | -| 高代码应用 | Python 编码 | 完全由代码定义逻辑 | 私有算法部署、深度定制 | - -### 1. 新版智能体编排 - -Agent 2.0 将知识库、MCP 统一抽象为「工具」，由智能体在每轮「规划—执行—反思」（ReAct）链路中自主决定调用顺序。关键配置： - -- **模型**：推荐 `千问-Max` 等工具调用能力强的模型；可配 `temperature`、`enable_thinking`、最长回复长度。 -- **系统提示词**：定义角色、行为指令、能力边界；支持 `/变量` 引用自定义变量。 -- **知识库**：作为工具被自主调用，可用标签限定查询范围。 -- **MCP / 插件**：以 MCP 协议接入外部工具，支持动态非固定顺序调用；插件可一键转 MCP。 -- **内置工具**：`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`，在隔离沙箱中默认关闭、按需开启。 -- **记忆**：短期 0–30 轮上下文。 -- **ReAct 最大轮次**：1–50，限制单次会话工具调用次数，超出自动退出并生成最终回复。 - -> 注意：旧版智能体与新版本架构不兼容，不支持升级、降级或版本切换。 - -### 2. 工作流编排 - -通过可视化节点将复杂任务拆解为有序步骤，逻辑确定。核心节点： - -- **开始 / 结束**：定义输入输出参数，开始节点预置 `query`、`historyList`、`imageList` 等。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆。 -- **意图分类节点**：按输入分流到不同下游分支。 -- **变量处理节点**：文本输出或变量加工。 -- **智能体群组节点**：把任务分解给多个已发布的子智能体协同完成，是工作流编排多智能体协作的核心。 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在画布右上角配置并在各节点引用。 - -### 3. 高代码编排 - -基于完整 Python 项目结构部署 AI 后端： - -- **部署形态**：Serverless Function（无状态、快速拉起、低成本）或 K8s（高性能、有状态、长程任务）。 -- **代码提交**：使用预置模板（基础对话 / 工具调用 / 深度研究 Agent）或上传本地 `.whl` 包。 -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务。 -- **网关**：应用稳定后建议开启网关，通过自定义域名和路由在生产环境访问。 - -## API 层编排 - -### Managed Agents API（托管运行时） - -平台托管会话、沙箱、工具执行与事件流，开发者通过 REST 或 DashScope SDK 管理 Agent、Environment、Skill、File、Session 等资源： - -- **Agent**：可复用的智能体配置（模型、系统提示词、工具包、技能）。 -- **Environment**：工具调用的执行沙箱与预装依赖，可被多会话复用。 -- **Session**：智能体的一次运行实例，绑定 Agent 与 Environment 快照。 -- **Skill**：以 zip 包封装的工具组合与文档，挂载时锁定版本。 -- **File**：独立文件资源，可挂载到沙箱供工具读写或作为消息内容。 - -调用流程四步：创建 Agent（一次创建长期复用）→ 创建 Session（每轮对话新建）→ 发送 Event（写入用户消息触发 `running`）→ 订阅 SSE 流式接收回复直至回到 `idle`。 - -### 应用调用 API - -将控制台编排好的应用集成进业务系统，两种接口： - -- **OpenAI 兼容 Responses API**：`POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，复用 OpenAI 生态，支持同步/异步、流式、多模态。 -- **DashScope 原生 API**：`POST /api/v1/apps/{APP_ID}/completion`，功能更全，新版智能体、工作流、旧版智能体均支持。 - -调用前需准备 APP ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）。多轮对话可通过 `session_id`（系统自动加载历史，有效期 1 小时、最多 50 轮）或自行维护 `messages` 数组实现。 - -## 选型建议 - -- 需要 AI 自主决策、动态规划 → 新版智能体。 -- 流程固定、要求稳定可复现 → 工作流，复杂协作用智能体群组节点。 -- 深度定制、私有算法 → 高代码应用。 -- 需要平台托管调度与执行基础设施 → Managed Agents API。 -- 仅把控制台应用接入业务系统 → Application.call / completion。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents api](../api/managed-agents-api.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md b/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md deleted file mode 100644 index 8fbe5ce1..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md +++ /dev/null @@ -1,93 +0,0 @@ -# API Key 鉴权 - -API Key 是阿里云百炼平台调用模型与应用的核心鉴权凭证：请求通过在 HTTP 头 `Authorization: Bearer ` 中携带该密钥完成身份校验，无需为不同模态单独申请。 - -## 获取与配置 - -- **创建**：需主账号或具备 `管理员` / `API-Key` 页面权限的子账号，在[阿里云百炼控制台](https://bailian.console.aliyun.com/)对应地域的 **API Key** 页面创建。 -- **归属业务空间**：决定 Key 的调用范围。默认业务空间的 Key 可调用所有标准模型及默认空间应用；子业务空间的 Key 只能调用已授权的模型及本空间应用，同一空间内 Key 权限相同。 -- **权限**：可选 **全部**，或 **自定义**（配置 IP 白名单最多 20 个 IPv4/IPv6 地址/网段，并限定可访问的模型/应用范围）。 -- **环境变量**：推荐配置到 `DASHSCOPE_API_KEY`，避免硬编码泄漏（各系统方式：`~/.bashrc`、`~/.zshrc`、`~/.bash_profile`，或 Windows 系统属性 / `setx` / PowerShell）。 -- **服务端点**：调用时除 Key 外还需指定 `base_url`（即创建弹窗中的 API Host），OpenAI 兼容协议与 Anthropic 兼容协议的 `base_url` 不同，且随地域变化。 - -> 安全升级后新建的按量付费 Key 以 `sk-ws` 开头，**仅创建时明文展示一次**，务必立即复制保存；升级前 `sk-` 开头的旧 Key 仍可使用。 - -## Key 的类型（互不通用） - -不同计费方案使用格式不同、完全隔离的 Key，混用会导致 401/403 或意外走按量计费扣费： - -| 前缀 | 类型 | 说明 | -| --- | --- | --- | -| `sk-` | 按量付费通用 Key（旧） | 升级前创建，仍可用 | -| `sk-ws` | 按量付费通用 Key（新） | 安全升级后创建，明文仅展示一次 | -| `sk-sp-` | Token Plan / Coding Plan 专属 Key | 订阅套餐专用，与通用 Key 隔离 | -| `st-` | 临时 API Key | 由永久 Key 生成，有时效 | - -各地域（北京 / 新加坡 / 弗吉尼亚）的 API Key 不互通，请求时须使用对应地域的 Endpoint 与 Key。 - -## 典型使用场景 - -### 1. SDK / 直接调用模型 - -配置好 `DASHSCOPE_API_KEY` 后，即可用官方 DashScope SDK（Python、Java）或 OpenAI 兼容 SDK（Python、Java、Node.js、Go）发起文本、图像、视频、语音、向量等调用。 - -### 2. 百炼 CLI（`bl` / `bailian`） - -支持多种可组合的认证方式： - -| 方式 | 命令 | 场景 | -| --- | --- | --- | -| 控制台登录（推荐） | `bl auth login --console` | 模型调用 + 应用管理 | -| API Key | `bl auth login --api-key sk-xxx` | 模型调用，会校验有效性 | -| 环境变量 | 配置 API Key 环境变量 | CI/CD、无界面环境 | -| 配置文件 | `bl config set --key api_key --value sk-xxx` | 持久化，**不校验** | -| 临时传入 | `bl text chat --api-key sk-xxx ...` | 单次调用，不落盘 | - -### 3. 第三方 AI 工具接入 - -Claude Code、Cursor、Cline、Codex、Qoder、Dify 等工具统一以「Base URL + API Key + 模型 ID」接入，通过 OpenAI 兼容或 Anthropic 兼容协议访问百炼网关。务必保证 Key 与 Base URL 的计费方案、地域一致。 - -### 4. 开源框架集成 - -LlamaIndex（Python）、Spring AI Alibaba（Java）均以 API Key 鉴权。注意 Spring AI Alibaba 两篇文档约定的变量名不一致：应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`，关键是 `application.yml` 中 `${...}` 占位符与实际变量名匹配。 - -### 5. 子业务空间调用 - -按业务线隔离权限或分账时，必须使用子业务空间自身的 Key，并提前为该空间授予标准模型（如 `qwen-plus`）的调用权限；调优部署的模型仅能由所在空间的 Key 调用。 - -## 临时 API Key - -在浏览器、移动端等不可信环境中，应由后端用永久 Key 生成时效性临时凭证，避免永久 Key 泄露。 - -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` - -**关键要点**： - -- `expire_in_seconds`：有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 -- 响应返回 `token`（`st-` 开头的临时 Key）与 `expires_at`（过期 UNIX 时间戳）。 -- 临时 Key 继承永久 Key 的全部权限（含模型/知识库访问限制），到期自动失效，无法提前删除。 -- 各地域不互通；新加坡地域需将 Endpoint 中的 `WorkspaceId` 替换为实际值。 - -## 编程化管理 - -除控制台外，百炼提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）管理 Key，调用需使用阿里云账号 AccessKey 签名认证并具备相应 RAM 权限。 - -## 常见排错 - -- **401 / 403 鉴权失败**：多因误用了错误前缀的 Key，或 Key 与 Base URL 的地域 / 计费方案不匹配。 -- **意外扣费**：套餐场景误用通用 `sk-` Key 会走按量计费通道。 -- **`InvalidApiKey`**：临时 Key 接口常见错误码，检查永久 Key 是否有效、Endpoint 地域是否对应。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [more](../api/more.md) -- [more about models](../api/more-about-models.md) -- [frameworks](../api/frameworks.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [token plan guide](../guides/token-plan-guide.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md deleted file mode 100644 index 1edf5d93..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md +++ /dev/null @@ -1,64 +0,0 @@ -# 异步调用与任务轮询 - -异步调用是百炼平台为耗时较长的模型任务（图像生成、视频生成、3D 生成、复杂智能体应用等）提供的调用模式：客户端先「创建任务」拿到一个 `task_id`，随后通过「轮询查询」或事件回调获取最终结果，从而避免长请求超时。 - -## 适用场景 - -在百炼平台上，凡是单次处理通常需要数十秒到数分钟的能力，基本都以异步为主： - -- **图像生成**：文生图、图像编辑、扩图、虚拟模特等（通常 1-2 分钟）。部分新一代模型（如 `wan2.6-image`、`wan2.7-image`、`z-image-turbo`）额外提供 HTTP 同步调用。 -- **视频生成**：万相、PixVerse、Vidu、Kling 及人像驱动等模型（通常 1-5 分钟），统一走 `video-synthesis` 接口，全部为异步。 -- **3D 生成**：基于 Tripo 的文生 3D / 图生 3D，仅支持异步。 -- **智能体 / 工作流应用**：Responses API 通过 `background=true` 开启异步，用于生成报告、多步骤工具调用等耗时任务（DashScope API 暂不支持异步）。 - -## 调用流程 - -异步调用统一分为两步： - -1. **创建任务**：向对应模型的下发接口发送 `POST` 请求，成功后立即返回 `task_id`。 - - 大多数生成类接口必须携带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 - - 智能体应用（Responses API）则通过请求体参数 `background=true` 开启异步。 - - `task_id` 有效期为 **24 小时**，**请勿重复创建任务**，创建成功后轮询即可。 - -2. **轮询查询结果**： - ``` - GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} - ``` - 返回体中 `output.task_status` 标识当前状态，流转过程通常为 `PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED` / `FAILED`。仅当状态为 `SUCCEEDED` 时才返回结果内容（如图像 / 视频 / 模型下载 URL）。 - -## 任务状态枚举 - -| 状态 | 含义 | -| --- | --- | -| `PENDING` | 任务排队中 | -| `RUNNING` | 任务处理中 | -| `SUCCEEDED` | 任务成功，返回结果 | -| `FAILED` | 任务失败，返回 `code` / `message` | -| `CANCELED` | 任务已取消（仅 `PENDING` 状态可取消） | -| `UNKNOWN` | 任务不存在，或超过 24 小时有效期被清理 | - -## 通用任务管理接口 - -除单任务查询外，百炼提供三个通用异步任务管理接口，流量限制均为 **20 QPS**（主账号维度）： - -- **查询单个任务**：`GET .../api/v1/tasks/{task_id}` -- **批量查询任务状态**：`GET .../api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx`，可按时间范围、模型名、状态过滤，单次时间跨度不超过 24 小时。 -- **取消任务**：`POST .../api/v1/tasks/{task_id}/cancel`，仅能取消尚在 `PENDING` 状态的任务，已开始处理的无法取消。 - -## 关键要点与最佳实践 - -- **轮询间隔**：建议约 15 秒查询一次，避免过于频繁触发限流（查询接口默认 20 RPS）。 -- **事件通知替代轮询**：频繁轮询浪费资源且易触发限流。百炼已接入阿里云 EventBridge，支持任务完成后主动推送通知（HTTP 回调 URL 或 RocketMQ）。事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`，关键字段为 `data.task_status` 与 `data.task_id`；收到通知后只需查询一次即可拿到结果。 -- **结果时效**：任务结果（如图像/视频 URL）与 `task_id` 有效期一般为 24 小时；3D 生成的模型下载链接有效期更短，仅 **2 小时**，需及时下载。 -- **地域一致性**：模型、Endpoint、API Key 必须属于同一地域（如华北2·北京、新加坡等），跨地域调用会失败。3D 生成目前仅支持华北2（北京）。 -- **接口路径差异**：不同模型的下发路径并不统一（如 `.../aigc/video-generation/video-synthesis`、`.../aigc/image2video/video-synthesis`、`.../text2image/image-synthesis` 等），接入前请以对应模型文档为准。 - -## 关联主题页 - -- [3d generation](../api/3d-generation.md) -- [video generation api](../api/video-generation-api.md) -- [image generation](../api/image-generation.md) -- [application call](../api/application-call.md) -- [more about models](../api/more-about-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md deleted file mode 100644 index 6b73c92e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md +++ /dev/null @@ -1,85 +0,0 @@ -# 计费 - -计费是百炼平台围绕模型推理、模型训练、模型部署及增值服务所建立的费用体系，涵盖按量付费、订阅制（[Token](token.md) Plan / Coding Plan）、预留容量（TPM 预留 / PTU）等多种方式，开发者可根据用量规模和业务场景灵活组合，实现成本最优。 - -## 计费场景与方式 - -百炼平台的计费覆盖以下主要场景： - -| 场景 | 计费方式 | 说明 | -|------|---------|------| -| 模型推理调用 | 按量付费（[Token](token.md) 计价） | 按输入/输出 [Token](token.md) 分别计价，部分模型支持阶梯计费 | -| 模型训练 | 按训练 Token 计费 | （训练数据 Token + 混合训练数据 Token）× 循环次数 × 单价 | -| 模型部署 | PTU / 模型单元 / Token 用量 | 三种方式创建后不可更改，需下线重建才能切换 | -| 知识库 | 按规格计费 | 标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| TPM 预留 | 按 kTPM 预付费 | 按天计费，输入/输出分别定价 | -| Token Plan 团队版 | 坐席订阅制 | 按 Credits 抵扣，198 元/坐席/月起 | -| Coding Plan | 月度订阅制 | 按模型调用次数限额，200 元/月 | - -## 费用抵扣顺序 - -多种付费方式并存时，系统按以下优先级自动抵扣： - -**免费额度 → 资源包 → 其他模型节省计划 → AI 通用型节省计划 → 按量付费** - -Token Plan 和 Coding Plan 的套餐额度独立于按量计费体系，不参与上述抵扣链路。 - -## 免费额度 - -首次开通百炼时自动发放新人免费额度，有效期 30~90 天。关键限制： - -- 仅适用于华北2（北京）地域、中国内地服务部署范围的实时推理 -- 不支持抵扣 Batch 调用、模型调优、模型部署、自定义模型 -- 主账号与 RAM 子账号共享 -- 建议开启"免费额度用完即停"功能，防止额度耗尽后自动扣费 - -## 成本优化方案 - -### AI 通用型节省计划 - -承诺每月消费金额获取阶梯折扣，最高可享 5.3 折。覆盖阿里直供全部模型，承诺金额 1,000 元起，支持 3/6/12/24 个月周期。当月未用完的额度自动清零，不可累积。 - -### 资源包 - -预先购买具体 Token 数量，用于抵扣特定模型的实时推理用量。适合用量明确且集中在单一模型的场景。 - -### Batch 调用优惠 - -支持 Batch 调用的模型，输入和输出 Token 单价均按实时推理价格的 50% 计费。 - -### 上下文缓存折扣 - -部分模型支持上下文缓存，命中缓存的输入 Token 按折扣系数消耗额度（例如 deepseek-v4-pro 为 0.08，即按 8% 折算）。Batch 调用与上下文缓存不能同时生效。 - -## 多地域定价差异 - -同一模型在不同地域的价格可能不同。华北2（北京）使用人民币定价，新加坡等国际地域按国际价格结算，通常高于国内价格。[API Key](api-key.md) 必须与 Base URL 同一地域，否则会报 `401` 错误。 - -## 账单查询 - -- **费用概览**：控制台"用量 & 费用 > 费用概览"查看当月总消费，支持按模型或 [API Key](api-key.md) 筛选 -- **模型用量**：按[业务空间](workspace.md)维度统计，数据延迟约 1 小时，不支持查看 30 天以前的数据 -- **出账时间**：大模型推理分钟级出账（2~10 分钟），批量推理和训练小时级出账 -- **分账管理**：通过[业务空间](workspace.md)标签按部门或项目归集费用，T+1 天生效 - -## 欠费与停止计费 - -账户可用额度小于 0 时视为欠费。Token Plan 和 Coding Plan 的套餐额度独立于账户余额，欠费期间可继续使用。停止计费的方式： - -- **模型推理**：停止 API 调用，删除不再使用的 [API Key](api-key.md) -- **模型部署**：在控制台下线已部署模型；包月预付费需额外退订实例 -- **模型训练**：无进行中的训练任务即不产生费用 -- **订阅制**：关闭自动续费，到期自动停止 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) -- [knowledge base](../guides/knowledge-base.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md b/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md deleted file mode 100644 index 2c3d22ce..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md +++ /dev/null @@ -1,32 +0,0 @@ -# 上下文窗口 - -上下文窗口（Context Window）是模型在单次推理中能够接收并处理的输入与输出 Token 总长度的上限，单位通常为 Token，是衡量模型承载长文本、多轮对话、多模态输入能力的核心指标。在百炼平台，千问系列文本生成模型（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）及视觉理解模型普遍支持 1M（约百万）Token 的上下文窗口，第三方模型如 `deepseek-v4-pro` / `deepseek-v4-flash` 同样提供 1M 上下文。 - -## 在百炼中的使用场景 - -- **长文档处理**：1M 上下文可一次性容纳数十万字的中文文档或数十张高分辨率图像，适合合同审阅、代码仓库分析、长报告摘要等场景，避免分段截断导致信息丢失。 -- **多轮对话**：聊天与 Agent 应用需在请求中累积历史消息，上下文窗口决定可保留的对话轮数。超出窗口后早期消息会被截断，需结合检索或摘要策略压缩历史。 -- **多模态输入**：视觉理解模型同时接收图像、视频与文本，图像 Token 按 `高 × 宽 / (32 × 32) + 2` 估算，单张图最高约 1600 万像素；视频最长支持 2 小时 / 2GB（Qwen3.7/3.6/3.5 系列），1 小时 / 2GB（Qwen3.5-Omni 系列含音频）。多模态输入会快速占用上下文窗口，需在长文本与多图之间权衡。 -- **工具调用与思考模式**：开启 `enable_thinking` 的思考模式、Function Calling 的工具定义与中间结果都会计入上下文消耗，复杂 Agent 流程需为工具输出预留充足窗口预算。 - -## 关键参数与配置 - -- **模型选型**：上下文长度随模型档位而定，文本/视觉模型以 1M 为主；向量、[重排序](rerank.md)、语音等模型按各自规格独立设定，调用前应以模型广场标注为准。 -- **对话历史管理**：仅 OpenAI 兼容 Responses 接口由平台自动管理对话历史，无需手动拼接 `messages`；OpenAI Chat Completions、Anthropic Messages、DashScope 原生接口均需调用方自行维护上下文长度与轮次，避免超出窗口导致请求失败或内容截断。 -- **Token 计费与限流**：上下文窗口内的输入与输出 Token 均参与计费和限流计量，长上下文请求会显著提升单次调用成本与首字延迟，建议按需裁剪历史与文档片段。 -- **超时与域名**：长上下文推理耗时较高，生产环境推荐使用业务空间专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（请求超时 3600 秒、SLA 99.9%），避免中心化或试用域名的较短超时与限流影响。 - -## 注意事项 - -- 上下文窗口是模型能力上限，不代表稳定可用长度；接近上限时模型对长尾内容的注意力可能下降，关键信息宜放在输入首尾。 -- 跨接口迁移时需确认目标兼容接口是否完整透传长上下文相关参数，部分兼容接口为协议一致可能不暴露全部原生参数，最全参数请使用 DashScope 原生接口。 -- 各地域、各模型对上下文窗口的支持存在差异，API Key、模型列表与接入域名不能跨地域混用，调用前请在目标地域的模型广场核对规格。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [get started with models](../guides/get-started-with-models.md) -- [model experience](../guides/model-experience.md) -- [more about models](../api/more-about-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md deleted file mode 100644 index 2b0d787d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md +++ /dev/null @@ -1,112 +0,0 @@ -# 跨会话记忆 - -跨会话记忆是指将对话中提取的关键信息和用户画像持久化存储，并在后续会话中通过语义检索召回并注入 Prompt 的能力，用于解决大模型上下文窗口无法跨会话延续的问题。百炼平台通过「记忆库（Memory Library）」与「长期记忆 API」提供该能力。 - -## 在百炼平台中的使用场景 - -跨会话记忆在以下场景中发挥作用： - -- **个性化智能体**：在多轮对话之间保留用户偏好、习惯和重要事件（如「每天上午 9 点提醒我喝水」），让智能体在新会话中仍然理解用户历史。 -- **长期用户画像**：通过自定义画像模板提取结构化属性（年龄、职业、偏好等），在后续对话中以固定字段持久化注入，适用于需要稳定属性支撑的业务。 -- **Agent 自动捕获/召回**：OpenClaw Agent 通过记忆插件在 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子中与长期记忆 API 交互，实现零侵入的跨会话记忆。 -- **应用观测中的记忆追踪**：在应用观测的调用链路中，记忆的写入与检索会作为 RETRIEVER、EMBEDDING 等节点出现，便于开发者定位记忆相关调用的延时与 Token 消耗。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，请求 Header 携带 `Authorization: Bearer $DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 下次对话调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（自动从对话提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -```bash -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope API Key；`userId` 用于隔离不同用户记忆空间。所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -## 记忆内容类型 - -记忆库提供两类持久化内容，可独立或组合使用： - -- **记忆片段**：从对话中自动提取的关键事件和信息，支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。适用于大多数长期记忆场景。 -- **用户画像**：基于画像模板（profile schema）从对话中提取的结构化属性，适用于需要固定属性持久化存储的场景。属性字段应清晰具体，避免「姓名/名称/名字」等同义字段并存，且不应期望一次对话就提取全部信息。 - -## 关键参数 - -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | 与 `custom_content` 二选一 | 对话消息列表，每个消息含 `role`（user/assistant）和 `content`，最多 50 条 | -| `custom_content` | 与 `messages` 二选一 | 自定义内容，最大 512 个字符，传入后忽略 `messages` | -| `profile_schema` | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | 否 | 记忆库 ID，最大 32 个字符，不传则使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不传则使用指定记忆库的默认规则 | -| `meta_data` | 否 | 用户自定义信息 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 用户 ID，用于隔离记忆空间 | -| `messages` | 是 | 查询对话内容，系统据此做语义检索 | -| `top_k` | 否 | 返回的记忆片段数量 | - -## 使用限制 - -| 项目 | 限制 | -| --- | --- | -| 全部接口总计 | 3000 QPM（阿里云账号级别） | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -## 有效期说明 - -记忆有效期在不同入口存在差异：长期记忆 API 文档指出「生成的记忆片段与用户画像暂无失效日期」，而控制台默认记忆片段规则预置了「默认有效期 180 天」，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 注意事项 - -- OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 API Key。 -- 应用观测暂不支持通过 Assistant API 创建的智能体应用；对高代码应用，仅能观测到入口 CHAIN 节点，不支持追踪其内部调用链路。 -- 记忆的写入与检索在应用观测中会体现为 RETRIEVER、EMBEDDING 等节点，可用于定位记忆相关调用的延时与 Token 消耗。 - -## 关联主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [application monitoring](../guides/application-monitoring.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md deleted file mode 100644 index de46d3bd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md +++ /dev/null @@ -1,99 +0,0 @@ -# DashScope SDK - -DashScope SDK 是阿里云百炼平台官方提供的软件开发工具包，封装了模型与应用调用的原生（DashScope）接口，让开发者用少量代码即可接入通义千问、万相、Qwen-MT、Qwen-OCR 等模型能力以及智能体/工作流应用。相比 HTTP 直连和 [OpenAI 兼容接口](openai-compatible-interface.md)，DashScope 原生接口暴露的参数最完整、功能集最丰富。 - -## 适用语言与安装 - -DashScope SDK 主要提供 Python 与 Java 两种官方实现，部分场景也可用 HTTP（如 Node.js 借助 `axios`）替代： - -- **Python**：`python3 -m pip install -U dashscope` -- **Java**：通过 Maven / Gradle 引入 `com.alibaba:dashscope-sdk-java`，建议版本 `>= 2.12.0` -- **Node.js / 其他语言**：目前无官方 SDK，直接走 HTTP API（发起 POST 请求） - -> 注意：不同能力对 SDK 语言与地域的支持存在差异。例如 `qwen-deep-research` **仅支持 Python DashScope SDK，且仅限华北2（北京）地域**，暂不支持 Java SDK 与 [OpenAI 兼容接口](openai-compatible-interface.md)。 - -## 在不同场景中的使用 - -### 1. 调用文本生成模型（Qwen 系列） - -Qwen 系列可通过 OpenAI 兼容、Anthropic 兼容或 DashScope 原生三类接口调用。其中 DashScope 是百炼原生接口，**功能集最完整、参数支持最丰富**；当需要使用最全的采样参数、插件或业务字段而兼容接口未暴露时，应改用 DashScope 原生接口。 - -### 2. 调用专用模型（[more](../api/more.md) models） - -法律、意图理解、翻译、OCR 等专用模型大多支持 OpenAI 兼容或 DashScope 两种方式调用： - -- `farui-plus`（法律大模型）：通过 DashScope SDK（Python / Java）调用 -- `tongyi-intent-detect-v3` / `qwen-mt-plus` / `qwen3.5-ocr`：OpenAI 兼容或 DashScope 均可 -- `qwen-deep-research`（深度研究）：仅 Python DashScope SDK - -### 3. 调用图像生成与编辑模型 - -千问-图像（Qwen-Image）、万相（Wan/Wanx）、Z-Image 等图像模型均可通过 HTTP 或 DashScope SDK 调用，覆盖文生图、图像编辑、图像翻译、风格迁移等能力。 - -### 4. 调用智能体应用与工作流应用 - -已创建并发布的智能体应用、工作流应用可通过 DashScope SDK 集成到业务系统，二者调用方式一致： - -- Python：`from dashscope import Application`，调用 `Application.call(...)` -- Java：构造 `ApplicationParam` 后调用 `application.call(param)` -- HTTP 等价接口：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` - -响应统一为 `{"output": {...}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 关键参数与配置 - -### 鉴权（API Key） - -- SDK 通过 [API Key 鉴权](api-key.md)，**推荐将密钥写入环境变量 `DASHSCOPE_API_KEY`**，SDK 会自动读取，避免在代码中硬编码。 -- 调用特定地域（如华北2/北京）或子业务空间下的模型/应用时，需使用对应地域的 API Key，并按需提供 Workspace ID。 - -### 通用请求参数 - -- `model`（string，必选）：目标模型名称。 -- `messages`（array）：对话消息列表，按顺序排列，需由调用方维护上下文。 -- `app_id`（应用调用）：目标应用 ID，从控制台应用卡片复制。 -- `prompt` / `input`：用户输入内容。 -- `stream`（bool，可选）：是否[流式输出](streaming.md)；如 `qwen-deep-research` 的反问阶段需设为 `true`。 -- `session_id`（应用多轮对话）：由云端维护上下文，免去手动拼接历史。 - -### 模型专属参数示例 - -- **Qwen-MT（翻译）**：通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制 `source_lang`、`target_lang`、`terms`（术语干预）、`tm_list`（翻译记忆）、`domain_prompt`（领域提示）。 -- **Qwen-OCR**：`messages.content` 为[多模态](multimodal.md)数组，可设 `min_pixels` / `max_pixels` 控制图像像素阈值。 - -## Python 快速示例（应用调用） - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'code={response.status_code}, message={response.message}') -else: - print(response.output.text) -``` - -## 使用建议 - -- 优先使用环境变量管理 API Key，区分不同地域的密钥。 -- 需要最完整功能与参数时选 DashScope 原生接口；追求生态兼容、迁移成本最低时可选 [OpenAI 兼容接口](openai-compatible-interface.md)。 -- 接入前先对照具体模型的 API 参考，确认其支持的 SDK 语言、协议与地域。 - -## 关联主题页 - -- [image generation](../api/image-generation.md) -- [more models](../api/more-models.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md deleted file mode 100644 index 1334d698..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md +++ /dev/null @@ -1,54 +0,0 @@ -# 向量嵌入 - -向量嵌入（Embedding）是将文本、图像、视频等非结构化数据转换为固定维度的数值向量，使语义相近的内容在向量空间中距离也相近。百炼平台提供多类嵌入模型，支撑语义检索、RAG 召回、跨模态搜索、聚类推荐等下游任务。 - -## 在百炼平台的使用场景 - -- **RAG 知识库召回**：知识库创建时选择向量模型，对导入文档切片做嵌入入库；查询时对 Query 做同款嵌入，再与切片向量做相似度匹配，作为 RAG 流程的第一段召回。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；图片问答类固定使用 `multimodal-embedding-v1`（1024 维）；视觉理解场景自动切换为 `qwen3-vl-embedding`。 -- **本地 RAG 应用**：本地知识库方案默认调用百炼 embedding API 生成向量，也可替换为本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。受 embedding API 限流影响，单文件不建议超过 100 MB。 -- **跨模态检索**：多模态向量模型（如 `qwen3-vl-embedding`、`multimodal-embedding-v1`）将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。支持「独立向量」（逐项生成）与「融合向量」（多输入合并为 1 个向量）两种模式。 -- **框架集成**：LlamaIndex 路线将知识库部署在百炼云端，使用官方向量模型与智能切分，不支持自定义嵌入模型；如需灵活选择嵌入模型，应改用本地知识库方案。 - -## 模型与关键参数 - -通用文本向量模型当前推荐 `text-embedding-v4`（Qwen3-Embedding 系列，支持 100+ 语种）。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -请求参数： - -- `model`（必选）：模型名称。 -- `input`（必选）：字符串、字符串列表或文件。 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持。 -- `encoding_format`（可选）：当前仅支持 `float`。 - -调用方式支持 OpenAI 兼容接口（base_url：`https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。 - -## 批处理接口 - -大规模文本向量化可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行文本。需在 HTTP 请求头加 `X-DashScope-Async: enable` 启用异步模式，提交后通过 `task_id` 轮询结果。同时处理中任务不超过 50 个，并发运行上限 3 个，超出部分排队等待。 - -## 与 Rerank 的关系 - -向量嵌入负责「召回」，Rerank 模型负责「精排」。在知识库检索流程中，先由向量 + 关键词混合检索召回 TopK 切片，再由 `qwen3-rerank`、`qwen3-vl-rerank` 等排序模型对候选切片二次排序，最终按相似度阈值与最大召回数量返回。两者配合提升 RAG 命中准确率。 - -## 注意事项 - -- 知识库向量模型与维度在创建时选定，**维度不可更改**；Meta 抽取、多轮对话改写等索引配置在创建后也无法追加，需重建知识库。 -- 嵌入与 Rerank 是两类不同模型，不要混用接口：qwen3-rerank 走 `/compatible-api/v1/reranks`，qwen3-vl-rerank / gte-rerank-v2 走 `/api/v1/services/rerank/text-rerank/text-rerank`。 -- `gte-rerank` 系列将于 2026 年 5 月 30 日下线，建议迁移到 `qwen3-rerank`。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md deleted file mode 100644 index a303b8d6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md +++ /dev/null @@ -1,77 +0,0 @@ -# 评测体系 - -评测体系是百炼平台用于系统化衡量输出质量的一整套能力，覆盖**应用评测**（智能体/工作流应用）与**模型评测**（文本生成模型）两大场景，通过评测集、评测维度/评估器、评分方式与评测报告构成完整的评估闭环。 - -## 两类评测场景 - -百炼的评测能力按被评测对象分为两条相对独立的链路： - -- **应用评测**：评估已发布智能体应用、工作流应用的回答质量，支持自动评测（大模型基于知识库生成评测集并打分）和手动评测（人工标注打分）。当前存在新旧两套系统，新版以「评测任务 + 评估器 + 标签」组织。 -- **模型评测**：评估文本生成类模型的能力，支持自定义评测（自有数据集 + 自定义维度）和基线评测（C-Eval、MMLU、GSM8K、BBH 等公开数据集）。 - -两者共享一套核心思路：准备评测数据 → 定义评分规则 → 创建评测任务 → 查看报告。 - -## 评分方式 - -无论应用还是模型评测，评分方式基本归为三类，选型取决于是否有标准答案与是否需要语义理解： - -| 评分方式 | 原理 | 适用场景 | 费用 | -|----------|------|----------|------| -| 大模型评估（LLM/AI 自动评测） | 裁判模型按 Prompt 语义打分或分类 | 相关性、幻觉、内容安全等语义场景 | 产生 Token/裁判模型费用 | -| 规则评估（Code/自动化指标） | ROUGE、BLEU、Cosine、字符串匹配等算法 | 翻译、摘要、格式校验、精确匹配等确定性场景 | 无额外费用 | -| 人工评估 | 人工逐条标注 Pass/Fail 或打分 | 创意写作、专业领域主观判断 | 无裁判费用 | - -**选型路径**：有标准答案且格式固定 → 字符串匹配；有标准答案但表述多样 → 文本相似度；无标准答案需语义理解 → 大模型评估；需主观判断 → 人工评估。 - -## 评测集（数据基础） - -评测集是评测任务的数据输入，需在数据管理模块或评测系统中准备： - -- **应用评测集**：旧版分对话分析（`.xls/.xlsx`）与知识问答（`.jsonl`）；新版分智能体、工作流、自定义三类，具备版本管理，创建后类型不可改。 -- **模型评测集**：单轮对话，Excel 格式，每行含 **Prompt**（用户输入）与 **Completion**（期望输出）。参评模型对每条 Prompt 推理，评分参考 Completion。 - -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 - -## 评测维度与评估器（评分规则） - -评分规则是评测体系的核心组件，创建为模板后可被多个任务复用： - -- **模型评测的评测维度**提供五种评分器类型：大模型评估-数值型、大模型评估-分类型、规则评估-文本相似度、规则评估-字符串匹配、人工评估-分类型。维度类型创建后不可修改。 -- **应用评测的评估器**支持预置模板（通用质量、智能体、文本匹配、文本相似度、格式校验）、自定义（LLM 评估器 / Code 评估器）以及基于历史评测任务标注结果抽象生成。每个评测任务最多添加 10 个评估器，建议组合 3-5 个从不同维度评估。 - -## 关键参数与配置 - -- **评分范围**（数值型）：整数打分区间，默认 0-5，建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**：判定 Pass/Fail 的分界线，数值型步长 0.1，相似度型步长 0.01。 -- **评分器 Prompt**（大模型评估）：至少含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符；用于指导裁判模型如何打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色/行为规范，通常可留空，注意勿与评分器 Prompt 混淆。 -- **数据来源**：评测数据集（含 Prompt+Completion，产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP 等，按所选模型动态加载。 -- **标签类型**（应用评测）：分类、布尔值、数字、文本，用于标注评测与观测数据。 - -## 计费说明 - -费用主要由**被评测模型推理费用**与**裁判模型评分费用**两部分构成： - -- 使用推理结果集可免去推理费用。 -- 规则评估、人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -成本优化建议：先用 50-100 条小规模验证 → 保存推理结果集复用 → 确定性场景优先用规则评估。 - -## 限制与注意事项 - -- 模型评测当前仅支持文本生成类模型，且仅支持控制台操作，不提供公开 API/SDK。 -- 基线评测仅北京地域可用，不支持下载评测结果。 -- 数据管理与数据清洗/增强能力仅适用于华北2（北京）地域。 -- 任务提交后不可更换目标模型、维度类型创建后不可修改，选错需删除重建。 -- 应用自动评测仅面向已发布且已配置知识库的应用，须开通应用观测并具备相应权限。 -- LLM 评分器存在位置偏差与自我偏好偏差，1-3% 的分差通常为噪声，建议定期人工抽查校准。 - -## 关联主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md deleted file mode 100644 index 3098445e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md +++ /dev/null @@ -1,89 +0,0 @@ -# 模型微调与生产链路 - -模型微调（Fine-tuning）是在 Prompt 工程、插件调用等手段仍无法满足效果时，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数的深度定制手段。它是百炼平台「数据准备 → 模型调优 → 模型压缩（可选）→ 模型部署」这条模型生产链路的核心环节。 - -> **注意**：微调、压缩、部署与调用能力**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 - -## 完整生产链路 - -一条典型的自定义模型生产链路包含四个阶段，前后衔接： - -1. **数据准备**：创建、清洗、增强训练集与评测集（仅控制台，暂无数据处理 API）。 -2. **模型调优**：指定基础模型、训练集/验证集与超参数，提交微调任务，训练完成后得到自定义模型。 -3. **模型压缩（可选）**：对全精度微调模型做量化，降低部署所需 MU 规格、减少推理成本。 -4. **模型部署**：把微调或导入的模型发布为独立、资源专享的在线推理服务，再通过标准 API 调用。 - -## 场景一：模型调优 - -按模态划分，不同模型支持的训练方式差异明显： - -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异（如 Qwen3-32B、Qwen2.5 系列支持全部 5 种，部分新模型仅支持 `sft`）。 -- **视觉理解（千问 VL）**：支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像/视频生成（万相）**：仅支持 SFT-LoRA 高效微调。 -- **语音合成（CosyVoice）**：仅支持 `efficient_sft`，且当前只能通过 API 发起，控制台暂不支持。 - -文本生成推荐按递进顺序组合：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下 `chosen` / `rejected` 回答对 | - -训练模式分**全参训练**与**高效训练（LoRA）**，两者费用相同；官方建议模型支持全参时优先全参（效果更好、性价比更高），LoRA 适合训练时间/成本敏感或数据集较小的场景。 - -## 场景二：模型压缩（量化） - -百炼的模型压缩特指**量化**，不涉及剪枝或蒸馏。它把全精度微调模型转为低精度版本，在保持能力前提下降低部署 MU 规格。以 qwen3.5-flash-2026-02-23 为例，压缩前 MU1*2（108 元/小时）、压缩后 MU8*1（47 元/小时），成本节省约 56%。 - -> **注意**：压缩不可逆，压缩后模型不支持继续微调或二次压缩；仅支持百炼平台微调产出的自定义模型。 - -## 场景三：模型部署 - -部署提供三种互斥的计费方式，创建后无法更改： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产。支持 PTU 长输入（部分模型最高 200K token）与前缀缓存折扣，超额自动转按量计费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占，支持部分预置模型与所有调优后模型。 -- **按 Token 使用量**：不使用不计费，仅支持 SFT 高效训练后的自定义模型，主要用于效果验证。 - -此外可通过**我的模型**从 OSS 导入本地训练的 **LoRA** 模型（不支持全参微调模型），rank 须为 8/16/32/64 之一，必需 `adapter_model.safetensors` 与 `adapter_config.json`，且不得修改 vocab 或 chat_template。 - -## 关键参数与配置 - -文本生成调优的常用超参： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议 3~5，>10000 建议 1~2。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型最大值；SFT **丢弃**超长数据，DPO **截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：不同文档默认学习率取值不一致（控制台面板显示 `3e-4`、API SFT 全参示例为 `1.6e-5`）。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频、CosyVoice 各有独立超参集（如万相 `max_steps`/`generation_type`，CosyVoice 分 `lm_*` 韵律与 `fm_*` 音色两组网络的参数，8 个子字段全部必填）。 - -## 数据格式要点 - -- **SFT（文本）**：ChatML，每行一个 `{"messages":[...]}`，所有 assistant 输出都会被训练；不支持 OpenAI 的 `name`/`weight`。 -- **SFT 思考模型**：仅训练最后的 assistant 输出，思考内容用 `` 标签包裹并保留前后换行。 -- **DPO**：ChatML 加 `chosen` / `rejected` 字段。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解/图生视频**：需按字段规范提供图片、视频路径或帧列表。 - -## 面向开发者的使用方式 - -- **控制台（推荐入门）**：模型调优页面创建任务 → 选训练方式与模型 → 配置训练集/验证集与 Checkpoint → 训练 → 部署 → 评测。 -- **API / 命令行**：统一四步流程——上传数据集（`POST /api/v1/files`）→ 创建调优任务 → 轮询任务状态与训练指标 → 部署为在线服务并调用推理端点。 - -若调优后评测效果不佳，最简单的改进办法是收集更多高质量数据继续训练；压缩免费期内可对同一模型尝试多个量化模板，用业务测试集验证后再上线。 - -## 关联主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md index 7e8ba43d..95adc3cb 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md @@ -1,45 +1,47 @@ -# 函数调用（Function Calling） +# 函数调用 -函数调用（Function Calling）是指大模型在推理过程中，根据用户输入、工具名称与工具描述判断是否需要调用外部工具（函数/API），并生成结构化的调用参数，再将工具返回结果并入上下文以生成最终回复的能力。它是弥补大模型在实时信息获取、精确计算、外部系统操作等原生局限的核心机制。 +函数调用（Function Calling）是百炼平台中模型主动识别用户意图、生成结构化工具调用请求，并交由外部系统执行的能力。它使大模型能突破自身知识与能力边界，安全、可控地接入实时搜索、代码执行、图像生成、OCR解析、GUI操作等外部服务，实现“思考→规划→调用→整合”的闭环推理。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼平台在多个层面暴露和使用函数调用能力： +函数调用并非单一接口，而是贯穿多类模型与交互范式的统一能力机制，具体体现为以下三种典型模式： -- **文本生成模型**：所有通用文本模型（如 `qwen3.7-plus`、`qwen3.6-flash`、`qwen3.7-max`）均支持 Function Calling。此外平台还提供免复杂配置的内置工具（联网搜索、代码解释器、网页抓取），开箱即用。 -- **API 接口层**：函数调用通过不同接口协议暴露，字段约定各有差异： - - **OpenAI 兼容 Chat Completions**：以 OpenAI 的 `tools` / `tool_calls` 字段约定描述工具与调用结果，迁移成本最低。 - - **OpenAI 兼容 Responses**：内置联网搜索、代码解释器、网页内容提取等工具，并自动管理对话历史，无需手动维护上下文。 - - **Anthropic 兼容 Messages**：以 tool use 形式支持工具调用，适合 Anthropic 生态应用。 - - **DashScope 原生接口**：功能与参数最完整，是使用平台全部工具能力时的首选。 -- **实时多模态（Qwen-Omni-Realtime）**：基于 WebSocket 的实时 API 同样支持工具调用（Function Calling）。模型触发调用后通过 `response.function_call_arguments.done` 服务端事件返回调用参数，客户端执行工具后再用 `conversation.item.create` 事件将结果回传给模型。 -- **应用构建（智能体 / 工作流）**： - - **智能体应用（Agent）/ Assistant API**：模型根据输入、工具名称与描述自主判断是否调用，动态选择并规划调用顺序；新版智能体（Agent 2.0）把知识库、MCP、插件统一为“工具”交由模型自主编排。 - - **工作流应用**：工具作为编排节点按预定义流程执行，调用顺序由用户编排而非模型规划。 -- **插件（Plug-in）**：调用插件的本质就是通过函数调用触发插件下的工具（如 `code_interpreter`、`calculator`、`quark_search` 等）。模型依据工具描述决定调用哪个工具。 +- **通用模型的自主工具调用**：`qwen3.7-plus`、`qwen3.6-flash`、`qwen3.5-omni-plus-realtime` 等主流模型在启用 `tools` 参数后，可基于用户输入自动决策是否调用、调用哪个工具、传入哪些参数，并返回标准化的 `tool_calls` 响应（含 `tool_id` 和 `arguments`）。该过程完全由模型内部推理完成，开发者只需提供工具定义（名称、描述、参数 schema），无需编写调度逻辑。 -## 关键参数与配置 +- **专用意图模型的显式决策**：`tongyi-intent-detect-v3` 是专为函数调用设计的轻量级模型，不生成自然语言回复，而是直接输出结构化意图标签（如 `"search"`、`"calculate"`）或完整工具调用指令（`INTENT_MODE` 模式）。适用于需强确定性、低延迟的路由/分发场景，常作为智能体前置网关。 -- **工具定义**：需为每个工具提供名称、功能描述与参数 Schema。工具描述质量直接影响模型判断是否/如何调用。 -- **模型能力要求**：应选用具备强工具调用能力的模型。插件调用目前支持 `qwen-turbo`、`qwen-plus`、`qwen-max`、`qwen-vl-max`、`qwen-vl-plus` 等；智能体推荐千问-Max 系列。 -- **思考模式**：可通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制），配合工具调用完成“规划-执行-反思”链路。 -- **ReAct 最大轮次**：智能体中取值 1-50，限制单次会话内工具调用的最大次数。 -- **实时 API 事件**：VAD/Manual 模式下通过 `session.update` 配置工具；调用参数由 `response.function_call_arguments.done` 返回，结果经 `conversation.item.create` 回传。 -- **调用约束**：每个智能体应用最多可添加 10 个工具；旧版智能体的自定义插件有 5 秒超时限制。 +- **垂直领域模型的内嵌工具链**：`gui-plus-2026-02-26` 通过 `computer_use` 工具实现 GUI 自动化；`qwen3.5-ocr` 在图文混合输入下自动触发结构化解析；`qwen-deep-research` 在研究流程中隐式调用检索与报告生成子模块。这些模型将函数调用深度集成至业务逻辑，对外表现为端到端能力，而非显式 `tool_calls` 字段。 -## 开发者实践建议 +> ⚠️ 注意：并非所有模型均支持函数调用。例如 `qwen-long`（10M上下文）明确不支持；`qwen3.7-max` 不支持结构化输出，因而无法返回合规的 `tool_calls`；`qwen-omni-turbo-realtime` 系列虽支持 `tools` 参数，但文档未确认其完整调用流程，建议优先选用 `qwen3.5-omni-realtime` 系列。 -- 迁移已有 OpenAI/Anthropic 应用时，优先选用对应生态的兼容接口，注意核对不同接口的工具字段映射差异。 -- 需要联网搜索、代码解释器等内置工具时，使用 Responses 接口而非普通 Chat Completions。 -- 需要最完整的工具能力与参数控制时，使用 DashScope 原生接口。 -- 编写清晰、无歧义的工具描述，是提升模型正确触发函数调用的关键。 +## 关键参数和配置 + +| 参数 | 类型 | 说明 | 必填 | 示例 | +|------|------|------|------|------| +| `tools` | `array` | 工具定义列表，每个元素包含 `tool_id`（字符串）、`description`（功能描述）、`parameters`（JSON Schema，定义必选/可选字段及类型） | 是（启用函数调用时） | `[{"tool_id": "calculator", "description": "执行数学计算", "parameters": {"type": "object", "properties": {"expression": {"type": "string"}}, "required": ["expression"]}}]` | +| `tool_choice` | `string` 或 `object` | 控制调用策略：
`"auto"`（默认，模型自主决定）
`"none"`（禁用调用）
`{"type": "function", "function": {"name": "xxx"}}`（强制指定工具） | 否 | `"auto"` | +| `enable_search` | `boolean` | **仅 `qwen3.5-omni-realtime` 系列支持**，启用内置联网搜索（与 `tools` 互斥） | 否 | `true` | +| `result_format` | `string` | 必须设为 `"message"`（推荐），确保响应中包含 `tool_calls` 字段；设为 `"text"` 将丢失结构化调用信息 | 是（推荐） | `"message"` | + +- **工具 ID 命名规范**：必须全局唯一、语义清晰（如 `quark_search`, `code_interpreter`），避免空格/特殊字符；官方插件 ID 可在控制台插件详情页复制。 +- **参数 Schema 要求**：`parameters` 必须为合法 JSON Schema，`required` 数组需准确声明必填字段；`Object` 类型参数仅支持 `POST` 请求，`GET` 请求中禁止使用。 +- **响应解析要点**：成功调用后，模型响应 `message` 中 `role` 为 `"assistant"`，`content` 为空或为中间思考，`tool_calls` 数组包含调用详情；后续需开发者自行执行工具并以 `tool_result` 角色提交结果，继续对话。 + +## 面向开发者，简洁实用 + +- ✅ **快速验证**：用 `qwen3.7-plus` + `calculator` 工具，发送 `"123 * 456 = ?"`，观察是否返回 `tool_calls`。 +- ✅ **调试技巧**：若模型未触发调用，检查 `description` 是否足够清晰、`parameters.required` 是否遗漏关键字段、`messages` 中是否提供足够上下文。 +- ✅ **生产建议**： + - 对高可靠性场景（如金融计算），优先使用 `tongyi-intent-detect-v3` 做意图路由，再交由专用工具执行； + - 实时语音对话中，`qwen3.5-omni-plus-realtime` 支持流式 `tool_calls` 事件，可边听边规划； + - 自定义插件务必完成在线调试并发布为“已发布”状态，否则调用失败且错误码不直观（常见 `130040`）。 +- ❌ **避坑提醒**：`tools` 与 `enable_search` 不能同时启用；`qwen-long` 等超长上下文模型不支持该能力；流式响应（`stream=true`）中 `tool_calls` 仅在最终 chunk 返回，勿在中间 chunk 解析。 ## 关联主题页 -- [omni realtime api](../api/omni-realtime-api.md) - [model experience](../guides/model-experience.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [llm application](../guides/llm-application.md) +- [omni realtime api](../api/omni-realtime-api.md) - [plug in](../guides/plug-in.md) +- [more models](../api/more-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md deleted file mode 100644 index 3062222e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md +++ /dev/null @@ -1,87 +0,0 @@ -# 知识库 - -知识库是阿里云百炼平台基于 RAG（[检索增强生成](rag.md)）技术构建的私有数据管理能力，用于为大模型补充私有数据与最新信息，使应用能够准确回答特定领域问题。知识库仅支持在中国站华北2（北京）地域开通和使用，提供标准版与旗舰版两种规格，并配套日志监控、API、效果优化与计费体系。 - -## 核心能力 - -知识库对私有数据或文件进行语义检索，可找出语义相同或相近的内容，即使关键词匹配度极低甚至为零。检索结果可作为上下文喂给大模型生成回答，也可仅作为检索能力单独使用。支持挂载到[智能体应用](agent-application.md)、工作流应用，或通过阿里云百炼 SDK、LlamaIndex、Spring AI Alibaba 等框架集成到外部应用。 - -知识库类型分为四类，单一知识库不支持同时选择多个类型： - -- 文档搜索：对非结构化文档做语义检索，细分为基础文档问答、图文并茂回复、视觉理解（富文本文档）、极速问答四种使用场景。 -- 数据查询：对结构化表格数据做检索。 -- 图片问答：基于 multimodal-embedding-v1（1024 维）对图片做检索。 -- 音视频搜索：对音频、视频内容按时间轴结构化检索。 - -## 在百炼平台中的使用场景 - -### [智能体应用](agent-application.md) - -在 Agent 2.0 架构中，知识库作为工具由智能体自主规划调用，与 MCP 等外部工具统一调度；在旧版（Agent 1.0）中，则先检索知识库再决策是否调用其他工具。可在应用配置页文档知识库右侧点击「+ 添加知识库」接入，并设置相似度阈值与权重，还可通过标签限定查询范围以提升准确性。开启「展示回答来源」后，回答会以角标形式展示知识来源与源文件/源网页地址。 - -### 工作流应用 - -将知识库节点拖入画布，配置输入变量（query）、选择固定知识库或动态引入、设置 TopK，用于在预定义流程中精确控制检索环节。 - -### 外部应用集成 - -- SDK：通过阿里云百炼 SDK 调用检索能力，子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](workspace.md)。 -- LlamaIndex（Python 3.9+）：将知识库部署在云端，使用默认智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活切分应改用本地知识库方案。 -- Spring AI Alibaba（Spring Boot 3.x，JDK 17+）：可调用百炼智能体/工作流应用并检索百炼知识库，应用集成推荐变量名 `DASHSCOPE_API_KEY`，知识库检索推荐 `AI_DASHSCOPE_API_KEY`。 -- 本地 RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。 - -### API 编排 - -通过百炼 OpenAPI（`bailian/2023-12-29`，ROA 风格）可程序化管理知识库全链路：申请上传租约 → 上传文件 → 添加文件到类目 → 轮询文件解析状态（INIT/PARSING/PARSE_SUCCESS）→ 初始化知识库 → 提交索引任务 → 轮询任务状态直至 COMPLETED。每[业务空间](workspace.md)最多创建 500 个类目，类目类型目前仅支持 `UNSTRUCTURED`。 - -## 关键参数与配置 - -### 规格与并发 - -| 规格 | 最高检索并发 | 平台存储空间 | 价格 | -| --- | --- | --- | --- | -| 标准版 | 1 QPS（固定，不可调） | ≤ 100 GB | 0.03 元/知识库/小时 | -| 旗舰版 | 50–10,000 QPS（可调，对应 1–200 RCU） | ≤ 9,999 GB | 0.2 元/RCU/小时 | - -RCU（Retrieval Compute Unit）是检索并发能力度量单位，1 RCU 约支撑最高 50 QPS，所需 RCU = 向上取整（检索峰值 QPS ÷ 50）。变配按发生时间分段计费，同一知识库 1 个自然日内最多变配 1 次。 - -### 向量与切片 - -- 向量模型：文档搜索、数据查询、音视频搜索类支持 text-embedding-v4、text-embedding-v3（均为 512 维）；图片问答类仅支持 multimodal-embedding-v1（1024 维）。向量维度不支持更改。 -- 文本切片长度上限：单个切片 6,000 [Token](token.md)；编辑切片（UpdateChunk）长度限制为 10–6,000 字符；删除切片（DeleteChunk）单次最多 10 个。 -- 召回文本切片数量：单次查询最多召回 20 个切片。 -- 切片方式：智能切分（保留语义完整性）或按长度切分。**知识库一旦创建，无法再配置 metadata 抽取，也无法更改文档切分 chunk**，需在创建时一次性规划好元数据与切片策略。 - -### 检索参数 - -- 相似度阈值：仅语义相似度高于此阈值的文本切片才会被召回。阈值过高（如 0.60）可能导致无召回结果或丢弃相关切片。 -- 召回片段数（K 值）：取值范围 1–20，调大可提升完整性但增加 [Token](token.md) 消耗；拼装后总长度超出大模型输入限制会被截断，并非越大越好。 -- 初步向量检索 TopK / 初步关键词检索 TopK：默认 50，取值范围 10–100，影响送入排序模型的切片数量与成本。 -- 权重：仅在**同类型知识库之间生效**，用于干预多知识库召回顺序。 - -### 解析方式 - -导入文件时可选择解析方式：电子文档解析（不支持插图与图表）、文档智能解析（提取插图文本与摘要）、大模型文档解析（支持对插图和图表提问，需配合选择模型）、Qwen VL 解析（仅图片，可传入 Prompt）、音视频解析（语音识别、视频帧提取、剧情解析按时间轴结构化对齐）。 - -## 效果优化 - -当出现召回不完整或内容不准确时，建议先建立评测基线（至少 100 组问题，覆盖事实型/比较型/教程型/分析型），再按 RAG 三阶段诊断改进： - -1. 建立索引：优化源文件排版（优先 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写（创建时开启，后续无法补开）。 -2. 检索召回：为文件添加标签过滤、配置元数据做结构化搜索、采用智能切分保留语义完整性、调整相似度阈值与召回片段数。 -3. 生成答案：更换为能力更强的商业模型（如通义千问 Max/Plus/QwQ）、优化提示词模板（限定输出、少样本提示、内容分隔标记且 `${documents}` 只出现一次）。 - -## 日志监控 - -检索日志由日志服务（SLS）承载，首次使用需授权角色 `AliyunServiceRoleForSFMAccessSLS` 并创建 LogStore。每条日志 topic 为 `log_dispatch`，包含 request_id、pipeline_id、workspace_id、latency、response_status_code、response_code 等字段。建议搭建调用量趋势、TopN 知识库排名、业务错误率与 HTTP 5xx 错误率等监控。 - -## 关联主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [frameworks](../api/frameworks.md) -- [llm application](../guides/llm-application.md) -- [application use cases](../guides/application-use-cases.md) -- [data connection overview](../guides/data-connection-overview.md) -- [application component api reference](../api/application-component-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md deleted file mode 100644 index f1f996e6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md +++ /dev/null @@ -1,45 +0,0 @@ -# 长上下文 - -长上下文指模型在单次调用中能够接收并处理的输入 token 上限（部分模型还包含输出）。上下文窗口越大，单次请求可携带的对话历史、文档、图像、视频等数据越多，能减少多轮拼接与外部检索的复杂度。 - -## 在百炼平台中的使用场景 - -- **文本生成**：Qwen3.7-max / Qwen3.7-plus / Qwen3.6-flash 以及 deepseek-v4 系列均提供 1M token 上下文窗口，可在单轮内喂入长文档、长对话历史或多文件做总结、问答与代码调试。 -- **视觉理解**：Qwen3.7-plus 等模型支持 1M 上下文，可接收最长 2 小时视频（2GB）或大段图文混合输入，进行视频内容分析与 OCR。 -- **PTU 部署**：预置吞吐部署支持长输入（部分模型最高 200K token），通过阶梯容量系数折算 TPM，超出额度或输入超过模型上限时自动转为按量计费，业务不中断。 -- **对话历史管理**：OpenAI 兼容 Responses 接口由平台自动维护对话历史，无需在请求中手动拼接 messages；其他接口需由调用方维护上下文长度与轮次，避免超出模型上下文窗口。 - -## 关键参数与配置 - -- **上下文长度**：按模型不同，常见档位为 32K / 64K / 128K / 200K / 1M。选型时确认目标模型在对应兼容接口下的实际上下文上限。 -- **PTU 长输入阶梯系数**：超出 32K 的输入按更高系数折算 TPM。以 glm-5.1 为例，`[0, 32K)` 系数 1.0，`[32K, 200K]` 输入系数 1.33 / 输出 1.17。 -- **前缀缓存折扣**：命中缓存的输入 token 按模型对应折扣折算容量（如 glm-5.1 为 0.2，deepseek-v4-pro 为 0.08）。 -- **自动溢出阈值**：千问 128K、DeepSeek 64K 为模型上限阈值，超过即转按量计费，响应头返回 `x-dashscope-ptu-overflow: true`。 -- **图像 token 计算**：视觉模型每张图片 token 数按 `h x w / (32 x 32) + 2` 估算，叠加在上下文总额内。 - -## 容量与额度评估 - -长输入场景下建议先用控制台的 PTU 容量计算器评估额度：根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推算输入 TPM 和输出 TPM，避免意外转为按量计费。PTU 部署的响应包含以下与额度相关的字段： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。OpenAI Chat 兼容为 `usage.prompt_tokens_details.cached_tokens`；OpenAI Responses 为 `usage.input_tokens_details.cached_tokens`；Anthropic 兼容暂不返回。 - -## 限制与注意事项 - -- **功能完整度**：兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全采样参数与插件能力，建议改用 DashScope 原生接口。 -- **上下文窗口 ≠ 输出上限**：1M 上下文主要指输入侧，单次输出长度仍受模型自身限制，长输入场景需评估输出截断风险。 -- **历史轮次管理**：迁移到不自动管理历史的接口时，需自行裁剪历史消息，避免累积超出窗口导致请求失败或尾部被截断。 -- **跨接口迁移**：从 OpenAI / Anthropic 迁移时应先确认目标 Qwen 模型在对应兼容接口下是否支持所需上下文长度与参数（如 `temperature`、`tools`、`stream` 等），再决定接口选型。 -- **地域差异**：各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，长上下文模型的可用性可能与地域相关，部署类操作仅适用于华北2（北京）地域。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [more about models](../api/more-about-models.md) -- [model experience](../guides/model-experience.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md new file mode 100644 index 00000000..f41f9f1e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md @@ -0,0 +1,48 @@ +# 长期记忆 + +长期记忆是百炼平台提供的结构化、持久化上下文管理能力，用于跨会话、跨对话地存储和检索用户意图、偏好、事件、计划等语义化信息，突破大模型单次推理的上下文窗口限制，支撑个性化、连贯的智能体体验。 + +## 在百炼平台的不同场景中如何使用 + +- **智能体（Agent）应用**：当前新版智能体（Agent 2.0）**不原生支持长期记忆**，仅提供短期记忆（最多30轮对话历史）。如需长期记忆能力，需通过 SDK 或 API 主动调用 `AddMemory` / `SearchMemory`，在 Agent 工具链中集成记忆读写逻辑（例如：在 `system_prompt` 中提示“请先检索用户历史偏好”，再调用 `SearchMemory` 工具注入上下文）。 + +- **工作流（Workflow）应用**：可通过节点间传递 `user_id`，在关键节点（如“初始化”或“响应生成”前）调用 `SearchMemory` 注入个性化上下文；也可在用户输入处理节点后调用 `AddMemory` 持久化新信息。配合会话变量（`historyList`）实现短期+长期双层记忆协同。 + +- **高代码应用**：完全由开发者自主控制。推荐在 Python 应用中使用 `agentscope-runtime` SDK 封装的异步工具类（`AddMemory`, `SearchMemory`, `ListMemory`, `DeleteMemory`），结合业务逻辑实现记忆生命周期管理（如注册回调、触发更新、设置过期策略）。 + +- **OpenClaw 等框架集成**：通过官方插件 `modelstudio-memory-for-openclaw` 开箱启用全自动机制：`autoCapture`（对话结束自动提取并写入）、`autoRecall`（对话开始前按 `user_id` 自动检索 Top-K 记忆），并暴露 `memory_search` / `memory_store` 工具供 Agent 主动调用。 + +- **用户画像构建**：需预先调用 `CreateProfileSchema` 定义结构化字段（如 `"age": "整数，用户年龄"`），并在 `AddMemory` 请求中传入 `profile_schema` ID，平台将自动从对话中抽取并聚合属性，后续可通过 `GetUserProfile` 获取完整画像。 + +## 关键参数和配置 + +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `user_id` | string | 是 | 记忆归属主键（≤64 字符），用于严格隔离不同用户数据空间，所有接口均需传入。 | +| `memory_library_id` | string | 否 | 目标记忆库 ID（≤32 字符）；不传则使用账号默认记忆库（不可删除）。 | +| `project_id` | string | 否 | 记忆片段提取规则 ID；不传则使用对应记忆库的默认规则（控制台可配置有效期：7/30/180 天或永不过期）。 | +| `profile_schema` | string | 否 | 用户画像 Schema ID；仅当需触发结构化属性抽取时必填。 | +| `messages` / `custom_content` | array / string | 二选一 | `messages`: 对话数组（最多50条），用于自动提取语义记忆；`custom_content`: 最多512字符纯文本，绕过提取直接写入。 | +| `meta_data` | object | 否 | 自定义键值对（如 `{"category": "preference", "source": "onboarding"}`），支持后续按字段过滤或业务标记。 | +| `top_k` | integer | 否（`SearchMemory` 默认10，OpenClaw插件默认5） | 检索返回的最大记忆条数（1–100）。 | +| `min_score` | double | 否（默认0.3，控制台推荐0.5–0.7） | 向量相似度阈值 [0,1]，低于此值的结果被过滤。 | +| `expire_time` | integer (Unix timestamp) | 否 | 秒级时间戳，显式指定记忆过期时间；优先级高于 `project_id` 规则中的默认有效期。 | + +> ⚠️ 注意：`UpdateMemory` 仅更新内容与 `meta_data`，不改变向量索引时间点；`timestamp` 元字段为秒级 Unix 时间戳（非毫秒）。 + +## 面向开发者的实用建议 + +- **首选 SDK**：安装 `pip install agentscope-runtime>=1.1.5`，直接使用 `AddMemory`, `SearchMemory`, `ListMemory`, `DeleteMemory` 异步工具类，避免手动构造 HTTP 请求与认证头。 +- **调试先行**：首次集成务必用 cURL 验证基础流程（如 `curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add -H "Authorization: Bearer $DASHSCOPE_API_KEY" -d '{"user_id":"u123","messages":[...]}')`，再迁移到 SDK。 +- **限流应对**：阿里云账号级总限流 3000 QPM（`AddMemory` ≤120 QPM，`SearchMemory` ≤300 QPM），超限返回 `429`，需实现指数退避重试。 +- **时效性管理**：长期记忆**无自动失效机制**，业务侧必须主动维护生命周期 —— 建议在关键业务节点（如用户注销、偏好变更）调用 `DeleteMemory` 或设置 `expire_time`。 +- **错误排查**：所有 API 响应含 `request_id`，是定位问题的关键标识；结合控制台「API 调用日志」与文档中的错误码表快速诊断。 + +## 关联主题页 + +- [long term memory new](../api/long-term-memory-new.md) +- [memory library overview](../guides/memory-library-overview.md) +- [llm application](../guides/llm-application.md) +- [managed agents](../guides/managed-agents.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md deleted file mode 100644 index 0f2719f1..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md +++ /dev/null @@ -1,92 +0,0 @@ -# 模型部署与高速推理 - -模型部署是将百炼平台的预置模型、微调模型或导入模型发布为独立、资源专享的在线推理服务的过程；高速推理则是在部署之上（或独立）为调用提供容量刚性兑付与更高输出速度的一组能力。两者共同解决生产环境对高并发、低延迟和确定性吞吐的需求。 - -## 在百炼平台的使用场景 - -模型生产的完整链路为：**模型调优（Fine-tuning）→ 模型压缩（可选量化）→ 模型部署 → 推理调用**。开发者可通过调优 API 定制专属模型，用模型压缩降低部署规格与成本，再通过部署 API 发布为在线服务，最后调用端点进行推理。 - -围绕"部署 + 推理"，平台提供多种资源与容量形态，按业务诉求选型： - -- **需要私有推理服务、资源独占**：用模型部署（PTU / 模型单元 / 按 Token）创建专属服务。 -- **只需锁定容量、抵御公共限流**：用 **TPM 预留**，为指定模型预留专属吞吐量。 -- **只需更快出字、计费不变**：用**快速模式（Fast mode）**，把输出速度提升到标准 API 的 1.5~2 倍。 -- **需要降低部署成本**：先对微调模型做**模型压缩（量化）**，再部署。 - -## 部署的三种计费方式 - -计费方式在服务创建时选定，创建后不可更改（需下线后重新部署）： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常较按 Token 提升约 1.5~2.0 倍。适合流量稳定、需并发/延迟确定性的高负载场景。支持按小时后付费与按天预付费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占、性能可自定义，支持 PD 分离（拆分 Prefill/Decode 降低首 Token 延迟）。适合私有微调模型与长时任务。支持按分钟后付费与按月预付费。 -- **按 Token 使用量**：仅对 SFT 高效训练（LoRA）后的自定义模型开放，主要用于调优效果验证。 - -> 预付费无法提前退费，首月内提前退订按单价 1.2 倍计费；PTU 超出购买吞吐或输入超模型上限时，自动切换为按量付费（响应头 `x-dashscope-ptu-overflow:true`）。 - -## PTU 长输入与前缀缓存 - -- **长输入阶梯系数**：部分模型对超过 32K 的输入按更高系数折算 TPM（如 glm-5.1 在 [32K, 200K] 区间输入 1.33 / 输出 1.17），部分模型无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 低至 0.08），显著降低多轮对话与重复前缀场景成本。 -- **额度识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度）、`provisioned_tokens`（折算后实际消耗）、`cached_tokens`（缓存命中数）。这些字段在 OpenAI Chat 兼容、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下的 JSON 路径不同，需分别读取；Anthropic 兼容格式暂不返回 `cached_tokens`。 - -建议创建或扩容前用控制台**容量计算器**（依据 RPM、平均输入/输出长度、预估缓存命中率）估算所需 TPM，避免额度不足产生意外按量费用。 - -## 模型导入（LoRA） - -部署自训练模型前需从 OSS 导入 LoRA 微调版本，核心要求： - -- 仅支持 **LoRA**，不支持全参微调；必需文件 `adapter_model.safetensors` 与 `adapter_config.json`。 -- **rank** 必须为 8/16/32/64 之一，且同模型各 LoRA 层 rank 一致。 -- 不得新增 token、修改 vocab 或 chat_template，须与开源基础模型完全一致；VL 模型必须冻结 VIT（含 `visual` 权重则无法导入）。 -- **OSS 前提**：Bucket 需添加 `bailian-datahub-access` 标签（值 `read`）、文件须放子目录、不支持归档类存储；首次导入需完成服务关联角色授权。 - -支持基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列。 - -## 使用 API / 命令行部署 - -部署接口统一为 `POST https://dashscope.aliyuncs.com/api/v1/deployments`（仅华北2·北京，需先配置 `DASHSCOPE_API_KEY`），通过 `plan` 字段区分计费方式： - -- **PTU**：`"plan": "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`"plan": "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit` 等。 -- **按 Token（LoRA）**：`"plan": "lora"`，`capacity` 必填但无效，扩缩容需在控制台申请。 - -部署自定义模型时 `model_name` 使用**模型 ID**（在"我的模型"页面获取）。查询状态用 `GET /deployments/{id}`。 - -## 模型压缩（量化）降本 - -模型压缩特指量化（不含剪枝/蒸馏），将全精度微调模型转为低精度版本以降低部署所需 MU 规格。例如 qwen3.5-flash 微调模型压缩前 MU1*2（108 元/小时），压缩后 MU8*1（47 元/小时），成本节省约 56%。压缩不可逆，压缩后不支持继续微调或二次压缩，仅华北2（北京）可用。量化模板中 MU 编号越大规格越小、成本越低但精度损失可能越大；校准数据应选择与推理场景语义相近的数据集。 - -## 高速推理能力 - -### TPM 预留：锁定专属容量 - -- **专属模型 code**：创建预留后系统生成专属 `model` code，需将请求中的 `model` 替换为该 code 才命中预留容量。 -- **超额不中断**：超出预留自动降级为公共池按量计费，无需改代码，可在详情页查看超额降级统计。 -- **计费**：按 kTPM（1 kTPM = 1000 Tokens/分钟）预付费，部署成功即计费。 -- **管理**：支持在线扩缩容；退订不可恢复，缩容/退订退费按已用部分 1.5 倍系数结算；到期后 2 小时内可调用，2~14 小时转已停止（可续费），14 小时后删除。 - -### 快速模式（Fast mode）：更高输出速度 - -当前处于 **preview 阶段**，面向 AI 编程助手、Agent 多步推理、实时对话等对输出速度敏感的场景： - -- **高速输出**：TPS 达标准 API 的 1.5~2 倍（约 80~100 TPS）。 -- **计费不变**：仍按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 不立即限流，请求进入排队队列（区别于 TPM 预留的"超额降级按量"）。 -- **使用方式**：将 `model` 指定为支持快速模式的 model ID（如 `glm-5.2-fast-preview`），域名格式为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。`glm-5.2` 默认返回 `reasoning_content`，[流式输出](streaming.md)时思考与回答分别通过 `delta.reasoning_content` 与 `delta.content` 推送。 - -## 选型建议 - -- 追求成本优化且流量波动大：**按量付费**。 -- 流量可预估、不能接受公共限流：**TPM 预留**。 -- 高吞吐 + 高性能确定性：**PTU 专属部署**。 -- 只想更快出字、代码改动最小：**快速模式**。 -- 私有微调模型且要降本：先**模型压缩**再按 **MU** 部署。 - -## 关联主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md deleted file mode 100644 index 22db84ab..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md +++ /dev/null @@ -1,83 +0,0 @@ -# 模型选型 - -模型选型是指在百炼平台提供的自研千问与第三方大模型矩阵中，按业务场景、能力档位、成本与延迟需求匹配最合适的模型，并选定对应地域、接入域名与调用方式的过程。选型结果直接决定效果、吞吐、合规性与单位成本。 - -## 选型总体思路 - -1. **先定场景**：明确任务是文本生成、视觉理解、视频/图像/3D 生成、语音、向量检索，还是 RAG、调优、深度研究等专用领域。 -2. **再定能力档**：在同类模型中按「高能力档 / 平衡档 / 轻量低成本档」对位选择，效果稳定后再用低档降本。 -3. **最后定接入**：根据合规与并发要求选择地域、服务部署范围与接入域名，并确认调用协议（OpenAI 兼容 / DashScope SDK / WebSocket / 异步任务）。 - -## 文本生成场景选型 - -对话、抽取、改写、Agent 等文本任务推荐从 `qwen3.7-plus` 入手——能力、速度与成本均衡，支持 1M 上下文、完整 Function Calling 与内置工具；需要最强推理选 `qwen3.7-max`；快速响应的简单任务用 `qwen3.6-flash` 降本。 - -从闭源或第三方模型迁移时按能力档对位： - -| 能力档 | 千问 | 第三方对位 | -| --- | --- | --- | -| 高能力档 | `qwen3.7-max` | — | -| 平衡档 | `qwen3.7-plus` | `deepseek-v4-pro` / `glm-5.2` | -| 轻量低成本档 | `qwen3.6-flash` | `deepseek-v4-flash` / `MiniMax-M2.5` | - -> 注意：第三方模型（`deepseek-v4-pro`、`kimi-k2.7-code`、`MiniMax-M2.5`、`mimo-v2.5-pro` 等）通常不支持内置工具与结构化输出，迁移时需评估能力差异。 - -## [多模态](multimodal.md)场景选型 - -| 模态 | 推荐入口 | 备选 / 降本 | -| --- | --- | --- | -| 视觉理解 / OCR | `qwen3.7-plus`（1M 上下文、2h 视频） | `qwen3.6-flash` 降本；`qwen-vl-ocr` 专做文档/表格/手写提取 | -| 图像生成 | `wan2.7-image-pro` / `qwen-image-2.0-pro` | `z-image-turbo` | -| 视频生成 | `happyhorse-1.1-t2v`（有声 1080P） | `wan2.7-t2v/i2v/r2v` 系列（自定义音频、首尾帧、角色一致性） | -| 3D 生成 | `Tripo/Tripo-H3.1` / `Tripo/Tripo-P1.0` | 仅北京地域，异步任务 | -| 语音合成（TTS） | `cosyvoice-v3.5-plus` | `qwen3-tts-instruct-flash` | -| 语音识别（ASR） | `fun-asr-realtime` | `qwen3.5-omni-plus-realtime` | -| 语音转语音 / 全模态 | `qwen3.5-omni-plus-realtime` | `qwen3.5-livetranslate-flash-realtime` | -| 音乐生成 | `fun-music-v1` | 邀测，仅北京 | -| 向量与重排序 | `text-embedding-v4` / `qwen3-rerank` | `qwen3-vl-embedding`（[多模态](multimodal.md)向量） | - -视频、图像、3D、音乐等生成类模型统一走异步任务模式；语音类多走 WebSocket / HTTP；向量与重排序走 HTTP。 - -## 专用领域模型选型 - -除通用对话模型外，百炼提供面向特定任务的专用模型，选型时需关注地域与协议差异： - -- **法律**：`farui-plus`，法律咨询/文书生成，DashScope SDK。 -- **意图理解**：`tongyi-intent-detect-v3`，同时输出意图与[函数调用](function-calling.md)信息。 -- **深度研究**：`qwen-deep-research`，两阶段（反问确认 + 深入研究），**仅 Python DashScope SDK、仅华北2（北京）**，不支持 OpenAI 兼容接口。 -- **翻译**：`qwen-mt-plus`，支持术语干预、翻译记忆、领域提示。 -- **OCR 与结构化抽取**：`qwen3.5-ocr`，可设 `min_pixels` / `max_pixels` 控制图像分辨率与 token 消耗。 -- **界面交互**：`gui-plus-2026-02-26`，通过 `computer_use` 工具操控桌面 GUI，仅北京。 - -## 选型时的关键参数与能力差异 - -- **思考模式**：通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制深度），Qwen3 及以上支持，多为可按请求切换的混合模式。 -- **Function Calling 与内置工具**：自定义工具全系列通用模型支持；内置工具（联网搜索、代码解释器、网页抓取）仅部分模型支持，如 `qwen3.7-plus`、`qwen3.6-flash`、`glm-5.2`、`mimo-v2.5-pro`。 -- **结构化输出**：非思考模式下返回合法 JSON，用于字段抽取；第三方模型多不支持。 -- **上下文与[多模态](multimodal.md)限制**：视觉模型单张图片最高 1600 万像素，token 计算 `h × w / (32 × 32) + 2`；视频时长上限因模型而异（2 小时 / 2GB 或 1 小时 / 2GB）。 -- **[流式输出](streaming-output.md)**：`stream` 参数控制；`qwen-deep-research` 反问阶段必须设为 `true`。 - -## 地域、部署范围与接入域名 - -地域决定接入点与数据存储位置，服务部署范围决定推理执行位置，接入域名决定并发承载与隔离性。三者共同影响可选模型与稳定性。 - -- **地域**：华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）、日本（东京）。各地域模型列表、API Key、接入域名独立，**不能跨地域混用**。 -- **服务部署范围**：无合规需求选「全球」（资源池更大）；有合规需求选特定地理边界（中国内地/美国/欧盟/日本/国际）。北京、新加坡各仅支持一种范围，无需选择。 -- **接入域名**：生产环境推荐专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（99.9% SLA、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC）；共享域名 `dashscope.aliyuncs.com` 兼容存量；试用域名仅用于快速体验。 -- **模型后缀**：美国地域使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理，不带后缀默认全球推理。 - -## 选型实践建议 - -- 先在控制台「模型体验」中迭代 Prompt，效果稳定后再固化到代码；切换模型后应回归测试。 -- 穷尽 Prompt 工程与 RAG 方案后再投入模型调优（SFT/CPT/DPO），性价比最高。 -- 低延迟批量场景可使用批量推理降本；遇 429 限流时用指数退避 + 客户端令牌桶平滑流量。 -- 切换模型前关注上下文长度、[函数调用](function-calling.md)、[流式输出](streaming-output.md)、[计费](billing.md)方式与地域支持的差异。 - -## 关联主题页 - -- [model experience](../guides/model-experience.md) -- [get started with models](../guides/get-started-with-models.md) -- [use cases](../guides/use-cases.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md new file mode 100644 index 00000000..0af3e708 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md @@ -0,0 +1,80 @@ +# 多模态能力 + +多模态能力指百炼平台模型对文本、图像、视频、音频等多种数据类型进行联合理解、生成与推理的能力，支持跨模态信息融合与协同处理，是构建智能体、内容创作、工业质检等复杂AI应用的核心基础。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +多模态能力并非单一模型特性，而是贯穿多个能力域的底层技术范式，在以下典型场景中体现为具体能力组合： + +- **视觉理解与推理**：`qwen3.7-plus`、`qwen3.5-omni-plus` 等全模态大模型可同时接收文本指令 + 多张图像/视频片段，执行OCR识别、图表解析、视频事件定位、结构化输出（如JSON）及Function Calling（例如调用天气API后结合截图分析出行建议）。单请求最多支持2048张图片或64段视频，输入按统一Token规则计费（图像Token ≈ h×w/(32×32)+2）。 + +- **音视频端到端处理**：`qwen3.5-omni-plus-realtime` 支持语音输入→文本理解→工具调用→语音合成全流程，无需拆解ASR/TTS模块；S2S模型可直接处理带背景音的语音流，并在响应中保留语调、节奏等声学特征。 + +- **生成类多模态协同**： + - 图像生成中，`wan2.7-image-pro` 接收文本+参考图+风格图三重输入，实现精准可控的图文混合生成； + - 视频生成中，`wan2.7-t2v-2026-06-12` 支持“[prompt](../guides/prompt.md) + 自定义音频文件注入”，实现音画同步生成； + - 3D生成中，`Tripo/Tripo-H3.1` 允许文生3D、单图生3D或多图（前/左/后/右）联合重建，输入模态决定几何重建精度与纹理生成策略。 + +- **跨模态检索与重排序**：`qwen3-rerank` 可对图文混合结果集（如含标题、缩略图、描述的搜索项）进行联合语义重排序；`text-embedding-v4` 虽为文本模型，但其向量空间经多模态对齐训练，可与图像Embedding模型（如`qwen-vl-embed`）共用相似度计算，支撑跨模态检索。 + +> ⚠️ 注意：并非所有模型均具备完整多模态能力。例如 `qwen-long`（10M上下文）专注长文本处理，**不支持图像/视频输入**；`qwen3.7-max` **不支持结构化输出与Function Calling**，即使输入含多模态数据，也仅作单向理解，无法触发工具链。选型时请以[model experience](model-experience.md)中各模型的能力矩阵为准。 + +## 关键参数和配置 + +多模态能力的启用与控制依赖以下关键参数（均置于`parameters`对象内，部分需配合特定请求头）： + +| 参数 | 类型 | 说明 | 典型值/约束 | 适用模型示例 | +|------|------|------|-------------|--------------| +| `max_image_count` / `max_video_count` | integer | 单请求最大媒体数量 | `qwen3.7-plus`: 2048 / 64；`qwen3.5-omni-plus`: 256 / 512 | 全模态理解模型 | +| `enable_thinking` | boolean | 启用分步推理模式（Chain-of-Thought），提升复杂多模态任务准确率 | `true` / `false`（默认`false`） | `qwen3.7-plus`, `qwen3.6-flash` | +| `tool_choice` | string / object | 控制工具调用策略（自动/指定/禁用），影响多模态意图识别后的动作决策 | `"auto"`, `"required"`, `{"type": "function", "function": {"name": "get_weather"}}` | 支持Function Calling的全模态模型 | +| `X-DashScope-Async` | request header | **强制异步调用标识**，所有耗时多模态生成任务（视频/3D）必须设置为`"enable"` | `"enable"`（必填） | `happyhorse-1.1-t2v`, `Tripo/Tripo-H3.1` | +| `audio` | boolean | 视频生成中是否启用音频轨道合成 | `true`（默认）/ `false` | `wan2.7-t2v-2026-06-12`, `happyhorse-1.1-t2v` | +| `texture_quality` / `geometry_quality` | string | 3D生成中贴图精细度与网格面数控制，直接影响多图输入的重建保真度 | `"standard"` / `"detailed"`；`"standard"` / `"ultra"` | `Tripo/Tripo-H3.1` | + +> ✅ 实用提示： +> - 多模态输入必须通过`input`字段统一组织（非`messages`），格式为`{"messages": [...]}`（文本为主）或`{"media": [...], "prompt": "..."}`（媒体为主）； +> - 混合输入时，**图像/视频URL必须为公网可访问的HTTPS链接**，且需提前校验可用性（3D生成要求JPEG/PNG，分辨率20–6000像素）； +> - 所有异步多模态任务（视频/3D）的`task_id`有效期严格为24小时，结果URL（如`pbr_model_url`）仅保留2小时，请及时下载。 + +## 面向开发者，简洁实用 + +- **快速验证**：用`curl`测试`qwen3.7-plus`的图文理解能力： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "qwen3.7-plus", + "input": { + "messages": [ + { + "role": "user", + "content": [ + {"type": "text", "text": "这张图里有哪些商品？价格分别是多少？"}, + {"type": "image_url", "image_url": {"url": "https://example.com/product.jpg"}} + ] + } + ] + }, + "parameters": {"max_output_tokens": 1024} + }' + ``` + +- **避坑指南**： + - ❌ 不要对`qwen-long`传入图片——会返回`400 Bad Request`； + - ❌ 不要省略`X-DashScope-Async: enable`调用视频/3D接口——会报错`current user api does not support synchronous calls`； + - ✅ 优先使用业务空间专属域名（如`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），多模态请求延迟降低30%+； + - ✅ 生产环境务必显式设置`max_output_tokens`，避免长输出导致超时或计费激增。 + +多模态能力的本质是“统一接口、混合输入、协同输出”。开发者只需关注业务需求的数据组合（文+图？音+视？文+图+3D？），平台将自动调度最适配的模型与计算资源——你负责定义“做什么”，我们负责实现“怎么做”。 + +## 关联主题页 + +- [model experience](../guides/model-experience.md) +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) +- [qwen api reference](../api/qwen-api-reference.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md deleted file mode 100644 index d238b34c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md +++ /dev/null @@ -1,45 +0,0 @@ -# 多模态能力 - -多模态能力指模型同时理解或生成文本、图像、音频、视频、3D 等多种模态内容的能力。在百炼平台上，它既体现为「输入多模态」（如图文混合输入、音视频实时对话），也体现为「输出多模态」（如文生图、文生视频、语音合成、3D 资产生成）。 - -## 在百炼平台的主要场景 - -百炼按模态与任务把能力拆分到多个方向，开发者可按需组合： - -- **实时音视频对话**：Qwen-Omni-Realtime 系列通过 WebSocket 提供低延迟的语音输入/输出、图像输入、语音活动检测（VAD）、工具调用与联网搜索，适用于智能客服、语音助手等实时交互场景。 -- **图像生成与编辑**：覆盖文生图、图生图、局部重绘、扩图、背景生成、虚拟模特、AI 试衣、创意海报等，涉及千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等模型家族。 -- **视频生成与编辑**：聚合万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及人像驱动模型，支持文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑与数字人。 -- **3D 资产生成**：基于 Tripo 模型支持文生 3D、单图生 3D、多图生 3D，产出带 PBR 材质或无贴图的 GLB 模型。 -- **视觉理解与 OCR**：以 Qwen 旗舰多模态模型理解图像与长视频（最长约 2 小时），并提供专优的 OCR/文档提取能力。 -- **语音合成 / 识别 / 语音转语音**：TTS、ASR、声音复刻/设计、S2S 实时对话与同传翻译，以及音乐生成。 - -## 输入与协议 - -- **实时流式（WebSocket）**：延迟最低，适合实时交互。Omni-Realtime 通过 `input_audio_buffer.append`（PCM 音频，Base64）、`input_image_buffer.append`（JPG/JPEG，Base64）等事件送入多模态素材，`session.update` 的 `modalities` 控制输出模态（`["text"]` 或 `["text","audio"]`）。 -- **HTTP 同步**：新一代图像模型（如 `wan2.6-image`、`wan2.7-image`、`z-image-turbo`）支持一次请求返回结果，路径为 `.../aigc/multimodal-generation/generation`，请求体用 `messages` 结构，`content` 内混排 `text` 与 `image`。 -- **HTTP 异步**：图像、视频、3D 生成等耗时任务（约 1-5 分钟）统一采用「创建任务拿 `task_id` → 轮询查询」两步流程，创建时必须携带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。`task_id` 有效期 24 小时，切勿重复创建，轮询即可。 - -## 关键参数与配置 - -不同模态的请求体大多由 `model`、`input`、`parameters` 三部分组成： - -- **实时对话**（`session.update`）：`modalities`、`voice`（音色，因模型而异）、`input_audio_format` / `output_audio_format`（仅 `pcm`，输入 16kHz、输出 24kHz）、`turn_detection.type`（`server_vad` / `semantic_vad`）及 `threshold`、`silence_duration_ms` 等。 -- **图像生成**：`input.prompt` / `negative_prompt`（或新协议 `messages`），图像编辑用 `images` / `image_url` / `mask_image_url`；`parameters` 含 `size`（如 `1024*1024`、`1K`/`2K`/`4K`）、`n`、`aspect_ratio`、`watermark`、`prompt_extend` 等。 -- **视频生成**：`input.prompt` 描述画面镜头，`input.media` 承载 `first_frame` / `last_frame` / `reference_image` / `video` 等素材；`parameters` 含 `resolution`（`480P`/`720P`/`1080P`）、`duration`、`ratio` 等。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥；`parameters` 含 `texture_quality`、`geometry_quality`、`pbr`、`texture`。 - -## 注意事项 - -- **地域隔离**：模型、Endpoint URL 与 API Key 必须属于同一地域，华北2（北京）、新加坡、美国（弗吉尼亚）等地域各自独立、不可混用；部分能力（如 Tripo 3D、Fun-Music）仅在特定地域可用。推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）以获得更好性能与稳定性。 -- **产物有效期**：图像结果 URL 有效期 24 小时，3D 模型下载链接仅 2 小时，务必及时下载。 -- **计费**：图像等仅对成功生成的输出计费，输入与失败任务不计费；具体计费、上下文窗口等实时参数以模型广场为准。 - -## 关联主题页 - -- [omni realtime api](../api/omni-realtime-api.md) -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) -- [model experience](../guides/model-experience.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md new file mode 100644 index 00000000..e6a43738 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md @@ -0,0 +1,82 @@ +# OpenAI 兼容接口 + +OpenAI 兼容接口是百炼平台提供的一组标准化 API 协议层，严格遵循 OpenAI REST API 的路径、请求/响应结构、参数命名与语义规范（如 `/v1/chat/completions`），使开发者能直接复用 OpenAI SDK（如 `openai>=1.0.0`）或现有代码逻辑调用千问（Qwen）及第三方模型，实现零改造迁移。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +OpenAI 兼容接口不是单一接口，而是一套按能力分层的协议集合，覆盖多种模型类型与任务场景： + +- **通用对话生成**：通过 `Chat Completions` 接口（`POST /compatible-mode/v1/chat/completions`）调用 `qwen3.7-plus`、`deepseek-r1`、`kimi-k2.7-code` 等文本/代码模型，支持标准 `messages` 格式、流式响应（`stream=true`）和基础采样参数（`temperature`, `top_p`, `max_tokens`）。 + +- **智能体增强能力**：`Responses API`（同路径但启用特定模型如 `qwen3.7-max`）在兼容基础上内置联网搜索、网页提取、代码解释器等工具链，自动处理工具调用循环，无需客户端手动解析 `tool_calls` 并重发 `tool_result`。 + +- **多模态理解**：`Vision` 接口（`/v1/chat/completions`）支持 OpenAI 格式的 `image_url` 输入，兼容 `qwen-vl-plus`、`qwen3-vl-plus` 等视觉语言模型，可混合文本与图像消息。 + +- **向量化与排序**：`Embedding`（`/v1/embeddings`）和 `Rerank`（`/v1/rerank`）接口完全对齐 OpenAI Embedding/Rerank 规范，支持 `dimensions`、`input` 数组、`top_n` 等关键参数，无缝接入 RAG 流水线。 + +- **批量异步处理**：`Batch` 接口（`/v1/batch`）提供 OpenAI 风格的异步提交能力，适用于高吞吐文本处理，单次支持 256K tokens 上下文。 + +- **会话状态管理**：`Conversations` 接口（`/v1/conversations`）配合 Responses 使用，实现跨请求的上下文持久化，避免手动拼接 `messages`。 + +> ⚠️ 注意：并非所有百炼模型都支持 OpenAI 协议——例如 `qwen-audio`、`wan2.6-t2i`（文生图）仅提供 DashScope 原生接口；Qwen3 系列部分高级能力（如 `enable_thinking`）需通过原生 SDK 或 `extra_body` 显式传递，OpenAI 兼容接口默认不透传。 + +## 关键参数和配置 + +| 参数 | 类型 | 说明 | 必填 | +|------|------|------|------| +| `model` | string | 模型 ID，如 `qwen3.7-plus`、`text-embedding-v4`、`qwen3-rerank`。**必须与 Base URL 所在地域匹配**（如 `qwen3.7-plus-us` 仅限美国地域）。 | 是 | +| `base_url` | string | 服务端点，**必须使用业务空间专属域名**以保障性能与稳定性：
• 北京：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
• 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`
• 弗吉尼亚：`https://dashscope-us.aliyuncs.com/compatible-mode/v1`
• 法兰克福/东京：同北京格式，替换对应地域代码 | 是 | +| `api_key` | string | 百炼 API Key，**严格按地域与计费方案隔离**（Token Plan、Coding Plan、按量计费 Key 不互通）。需通过 [API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建并匹配 `base_url` 地域。 | 是 | +| `messages` | array | 对话消息列表，格式为 `[{"role": "user", "content": "..."}]`。`Responses API` 可自动注入历史，但生产环境建议显式传入以确保可控性。 | Chat Completions / Responses / Vision 等需对话场景下必填 | +| `stream` | boolean | 启用流式响应（`true`）。返回 `data: {...}` SSE 格式，每 chunk 含 `delta.content`。 | 否 | +| `stream_options` | object | 当 `stream=true` 时，设 `{"include_usage": true}` 可在末尾 chunk 返回 token 统计（`usage` 字段）。 | 否 | +| `temperature` / `top_p` | float | 控制生成随机性，二者互斥。`temperature ∈ [0, 2.0)`，`top_p ∈ (0, 1.0]`。 | 否 | +| `max_tokens` | integer | 最大输出 token 数，影响响应长度与计费。不同模型有硬上限（如 `qwen3.7-plus` 支持 32K 上下文，实际可用受系统 [prompt](../guides/prompt.md) 占用）。 | 否 | + +## 面向开发者，简洁实用 + +- ✅ **快速上手**：只需三步——获取 API Key → 构造 `base_url`（含 WorkspaceId）→ 用 OpenAI SDK 调用，无需修改业务代码。 +- ✅ **灵活切换**：同一套 SDK 可无缝切换 Qwen、DeepSeek、Kimi 等模型，仅需改 `model` 参数。 +- ✅ **生产就绪**：推荐使用业务空间专属 `base_url`（而非公共域名），获得更低延迟、更高并发与独立配额。 +- ❌ **避坑提示**： + - 不要混用地域 Key 与 URL（如北京 Key + 美国 URL → 401）； + - `qwen-audio`、`wan2.6-t2i` 等模型**不支持** OpenAI 协议，请查文档确认模型兼容性； + - 工具调用（`tools`）在 Chat Completions 中需客户端自行解析并重发，`Responses API` 才支持自动执行； + - `enable_search`、`enable_thinking` 等 Qwen 特有参数**不在 OpenAI 协议内**，需通过 DashScope SDK 或 `extra_body` 传递。 + +示例（Python）： +```python +from openai import OpenAI + +client = OpenAI( + api_key="sk-xxx", # 替换为你的百炼 API Key + base_url="https://your-workspace-id.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +) + +# 标准对话 +resp = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "用 Python 写一个快速排序"}] +) +print(resp.choices[0].message.content) + +# 流式响应 +for chunk in client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "讲个笑话"}], + stream=True, + stream_options={"include_usage": True} +): + if chunk.choices[0].delta.content: + print(chunk.choices[0].delta.content, end="") +``` + +## 关联主题页 + +- [qwen api reference](../api/qwen-api-reference.md) +- [get started with models](../guides/get-started-with-models.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) +- [vector and sort](../api/vector-and-sort.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md deleted file mode 100644 index 7105da05..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md +++ /dev/null @@ -1,79 +0,0 @@ -# OpenAI 兼容接口 - -OpenAI 兼容接口是百炼平台提供的一套与 OpenAI 官方 API 高度对齐的调用方式：开发者可直接复用官方 OpenAI 客户端库（Python/Node SDK）或 HTTP 请求，仅需替换 `api_key`、`base_url` 与 `model` 三项，即可将现有 OpenAI 应用平滑迁移到百炼，无需改动业务逻辑。 - -## 在百炼平台的使用场景 - -OpenAI 兼容接口是百炼最主流的接入入口，覆盖多种典型场景： - -- **迁移已有 OpenAI 应用**：追求最低迁移成本时的首选。已基于 OpenAI SDK 构建的应用，改动量极小。 -- **调用通义千问及三方模型**：支持 Qwen 商业版/开源版、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及 DeepSeek、Kimi、GLM、MiniMax 等三方模型（三方直供模型仅在中国内地地域可用，需先在控制台开通）。 -- **调用专用模型**：意图理解（`tongyi-intent-detect-v3`）、翻译（`qwen-mt-plus`）、OCR（`qwen3.5-ocr`）、界面交互（`gui-plus`）等多数专用模型均可通过 OpenAI 兼容接口调用。 -- **接入第三方工具**：Cursor、Cline、Cherry Studio、Chatbox、Dify 等聊天客户端与开发工具，统一以「Base URL + API Key + 模型 ID」的形式通过 OpenAI 兼容协议接入百炼网关。 - -> 注意：并非所有模型都支持。例如 Qwen-Audio、`qwen-deep-research` 仅支持 DashScope 协议（后者仅 Python SDK、仅北京地域）。使用前请对照具体模型的 API 参考确认。 - -## 接口家族 - -OpenAI 兼容体系不止 Chat Completions，还包含多个能力接口： - -- **Chat Completions（对话补全）**：最常用的接口，支持非流式、流式（`stream=True`，配合 `stream_options={"include_usage": True}` 返回 Token 统计）与 function call 工具调用。 -- **Responses（智能体原生接口）**：Chat Completions 的演进版本，内置联网搜索、网页抓取、代码解释器、文搜图/图搜图等工具；输入更灵活（可直接传字符串），并通过 `previous_response_id` 自动管理多轮上下文，无需手动拼接消息历史。 -- **Conversations（会话管理）**：提供会话的创建、查询、更新、删除及消息项管理，配合 Responses 可自动注入历史上下文。 -- **Completions（文本补全）**：面向代码补全/内容续写，当前仅支持 `qwen-coder-turbo`，且仅适用于中国内地（北京）地域，使用 `<|fim_prefix|>...<|fim_suffix|>...<|fim_middle|>` 模板。 -- **Embedding、文件、Batch** 等能力也提供对应的兼容接口，并可直接接入 LangChain / LangChain4j 等主流框架。 - -## 关键参数与配置 - -迁移的核心是配置以下三项： - -- **`api_key`**：替换为百炼 API Key。各地域的 API Key 相互独立、不能跨地域混用，切换地域时需同步更换。建议写入环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。 -- **`base_url`**：OpenAI SDK 统一使用以 `/compatible-mode/v1`（部分方案为 `/v1`）结尾的路径；HTTP 调用在其后追加资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。 -- **`model`**：替换为百炼支持的模型名称。 - -各地域 SDK `base_url`（`{WorkspaceId}` 为业务空间 ID，可在控制台业务空间详情页查看）： - -| 地域 | base_url | -| --- | --- | -| 华北2（北京） | `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | -| 新加坡 | `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 日本（东京） | `https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 德国（法兰克福） | `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` | -| 美国（弗吉尼亚） | `https://dashscope-us.aliyuncs.com/compatible-mode/v1` | - -最小调用示例（北京地域）： - -```python -import os -from openai import OpenAI - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen-plus", - messages=[{"role": "user", "content": "你是谁？"}], -) -print(completion.choices[0].message.content) -``` - -部分模型的专属参数需通过 OpenAI SDK 的 `extra_body` 传入，例如 Qwen-MT 的 `translation_options`（`source_lang` / `target_lang` / `terms` / `tm_list` / `domain_prompt`）、Qwen-OCR 的 `min_pixels` / `max_pixels`、GUI-Plus 的 `vl_high_resolution_images`。 - -## 注意事项 - -- **Base URL 与 API Key 必须同地域、同计费方案配套**，否则会报 401 或跨地域错误。 -- **推荐使用业务空间专属域名**（`{WorkspaceId}.{region}.maas.aliyuncs.com`）用于生产环境，具备更高并发、更低时延与流量隔离，请求超时 3600 秒；旧版 `dashscope.aliyuncs.com` 等域名仍可用但建议迁移，超时 600 秒。 -- **旧版路径将停止维护**：Responses 与 Conversations 的旧路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将下线，请迁移到新版 `/compatible-mode/v1/...`。 -- **依赖内置工具（联网搜索、代码解释器等）时须用 Responses 接口**，而非普通 Chat Completions。 -- 跨接口迁移时需核对参数映射：DashScope 原生接口参数最全，OpenAI/Anthropic 兼容接口以各自生态的字段约定为准。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [get started with models](../guides/get-started-with-models.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md b/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md new file mode 100644 index 00000000..2800f69d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md @@ -0,0 +1,56 @@ +# 插件机制 + +插件机制是百炼平台提供的核心能力扩展框架，通过标准化接口将外部工具（API、服务或计算能力）安全、可控地集成到大模型推理链路中，使模型在保持语言理解能力的同时，具备实时搜索、代码执行、图像生成、专业计算等超越纯文本推理的增强能力。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +插件机制并非单一技术实现，而是贯穿多个能力层的统一抽象，其具体形态和使用方式因场景而异： + +- **智能体应用（Agent）**：插件作为“可调用工具”被注入模型上下文。大模型基于用户输入语义自主规划是否调用、调用哪个插件及传入参数（如 `calculator` 计算 `237 × 48`），整个过程无需开发者编写调度逻辑。官方插件（如 `quark_search`）、三方插件（云市场服务）和自定义插件均可在此模式下启用。 + +- **工作流应用（Workflow）**：插件以显式节点形式存在，开发者手动拖拽、配置输入/输出连接与参数映射（如将上一节点提取的地址传给 `amap_weather` 工具）。此时插件不依赖模型决策，适用于确定性、多步骤、需精确控制的业务编排。 + +- **Assistant API 调用**：通过 `tools` 数组声明插件 ID 及结构化描述（含 `name`、`description`、`parameters` JSON Schema），由 SDK 或平台自动完成 function calling 的请求构造、响应解析与结果注入，实现与 OpenAI 兼容的工具调用范式。 + +- **MCP（Model Context Protocol）服务**：作为插件机制的协议升级形态，MCP 提供更严格的标准化通信契约（如 `streamableHttp` 协议、JSON Schema 输入校验、KMS 加密凭据管理），支持跨平台工具复用与统一治理。所有 MCP 服务（包括官方 Amap Maps、WebSearch 及自定义部署服务）在百炼中均以“插件”身份被发现、授权和调用。 + +> ⚠️ 注意：`Skill`（技能）虽常被类比为“插件”，但其本质不同——Skill 是预打包的、无网络调用的本地计算能力（如 CSV 清洗、PDF 解析），由百炼调度引擎基于 `description` 语义匹配触发，不涉及 HTTP 请求或外部服务授权，因此**不属于插件机制范畴**。 + +## 关键参数和配置 + +以下参数在自定义插件或 MCP 服务配置中必须准确设置，直接影响调用成功率与安全性： + +| 参数 | 说明 | 开发提示 | +|------|------|----------| +| **`tool_id`（工具 ID）** | 插件内唯一标识具体工具的字符串（如 `text_to_image`），用于 Assistant API 的 `tools` 声明或工作流节点选择。可在控制台插件详情页复制。 | 避免使用空格、中文或特殊符号；同一插件内不可重复。 | +| **`url` + `path`** | 插件服务根地址（如 `https://api.example.com`）与工具路径（如 `/v1/generate`），拼接后构成完整调用 URL。MCP 中对应 `url` 字段（`type=streamableHttp` 时必填）。 | 确保 URL 可公网访问且 HTTPS 启用；路径区分大小写。 | +| **`inputSchema`（JSON Schema）** | 定义输入参数结构，直接影响模型参数提取准确性。必须包含 `type`、`properties`，推荐使用 `required` 明确必填项。 | 示例：`{"type":"object","properties":{"prompt":{"type":"string"}},"required":["prompt"]}` | +| **鉴权方式** | 支持 `Header`（如 `Authorization: Bearer `）或 `Query`（如 `?api_key=xxx`）；MCP 强制要求 `Authorization` header 为 `Bearer `。 | 敏感密钥（如地图 API Key）**必须通过 KMS 加密存储**，禁止明文配置。 | +| **`output_parameters`（输出字段）** | 指定 API 返回 JSON 中哪些顶层字段供模型读取（如 `{"image_url": "string"}`），需扁平、非嵌套、非空。 | 避免返回大体积二进制或原始 HTML；仅保留模型生成回复所需的最小数据集。 | + +## 面向开发者，简洁实用 + +- **快速起步**：优先使用控制台「插件市场」添加官方插件（如 `code_interpreter`），无需配置即可在智能体中测试；确认 `AliyunServiceRoleForSFMAccessCloudAPI` 角色已授权（主账号一键授权，RAM 用户需 `ram:CreateServiceLinkedRole` 权限）。 + +- **自定义插件上线三步**： + 1. 在控制台创建插件 → 填写 `tool_id`、`url`、`path`； + 2. 定义 `inputSchema` 和 `output_parameters`（用 JSON Schema 校验器验证）； + 3. **在线调试通过并发布为“已发布”状态**（草稿/未启用 = 调用失败）。 + +- **避坑指南**： + - 所有插件调用均**只透传 `Authorization` header**，其他自定义 header 会被平台剥离； + - `Object` 类型输入参数在 `GET` 请求中不被支持（仅 `POST` 允许）； + - 智能体最多添加 10 个工具（含 Skill），MCP 服务上限为 5 个； + - 实际模型兼容性以控制台运行结果为准，文档列表可能滞后（如 `qwen2.5` 系列需实测）。 + +- **调试技巧**：开启 `stream=True` 查看模型思考过程（含 tool call 步骤）；检查返回错误码（如 `130040` = 参数描述缺失，`11200054` = MCP 协议解析失败），对照文档定位问题。 + +## 关联主题页 + +- [plug in](../guides/plug-in.md) +- [skill](../guides/skill.md) +- [model context protocol](../guides/model-context-protocol.md) +- [application support](../guides/application-support.md) +- [more about models](../api/more-about-models.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md deleted file mode 100644 index fac9783f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md +++ /dev/null @@ -1,59 +0,0 @@ -# 上下文缓存 - -上下文缓存（又称前缀缓存）是百炼平台提供的一种推理加速与成本优化能力：当多次请求共享相同的输入前缀时，平台会缓存该前缀的中间计算结果，后续命中缓存的部分按折扣价折算额度或计费，从而降低延迟与费用。 - -## 在百炼平台的使用场景 - -### 预置吞吐（PTU）部署中的前缀缓存 - -PTU 部署原生支持长输入与前缀缓存，是上下文缓存的主要应用场景。通过阶梯容量系数和缓存折扣管理额度消耗： - -- **缓存命中折扣**：命中缓存的输入 token 按模型对应折扣折算容量。例如 glm-5.1 折扣为 0.2，deepseek-v4-pro 为 0.08，显著降低 TPM 消耗。 -- **自动转按量计费**：当请求超出 PTU 额度或输入超过模型上下文上限（千问 128K / DeepSeek 64K）时，请求自动转为按量计费，响应头包含 `x-dashscope-ptu-overflow:true`，业务不中断。 - -### 显式缓存调用 - -百炼支持通过 OpenAI 兼容接口或 DashScope SDK 显式触发缓存。开发者可在请求中标识需要缓存的前缀（如系统提示词、[长上下文](long-context.md)文档），平台据此管理缓存命中与失效，适用于 RAG 应用、多轮对话等固定前缀反复出现的场景。 - -## 关键参数与响应字段 - -### 请求侧 - -- 在 PTU 部署下，缓存由平台自动管理，无需额外参数；显式缓存则按接口协议提供缓存控制字段。 -- 建议将稳定不变的内容（系统提示、知识库文档、few-shot 示例）置于请求前缀部分，动态内容置于后段，以提高缓存命中率。 - -### 响应侧（PTU 部署） - -PTU 部署的响应中包含额度相关字段，用于观测缓存命中情况： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。 - -`cached_tokens` 在不同 API 格式下的 JSON 路径有差异： - -| API 格式 | 字段路径 | -| --- | --- | -| OpenAI Chat 兼容 | `usage.prompt_tokens_details.cached_tokens` | -| OpenAI Responses | `usage.input_tokens_details.cached_tokens` | -| Anthropic 兼容 | 暂不返回 `cached_tokens` | - -## 容量评估建议 - -在创建或扩容 PTU 部署时，控制台提供容量计算器，可根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推荐输入 TPM 和输出 TPM。长输入场景下建议先用计算器评估额度，结合预期缓存命中率规划容量，避免意外转为按量计费。 - -## 注意事项 - -- 缓存命中要求前缀完全一致，任何字符变化（包括空格、换行、字段顺序）都可能导致缓存失效。 -- 缓存有有效期，长时间无请求后缓存会被淘汰，需重新预热。 -- 兼容接口（OpenAI / Anthropic）为保持协议一致，可能不暴露百炼原生的全部缓存控制参数；如需最完整的缓存能力，建议使用 DashScope 原生接口。 - -## 关联主题页 - -- [use cases](../guides/use-cases.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [qwen api reference](../api/qwen-api-reference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md deleted file mode 100644 index 6241b301..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md +++ /dev/null @@ -1,74 +0,0 @@ -# 提示词工程 - -提示词工程（Prompt Engineering）是通过设计、组织与优化 Prompt 来引导大模型生成符合预期结果的方法论。在百炼平台中，它贯穿智能体配置、[工作流](workflow.md)节点、模型直调与多模态生成等几乎所有 LLM 场景，并提供模板化管理、自动优化、样例库与反馈优化等成熟能力。 - -## 在百炼中的使用方式 - -### 1. 直接编写 System Prompt - -最基础的形态是在智能体应用的「系统提示词」中定义角色、行为指令与能力边界，支持通过 `/` 引用自定义变量。[工作流](workflow.md)应用的大模型节点同样通过「提示词 + 用户提示词」驱动推理。Prompt 越清晰、具体、无歧义，模型表现越稳定。 - -### 2. Prompt 模板（结构与变量分离） - -将固定结构与动态变量分离，统一管理、复用，是团队协作与版本一致性保障的推荐做法。入口位于控制台「应用开发 > 组件管理 > 提示词」。 - -- **预置模板**：平台提供，覆盖创意文案、办公助理等通用场景，效果稳定、不可修改。 -- **自定义模板**：用户自行设计，支持「自定义创建」（直接粘贴现成 Prompt，可选「优化 Prompt」润色）和「基于 Prompt 工程创建」（选择 ICIO / CRISPE / RASCEF 框架，按字段结构化填写）两种模式。 - -框架选型建议： - -| 框架 | 适用场景 | -| --- | --- | -| ICIO | 简单、明确的任务执行，如数据分析、内容生成、文本摘要 | -| CRISPE | 需要 AI 扮演特定角色的交互，如智能客服、创意写作 | -| RASCEF | 涉及多步骤的复杂业务流程，如项目规划、战略分析 | - -> 注意：Prompt 模板相关功能仅适用于华北2（北京）地域，使用前请确认业务空间所在地域。 - -### 3. Prompt 自动优化 - -当缺乏经验或手动编写耗时，可在「提示词 > 自动优化」页面输入原始 Prompt，由大模型进行结构重组、角色扮演引导、指令增强、安全与边界注入等重写，生成结构更优的新版本。该功能不计费，提交数据不会被存储或用于训练。优化失败常见原因：输入超长超出 Token 限制、触发内容审核、网络或服务临时不可用。 - -### 4. Prompt 样例库（Few-shot 检索） - -针对特定领域专业任务，从预定义的高质量问答对中检索相关样例注入上下文，引导模型生成更准确、风格更一致的回复。适用于智能客服、特定领域问答、格式化内容生成。注意该功能已不再维护，官方推荐迁移到 RAG 表格库。 - -### 5. Prompt 反馈优化 - -基于输入输出样例与评测数据，多轮自动评估、反思、优化 Prompt，涉及推理调用。适合对输出质量有持续提升需求的闭环场景。 - -## 关键参数与配置 - -### 智能体应用 - -- **系统提示词最大长度**：6144 字符（从模板「使用 [prompt](../guides/prompt.md) > 创建应用」时自动填充到此上限）。 -- **模型参数**：`temperature`、最长回复长度、`enable_thinking`（开启思考模式以提升反思效果，仅支持思考模式模型）。 -- **变量引用**：在系统提示词中通过 `/` 嵌入自定义变量，运行时由业务数据填充。 - -### 多模态生成 - -- **文生图**：`prompt`（正向）、`negative_prompt`（反向，描述不希望出现的内容）、`prompt_extend`（V2 专用，大模型智能改写，默认 `true`）。基础公式 `主体 + 场景 + 风格`，进阶公式追加镜头语言、氛围词与细节修饰。 -- **文生视频**：基础公式 `主体 + 场景 + 运动`，进阶公式 `主体描述 + 场景描述 + 运动描述 + 美学控制 + 风格化`；图生视频简化为 `运动 + 运镜`。wan2.7 起单/多镜头由提示词控制，不再使用 `shot_type`。 - -### 模板调用 API - -- **创建模板**：`CreatePromptTemplate`，需先获取 Workspace ID。 -- **获取模板**：`GetPromptTemplate`，传入 `workspaceId` 与 `promptTemplateId`，返回 `content`、`variables` 等字段，在代码中填充变量后调用模型。相比字符串拼接，可实现逻辑与内容分离、集中管理与版本一致。 - -## 设计要点 - -1. **结构化优先**：复杂任务优先采用 ICIO/CRISPE/RASCEF 框架或「背景-目的-风格-语气-受众-输出」六要素，避免笼统指令。 -2. **变量分离**：把不可变的结构与可变的业务数据拆开，通过模板或 `/` 变量注入，便于复用与协作。 -3. **样例引导**：对风格或格式敏感的任务，用 Few-shot 样例或反馈优化建立闭环，持续校准输出。 -4. **正反向结合**：图片与视频生成场景同时使用正向与反向 Prompt 精确控制画面内容。 -5. **迭代验证**：结合在线调试面板与评测能力，对 Prompt 变更做回归验证后再发布。 - -## 关联主题页 - -- [prompt](../guides/prompt.md) -- [start using](../guides/start-using.md) -- [llm application](../guides/llm-application.md) -- [use cases](../guides/use-cases.md) -- [application component api reference](../api/application-component-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md index f136ddf7..2a278f6b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md @@ -1,64 +1,56 @@ -# 检索增强生成（RAG） +# 检索增强生成 -检索增强生成（Retrieval-Augmented Generation，RAG）是指大模型在生成回答前，先从外部知识库检索相关内容并拼接进上下文，从而补充私有数据、最新信息，提升回答的准确性并降低幻觉。它是阿里云百炼平台知识库、智能问答与多种应用场景的核心底层能力。 +检索增强生成（Retrieval-Augmented Generation，RAG）是一种将大语言模型（LLM）的生成能力与外部知识源的精准检索能力相结合的技术范式。它通过在模型推理前动态召回相关上下文片段，并将其注入提示词（Prompt），使模型能在私有、领域专属或时效性强的知识基础上生成更准确、可溯源、抗幻觉的回答。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼围绕 RAG 提供从建库到问答、从控制台到 API、从云端到本地的多条落地路径： +在百炼平台中，RAG 不是单一接口，而是贯穿多个能力层的协同工作模式，核心围绕**知识库**这一基础设施展开，支持三种主流集成路径： -- **云端知识库（控制台）**：基于百炼知识库能力构建 RAG，先创建知识库（选类型、配数据源与索引参数），再关联到智能体或工作流应用。工作流应用中把「知识库」节点接在开始节点后，用内置变量 `query` 作输入，输出 `result` 传给大模型节点。知识库功能仅在中国站 **华北2（北京）** 地域开通使用。 -- **知识检索与知识问答服务**：平台在知识库之上提供两类独立服务。知识检索支持多知识库联合检索（最多 15 个），流水线为 Query 改写 → 向量+关键词混合检索 → Rerank 精排 → 加权返回；知识问答基于大模型结合检索生成自然语言回答，提供极速（单轮）与多轮智能（Agentic 规划搜索）两种模式，支持拒答、防泄漏、引用来源展示。 -- **HTTP REST API（DashScope 应用网关）**：知识检索接口 `POST /api/v1/indices/knowledge/search` 适合自定义生成流程（拿到排序切片后自行拼 [prompt](../guides/prompt.md) 调模型）；知识问答接口 `POST /api/v2/apps/knowledge/chat` 通过 SSE 流式返回规划、工具调用、生成三阶段，适合开箱即用。二者用 API Key Bearer 鉴权，Base URL 为 `https://{workspaceId}.cn-beijing.maas.aliyuncs.com`，默认 25 QPS。 -- **开源框架集成**：LlamaIndex（Python）用于读取本地文件上传建云端知识库并构建 RAG 应用；Spring AI Alibaba（Java）用于集成智能体/工作流应用并检索百炼知识库。均以 [API Key 鉴权](api-key.md)。 -- **本地知识库 RAG**：检索在本地执行、生成调用通义千问 API，适合需要灵活切分与自选嵌入模型的场景。 -- **应用接入渠道**：RAG 应用可通过 AppFlow 接入网站、企业微信、微信公众号、钉钉等渠道，为回答覆盖私域问题。 +- **应用内嵌 RAG（推荐用于生产级智能体/工作流）** + 在智能体或工作流应用配置页中直接绑定已创建的知识库，设置“相似度阈值”“权重”和调用策略（如“必定调用”）。工作流中拖入“知识库节点”，配置 `TopK` 和输入变量（如 `query`），再连接至大模型节点；模型提示词中通过 `{result}` 引用召回内容。该方式支持多轮对话改写、Agentic 规划搜索及引用溯源。 -## RAG 流水线与底层模型 +- **独立服务形态（适合快速验证与 API 集成）** + 通过控制台“知识检索”或“知识问答”标签页发布统一服务：可跨最多 15 个知识库联合检索，配置混合检索（向量+关键词）、Rerank 模型（如 `qwen3-rerank`）、拒答与防泄漏策略。发布后通过标准化 HTTP API 调用： + - 检索：`POST /api/v1/indices/knowledge/search` → 返回结构化切片； + - 问答：`POST /api/v2/apps/knowledge/chat` → 默认 SSE 流式响应，含 planning、tool calling、generation 三阶段输出。 -一次典型 RAG 调用可拆为三阶段，每阶段对应可优化的环节与模型： +- **框架集成（面向开发者快速构建）** + - **LlamaIndex**：调用 `DashScopeCloudIndex.from_documents()` 构建云端知识库，通过 `SimilarityPostprocessor` + `DashScopeRerank` 控制召回与重排，最终 `query_engine.query()` 触发端到端 RAG。 + - **Spring AI Alibaba**：使用 `DashScopeDocumentRetriever` 按 `INDEX_NAME` 检索上下文，并自动注入提示词交由 `qwen-max` 等模型生成；或通过 `DashScopeAgent` 调用已发布的智能体应用（隐式封装 RAG 逻辑）。 -1. **建立索引**：文档切分（智能切分）、Meta 信息抽取、生成向量。文本向量推荐 text-embedding-v4（Qwen3-Embedding 系列），支持 2048/1536/1024/768/512 等多种维度；多模态检索用 qwen3-vl-embedding / multimodal-embedding-v1。 -2. **检索召回**：向量检索 + 关键词检索混合，再经 Rerank 精排。排序模型推荐 qwen3-rerank（文本，支持 100+ 语种）与 qwen3-vl-rerank（多模态），`gte-rerank` 将于 2026-05-30 下线。 -3. **生成答案**：将召回切片与用户提问一并送入大模型（如 qwen-max、qwen3.6-plus/qwen3.7-plus 等）生成回答。 +> ⚠️ 注意：所有 RAG 路径均依赖知识库预置——知识库必须部署在华北2（北京）地域，且需完成文档上传、解析、向量化与索引构建；不支持裸知识库 ID 直接调用，问答接口必须传入已绑定知识库的 `app_id`。 -## 关键参数与配置 +## 关键参数和配置 -云端知识库侧： +| 参数 | 所属层级 | 类型 | 说明 | 典型值 | 备注 | +|------|----------|------|------|--------|------| +| `retrieval_top_k` / `top_k` | 应用层 / API 层 | integer | 初始召回切片数量（向量/关键词双路） | `3`–`10` | 过高增加 Token 开销，过低影响召回完整性；最大支持 100 | +| `similarity_threshold` | 应用层 / 知识库层 | float (0.01–1.0) | Rerank 后过滤阈值，仅保留得分高于此值的切片 | `0.4`–`0.6` | 值过高易漏召，过低引入噪声；纯文本知识库专用 | +| `max_retrieved_chunks` | 应用层 | integer | 最终传递给大模型的上下文切片总数 | `1`–`20` | 控制 Prompt 长度与成本，建议 ≤10 | +| `weight` | 多知识库混排 | float | 知识库在联合检索中的相对优先级 | `1.0`, `2.0` | 权重越高，同 Query 下该库切片排序越靠前 | +| `tags` | 知识库层 | string array | 单文件最多 32 个标签，用于精准范围过滤 | `["finance", "2024Q3"]` | 支持 `AND` 语义匹配，提升高干扰场景精度 | +| `metadata` 字段 | 索引层 | key-value | 在切片索引时注入结构化信息（如 `filename`, `date`, `author`） | — | 实现“先过滤、再检索”，降低误召率 | -- **相似度阈值**：仅语义相似度高于阈值的切片才被召回，阈值过高会导致全部被过滤（如调到 0.60 可能无召回）。 -- **召回片段数 / 最大召回数量（TopK/K）**：取值 1–20，复杂问题可适当增大以补全答案，但会增加 Token 消耗，推荐「按拼装长度」策略。 -- **权重**：多知识库召回时按信息源重要性分配，**仅在同类型知识库之间生效**。 -- **Meta 信息抽取**：以 key-value 附加到切片，提升检索准确性并降低 Token 消耗；支持常量、变量、大模型、正则、关键词五种取值方式。**知识库创建后无法再配置 metadata 抽取**。 -- **智能切分 / 多轮对话改写**：均在创建知识库时配置，创建时未开启则后续无法补开（除非重建）。 -- 检索服务全局参数：知识库路由、混排模型（qwen3-rerank / qwen3-rerank(hybrid) / qwen3-vl-rerank）、混排模式（问答/相似/自定义）；每库可独立配置向量/关键词 TopK（1–100）、排序模型、相似度阈值、标签过滤。 +> ✅ 提示：参数生效位置不同——`similarity_threshold` 和 `weight` 在知识库或应用配置页设置；`top_k` 和 `stream` 在 API 请求 Body 或 Header 中指定；`chunk_size`/`chunk_overlap` 仅在知识库创建时一次性配置，不可修改。 -LlamaIndex 侧关键参数： +## 面向开发者，简洁实用 -- `Settings.llm = DashScope(model_name="qwen-max")`：生成回答调用的大模型。 -- `similarity_top_k`：返回相似度最高的检索结果数（示例 5）。 -- `similarity_cutoff`：过滤检索结果的最低相似度阈值（示例 0.4）。 -- `top_n`：重排后返回的结果数（示例 1）。 -- 后处理 `node_postprocessors`：`SimilarityPostprocessor`（按阈值过滤）、`DashScopeRerank(model="gte-rerank")`（重排）、`response_mode="tree_summarize"`（响应聚合方式）。 +- **起步最快**：控制台创建知识库 → 上传 PDF/DOCX/TXT → 发布“知识问答”服务 → 调用 `/api/v2/apps/knowledge/chat`，只需 `workspaceId` + `API Key` + `app_id` + `messages`。 +- **调试关键**：开启 SLS 日志监控，关注 `data.nodes[]` 字段确认召回质量；流式响应需按 `event: chunk` 解析，末尾 `event: done` 包含完整结果与 `docReferences`。 +- **避坑指南**： + - 域名必须含 `workspaceId`，API Key 必须归属该 workspace，否则 `401`； + - 知识库类型（文档/图片/表格）决定可用模型：纯文本仅支持 `qwen3-rerank`，多模态知识库才可用 `VL-Max`； + - 文件上传后需等待解析完成（1–6 分钟），未就绪时检索返回空； + - 免费额度覆盖全部 RAG 场景，但向量模型与 Rerank 模型按 Token 单独计费，非包含在知识库规格费中。 -本地 RAG 侧关键参数：模型选择、温度、最大回复长度、携带上下文轮数；召回片段数、相似度阈值（为 0 时不剔除）；嵌入模型默认用百炼 embedding API，也可换本地 GTE 向量模型；受限流约束，单文件建议不超过 100 MB。 - -## 效果优化建议 - -RAG 效果由建立索引、检索召回、生成答案三阶段共同决定。建议先建立至少 100 组问题的评估基线（覆盖事实型/比较型/教程型/分析型），再针对失败用例（大模型打分 < 4）逐项诊断： - -- 检索无效 → 补充知识、优化排版、统一实体、开启多轮对话改写； -- 召回不相关 → 标签过滤、元数据结构化搜索； -- 切片不完整 → 智能切分 + 人工修正； -- 重排不佳 → 调整相似度阈值与召回片段数； -- 模型理解有误 → 更换为参数更多的商业模型。 +RAG 的本质是“让模型知道它该知道的”。在百炼，你只需聚焦业务知识——平台负责高效检索、可信增强、稳定生成。 ## 关联主题页 +- [knowledge](../api/knowledge.md) - [knowledge base](../guides/knowledge-base.md) - [frameworks](../api/frameworks.md) -- [knowledge](../api/knowledge.md) - [application use cases](../guides/application-use-cases.md) -- [use cases](../guides/use-cases.md) -- [vector and sort](../api/vector-and-sort.md) +- [data connection overview](../guides/data-connection-overview.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md deleted file mode 100644 index c43445dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md +++ /dev/null @@ -1,71 +0,0 @@ -# 限流与配额 - -限流与配额是百炼平台对模型推理调用施加的吞吐量与调用频率约束，用以保障公共推理资源的公平使用和整体稳定性；当业务流量超过共享上限或需要刚性容量保障时，可通过专属域名、TPM 预留、PTU 部署或异步通知等手段规避限流影响。 - -## 限流的产生场景 - -百炼模型调用主要在以下维度受到约束： - -- **TPM（Tokens Per Minute）**：每分钟可消耗的 Token 数上限，按主账号 + 业务空间 + 模型维度计算，超出会返回 `429` 限流错误。 -- **RPM（Requests Per Minute）**：每分钟请求数，RPM 越大建议 TPM 同比增大。 -- **QPS 限制**：部分通用接口有独立的 QPS 限制。例如异步任务管理（查询、批量查询、取消）三个接口统一限制为 20 QPS，按主账号 + 子账号维度计算。 -- **公共池共享**：按量付费方案无专属容量，调用进入公共共享池，受公共限流波动影响，业务高峰期可能被限流。 - -监控页面的「错误」指标中专门提供**限流错误次数（429）**，可用于定位限流问题；「性能」指标中的 RPM、TPM 则帮助评估容量是否接近上限。 - -## 不同方案下的限流行为 - -| 方案 | 容量保障 | 超额处理 | 接入改动 | -| --- | --- | --- | --- | -| 按量付费 | 无（共享公共池） | 自动服务，受公共限流 | 无需改动 | -| 资源包 / 节省计划 | 承诺用量折扣（非专属容量） | 超出转按量 | 无需改动 | -| TPM 预留 | 专属容量刚性兑付 | 超出自动降级公共池按量，不中断 | 替换 `model` 参数为专属模型 code | -| PTU（模型部署） | 专属部署实例 | 超出转按量 | 替换 `model` 参数 | - -当业务流量可预估且不能接受限流时，优先选择 TPM 预留；对极致性能与隔离有更高要求时，可考虑 PTU 专属部署。 - -## 接入域名对限流的影响 - -接入域名直接影响并发上限与超时表现： - -- **业务空间专属域名** `https://{WorkspaceId}.{region}.maas.aliyuncs.com`：生产推荐，请求超时 3600 秒、SLA 99.9%，提供更高吞吐与时延隔离。 -- **Dashscope 中心化域名** `https://dashscope.aliyuncs.com`：存量业务兼容，可跨业务空间调用。 -- **试用域名** `https://trial.{region}.maas.aliyuncs.com`：仅限快速验证，限流小，不建议生产。 - -各地域的 API Key、模型列表、接入域名不能跨地域混用；美国（弗吉尼亚）暂不支持业务空间专属域名，需使用 `dashscope-us.aliyuncs.com`。 - -## TPM 预留的关键参数 - -创建 TPM 预留时需指定输入 TPM 与输出 TPM，单位为 kTPM（1 kTPM = 1,000 Tokens/分钟），起步和步长因模型而异。容量计算受以下参数影响： - -- **每分钟请求数（RPM）**：业务高峰期每分钟请求数。 -- **平均输入/输出长度（token）**：输入越长，阶梯系数越大，所需输入 TPM 越高。 -- **预估缓存命中率（%）**：命中率越高，输入容量消耗越慢，所需输入 TPM 越低；仅影响输入 TPM。 - -部分模型支持长输入阶梯系数和缓存折扣（如 glm-5.1 在 [32K, 200K] 区间输入系数 1.33、输出 1.17；deepseek-v4-pro 缓存命中部分按 8% 折算），TPM 容量计算器会自动应用这些参数。 - -## 规避异步任务轮询限流 - -图像/视频生成等耗时模型采用异步机制，频繁轮询结果接口会浪费资源并可能触发 20 QPS 限流。百炼已接入事件总线 EventBridge，任务完成（无论成功或失败）后主动上报 `dashscope:System:AsyncTaskFinish` 事件，可推送到 HTTP 回调 URL 或 RocketMQ 消息队列。通知方案不限流、实时性高，适合高并发、大规模或对实时性要求高的任务。 - -## 免费额度与配额停用 - -免费额度页面提供「免费额度用完即停」开关：开启后免费额度用尽时服务自动停止，返回 `403 AllocationQuota.FreeTierOnly`，避免产生免费额度以外的费用。仅在账户内仍有未消耗的免费额度时才能开启；关闭需等免费额度完全消耗后进行。 - -## 应对限流的实践建议 - -1. 生产环境优先迁移到业务空间专属域名，获得更高吞吐与时延隔离。 -2. 流量可预估且不能接受限流的核心业务，使用 TPM 预留锁定专属容量。 -3. 通过模型监控的限流错误次数（429）与 RPM/TPM 指标评估容量是否接近上限。 -4. 异步任务改用 EventBridge 通知方案，避免轮询触发 20 QPS 限制。 -5. 需要严格控制成本的场景，可开启「免费额度用完即停」避免超额消费。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [model high speed inference](../guides/model-high-speed-inference.md) -- [model monitoring](../guides/model-monitoring.md) -- [more about models](../api/more-about-models.md) -- [use cases](../guides/use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md b/skills/bailian-docs-llm-wiki/wiki/concepts/region.md deleted file mode 100644 index 3006794c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md +++ /dev/null @@ -1,80 +0,0 @@ -# 地域与可用区 - -地域（Region）是百炼平台在物理地理上划分的部署区域，决定接入点位置、数据存储位置与推理执行位置；可用区是地域内相互隔离的物理设施单元。在百炼中，"地域"主要与**服务部署范围**和**接入域名**共同决定调用路径、数据驻留约束与服务保障等级。 - -## 可用地域 - -百炼当前支持的地域及其 Region ID： - -| 地域 | Region ID | 数据驻留约束 | -| --- | --- | --- | -| 华北2（北京） | `cn-beijing` | 数据不出中国内地 | -| 新加坡 | `ap-southeast-1` | 数据不经过中国内地 | -| 美国（弗吉尼亚） | — | 数据不出美国 | -| 德国（法兰克福） | `eu-central-1` | 数据不出欧盟 | -| 日本（东京） | `ap-northeast-1` | 数据不出日本 | - -若对驻留无限制但追求更大推理资源池，可选择美国/德国/日本的全球部署范围。 - -## 三个关键概念 - -调用前需同时确认**地域**、**服务部署范围**、**接入域名**： - -- **地域**：决定接入点与数据存储位置。 -- **服务部署范围**：决定推理执行位置（数据是否跨域流转）。 -- **接入域名**：影响并发上限、超时、SLA 等服务保障。 - -三者在控制台或 Base URL 中体现，**各地域的接入域名、API Key、模型列表不能跨地域混用**。 - -## 接入域名类型 - -| 域名类型 | 格式 | 适用场景 | 关键指标 | -| --- | --- | --- | --- | -| 专属域名 | `{WorkspaceId}.{region}.maas.aliyuncs.com` | 生产环境 | SLA 99.9%、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC | -| 共享域名 | `dashscope.aliyuncs.com` | 存量兼容 | — | -| 试用域名 | `trial.{region}.maas.aliyuncs.com` | 快速体验 | 不建议生产 | - -从共享域名迁移到专属域名只需替换 Base URL 中的域名部分，业务逻辑代码无需修改。 - -## 各地域调用差异 - -- **华北2（北京）**：默认推荐地域，专属域名 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。知识库、[模型部署](model-deployment.md)、数据管理、Prompt 模板、模型调优等大量能力仅在此地域开放。 -- **新加坡**：专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`。 -- **美国（弗吉尼亚）**：使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理；共享接入域名 `https://dashscope-us.aliyuncs.com/compatible-mode/v1`。 -- **德国（法兰克福）/日本（东京）**：通过业务空间区分全球/欧盟或全球/日本部署范围，调用前需先在业务空间管理页面创建并选择对应业务空间。 - -## 业务空间与 WorkspaceId - -使用北京、新加坡、日本、德国等地域时，需在 Base URL 中填入 WorkspaceId。WorkspaceId 可在控制台「业务空间管理」页面查看。子账号需加入对应业务空间并获得相应策略后方可操作该空间下的资源。 - -## 地域相关能力约束 - -部分能力仅在特定地域可用： - -| 能力 | 可用地域 | -| --- | --- | -| 知识库（标准版/旗舰版） | 仅华北2（北京） | -| [模型部署](model-deployment.md)（资源专享推理） | 仅华北2（北京） | -| 数据管理（训练集/[评测](evaluation.md)集/数据流） | 仅华北2（北京） | -| Prompt 模板 | 仅华北2（北京） | -| `qwen-deep-research` | 仅华北2（北京），且仅支持 Python DashScope SDK | -| `qwen-mt-plus` / `qwen3.5-ocr` | 北京、新加坡、美国（弗吉尼亚） | -| `gui-plus` | 华北2（北京） | - -## 选择建议 - -1. 明确数据驻留要求（是否允许数据出境）。 -2. 确认所需模型/能力是否在目标地域开放。 -3. 生产环境统一使用专属域名，避免使用试用域名。 -4. 同一业务的所有调用、API Key、模型列表保持在同一地域，不要混用。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [knowledge base](../guides/knowledge-base.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [prompt](../guides/prompt.md) -- [model data overview](../guides/model-data-overview.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md deleted file mode 100644 index c08e584a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md +++ /dev/null @@ -1,61 +0,0 @@ -# 重排序 - -重排序（Rerank）是检索流程中的二次精排环节：在向量或关键词召回得到候选文档后，由排序模型结合查询与文档的相关性重新打分并排序，把最相关的切片置顶，再交由后续生成或返回流程使用。 - -## 在百炼平台的使用场景 - -重排序贯穿于平台的检索增强与排序能力两条主线： - -- **知识库 RAG 流水线**：知识检索服务的流程为 Query 改写 → 向量 + 关键词混合检索 → Rerank 排序 → 加权返回。排序模型对召回阶段返回的 TopK 切片做精排，再按相似度阈值过滤与最大召回数量截断，决定最终送入生成模型的上下文。知识问答服务在多轮智能模式下会反复触发该精排环节。 -- **独立文本排序 API**：作为向量与排序模型能力的一部分，开发者可直接调用 rerank 接口，对自定义的候选文档列表按查询相关性排序，用于语义搜索、推荐系统、聚类分类等下游任务，无需绑定知识库。 - -## 模型选型 - -| 模型 | 最大文档数 | 单条最大 Token | 请求最大 Token | 特点 | -| --- | --- | --- | --- | --- | -| qwen3-vl-rerank | 文本 100 / 图片 40 / 视频 4 | 8,000 | 120,000 | 多模态，支持图文视频排序 | -| qwen3-rerank | 500 | 4,000 | - | 100+ 语种，高性能文本排序 | -| gte-rerank-v2 | - | - | 30,000 | 50+ 语种（2026-05-30 下线，建议迁移到 qwen3-rerank） | - -知识库内置的排序模型选项包括 `qwen3-rerank`、`qwen3-rerank(hybrid)` 与 `qwen3-vl-rerank`；多模态知识库只能选 `qwen3-vl-rerank`。 - -## 关键参数 - -独立调用 rerank API 时的核心参数： - -- `model`（必选）：模型名称 -- `query`（必选）：查询内容，qwen3-vl-rerank 支持文本与图片两种查询模态 -- `documents`（必选）：待排序文档列表 -- `top_n`（可选）：返回排序后的前 N 个文档 -- `instruct`（可选）：自定义排序任务说明，可指导模型采用不同排序策略（仅 qwen3-rerank 与 qwen3-vl-rerank） - -知识库检索流程中的相关配置： - -- **初步向量检索 TopK / 关键词检索 TopK**：1–100，默认 50，控制进入精排的候选规模 -- **排序模式**：问答模式按 QA 匹配度排序，相似模式按语义相似度排序，自定义高级模式可自行调参 -- **相似度阈值**：0.01–1.0，用于过滤精排后的低分切片，过高会丢弃全部结果 -- **最大召回数量**：1–20，最终返回的切片数 - -## API 接口 - -不同模型走不同的专属接口，Base URL 均为业务空间 ID 拼接的专属域名： - -- qwen3-rerank：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks` -- qwen3-vl-rerank / gte-rerank-v2：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank` - -请求需携带 `Authorization: Bearer `，API Key 在控制台 API Key 页面获取，业务空间 ID 在业务空间管理页面获取。 - -## 调优建议 - -- 召回结果不理想时，先建立至少 100 组评测基线，再按失败类型针对性调整。 -- 对「列举 / 总结 / 比较」类问题适当提高最大召回数量 K（1–20），并优先选按拼装长度避免超长截断。 -- 通过命中测试反复调整相似度阈值与召回片段数，找到精排质量与上下文长度的平衡点。 -- 多知识库召回时，相似度相同的切片优先返回权重高的库，但权重仅在同类知识库之间生效。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md deleted file mode 100644 index 53a3da52..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md +++ /dev/null @@ -1,54 +0,0 @@ -# 软件开发工具包 - -软件开发工具包（SDK，Software Development Kit）是百炼平台为开发者提供的封装库，屏蔽底层 HTTP 接口与鉴权细节，使开发者能在 Python、Java、Node.js、Go 等语言中通过少量代码即可调用大模型、应用与实时多模态能力。SDK 通常随官方迭代持续升级，建议通过包管理器保持最新版本以获得最新功能与稳定性修复。 - -## 接入路径 - -百炼提供两条接入路径，开发者可按工程栈与迁移成本选择： - -1. **阿里云官方 DashScope SDK**：当前仅覆盖 Python 与 Java，封装了百炼原生协议（含智能体/工作流应用调用、异步任务管理、实时多模态交互等能力），适合需要使用百炼独有特性（如 Qwen-Audio、Omni Realtime）的场景。 -2. **OpenAI 兼容 SDK**：通过 OpenAI 多语言 SDK（Python/Java/Node.js/Go）调用百炼的 OpenAI 兼容接口，只需调整 `api_key`、`base_url`、`model` 三处参数即可复用现有 OpenAI 代码与生态工具，迁移成本最低。 - -## 语言与版本要求 - -| 语言 | DashScope SDK | OpenAI 兼容 SDK | -| --- | --- | --- | -| Python | `pip install -U dashscope`，需 Python ≥ 3.8 | `pip install -U openai` | -| Java | Maven/Gradle 引入 `com.alibaba:dashscope-sdk-java`（建议 ≥ 2.12.0） | 引入 `com.openai:openai-java`，需 Java 8+，推荐 3.5.0+ | -| Node.js | 暂无官方 DashScope SDK，可改用 HTTP/`axios` 或 OpenAI SDK | `npm install --save openai` | -| Go | 暂无官方 DashScope SDK | 需 Go 1.22+，`go get github.com/openai/openai-go/v3` | - -安装失败时可临时切换镜像源：Python 用国内 PyPI 镜像；Node.js 执行 `npm config set registry https://registry.npmmirror.com/`；Go 执行 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 - -## 鉴权与配置 - -- **API Key 注入**：所有 SDK 均读取环境变量 `DASHSCOPE_API_KEY`，避免在代码中硬编码密钥。临时 API Key（有效期 1–1800 秒，前缀 `st-`）适用于浏览器、移动 App 等不可信环境。 -- **base_url**：使用业务空间专属域名 `https://{WorkspaceId}..maas.aliyuncs.com/compatible-mode/v1`，新加坡地域从旧域名 `dashscope-intl.aliyuncs.com` 迁移至专属域名可获得更好性能与稳定性；美国（弗吉尼亚）使用固定域名 `dashscope-us.aliyuncs.com`，不带 `{WorkspaceId}`。 -- **子业务空间**：若应用或知识库位于子业务空间，需额外配置业务空间 ID 环境变量（如 Spring AI Alibaba 应用集成用 `WORKSPACE_ID`，知识库检索用 `AI_DASHSCOPE_WORKSPACE_ID`）。各地域 API Key 不互通，切换地域需同步更换 Key。 - -## 典型使用场景 - -- **模型调用**：通过 Chat Completions、Responses、Embeddings 等兼容接口调用 Qwen 系列及三方直供模型；OpenAI SDK 复用现有代码，DashScope SDK 适配百炼独有协议。 -- **应用调用**：DashScope SDK 提供 `Application.call`（Python）/ `Application.call(param)`（Java）封装 `POST /apps/{app_id}/completion`，调用智能体应用与工作流应用方式一致，统一返回 `output.text` 供业务消费。 -- **实时多模态交互**：Qwen-Omni-Realtime 通过 WebSocket 长连接驱动，Python/Java SDK 封装了客户端事件（`session.update`、`input_audio_buffer.append`、`response.create` 等）与服务端事件，支持 VAD 与 Manual 两种交互模式。 -- **框架集成**：LlamaIndex（Python 3.9+）构建云端知识库 RAG 应用；Spring AI Alibaba（Spring Boot 3.x + JDK 17+）通过 `spring-ai-alibaba-starter-dashscope` 集成智能体/工作流应用并检索知识库。 -- **异步任务**：图像/视频生成等耗时模型采用异步机制，SDK 封装创建任务、查询结果、取消任务接口（限流 20 QPS）；建议结合 EventBridge 事件通知避免轮询。 - -## 选用建议 - -- 已有 OpenAI 代码栈或需要多语言覆盖 → 优先 OpenAI 兼容 SDK，仅改三处参数即可迁移。 -- 需要百炼独有能力（Qwen-Audio、Omni Realtime、智能体应用封装、异步任务管理）→ 选用 DashScope SDK（Python/Java）。 -- 仅做一次性脚本或跨语言验证 → 可直接 HTTP 调用，跳过 SDK 安装。 - -> **注意**：使用 OpenAI 兼容接口时，新加坡与北京地域的 API Key 不同，切换地域需同步更换；Qwen-Audio 不支持 OpenAI 兼容协议，必须走 DashScope 协议。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [more about models](../api/more-about-models.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md index b22c24dd..efba1ac1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md @@ -1,62 +1,63 @@ # 流式输出 -流式输出（Streaming Output）指模型在生成过程中将结果以分块方式逐步返回给客户端，而非等待整体生成完成后一次性返回。开发者可在首 token 产出时即开始处理，从而显著降低首字延迟、改善交互体验，并支持边生成边消费的实时场景。 +流式输出（Streaming Output）是百炼平台提供的一种实时响应机制，允许模型在生成结果的过程中，将输出内容分块、逐步推送至客户端，而非等待全部内容生成完毕后一次性返回。该机制显著降低端到端延迟，提升用户交互体验，尤其适用于对话类、语音合成、长文本生成等对实时性敏感的场景。 -## 在百炼平台的使用方式 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼的流式输出按接入协议与生成任务类型分为三种形态： +流式输出在百炼平台中并非统一协议，而是根据接口类型和底层能力采用不同传输机制，开发者需按场景适配解析方式： -### 1. 文本生成流式（SSE） +- **HTTP SSE（Server-Sent Events）流式** + 主要用于 `knowledge/chat`（知识问答）和 `application-call/responses`（OpenAI 兼容 Responses API）等 REST 接口。服务端以 `text/event-stream` MIME 类型响应，每条消息格式为： + ``` + event: chunk + data: {"output":{"text":"你好"}} + + event: chunk + data: {"output":{"text":"，很高兴为您服务。"}} + + event: done + data: {"output":{"text":"，很高兴为您服务。"},"usage":{...}} + ``` + 客户端需监听 `chunk` 事件持续拼接文本，并在收到 `done` 事件后处理最终结果与统计信息。 -Qwen 系列文本模型通过 OpenAI 兼容、Anthropic 兼容或 DashScope 原生接口调用时，统一以 **Server-Sent Events (SSE)** 形式增量返回。在请求体中将 `stream` 设为 `true`，响应体按 `data: {chunk}\n\n` 逐 chunk 推送 `delta` 内容，最后一个 chunk 携带 `finish_reason` 与 usage 统计。 +- **WebSocket 实时流式** + 专用于 `Qwen-Omni-Realtime` 系列 API。通过双向 WebSocket 连接，服务端主动推送结构化事件（如 `response.text.delta`、`response.audio.delta`），支持文本、音频、工具调用状态等多模态增量输出，适用于语音助手、实时字幕等低延迟场景。 -- OpenAI 兼容接口：`POST /compatible-mode/v1/chat/completions`，字段与 OpenAI 客户端一致，可直接复用 SDK 的 stream 读取逻辑。 -- Anthropic 兼容接口：`POST /compatible-mode/v1/messages`，沿用 Anthropic `message_delta` / `content_block_delta` 事件结构，支持 thinking 与工具调用增量。 -- DashScope 原生接口：使用 `X-DashScope-SSE` 请求头启用，输出体为 SSE 流，参数集最完整。 +- **DashScope 原生 HTTP 流式** + 在 `Generation.call` 等原生接口中，通过 `stream=true` 启用，返回标准 SSE 格式；若需更细粒度控制（如仅增量返回新 token），可配合 `incremental_output=true` 参数，此时 `output.text` 字段为本次增量内容，而非累计全文。 -### 2. 实时多模态流式（WebSocket 双向消息） +- **非流式回退兼容** + 所有支持流式的接口均提供 `stream=false` 选项（默认值因接口而异），此时返回单次完整 JSON 响应，结构与流式末尾 `event: done` 的 `data` 字段一致，便于快速调试或轻量集成。 -Qwen-Omni-Realtime 通过 WebSocket 长连接实现低延迟语音/视频对话，与文本 SSE 不同，它是双向事件流： +> ⚠️ 注意：流式能力依赖服务端配置。例如工作流应用需在「结束节点」显式开启“流式输出”开关并重新发布；Omni Realtime 模型需在 `session.update` 中设置 `modalities: ["text", "audio"]` 才能触发音频流。 -- 客户端通过 `input_audio_buffer.append` 持续推送 Base64 音频片段； -- 服务端流式返回 `response.audio.delta`、`response.audio_transcript.delta` 等增量事件，实现"边听边说"。 -- 配合 VAD 模式可自动检测语音起止，配合 Manual 模式则由客户端显式 `response.create` 触发生成。 +## 关键参数和配置 -声音复刻、工具调用、联网搜索等能力均在该流式通道内完成，无需额外端点。 +| 参数 | 类型 | 说明 | 所属接口/场景 | +|------|------|------|----------------| +| `stream` | `boolean` | 全局开关，启用流式传输（SSE 或 WebSocket）。设为 `true` 时，响应头含 `Content-Type: text/event-stream`（HTTP）或建立 WebSocket 连接（Realtime）。 | 所有支持流式的接口（`knowledge/chat`, `responses`, `Generation`, `Application.call` 等） | +| `incremental_output` | `boolean` | **仅 DashScope 原生接口有效**。当 `stream=true` 时，若设为 `true`，则 `output.text` 返回本次增量内容；若为 `false`（默认），则返回当前累计全文。 | `dashscope.Generation` / 原生 `/generation` 接口 | +| `modalities` | `string[]` | **仅 Omni Realtime API 有效**。指定输出模态组合，如 `["text"]` 或 `["text","audio"]`，决定是否触发对应流式事件。 | `qwen3.5-omni-realtime` 等 WebSocket 接口 | +| `session_id`（流式上下文） | `string` | 在支持会话的流式调用中（如 `Application.call`），`session_id` 用于关联多轮流式响应，确保上下文连续性。注意：`Responses API` 异步模式（`background=true`）不支持流式。 | `Application.call`, `responses` 同步流式 | -### 3. Managed Agents 事件流（SSE） +## 面向开发者，简洁实用 -托管智能体在 Session 写入用户消息后，通过 SSE 端点 `GET /sessions/{session_id}/events/stream` 持续接收 Agent 的运行事件（思考、工具调用、回复文本等），直至 Session 回到 `idle`。该流式通道承载整个 Agent 执行生命周期，适合构建交互式 Agent UI。 +- ✅ **首选 SDK 调用**：Python 使用 `dashscope` SDK 的 `stream=True` 参数（如 `Generation.call(..., stream=True)`），SDK 自动处理 SSE 解析与事件分发，避免手动解析 `event:` 行。 +- ✅ **HTTP 调试建议**：用 `curl -N` 或浏览器 DevTools 的 Network → EventStream 查看原始流数据；生产环境务必设置超时（如 `timeout=120s`），防止连接挂起。 +- ✅ **错误处理要点**：流式请求失败时，可能已部分接收数据。请检查 HTTP 状态码（如 `401`, `429`）及首个 `event: error` 消息，不要仅依赖 `event: done`。 +- ❌ **避坑提示**： + - 工作流应用未在结束节点开启“流式输出” → 即使传 `stream=true` 也返回非流式响应； + - Omni Realtime 使用 `qwen-omni-turbo-realtime` 模型 → 不支持 `temperature`/`max_tokens` 等参数，但流式功能正常； + - `knowledge/chat` 接口 `stream=false` 时，响应结构与流式 `done` 事件 payload 完全一致，可复用同一解析逻辑。 -### 4. 异步任务的非流式约定 - -注意：图像、视频生成类接口（万相、HappyHorse、Kling 等）**不支持流式输出**，统一采用「创建任务得 `task_id` → 轮询 `GET /tasks/{task_id}`」的异步模式。视频任务通常耗时 1–5 分钟，需按建议间隔轮询，不应将其与流式输出混淆。 - -## 关键参数与配置 - -| 参数 / 头部 | 适用接口 | 作用 | -| --- | --- | --- | -| `"stream": true` | OpenAI / Anthropic 兼容、DashScope 原生 Chat | 启用 SSE 流式返回 | -| `X-DashScope-SSE: enable` | DashScope 原生接口 | 启用流式（部分原生接口必需） | -| `stream_options.include_usage` | OpenAI 兼容 | 在末尾 chunk 携带 token usage 统计 | -| WebSocket `session.update` | Omni Realtime | 配置 VAD、音色、工具等会话级参数 | -| `idle_timeout_ms` | `qwen3.5-omni-plus-realtime` 等 | 静默超时后模型主动引导对话 | -| `Accept: text/event-stream` | Managed Agents events/stream | 声明接收 SSE 事件流 | - -## 开发者注意事项 - -- **首字延迟 vs 总延迟**：流式不缩短总生成时间，但显著缩短首 token 等待，适合聊天、Agent 回复等交互场景。 -- **断连与重试**：SSE 与 WebSocket 均为长连接，网络抖动会中断流；建议记录已消费的 chunk 序号以便续接，或降级为非流式重试。 -- **工具调用**：流式下工具调用参数也是分块到达，需按 `tool_call_delta` 累积拼接后再执行。 -- **地域一致性**：实时多模态与托管 Agent 必须使用与 API Key 同地域的专属域名（如 `ws_xxx.cn-beijing.maas.aliyuncs.com`），跨地域会失败。 -- **[上下文窗口](context-window.md)**：流式不改变上下文限制，长对话仍需调用方截断或依赖 Responses 接口的自动历史管理。 +流式输出是构建高性能 AI 应用的关键能力。正确启用并解析它，能让您的产品获得接近本地响应的流畅体验。 ## 关联主题页 -- [qwen api reference](../api/qwen-api-reference.md) -- [image generation](../api/image-generation.md) +- [knowledge](../api/knowledge.md) +- [application call](../api/application-call.md) - [omni realtime api](../api/omni-realtime-api.md) -- [video generation api](../api/video-generation-api.md) -- [managed agents api](../api/managed-agents-api.md) +- [qwen api reference](../api/qwen-api-reference.md) +- [bailian application calling](../guides/bailian-application-calling.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md deleted file mode 100644 index 82223a7f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md +++ /dev/null @@ -1,47 +0,0 @@ -# 流式输出 - -流式输出（Streaming）是指服务端在生成内容的过程中，以增量分片（delta）的方式持续返回结果，而非等待全部内容生成完毕后一次性返回。它能显著降低首字延迟、改善实时交互体验，广泛用于对话、语音助手等场景。 - -## 在百炼平台的使用场景 - -百炼平台在多类接口中都支持流式输出，核心场景包括： - -- **应用调用（Application Call）**：无论是 OpenAI 兼容的 Responses API 还是 DashScope 原生 API，调用智能体或工作流应用时均支持流式输出，可用于需要边生成边展示的实时交互场景。 -- **文本生成模型 API**：OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages 以及 DashScope 原生接口均可开启流式返回，适合聊天补全类应用逐字/逐段渲染输出。 -- **实时多模态交互（Omni-Realtime API）**：基于 WebSocket 协议，流式是其原生工作方式。服务端通过一系列增量事件持续推送音频与文本，天然适配低延迟的语音对话场景。 - -## 关键参数与配置 - -### HTTP / SDK 接口 - -- **`stream`**：布尔值，控制是否开启流式输出，默认 `false`。设为 `true` 后，服务端以 SSE（Server-Sent Events）方式逐片返回结果。 - - 适用于 OpenAI 兼容 Responses API、DashScope API 等应用调用与模型调用接口。 -- 使用 SDK 时，开启 `stream=true` 后通过迭代响应对象逐步获取增量内容；部分接口可配合 `stream_options` 等参数控制是否返回用量统计等附加信息（以对应接口文档为准）。 - -### Omni-Realtime（WebSocket） - -实时接口不使用 `stream` 参数，而是以事件流的形式天然流式返回。关键的增量事件包括： - -| 事件 | 含义 | -| --- | --- | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | -| `response.done` | 本轮响应流结束 | - -此外，可通过 `session.update` 事件中的 `smooth_output`（部分模型支持）等参数调节流式输出的平滑度。 - -## 开发建议 - -- **注意结束标志**：SSE 流需处理到结束标记（如 `[DONE]`）或 `response.done` 事件后再收尾，避免内容截断。 -- **增量拼接**：客户端需将各 delta 分片按序拼接，才能得到完整结果。 -- **错误处理**：流式过程中仍可能收到错误事件，需在读取流的循环中做好异常捕获与连接重试。 -- **跨接口差异**：不同接口的分片结构与字段命名不同（OpenAI/Anthropic 兼容接口以对应生态约定为准，DashScope 参数最全），跨接口迁移时需核对字段映射。 - -## 关联主题页 - -- [application call](../api/application-call.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md deleted file mode 100644 index 169b6401..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md +++ /dev/null @@ -1,69 +0,0 @@ -# Token 与计费 - -Token 是百炼平台衡量模型处理文本量的基本单位，也是绝大多数计费、限流和用量统计的计量基础；平台的费用则围绕 Token 消耗，通过按量付费、免费额度、节省计划与资源包等机制综合结算。理解 Token 如何被计量与抵扣，是控制大模型使用成本的前提。 - -## Token 在计费中的角色 - -对大语言模型、全模态模型和向量模型，用量与费用均按 **Token** 计量；图像生成按「张」、视频生成按「秒」、语音模型按「秒/字符/Token」（视模型而定）。因此谈「计费」时，Token 主要针对文本类调用。 - -模型调用按**输入 Token** 和**输出 Token** 分别计费，单价以「每百万 Token」为单位。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档定价，落入某一区间后该请求全部 Token 按对应档位结算（例如 `qwen3-max` 华北2·北京划分为 `0 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** - -- **免费额度**：首次开通时各模型自动发放（通常每模型 100 万 Token），仅抵扣实时推理，不抵扣 Batch、调优、部署；不同模型（含同一模型不同快照版本）额度相互独立。开启「免费额度用完即停」后，额度耗尽会停止响应并返回 `AllocationQuota.FreeTierOnly`（或 403），避免意外扣费。 -- **折扣**：Batch 调用输入/输出单价按实时推理价的 50% 计费；支持上下文缓存的模型仅输入 Token 享折扣，两者不能同时生效。 -- **地域差异**：同一模型在不同地域单价不同，境外地域通常无免费额度。 - -### 2. 模型训练（调优） - -按训练 Token 计费。文本模型公式为 `(训练数据 Token + 混合数据 Token) × 循环次数 × 训练单价`；图像/视频生成模型的训练 Token 总量由 `max_steps`、`max_pixels`、`n_epochs` 等超参决定。免费额度和节省计划**均不抵扣**训练费用。 - -### 3. 模型部署 - -免费额度和节省计划同样**不抵扣**部署费用。三种计费方式围绕 TPM（每分钟 Token 数）： - -- **预置吞吐（PTU）**：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`。PTU 下超出购买吞吐或输入超上限时自动转按量付费（响应头 `x-dashscope-ptu-overflow:true`）。长输入按阶梯系数折算 TPM 消耗，命中前缀缓存的 Token 按折扣系数消耗额度。 -- **模型单元（MU）**：`费用 = 使用时长 × 模型单元数量 × 模型单元单价`。 -- **按 Token 使用量**：仅对 LoRA 微调后的自定义模型开放，用于调优效果验证。 - -### 4. 订阅套餐（Token Plan / Coding Plan） - -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，面向团队协作。 -- **Coding Plan**：按模型调用次数计量，面向个人开发。 - -两者均使用 `sk-sp-` 前缀的**专属 API Key**，**不消耗新人免费额度**，且 Base URL 与按量付费端点完全隔离，混用会导致意外扣费或 401/403 鉴权失败。 - -### 5. 监控与用量统计 - -模型监控将 Token 消耗归入「成本」类指标，并提供首 Token 延时、RPM/TPM 等性能指标。应用观测则可查看每次调用的输入/输出/平均 Token 量与平均首 Token 耗时。开通推理日志后可查看单次调用的 Token 消耗，用于排查与审计。 - -## 关键参数与配置 - -- **`max_tokens`**：限制单次生成的最大输出 Token 数，是控制输出成本和防止过度生成的首要手段。 -- **TPM / RPM 限流**：部署时可配置 `tpm_limit`、`rpm_limit`；触发限流后等待时间取决于限流值。 -- **PTU 计费识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度，`default` 或缺失表示按量）、`provisioned_tokens`（折算后实际消耗额度）、`cached_tokens`（缓存命中数，Anthropic 兼容格式暂不返回）。这些字段在 OpenAI Chat、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下 JSON 路径不同，需按协议读取。 - -## 成本优化与出账 - -- **节省 Token**：优化 Prompt 减少输入 Token、简单任务选用轻量级模型、非实时任务走批量推理、合理设置 `max_tokens`。 -- **预付费方案**：AI 通用型节省计划（承诺月消费换阶梯折扣，最高 5.3 折，月额度不可跨月累积）、其他模型节省计划、资源包（预购具体 Token/图片数量，仅抵扣单个模型超免费额度后的实时推理）。 -- **出账时效**：大模型推理分钟级出账（约 2～10 分钟）；批量推理、模型训练等小时级出账。用量统计数据约 1 小时延迟，且不支持查看 30 天前数据。 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) -- [support](../guides/support.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md index 0b4e6e3c..af5d22b3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md @@ -1,65 +1,51 @@ -# Token 与计量计费 +# Token 计量与管理 -Token 是大语言模型处理文本的最小计量单位，百炼平台以 Token 为核心，对模型的输入、输出、训练用量进行计量、计费与监控。理解 Token 的产生方式与计费规则，是控制模型调用成本、优化应用性能的基础。 +Token 计量与管理是百炼平台对大模型调用资源消耗进行标准化度量、实时追踪、精准计费与精细化治理的核心机制。它以 **Token** 为最小计量单元，覆盖输入、输出、缓存等全链路消耗，并统一映射到 Credits（Token Plan）或按量账单（Pay-as-you-go），支撑成本控制、用量分析与性能优化。 -## 什么是 Token +## 在百炼平台的不同场景中，这个概念如何使用 -Token 是模型在处理文本时切分出的基本片段（一个汉字、单词或子词可能对应一个或多个 Token）。在百炼平台中，Token 既是**用量计量单位**，也是**大部分计费的核算基准**： +- **Token Plan 订阅服务**：以 Credits 为计费单位，按实际消耗的 `input_tokens`、`output_tokens` 和 `cache_tokens` 动态抵扣；模型白名单严格限定可计量范围，非白名单模型调用不计入 Credits，可能触发按量扣费。 +- **模型监控（Model Monitoring）**：提供分钟级/小时级 `model_usage` 指标（含 `input_tokens`/`output_tokens`/`total_tokens` 等 `usage_type` 维度），支持按 `model`、`apikey_id`、`workspace_id` 等标签过滤，用于成本归因与异常排查（北京地域支持单次请求级 Token 查看）。 +- **应用观测（Application Monitoring）**：在 Span 级别精确统计 `Input Tokens` 与 `Output Tokens`，关联至具体节点（如 `LLM`、`EMBEDDING`），支持按链路深度、节点类型、状态筛选，是智能体/工作流成本拆解与性能瓶颈定位的关键依据。 +- **模型评测（Model Evaluation）**：评测任务执行时，被评测模型推理和裁判模型评分均产生 Token 消耗——前者计入被评测模型用量，后者计入裁判模型用量，二者独立计量、分别计费。 +- **应用支持（Application Support）**：插件调用、RAG 检索、[流式输出](streaming-output.md)等能力本身不额外计 Token，但其触发的模型调用（如 LLM 生成响应、Embedding 向量化）仍遵循标准 Token 计量规则；`incremental_output=True` 不改变总 Token 数，仅影响传输方式。 -- 文本生成模型按**输入 Token** 和**输出 Token** 分别计量，思考模式下的输出 Token 同时包含「思维链 + 回答」两部分。 -- 不同模型类型的计量单位不同：大语言模型 / 全模态模型 / 向量模型按 **Token** 计量，图像生成按**张**，视频生成按**秒**，语音模型按**秒 / 字符 / Token**（视模型而定）。 +## 关键参数和配置 -## 在各场景中的使用 +| 参数 | 说明 | 注意事项 | +|------|------|----------| +| `input_tokens` | 模型接收到的 Prompt、上下文、工具描述、图片 Base64 编码等输入内容所占 Token 数 | 图片按分辨率折算（如 `qwen-image-2.0` 使用固定 token 开销 + 可变视觉 token）；系统自动去除冗余空格与换行，但不压缩语义 | +| `output_tokens` | 模型实际生成的文本或结构化响应（含 function call 参数）所占 Token 数 | 流式响应中累计计数，`incremental_output` 不影响总量；截断（`max_tokens`）会限制此值上限 | +| `cache_tokens` | 模型缓存 Prompt 或历史对话产生的额外开销（如 KV Cache 预分配） | 当前仅部分模型（如 `qwen3.7-plus`）在启用缓存优化时显式计量，多数场景隐含在 `input_tokens` 中 | +| `total_tokens` | `input_tokens + output_tokens`（缓存通常不单独计入） | 计费与监控口径统一以此为准 | +| `usage_type` | Prometheus 指标 `model_usage` 的关键 label | 必须显式指定 `input_tokens`/`output_tokens`/`total_tokens` 才能正确聚合 | -### 1. 按量付费与免费额度 +> ⚠️ 重要约束： +> - Token 计量基于模型实际 tokenizer 行为，**不接受客户端预估**；开发者不可自行计算并传入 `token_count` 参数。 +> - 所有计量均发生在服务端，调用返回的 `usage` 字段（如 OpenAI 兼容 API 中的 `"usage": {"prompt_tokens": ..., "completion_tokens": ...}`）为唯一可信来源。 +> - 图像生成（`multimodal-generation` API）与视觉理解（多模态输入）的 Token 计算逻辑与纯文本模型不同，需查阅对应模型文档确认细则。 -- 首次开通时，各模型会发放新人专属免费额度（通常各 100 万 Token），仅抵扣**实时推理**费用，且不同模型（含同一模型不同快照版本）额度相互独立、不可合并。 -- 免费额度耗尽后默认转为**按量付费**，按输入 / 输出 Token 计费。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档（如 `qwen3-max` 分 0–32K / 32K–128K / 128K–256K 三档），落在哪一档，该请求全部 Token 均按该档单价结算。 -- 同一模型在不同地域（北京、弗吉尼亚、新加坡、法兰克福、东京）单价不同。 +## 面向开发者，简洁实用 -### 2. 订阅制套餐 +- ✅ **必查返回值**：每次成功调用后，务必解析响应中的 `usage` 字段，用于本地日志记录、预算预警或用量上报。 +- ✅ **善用监控工具**：在北京地域部署关键应用时，开启模型监控的「高级监控」并配置告警，当 `model_usage{usage_type="total_tokens"}` 异常飙升时快速定位问题模型或恶意请求。 +- ✅ **成本优化实践**： + - 对长上下文场景，优先启用 `cache_tokens` 支持的模型（查看模型文档支持列表）； + - 评测任务中，用规则评估替代大模型评估可规避裁判模型 Token 费用； + - Token Plan 用户应定期检查控制台「用量分析」，识别高消耗模型/成员，及时调整分配策略。 +- ❌ **避免踩坑**： + - 不要复用通用 API Key（`sk-`）调用 Token Plan 模型——将导致 401 错误或意外按量扣费； + - 不要尝试通过修改 `max_tokens` 或 [prompt](../guides/prompt.md) 格式“欺骗”Token 计量——平台按真实 tokenizer 输出计费； + - 不要依赖前端渲染逻辑（如 Markdown 解析）估算 Token——实际消耗由模型侧 tokenizer 决定。 -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，工具调用（联网搜索、代码解释器等内置工具）产生的 Token 同样从套餐 Credits 抵扣，不额外收费。 -- **Coding Plan**：按模型调用**次数**计费（而非 Token），面向个人开发场景。 - -### 3. 模型训练与部署 - -- **训练**按训练 Token 计费，文本模型公式为 `（训练数据 Token + 混合训练数据 Token）× 循环次数 × 训练单价`；图像 / 视频模型的训练 Token 由 `max_steps`、`max_pixels`、`n_epochs` 等超参数推算。 -- **部署**（预置吞吐 TPM）按输入 / 输出 TPM 单价与时长计费，与 Token 用量间接相关。训练与部署**不能**用免费额度或节省计划抵扣。 - -### 4. 监控与观测 - -- 应用观测可查看每次调用的 Token 量，监控统计提供 Token 总量（全部 / 输入 / 输出）、平均单次请求 Token 量、平均首 Token 耗时等指标。 -- 模型监控将 Token 消耗归入「成本」类指标，首 Token 延时归入「性能」类指标，支持按分钟 / 小时 / 天聚合，并可配置告警。 -- 用量统计按业务空间维度归集，数据延迟约 1 小时。 - -## 关键参数与配置 - -| 项 | 说明 | -| --- | --- | -| `max_tokens` | 限制单次生成的最大输出 Token 数，用于控制成本、防止过度生成，也是降低幻觉的手段之一 | -| 上下文缓存 | 命中缓存的**输入** Token 享折扣；缓存折扣与 Batch 折扣不可同时生效，价格表输入单价不含缓存单价 | -| Batch 调用 | 支持的模型输入 / 输出单价按实时推理价的 50% 计费 | -| 免费额度用完即停 | 开启后额度耗尽即停服（返回 403），避免意外扣费，但也会阻断节省计划继续抵扣 | - -## 成本优化建议 - -- **优化 Prompt**：简洁清晰的 Prompt 可减少不必要的输入 Token 消耗。 -- **控制输出长度**：合理设置 `max_tokens`，避免冗长输出。 -- **按任务选模型**：分类、摘要等简单任务优先使用轻量级模型。 -- **批量推理**：非实时大批量任务走 Batch 接口，Token 单价更低。 -- **善用抵扣顺序**：`免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费`，据此规划预付费方案。 - -## 账单中的 Token - -大模型推理为分钟级出账（通常 2～10 分钟），批量推理与训练为小时级。账单「实例 ID」以英文分号分隔，包含 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此区分输入 / 输出 Token 的费用来源与调用渠道（`app` 代码调用、`bmp` 控制台体验、`assistant-api`）。 +Token 计量是百炼平台资源治理的基石。理解它，就是掌握成本、性能与合规的主动权。 ## 关联主题页 - [token plan guide](../guides/token-plan-guide.md) -- [test 1](../guides/test-1.md) -- [application monitoring](../guides/application-monitoring.md) - [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) +- [application monitoring](../guides/application-monitoring.md) +- [model evaluation introduction](../guides/model-evaluation-introduction.md) +- [application support](../guides/application-support.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md deleted file mode 100644 index 195756dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md +++ /dev/null @@ -1,57 +0,0 @@ -# 向量与嵌入 - -向量与嵌入（Embedding）是指将文本、图像、视频等非结构化内容映射为固定维度的数值向量，使语义相近的内容在向量空间中距离也更近，从而支持语义检索、聚类、推荐与 RAG 召回等下游任务。在百炼平台中，向量化是知识库检索与跨模态搜索的基础环节，与 Rerank 排序模型共同构成检索链路。 - -## 在百炼平台的使用场景 - -- **知识库检索召回**：文档搜索、数据查询、音视频搜索类知识库在创建时会对接向量模型，将入库切片向量化并写入索引；查询时对用户问题做同样的向量化，再执行向量 + 关键词混合检索。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；视觉理解场景自动切换为 `qwen3-vl-embedding`；图片问答类仅支持 `multimodal-embedding-v1`（1024 维）。 -- **RAG 应用构建**：无论是云端知识库还是本地 RAG 方案，嵌入模型都决定了召回上限。云端方案使用百炼官方嵌入模型，不支持自定义切分与嵌入；本地 RAG 可改用自部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`），以灵活控制切分与嵌入。 -- **跨模态检索**：多模态向量模型将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。`qwen3-vl-embedding` 默认维度 2560，支持「独立向量」（每个输入各生成一向量，用于逐项对比）与「融合向量」（所有输入融合为 1 个向量，用于综合理解）两种类型。 -- **框架集成**：通过 LlamaIndex 构建 RAG 应用时，云端知识库默认使用百炼官方向量模型，配合 `similarity_top_k`、`similarity_cutoff`、`top_n` 等参数控制召回与重排。 - -## 文本向量模型 - -通用文本向量模型将文本转换为数值向量，当前推荐 `text-embedding-v4`（属 Qwen3-Embedding 系列），支持 100+ 主流语种。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -**关键参数：** - -- `model`（必选）：模型名称 -- `input`（必选）：字符串、字符串列表或文件 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持 -- `encoding_format`（可选）：当前仅支持 `float` - -**调用方式**：支持 OpenAI 兼容接口（`base_url: https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。对于大规模文本向量化，可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行，通过 `X-DashScope-Async: enable` 请求头启用异步模式，提交后用 `task_id` 轮询结果；同时处理中任务不超过 50 个，并发上限 3。 - -## 多模态向量模型 - -多模态向量模型将文本、图像、视频统一映射到同一语义空间，支持跨模态检索。向量类型分「独立向量」（每个输入分别生成向量，适用于逐项对比）与「融合向量」（将所有输入融合为 1 个向量，适用于综合理解多模态内容）。`qwen3-vl-embedding` 默认维度 2560，支持独立与融合两种模式；`multimodal-embedding-v1` 默认 1024 维，用于图片问答类知识库。 - -## 知识库中的嵌入配置要点 - -- 文档搜索、数据查询、音视频搜索类知识库统一使用 512 维 `text-embedding-v4` / `text-embedding-v3`，维度不可更改。 -- 视觉理解场景自动切换为 `qwen3-vl-embedding`，无需手动指定。 -- 嵌入模型与切片策略共同决定召回质量：推荐使用智能切分（基于语义自适应选择切片点，单切片 Token 上限 6,000），切片过大或过小都会影响向量匹配效果。 -- 本地 RAG 场景若需自定义嵌入模型，可改用自部署 GTE 文本向量模型，但需注意 embedding API 限流，不建议传入超过 100 MB 的文件。 - -## 与 Rerank 的协作 - -向量检索负责语义召回（初步 TopK 默认 50，可设 1–100），Rerank 模型负责二次精准排序。知识库中可选 `qwen3-rerank` / `qwen3-rerank(hybrid)` / `qwen3-vl-rerank`（多模态库只能选 vl-rerank），排序后再按相似度阈值（0.01–1.0）与最大召回数量（1–20）裁剪。向量召回质量直接影响 Rerank 上限，调优时通常先确认嵌入模型与切片合理，再调整 Rerank 阈值与 K 值。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md deleted file mode 100644 index 523b5b79..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md +++ /dev/null @@ -1,55 +0,0 @@ -# 私网访问 - -私网访问（VPC Private Access）是指阿里云百炼平台的模型调用、[模型部署](model-deployment.md)、异步任务通知等能力通过 VPC 内网或私网链路访问，避免公网暴露、提升传输安全性与网络稳定性的接入方式。它是百炼安全合规体系中"传输加密—私网访问"环节的核心抓手，与身份权限、内容安全、合规备案、安全存储共同构成端到端安全链路。 - -## 在百炼平台中的使用场景 - -### 1. 模型调用的私网接入 - -百炼模型调用默认通过公网（`dashscope.aliyuncs.com`）发起，对安全要求高或需统一出口的企业可走 VPC 私网。临时 API Key 可在不可信客户端环境使用，配合 VPC 私网访问可在内网完成调用链路，避免永久 Key 在公网流转。 - -> 各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，临时 Key 与生成它的永久 Key 必须属于同一地域，私网链路也需在对应地域内打通。 - -### 2. 异步任务完成通知的 VPC 回调 - -图像/视频生成等耗时任务接入事件总线 EventBridge 后，任务完成事件由事件总线推送。HTTP 回调 URL 接收端**支持公网或 VPC 访问**，对安全敏感场景建议使用 VPC 类型的 HTTP 接口作为事件目标，使通知链路全程在内网闭环。 - -- 事件源：`acs.dashscope` -- 事件类型：`dashscope:System:AsyncTaskFinish` -- 事件总线默认为 `default`（北京地域云服务专用总线） - -RocketMQ 消息队列方案同样可部署在 VPC 内，由事件总线投递后业务方在内网消费，支持消息无丢失与失败重试，可靠性更高。 - -### 3. [模型部署](model-deployment.md)与模型导入的私网通道 - -[模型部署](model-deployment.md)（PTU、模型单元、按 Token 用量）仅适用于华北二（北京）地域，部署后推理服务可在 VPC 内调用，满足高并发、低延迟及不出公网的诉求。 - -部署自定义 LoRA 模型时，需将本地训练的 LoRA 模型从阿里云 OSS 导入百炼。OSS Bucket 支持**私有 Bucket**与**内容加密**，并需添加 `bailian-datahub-access` 标签（值为 `read`）。通过 VPC 终端节点或私网访问 OSS，可避免训练产物经公网传输，保障数据安全。 - -## 关键参数与配置 - -| 维度 | 关键约束 | -| --- | --- | -| 地域一致性 | API Key、临时 Key、私网链路须同地域；模型部署仅北京 | -| 临时 Key 有效期 | 默认 60 秒，可通过 `expire_in_seconds` 设置，范围 [1, 1800] 秒 | -| 异步任务接口限流 | 20 QPS（按主账号 + 子账号维度）；通知方案不限流 | -| 任务保留时长 | 完成后约 24 小时自动清理 | -| OSS 导入标签 | Bucket 须加 `bailian-datahub-access=read` 标签 | -| OSS 存储类型 | 不支持归档/冷归档/深度冷归档；须使用子目录，非根目录 | -| 授权角色 | OSS 导入需开通 `AliyunServiceRoleForSFMDataHubOSSImport` 服务关联角色 | - -## 开发者实践建议 - -- **生产隔离**：按环境（dev/test/prod）划分业务空间，prod 流量走 VPC 私网，降低公网泄露面。 -- **回调优先 VPC**：异步任务通知的 HTTP 回调端点部署在 VPC 内，事件总线直推即可，无需公网暴露。 -- **OSS 导入走私网**：LoRA 模型导入使用私有 Bucket + 内容加密，并通过 VPC 终端节点访问 OSS。 -- **避免跨地域**：规划业务空间时明确地域，避免因 Key 与链路跨地域导致调用失败。 -- **PTU 部署内网化**：高负载生产环境优先 PTU 部署并经 VPC 调用，兼顾稳定吞吐与传输安全。 - -## 关联主题页 - -- [security and compliance](../guides/security-and-compliance.md) -- [more about models](../api/more-about-models.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md deleted file mode 100644 index 0badc9bf..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md +++ /dev/null @@ -1,75 +0,0 @@ -# 工作流 - -工作流（Workflow）是百炼平台三种核心应用构建模式之一，通过可视化节点编排将复杂任务拆解为有序步骤，逻辑确定、稳定可复现，适合流程固定、要求可复现的业务场景。 - -## 在百炼平台中的定位 - -百炼提供智能体、工作流、高代码应用三种互补的应用构建模式。工作流对应"可视化节点编排（低代码）"路线，由预定义节点精确控制流程，适合 IT 运维、业务分析师构建报告生成、订单处理、审批流等场景。与由大模型自主规划的智能体不同，工作流强调开发者对流程的显式控制，输出可复现、可调试。 - -## 核心节点 - -工作流画布由以下节点类型组成： - -- **开始 / 结束节点**：定义输入与输出参数。开始节点预置 `query`（用户输入）、`historyList`（对话历史）、`imageList`（图片）等变量，可在下游节点中引用。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆等。 -- **意图分类节点**：根据输入分流到不同下游分支，支持多意图。 -- **变量处理节点**：用于文本输出或变量加工。 -- **智能体群组节点**：将任务分解给多个已发布的子智能体协同完成。 - -## 会话变量 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在各节点中引用，在画布右上角配置。适合跨节点传递状态、累积上下文信息。 - -## 典型案例 - -- **诈骗信息识别**：开始 → 大模型（提示词判定诈骗嫌疑）→ 结束。 -- **智能导购**：意图分类节点将输入分流到手机 / 电视 / 冰箱等大模型分支，未命中分支走变量处理节点。 -- **日程管理助手**：通过智能体群组节点串联"信息收集"与"数据整理"两个子智能体。 - -## 创建与发布流程 - -1. 控制台 → 应用管理 → 创建应用 → 工作流应用。 -2. 在画布上拖拽编排节点，配置各节点的模型、提示词、变量引用关系。 -3. 右侧对话框调试。 -4. 右上角"发布"——发布是后续 API 调用与集成的前提。 - -## API 调用 - -工作流应用与智能体应用共用同一套调用接口，区别仅在应用内部的编排逻辑。发布后通过 `APP_ID` 调用： - -- **DashScope 原生 API**：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`，请求体为 `{"input": {"prompt": "..."}, "parameters": {}, "debug": {}}`，响应中业务侧主要消费 `output.text`。 -- **OpenAI 兼容 Responses API**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，可复用现有 OpenAI 生态代码库，支持同步 / 异步、流式、多模态。 -- **SDK**：Python 使用 `dashscope.Application.call`，Java 使用 `com.alibaba.dashscope.app.Application`，Node.js 可直接以 `axios` 发起 POST。 - -调用前需在控制台获取应用 ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）；若应用位于子业务空间，还需提供 Workspace ID。 - -## 多轮对话 - -工作流应用支持多轮对话，有两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮。 -- **自行管理 `messages`（推荐）**：手动维护消息列表，灵活性更高，不受 session 有效期限制。 - -## 评测 - -工作流应用可作为评测任务的被测对象。在新版应用评测中，评测集支持智能体、工作流、自定义三种类型，按所选应用的出入参形式自动生成数据模板。评测任务可关联工作流应用，由 LLM 评估器或 Code 评估器自动评分，也可人工标注。 - -## 发布与分享 - -工作流应用支持多种发布渠道：UI 应用（通过 UI 设计器构建自定义界面）、钉钉机器人、微信公众号、组件化复用（将工作流作为其他智能体或工作流的子组件）、音视频实时互动（仅限图文对话类）。需要注意的是，官方网页版分享当前只支持智能体应用，不支持工作流应用。 - -## 关键限制与注意事项 - -- 工作流应用若配置了文件类型的自定义参数，在 UI 设计器中需指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），否则应用无法正确读取用户上传的文件。 -- 旧版"智能体编排"应用已被工作流应用替代，新建应用请直接选择工作流。 -- 工作流与智能体调用接口一致，但可附加的扩展能力（如自定义参数传递）取决于应用内部编排逻辑。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application evaluation](../guides/application-evaluation.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md deleted file mode 100644 index 0dc4ae49..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md +++ /dev/null @@ -1,63 +0,0 @@ -# 业务空间（Workspace） - -业务空间（Workspace）是阿里云百炼平台中进行精细化权限管理、资源隔离与阿里云账单分账的**最小管理单元**。平台按地理区域划分资源，单个业务空间不能跨地域存在——即使是各地域的默认业务空间，也彼此独立、互不相同。 - -## 核心定位 - -- **权限最小单元**：模型调用、模型训练、模型部署、用户页面权限等授权都以业务空间为边界。 -- **资源与成本隔离**：不同业务空间的数据、应用、API Key 相互隔离，可作为账单分账的粒度。 -- **地域绑定**：每个业务空间归属唯一地域（如华北2（北京）、新加坡），拥有各自的 `WorkspaceId`。 - -## 在不同场景中的使用 - -### 1. OpenAPI 调用 - -百炼应用组件 API（`bailian/2023-12-29`，数据连接、知识库、Prompt 模板、长期记忆等）的**所有接口都需传入 `WorkspaceId`**。RAM 子账号必须先获得对应权限策略并加入目标业务空间，才能发起调用。类目、文件等资源也在业务空间内计数（如每空间最多 500 个类目）。 - -### 2. 权限与身份管理 - -权限体系围绕三种角色展开，权限范围自上而下递减： - -- **超级管理员**：阿里云主账号，或拥有 `AliyunBailianFullAccess` 系统策略的 RAM 用户，可跨空间统一管理用户权限、可用模型、限流和 API Key。 -- **业务空间管理员**：拥有某业务空间「权限管理」页面访问权的 RAM 用户，管理该空间内的用户与资源，权限包含访问该空间所有页面。 -- **普通用户**：按分配的权限使用被授权的空间、页面与资源。 - -> OpenAPI 接口权限不通过业务空间角色授予，必须由阿里云主账号在 RAM 控制台添加专用系统策略（如 `AliyunBailianDataFullAccess` / `AliyunBailianDataReadOnlyAccess`）。 - -### 3. 模型级精细化授权 - -在非默认业务空间内，可对模型进行三类控制（**默认业务空间无法设置这些限制，所有模型均可调用、调优、部署**）： - -| 权限项 | 控制范围 | 配置入口 | -| --- | --- | --- | -| 限制模型调用 | 是否可调用（控制台 & API）+ 请求数限流 + Token 限流 | 模型列表 → 模型调用列开关 + 当前空间限流列 | -| 限制模型训练 | 是否可调优及调优后部署 | 模型列表 → 模型授权 → 模型训练列 | -| 限制模型部署 | 是否可直接部署 | 模型列表 → 模型授权 → 模型部署列 | - -### 4. API Key 归属 - -单个 API Key 只能归属一个地域内的**一个业务空间和一个用户**，且不能转移。其可调用功能与模型限流始终与**归属业务空间**的权限保持一致，不受用户控制台权限影响。将 RAM 账号移出业务空间会使其 API Key 失效（重新加入后恢复），在 RAM 控制台删除账号/角色则永久失效。 - -### 5. 应用观测与数据管理 - -- **应用观测**：端到端查看业务空间内应用（智能体应用、工作流应用、高代码应用）的调用链路与延时、Token 量等指标。若应用未出现在观测列表，常因未发布或不属于当前业务空间。 -- **数据管理**：统一管理业务空间下的训练集与评测集。注意数据管理与数据处理能力**仅适用于华北2（北京）地域**。 - -## 关键要点与配置 - -- **`WorkspaceId`**：业务空间唯一标识，是几乎所有应用组件 API 的必填参数。 -- **默认业务空间的限制**：无法设置模型调用/训练/部署限制，也无法限流。 -- **地域隔离**：跨地域需分别在对应地域的业务空间下管理资源与权限。 -- **生产实践**： - - **空间规划**：推荐按环境（dev/test/prod）或按业务线划分业务空间，实现隔离与成本管理。 - - **限流分配**：将主账号总配额按比例下发到各业务空间并预留缓冲。例如总配额 1000 QPM，可分配 prod 600 / test 200 / dev 100，预留 100。 - -## 关联主题页 - -- [application component api reference](../api/application-component-api-reference.md) -- [application permission management](../guides/application-permission-management.md) -- [application monitoring](../guides/application-monitoring.md) -- [security and compliance](../guides/security-and-compliance.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md index 0db8c3d7..c409420b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md @@ -1,152 +1,47 @@ # application evaluation -阿里云百炼提供完整的应用[评测体系](../concepts/evaluation.md)，支持对[智能体应用](../concepts/agent-application.md)和工作流应用的输出质量进行系统化评估。平台同时提供自动评测与手动评测两种模式，并通过评测集、评估器和标签三大组件构建多维度的评测闭环。当前平台存在新旧两套评测系统，新版在评测任务管理、评估器和标签体系上做了较大升级。 +应用评测是百炼平台用于系统化评估智能体/工作流应用输出质量的核心能力，支持自动与手动两种评测范式。自动评测基于大模型与知识库自动生成评测集并完成端到端评分，适用于快速迭代与横向对比；手动评测则依赖人工构建评测集与标注，适用于高精度、强主观性或需深度归因的场景。两类评测均围绕评测集、评测任务、评估器与标签四大核心组件展开，形成可配置、可复用、可追溯的质量保障闭环。 -## 评测模式 +## 支持的模型/功能 -百炼支持两种评测模式，分别适用于不同场景： +- **自动评测**：仅支持 `qwen-max` 和 `qwen-plus` 两种模型用于评测集生成与最终评分，不支持其他模型（如 `qwen-turbo` 或 `qwen2` 系列）[原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。该限制同样适用于评测规则配置阶段。 +- **评估器类型**：新版评测体系支持 LLM 评估器（调用大模型进行语义评分）、Code 评估器（执行 Python 脚本进行规则校验）及基于历史评测任务自动生成的 LLM 评估器 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。LLM 评估器默认限时免费，但实际调用仍产生 Token 费用。 +- **评测集类型**：当前存在两套并行体系： + - 旧版仅支持 **对话分析**（`.xls`/`.xlsx`）和 **知识问答**（`.jsonl`）两类，分别用于手动评测与自动评测 [原文标题](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)； + - 新版扩展为 **智能体**、**工作流** 和 **自定义** 三类，支持按应用出入参结构自动生成模板，并引入版本管理与表结构编辑能力 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)。 +> **注意**：文档 3 与文档 4 对评测集类型的定义存在明显差异——前者限定为“对话分析”和“知识问答”，后者升级为“智能体/工作流/自定义”。这反映平台已从单一 RAG 场景向通用应用评测演进，**旧版类型已逐步被新版覆盖，新建评测应优先采用新版评测集**。 -### 自动评测 +## 关键参数 -[自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)利用大模型基于应用关联的知识库自动生成评测集，并对智能体的回答进行自动评分，生成评测报告与调优建议。支持两种子模式： +- **评测集字段映射**：所有评估器（尤其是预置模板）对输入字段有明确要求。例如，“问答相关性”评估器必需 `query` 和 `response` 字段；若评测集字段名为 `Prompt`/`Completion`，必须在参数映射中显式绑定 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +- **分类采样数**：自动评测中，需为每种任务类型（事实型、教程型等）单独设置采样数量，直接影响评测覆盖面与 Token 消耗 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **评估器评分范围与阈值**：LLM/Code 评估器均需配置 `评分范围`（如 `0-1` 或 `1-5`）和 `通过阈值`（如 `0.8` 或 `4`），二者共同决定 Pass/Fail 判定逻辑，且必须在 Prompt 中保持语义一致 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +- **标签类型约束**：标签创建时需指定类型（分类/布尔值/数字/文本），不同类型对应不同筛选条件与标注方式，影响后续指标统计维度 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)。 -- **单应用评测**：深度评估单个[智能体应用](../concepts/agent-application.md)的表现，生成包含评分、错误分析和优化建议的详细报告。 -- **多应用横向评测**：在同一评测基准下对比最多 8 个应用（或同一应用的不同版本），用于选型决策或版本迭代效果验证。 +## 使用方式 -前提条件： +1. **准备数据基础**： + - 创建评测集：可选择自动生成（仅限知识问答类型，依赖知识库）或手动上传（支持 `.xls`/`.xlsx`/`.jsonl` 格式，单文件 ≤20MB）[原文标题](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)； + - 发布评测集：草稿状态不可用于评测，必须点击“发布”使其生效； + - （可选）创建标签与评估器：按业务需求定义多维标注体系与自动化评分规则。 -1. 仅面向**已发布**的[智能体应用](../concepts/agent-application.md)，且应用须已配置知识库。 -2. 须开通**应用观测**功能，并将待评测应用添加到观测列表。 -3. 子账号需获取`管理员`或`应用评测-操作`权限。 -4. 多应用横向评测时，所有被选应用必须关联至少一个相同的知识库。 +2. **发起评测任务**： + - **自动评测**：进入控制台自动评测页面 → 选择已发布且配置知识库的智能体应用 → 选择知识库 → 生成或选用评测集 → 设置采样数与评测模型 → 发起任务 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)； + - **手动评测**：上传评测集 → 进入手动评测页 → 选择应用与已发布评测集 → 配置评测维度 → 开始评测 → 人工打标（较差/一般/较好 或 1–5 分）→ 提交结果； + - **新版评测任务**：支持“不关联应用”（纯人工标注）、“智能体”或“工作流”关联模式，并可同时添加最多 10 个评估器与任意标签 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 -自动评测流程分四步：创建评测任务 → 设置评测集 → 配置评测规则 → 执行评测。评测集生成和评估模型当前仅支持 `qwen-max` 和 `qwen-plus`。 +3. **分析与迭代**： + - 查看报告：自动评测提供总正确率、BadCase 归因（模型理解/重排/检索/切片/未获取知识）、RAG 各类型得分；手动评测提供人工标注汇总； + - 使用标签筛选 BadCase，结合评估器结果定位问题根因； + - 基于归因建议优化 Prompt、知识库切分策略或检索配置，发布新版本后复用同一评测集验证效果。 -### 手动评测 +## 限制和注意事项 -[手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)通过人工构建评测集，对应用回答进行人工分析与评分。流程为：准备评测集（下载模板填充数据）→ 上传评测集 → 创建评测任务 → 人工标注打分 → 查看评测报告。 - -手动评测适用于需要领域专家主观判断的场景，评测维度支持使用内置模板或自定义评测维度模板。 - -> **注意**：手动评测属于旧版评测系统的功能。新版评测系统通过「评测任务 + 标签」的组合同样支持人工标注场景，且功能更灵活。 - -## 评测集 - -评测集是评测任务的数据基础，用于存储和管理评测数据。 - -### 旧版评测集 - -[旧版评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)支持两种类型： - -| 类型 | 文件格式 | 适用场景 | -|------|----------|----------| -| 对话分析 | `.xls` / `.xlsx` | 人工评测，包含 Prompt、Completion、SessionId 字段，支持多轮对话 | -| 知识问答 | `.jsonl` | 自动评测，包含 query、queryType、referenceAnswer、fineKeywords、coarseKeywords 字段 | - -创建方式：自动生成（基于知识库，仅知识问答类型）或手动上传。单次上传最多 10 个文件，单个文件不超过 20MB。 - -### 新版评测集 - -[新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)支持三种类型： - -| 类型 | 说明 | -|------|------| -| 智能体 | 根据选中智能体应用的出入参形式定义评测集 | -| 工作流 | 根据选中工作流应用的出入参形式定义评测集 | -| 自定义 | 任意定义评测集表结构，适用于特殊评测场景 | - -新版评测集支持手动上传和从应用观测导入两种创建方式，并具备版本管理能力，每次发布生成新版本。创建后类型不可修改。 - -## 评估器 - -[评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)是新版[评测体系](../concepts/evaluation.md)的核心组件，用于自动评估应用输出质量。支持三种创建方式： - -### 基于预置模板 - -百炼提供多种预置评估器模板，覆盖以下分类： - -- **通用质量**：评估回答的基本质量指标 -- **智能体**：专门用于评测智能体应用 -- **文本匹配**：精确规则文本匹配 -- **文本相似度**：计算文本相似度得分 -- **格式校验**：验证输出格式规范性 - -### 自定义评估器 - -| 类型 | 评估方式 | 适用场景 | 成本 | -|------|----------|----------|------| -| LLM 评估器 | 大模型语义理解 | 相关性、有害性、幻觉检测 | 产生 [Token](../concepts/token.md) 费用 | -| Code 评估器 | Python 代码规则判断 | 格式校验、数值计算、精确匹配 | 无额外费用 | - -### 基于评测任务创建 - -通过历史评测任务的标注结果自动抽象为新的 LLM 评估器，适用于将人工标注经验固化为自动化评估规则的场景。需选择已完成评估的评测任务，并配置 query、response、label_score 的字段映射。 - -每个评测任务最多支持添加 10 个评估器。建议组合 3-5 个评估器从不同维度评估应用质量，例如：相关性评估器（LLM）+ 格式校验评估器（Code）。 - -## 标签管理 - -[标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)用于对评测数据和应用观测数据进行自定义标注，支持四种标签类型： - -| 标签类型 | 数据类型 | 适用场景 | -|----------|----------|----------| -| 分类 | 预定义选项（最多 20 个） | 回答质量分级、错误类型分类 | -| 布尔值 | True / False | 是否正确、是否存在幻觉 | -| 数字 | Double 数值 | 评分（1-5）、相关性得分（0-1） | -| 文本 | 自由文本 | 错误原因说明、改进建议 | - -标签可同时用于评测任务的人工标注和应用观测的数据标注，支持基于标签的数据筛选和指标统计。 - -## 评测任务（新版） - -[新版评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)支持智能体和工作流应用的评测，核心配置包括： - -- **选择评测集**：从已发布的评测集列表中选择评测集和版本 -- **关联应用**：支持不关联应用（纯人工标注）、关联工作流或关联智能体三种方式 -- **添加评估器**：配置自动评分规则及参数映射 -- **添加标签**：配置人工标注维度 - -任务详情页提供数据明细和指标统计两个视图，支持普通模式和快速标注两种标注方式。 - -## 评测报告与归因分析 - -自动评测完成后生成评测报告，包含以下维度： - -- **总正确率**：得分 >= 4 分的回答占比（评分范围 1-5 分） -- **BadCase 分析**：按分数从低到高展示错误评测条目 -- **调优建议**：基于归因分析提供 Prompt、检索配置或知识库切片的具体优化建议 -- **RAG 智能体评价**：按问题类型展示单项得分 - -归因分析将 BadCase 定位到 RAG 流程的具体环节： - -| 归因类型 | 含义 | 优化方向 | -|----------|------|----------| -| 模型理解有误 | 已获取正确知识但推理错误 | 优化提示词或切换更强模型 | -| 重排不佳 | 正确切片排序靠后 | 调整重排配置或增加切片数量 | -| 检索无效 | 召回切片过多或过少 | 调整检索策略 | -| 切片不完整 | 语义单元被分割到多个切片 | 增大切片长度或启用语义切分 | -| 未获取知识 | 知识库缺少相关内容 | 补充知识库内容 | - -## 最佳实践 - -### 建立持续评测机制 - -以下场景建议触发评测：知识库更新后、调整 Prompt 后、更换或升级模型后、调整检索/重排策略后、定期回归（每周或每月）。 - -### 优化闭环 - -识别 BadCase → 分析归因定位问题 → 实施针对性优化 → 发布新版本再次评测 → 对比结果确认改进。若效果未达预期则继续迭代。 - -## 计费说明 - -评测任务调用大模型产生的 [Token](../concepts/token.md) 费用正常计费。自动评测的评测集生成和评估均会消耗 [Token](../concepts/token.md)，预估平均消耗仅为参考值，最终以实际账单为准。评估器模型当前限时免费。 - -## 常见问题 - -- **评测集生成进度长时间保持 0%**：评测集生成和应用评测为离线任务，需后台排队执行，排队期间进度保持 0%，任务开始后自动更新。 -- **评测期间能否关闭应用观测**：不可以，否则可能导致评测任务失败或数据丢失。 -- **评测报告中用例数量与设置不符**：自动评测可能部分失败，报告仅展示成功完成的用例。 -- **评测任务创建后可否修改**：任务配置（应用、评测集）不可修改，但可随时添加人工标签。如需不同配置请创建新任务。 +- **权限与前提**：自动评测要求子账号具备 `管理员` 或 `应用评测-操作` 权限，且目标应用必须已发布、配置知识库、并加入应用观测列表 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **数量限制**：单次自动评测最多支持 8 个应用横向对比；单个评测任务最多添加 10 个评估器；单次上传评测集文件不超过 10 个，单文件 ≤20MB。 +- **Token 消耗**：所有调用大模型的操作（评测集生成、自动评分、LLM 评估器）均产生 Token 费用，预估消耗仅为参考，实际以账单为准；试运行也会消耗少量 Token [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **评测失败处理**：自动评测中失败用例不计入正确率计算；手动评测中未完成打标的条目不影响已完成部分的统计。 +- **兼容性提示**：新版评测任务（文档 5/6/7）与旧版自动/手动评测（文档 1/2/3）共存，但二者数据模型与流程不互通。**新建项目应统一使用新版体系**，旧版功能仅维持兼容，不再新增特性。 ## 来源文档 @@ -154,15 +49,8 @@ - [手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - [新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) -- [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) +- [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md index b636ebef..28d20d23 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md @@ -1,127 +1,64 @@ # application monitoring -阿里云百炼提供**应用观测**功能，用于端到端查看[业务空间](../concepts/workspace.md)内应用（[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用、高代码应用）的处理流程，并获取延时、[Token](../concepts/token.md) 量等关键指标，指标更新频率为分钟级。该功能可帮助开发者追踪应用内部调用链路、查看模型响应延时与思考过程，进而优化运营效果与成本。详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 - -## 支持的应用范围 - -应用观测支持以下三类应用： - -- **[智能体应用](../concepts/agent-application.md)**（AgentApp） -- **[工作流](../concepts/workflow.md)应用**（WorkflowApp） -- **高代码应用**（FullCodeApp） - -> **注意**：应用观测暂不支持通过 Assistant API 创建的[智能体应用](../concepts/agent-application.md)；对高代码应用，目前不支持追踪其内部调用链路，仅能观测到入口 CHAIN 节点。应用观测本身也没有 API，只能通过控制台操作。 - -## 前提条件与开通 - -首次使用需在应用观测页面右上角完成**应用观测配置**，依次执行：授权可观测链路 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore。 - -- 推荐使用**主账号**操作，开通后通常分钟级生效，高峰期可能略有延迟。 -- 如需**子账号**开通，主账号需为其配置 `AliyunBailianFullAccess` 全局权限、`应用观测-操作`（或 `管理员`）页面权限，并额外授予 `ram:CreateServiceLinkedRole` 系统策略（用于创建服务关联角色）。 - -> 子账号权限若未配置完整，开启应用观测时会失败。配置完成后需返回应用观测界面再次尝试开启。 +应用观测（Application Monitoring）是阿里云百炼平台提供的端到端可观测能力，用于追踪和分析智能体、工作流及高代码类应用的内部执行链路。它支持查看调用延时、Token 消耗、模型思考过程及各节点状态，并提供分钟级指标聚合与原始 Span 数据导出能力。该功能基于 OpenTelemetry 构建，需依赖可观测链路服务，**当前不提供 API 接口** [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)。 + +## 支持的模型/功能 + +- **支持的应用类型**：智能体应用、工作流应用、高代码应用（但高代码应用仅上报 `CHAIN` 根节点，**不支持内部链路追踪**） +- **核心可观测维度**： + - 调用链路（Root Span / All Span / Model Span 三种视图） + - 延时（含平均首 Token 耗时、平均调用时长） + - Token 统计（输入/输出/总量） + - 状态（正常/错误，含错误类型细分） + - 节点类型与嵌套关系（如 `LLM`、`RETRIEVER`、`EMBEDDING`、`GUARDRAIL` 等） +- **扩展能力**： + - 数据标注（布尔值、分类、数字、文本四类标签） + - 批量导出（JSONL / Excel） + - 添加 Span 到评测集（支持字段映射与导入策略配置） +- **不支持场景**：通过 Assistant API 创建的智能体应用 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)；[长期记忆](../concepts/long-term-memory.md)中的检索过程；高代码应用内部节点 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)。 + +## 关键参数 + +| 参数 | 说明 | 备注 | +|------|------|------| +| `Request ID` / `Trace ID` / `Span ID` | 用于精准定位单次调用或子链路 | 可在节点详情页点击「查看 ID」获取 | +| `Span Name` | 节点逻辑名称（如 `AgentApp`, `TextRetriever`, `LLM`） | 支持模糊匹配筛选 | +| `Status` | `normal` 或 `error`，错误时可进一步区分类型（如 `GuardrailBlocked`, `LLMTimeout`） | — | +| `Latency (ms)` | 节点执行耗时（含网络与模型推理时间） | `LLM` 节点延时包含流式响应全过程 | +| `Input Tokens` / `Output Tokens` | Embedding 或 LLM 调用的 Token 数量 | 定义见 [附录](#f0ed9407canlv)（原文档） | +| `Label` | 用户自定义标注字段（类型强约束） | 与评测系统共享标签管理 | + +> **注意**：`TextRetriever` 和 `VectorRetriever` 默认返回 100 个切片，且**暂不支持调整数量**；此限制在文档中被多次强调，属设计约束而非临时限制。 ## 使用方式 -### 1. 选择被观测的应用 - -在应用观测页面单击「选择被观测的应用」>「添加」。若列表中看不到已创建的应用，通常是因为该应用尚未发布，或应用不属于当前[业务空间](../concepts/workspace.md)。 - -### 2. 开始观测 - -添加完成后，应用会出现在观测列表中。此后所有输入该应用的 Prompt 及相关数据、指标会被自动追踪并以分钟级频率同步。单击「关闭观测」可停止同步，重新添加后仅同步新增数据。 - -在「查看详情」中可查看最长 30 天内的调用记录，包括 Prompt 内容、输出、延时、调用时间和 [Token](../concepts/token.md) 量，并支持按 Request ID / Trace ID / Span ID 检索和按时间范围筛选。单击节点名称可查看详情、原始数据和标注记录。 - -> 列表中的 **CHAIN** 节点表示一次完整的应用内部调用追踪，支持展开。状态分为「正常」与「错误」两类。 - -### 3. 导出数据 - -在应用详情页的 Trace 列表页签右上角单击「导出数据」，可将当前筛选条件下的数据导出为 **JSONL** 或 **EXCEL** 格式。 - -### 4. 查看监控统计 - -「监控统计」页签提供性能监控图表：调用次数（含失败次数与失败率）、[Token](../concepts/token.md) 总量（全部/输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时（流式场景）、平均调用时长。支持按时间范围（最长 30 天）和聚合粒度（分钟/小时/天）查看，每个图表可放大、下载、复制。 - -## 数据筛选与标注 - -### Span 筛选模式 - -- **Root Span**：仅显示根节点（默认） -- **All Span**：平铺展示所有 Span -- **Model Span**：仅显示包含模型调用的 Span - -### 过滤器 - -支持按状态（正常/错误，可按错误类型细分）、Span Name、输入、输出、延时、[Token](../concepts/token.md) 总量、输入 [Token](../concepts/token.md)、输出 [Token](../concepts/token.md)、标签等字段添加多个筛选条件，条件之间组合应用。 - -### 数据标注 - -支持对 Span 数据添加标签（布尔值/分类/数字/文本四种类型），标签与应用[评测](../concepts/evaluation.md)的标签管理共享、统一管理。标注内容自动保存，并可在 Span 列表页「标签」列查看。 - -### 添加到[评测](../concepts/evaluation.md)集 - -应用观测支持将 Span 数据直接加入[评测](../concepts/evaluation.md)集，将真实线上调用作为评测样本。配置时需选择目标评测集、导入方式（追加数据或全量覆盖）并完成字段映射。每个评测集最多支持 50 个字段映射。 - -## 节点类型 +### 前置配置（仅首次使用需执行） +1. 使用主账号（或已授权子账号）进入 [应用观测配置](https://bailian.console.aliyun.com/tab=app?tab=app#/app-observe) 页面； +2. 授权 `AliyunServiceRoleForOpenTelemetry` 服务关联角色； +3. 开通可观测链路 OpenTelemetry 服务并初始化 LogStore。 -被观测应用在调用过程中会按操作单元生成不同类型的**节点**，节点之间可形成嵌套关系。仅在被触发或调用时才展示对应节点。完整节点类型与说明见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +> 子账号需额外配置 `CreateServiceLinkedRole` 权限策略，详见 [常见问题](#cd0f1152d50hj)（原文档锚点）。 -### [智能体应用](../concepts/agent-application.md)节点 +### 日常操作流程 +1. **添加应用**：在应用观测列表中点击「添加」，仅支持已发布且归属当前业务空间的应用； +2. **查看数据**： + - 在 Span 列表页切换筛选模式（Root/All/Model Span）； + - 使用过滤器按 `Status`、`Span Name`、`Input`、`Output`、`Latency`、`Tokens` 或 `Label` 组合筛选； + - 单击节点名称展开详情，查看原始请求/响应、标注记录、子节点等； +3. **导出与复用**： + - 点击「导出数据」下载 JSONL 或 Excel； + - 选中 Span 后点击「添加到评测集」，完成字段映射与导入策略配置。 -| 节点 | 说明 | -| --- | --- | -| CHAIN | 连接大模型节点与其他节点，处理复杂任务；作为根节点时名称为 AgentApp 或 WorkflowApp | -| AGENT | 对智能体的调用 | -| RETRIEVER | 检索操作；KnowledgeRetriever 表示在[知识库](../concepts/knowledge-base.md)中检索。子节点名称含 TextRetriever（改进 BM25，默认返回 100 个切片）、VectorRetriever（向量检索，默认返回 100 个切片） | -| REWRITER | 基于会话上下文调整原始 Prompt 以提升检索效果 | -| EMBEDDING | 将 Prompt 转为向量，[Token](../concepts/token.md) 量为本次向量化的 [Token](../concepts/token.md) 数 | -| RERANKER | 计算文本切片相似度分数并降序排列 | -| LLM | 大模型推理/文本生成，[Token](../concepts/token.md) 量 = 输入 + 输出；延时包含输出回复过程 | -| TOOL | 插件调用（官方或自定义） | -| GUARDRAIL | 阿里绿网调用；ManualIntervention 为用户干预规则，SystemIntervention 为系统干预规则 | +## 限制和注意事项 -> 目前暂不支持观测长期记忆中的检索过程；TextRetriever 与 VectorRetriever 默认返回 100 个切片，暂不支持调整数量。 - -### [工作流](../concepts/workflow.md)应用节点 - -除上述 CHAIN、RETRIEVER、REWRITER、EMBEDDING、RERANKER、LLM、GUARDRAIL 外，还包含工作流专属节点：START（开始）、END（结束）、API、CLASSIFIER（意图分类）、TEXT_CONVERTER（文本转换）、SCRIPT（脚本转换）、CONDITION（条件判断）、FUNCTION_COMPUTE（函数计算）、APP_FLOW。 - -### 高代码应用节点 - -仅有 CHAIN（FullCodeApp）作为入口节点，目前不支持追踪其内部调用链路。若已开启观测却看不到调用量等统计数据，需排查：代码中是否使用 AgentScope-AI 的 Tracing 模块定义上报信息，以及部署时是否添加 `--telemetry enable` 参数。 - -## [计费](../concepts/billing.md)说明 - -应用观测功能本身**不收费**，但观测数据需存储在可观测链路 OpenTelemetry 服务中，相关存储费用由 OpenTelemetry 服务收取。 - -## 关键指标说明 - -- **延时（调用时长）**：对 LLM 节点，包含输出回复的完整过程。 -- **[Token](../concepts/token.md) 量**：Embedding 节点为本次向量化 [Token](../concepts/token.md) 数；LLM 节点为输入 [Token](../concepts/token.md) + 输出 [Token](../concepts/token.md)。 -- **数据时效**：指标更新频率为分钟级，调用记录最长可查 30 天。 -- **应用总量 / 平均延时**：用于评估应用运营效果与成本，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +- **无 API 支持**：应用观测为纯控制台功能，不开放 SDK 或 RESTful 接口 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)； +- **数据延迟**：指标同步频率为**分钟级**，不适用于实时告警场景； +- **存储计费**：功能本身免费，但底层 OpenTelemetry 存储费用需单独承担； +- **高代码应用限制**：即使开启观测，也仅上报 `FullCodeApp` 根节点，无法观测其内部[函数调用](../concepts/function-calling.md)或自定义逻辑——若需细粒度追踪，必须在代码中集成 `AgentScope-AI` 的 Tracing 模块并部署时启用 `--telemetry enable` 参数； +- **权限要求**：子账号开通需满足三重权限（`AliyunBailianFullAccess` + 页面写入权限 + `CreateServiceLinkedRole` 策略），缺一不可。 ## 来源文档 - [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md index 6d21a266..04de291f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md @@ -1,136 +1,60 @@ # application permission management -阿里云百炼支持基于控制台页面级、模型级的多维度权限控制，满足多地域、多用户的复杂组织架构需求。单个[业务空间](../concepts/workspace.md)是进行精细化权限管理（模型、用户）和阿里云账单分账的最小管理单元，权限管理围绕超级管理员、[业务空间](../concepts/workspace.md)管理员、普通用户三种角色展开。详细的角色定义与权限矩阵见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +百炼平台的权限管理以“业务空间”为最小单元，支持跨地域、多角色的精细化控制，覆盖模型调用/调优/部署、用户页面访问、API Key 管理及 OpenAPI 接口调用等核心场景。权限策略严格遵循阿里云 RAM 体系，需结合控制台操作与 RAM 策略协同配置。详细设计逻辑请参见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 -## 角色体系 +## 支持的模型/功能 -百炼的身份管理基于以下三种角色，权限范围自上而下递减： +- **模型级管控**：支持对单个模型在指定业务空间内独立设置： + - 调用权限（含控制台 & API） + - 调优（训练）权限 + - 部署权限 +- **资源维度隔离**：业务空间按地域物理隔离，同一地域内可创建多个业务空间，但**单个业务空间不能跨地域存在**（详见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 +- **角色能力矩阵**： + | 功能 | 超级管理员 | 业务空间管理员 | 普通用户 | + |---|---|---|---| + | 模型调用 & 限流 | ✅ | ❌ | ❌ | + | 模型调优 | ✅ | ❌ | ❌ | + | 模型部署 | ✅ | ❌ | ❌ | + | 用户管理 | ✅ | ✅ | ❌ | + | 页面权限管理 | ✅ | ✅ | ❌ | + | API Key 管理 | ✅ | ✅ | ❌ | + | OpenAPI 接口权限 | ❌（仅主账号可开通） | ❌ | ❌ | -- **超级管理员**：可跨空间统一管理用户权限、空间可用模型、空间模型限流和 [API Key](../concepts/api-key.md)。包含两类账号：阿里云主账号，以及拥有 `AliyunBailianFullAccess`（百炼管理员）系统策略的 RAM 用户。超级管理员可通过百炼全局管理菜单（北京 / 新加坡 / 弗吉尼亚）为任意 RAM 用户授权任意地域、任意空间的几乎所有权限，仅 OpenAPI 接口权限需阿里云主账号添加。 -- **[业务空间](../concepts/workspace.md)管理员**：拥有访问某个[业务空间](../concepts/workspace.md)「权限管理」页面的 RAM 用户，只负责该特定[业务空间](../concepts/workspace.md)内的用户权限和资源管理。管理员权限包含可访问该[业务空间](../concepts/workspace.md)下所有页面的权限。 -- **普通用户**：根据分配的权限使用资源，可访问/使用被授权的空间、页面、资源。 +> **注意**：文档中多次强调“默认业务空间无法设置模型调用/调优/部署限制”，但未明确说明该限制是否适用于所有地域。实际配置时请以控制台实时提示为准，避免依赖默认空间进行生产环境权限隔离 —— 此点与 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) 中“应用于生产环境”章节推荐的按环境划分空间策略存在隐含冲突。 -### 权限矩阵 +## 关键参数 -| [业务空间](../concepts/workspace.md)权限 | 超级管理员 | [业务空间](../concepts/workspace.md)管理员 | 普通用户 | -| --- | --- | --- | --- | -| 允许特定模型调用 & 限流 | 支持 | 不支持 | 不支持 | -| 允许特定[模型调优](../concepts/fine-tuning.md) | 支持 | 不支持 | 不支持 | -| 允许特定[模型部署](../concepts/model-deployment.md) | 支持 | 不支持 | 不支持 | -| 用户管理 | 支持 | 支持 | 不支持 | -| 用户可用页面管理 | 支持 | 支持 | 不支持 | -| [API Key](../concepts/api-key.md) 管理 | 支持 | 支持 | 不支持 | -| 访问/使用被授权的空间、页面、资源 | 支持 | 支持 | 支持 | -| OpenAPI 接口权限 | 不支持 | 不支持 | 不支持 | +- **业务空间 ID（Workspace ID）**：API 调用必需参数，用于标识资源归属空间，获取方式见 [获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +- **API Key 归属约束**：单个 API Key 仅绑定**一个地域 + 一个业务空间 + 一个 RAM 用户**，不可迁移；其可用模型与限流策略完全继承自归属业务空间的配置（[API-Key 权限](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 +- **限流粒度**：支持 QPM（每分钟请求数）和 Token 限流两种模式，均在业务空间维度配置。 +- **OpenAPI 权限策略**：必须由阿里云主账号在 RAM 控制台显式授予 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess`，RAM 用户默认无权调用应用、知识库、Prompt 工程等核心 OpenAPI（[OpenAPI 接口权限](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 -> **注意**：OpenAPI 接口权限不通过[业务空间](../concepts/workspace.md)角色授予，必须由阿里云主账号在 RAM 控制台为 RAM 用户添加专用系统策略。 +## 使用方式 -## [业务空间](../concepts/workspace.md)权限管理 +1. **角色初始化**： + - 超级管理员：主账号或拥有 `AliyunBailianFullAccess` 策略的 RAM 用户，通过全局管理菜单（[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) / [新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/business_management) / [弗吉尼亚](https://modelstudio.console.aliyun.com/us-east-1?tab=globalset#/efm/business_management)）统一配置。 + - 业务空间管理员：由超级管理员或同空间管理员在控制台「权限管理」页签中为 RAM 用户授予「管理员」角色。 -百炼按地理区域划分资源和[业务空间](../concepts/workspace.md)，**单个[业务空间](../concepts/workspace.md)不能跨地域存在**，即使是各地域的默认[业务空间](../concepts/workspace.md)，也是不同的空间。[业务空间](../concepts/workspace.md)是精细化权限管理的最小单元，可管理以下维度（默认[业务空间](../concepts/workspace.md)无法设置这些限制）： +2. **模型权限开通（必需前置步骤）**： + - 超级管理员需先在全局管理菜单中为业务空间启用目标模型的**调用、调优或部署权限**（默认业务空间自动全开，但不支持限流）。 -- **限制模型调用**：管理某个模型可否在该[业务空间](../concepts/workspace.md)调用（控制台 & API），并设置该模型的请求数限流和 [Token](../concepts/token.md) 限流。默认[业务空间](../concepts/workspace.md)所有模型均可调用且无法限流。 -- **限制模型训练**：管理某个模型可否在该[业务空间](../concepts/workspace.md)进行调优和调优后部署。默认[业务空间](../concepts/workspace.md)所有支持调优的模型均可调优及部署。 -- **限制[模型部署](../concepts/model-deployment.md)**：管理某个模型可否在该[业务空间](../concepts/workspace.md)直接部署。默认[业务空间](../concepts/workspace.md)所有支持部署的模型均可部署。 -- **用户控制台权限管理**：管理某个 RAM 用户是否能使用该业务空间控制台的功能及能使用哪些功能，但无法限制归属该用户的 [API Key](../concepts/api-key.md) 的调用。阿里云主账号无须设置，可访问所有业务空间的所有页面。 +3. **用户权限分配**： + - 控制台操作：在业务空间「权限管理」页签中，为 RAM 用户勾选对应功能模块权限（如「模型体验-操作」「模型调优-操作」「批量推理-操作」等）。 + - API 调用：为用户在目标业务空间创建 API Key，Key 的能力范围由该空间模型权限决定，**不受用户控制台权限影响**。 -关于业务空间的地域隔离与限流细节，可进一步参考 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +4. **OpenAPI 授权**： + - 主账号登录 RAM 控制台 → 找到目标 RAM 用户 → 添加 `AliyunBailianDataFullAccess`（读写）或 `AliyunBailianDataReadOnlyAccess`（只读）系统策略。 -## API-Key 权限 +## 限制和注意事项 -单个 [API Key](../concepts/api-key.md) 只能归属一个地域内的一个业务空间和一个用户，且不能转移。[API Key](../concepts/api-key.md) 可调用的功能和模型限流与**归属业务空间**的权限保持一致，不受用户控制台权限管理的影响，也无需为不同模型（如文生文、文生图、语音合成）创建不同的 [API Key](../concepts/api-key.md)。 - -[API Key](../concepts/api-key.md) 的状态随归属用户操作变化： - -| 触发操作 | 主账号的 [API Key](../concepts/api-key.md) | RAM 账号的 [API Key](../concepts/api-key.md) | -| --- | --- | --- | -| 主动删除 [API Key](../concepts/api-key.md) | 失效，不可恢复 | 失效，不可恢复 | -| 将账号移出业务空间 | — | 失效（重新加入后恢复生效） | -| 在 RAM 控制台删除账号/角色 | — | 失效，不可恢复 | -| 为 [API Key](../concepts/api-key.md) 设置 IP 访问白名单 | 华北2（北京）地域支持 | 华北2（北京）地域支持 | - -> **注意**：自 2026 年 3 月 25 日起，华北2（北京）地域的所有新创建的 [API Key](../concepts/api-key.md) 均归属主账号。 - -可通过百炼控制台左侧导航栏「权限管理」页签为 RAM 用户添加 API-Key 权限，赋予其创建、删除、查看该空间下所有 API-Key 的权限。 - -## OpenAPI 接口权限 - -RAM 用户默认无权调用百炼应用的数据、[知识库](../concepts/knowledge-base.md)、Prompt 工程及长期记忆等功能的 Open API。需由阿里云主账号在 RAM 控制台为 RAM 用户添加以下权限之一： - -- `AliyunBailianDataFullAccess`：可调用百炼应用 API 目录下的所有 API。 -- `AliyunBailianDataReadOnlyAccess`：可调用百炼应用 API 目录下的只读类 API，如 `DescribeFile`、`GetIndexJobStatus` 等。 - -## 账单与预付费权限 - -RAM 用户默认无权查看阿里云账单和购买预付费产品，需在 RAM 控制台添加特定权限。这两项权限会授予 RAM 用户查看**所有产品**账单或购买**所有预付费产品**的权限，请谨慎授权。 - -- 查看阿里云账单：添加 `AliyunBSSReadOnlyAccess`。 -- 购买阿里云预付费产品：添加 `AliyunBSSOrderAccess`。 - -## 常用配置流程 - -### 设置超级管理员 - -需要阿里云主账号或具备 `AliyunRAMFullAccess` 系统策略的 RAM 用户操作。前往 RAM 控制台为 RAM 用户添加 `AliyunBailianFullAccess` 和 `AliyunBSSOrderAccess` 权限后，即可通过百炼全局管理菜单授权任意地域、空间的权限并购买预付费产品。 - -### 设置业务空间管理员 - -需超级管理员或业务空间管理员操作。在百炼控制台左侧导航栏「权限管理」页签内为 RAM 用户添加「管理员」权限。 - -### 设置模型调用权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的模型调用权限。 -2. 通过控制台调用时，需由超级管理员或业务空间管理员为 RAM 用户添加：**模型体验-操作**（控制台调用模型）、**批量推理-操作**（支持批量推理）、**模型观测-操作**（查看 [Token](../concepts/token.md) 消耗量）。 -3. 通过 API 调用时，需为 RAM 用户在对应业务空间创建或分配 API Key。 - -### 设置[模型调优](../concepts/fine-tuning.md)权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的[模型调优](../concepts/fine-tuning.md)（训练）权限。 -2. 在「权限管理」页签内为 RAM 用户添加以下权限：**模型体验-操作**、**模型调优-操作**、**我的模型-操作**（管理调优后模型快照）、**[模型部署](../concepts/model-deployment.md)-操作**（部署调优后的模型）、**模型[评测](../concepts/evaluation.md)-操作**、**数据管理-操作**（管理调优数据集）、**模型观测-操作**。 -3. 通过 API 调优时，无需额外控制台权限，只需为 RAM 用户分配 API Key 即可。 - -完整的权限配置流程与截图说明见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 - -## 生产环境实践 - -- **空间规划策略**：推荐按环境划分（开发 `project-dev-workspace`、测试 `project-test-workspace`、预发与生产 `project-prod-workspace`），实现严格的环境隔离；也可按业务线划分（如 `marketing-team-workspace`、`customer-team-workspace`），便于权限和成本管理。 -- **限流策略**：将主账号总配额按比例分配给各业务空间并预留缓冲。例如账号总配额 1000 QPM，可分配生产 600 QPM（60%）、测试 200 QPM（20%）、开发 100 QPM（10%）、预留缓冲 100 QPM（10%），以应对突发流量。 - -## 限制与注意事项 - -- 业务空间不能跨地域存在；不同地域的默认业务空间也是不同空间。 -- 默认业务空间无法设置模型调用、调优、部署限制，所有模型均按默认策略可用且无法限流。 -- API Key 不可跨业务空间或跨用户转移；账号移出业务空间后其 API Key 失效，重新加入后恢复。 -- OpenAPI 接口权限、账单与预付费权限必须由阿里云主账号在 RAM 控制台授权，业务空间管理员无法授予。 -- `AliyunBSSReadOnlyAccess` / `AliyunBSSOrderAccess` 为全产品级权限，授权范围远超百炼本身，需谨慎。 -- 开通 AI 安全护栏、模型监控、应用观测等功能，建议使用阿里云主账号在控制台一次性授权开通。 - -## 常见问题 - -- **如何获取业务空间 ID**：参考应用开发的「获取 Workspace ID」文档。 -- **如何使用子业务空间调用模型**：无需特殊设置，使用子业务空间的 API Key 即可。 -- **如何使用特定业务空间的应用**：使用 API 管理和调用特定业务空间的应用时，需同时设置 APP ID 和 Workspace ID。 +- **地域强绑定**：业务空间与地域一一对应，API Key、模型限流、用户权限均不可跨地域复用。 +- **默认空间限制**：默认业务空间无法配置模型调用/调优/部署限制，且不支持限流，**严禁用于生产环境**（[权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) 明确建议按环境或业务线新建独立空间）。 +- **API Key 生命周期**：RAM 用户被移出业务空间后，其 API Key **立即失效**（重新加入后恢复）；若在 RAM 控制台删除该用户，则 Key **永久失效**。 +- **账单与预付费权限**：RAM 用户需额外授予 `AliyunBSSReadOnlyAccess`（查看账单）或 `AliyunBSSOrderAccess`（购买预付费）策略，且这些权限作用于**全部阿里云产品**，非百炼专属，授权需谨慎。 +- **IP 白名单支持范围**：仅华北2（北京）地域的 API Key 支持设置 IP 访问白名单。 ## 来源文档 - [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md index 110f5e44..0df1e08f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md @@ -1,79 +1,46 @@ # application publishing and sharing -阿里云百炼支持将已构建并发布的应用以多种渠道对外分享，或封装为可复用的模块化组件供其他应用接入。本页汇总了[智能体应用](../concepts/agent-application.md)的分享渠道、组件化发布与接入方式，以及基于魔笔能力的 UI 设计器发布流程，面向需要将百炼应用集成到实际业务的开发者。 +百炼平台支持将已发布的智能体应用（Agent 1.0）或工作流应用以多种方式对外共享与集成，包括生成可访问的 UI 应用、发布为跨平台机器人（钉钉/微信）、封装为可复用组件、以及接入音视频实时互动场景。所有发布行为均需基于已上线的应用，并受 Agent 版本、权限空间和计费模型约束。 -## 适用范围与版本限制 +## 支持的模型/功能 -分享渠道（魔笔/UI 应用、钉钉、微信、组件、音视频实时互动）均为 **Agent 1.0** [智能体应用](../concepts/agent-application.md)的功能。 +- **仅限 Agent 1.0**：魔笔分享渠道、钉钉机器人、微信公众号、组件发布、音视频实时互动等功能**全部仅支持 Agent 1.0 智能体应用**；Agent 2.0 应用不支持上述任何发布渠道，仅可通过 API 调用 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **UI 应用支持范围更广**：UI 设计器支持集成**智能体应用（Agent 1.0/2.0）和工作流应用**，但前提是二者与 UI 所属业务空间一致 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **组件来源多样**：智能体应用和工作流应用均可发布为组件，且组件可在智能体或工作流中被引用 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 -> **注意**：**Agent 2.0** [智能体应用](../concepts/agent-application.md)仅支持通过 API 调用，**不支持**上述任何分享渠道。若需分享，请确认应用版本。 +> **注意**：文档 1 明确限定“分享渠道均为 Agent 1.0 功能”，而文档 3 在“准备工作”中指出 UI 设计器支持“智能体应用或工作流应用”，未限定 Agent 版本；结合控制台实际能力，UI 集成对 Agent 2.0 的支持是例外情形，但组件发布、钉钉/微信等渠道严格不兼容 Agent 2.0。 -前提条件是已有构建好且**已发布**的智能体应用。所有分享渠道通过百炼控制台 **应用管理 → 目标应用卡片 → 发布** 进入。详见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +## 关键参数 -## 分享渠道 +| 参数 | 说明 | 约束 | +|------|------|------| +| `API Key` | 用于身份认证与调用鉴权，必须与应用、UI 同属一个业务空间 | 缺失时需在发布流程中创建或管理；钉钉/微信/音视频配置均依赖此密钥 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | +| `query` / `imageList` | 组件预设系统参数：`query`（String，必填）传递用户文本输入；`imageList`（Array，非必填）传递图像公网地址 | 预设参数不可删除，无需显式定义；若组件不处理图像，应将 `imageList` 设置为“是否可见 = 否” [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) | +| `传参方式`（业务透传 / 模型识别） | 决定参数值由调用方提供（业务透传）还是由大模型从上下文推断（模型识别） | **工作流中模型识别无效**：即使配置为“模型识别”，仍需上游节点明确传入值 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | -智能体应用（Agent 1.0）支持四种分享或发布方式，以及音视频实时互动： +## 使用方式 -- **UI 应用 / 魔笔分享渠道**：进入 UI 设计器编辑并发布界面，在 **环境部署** 中获取应用地址后分享。持有链接的阿里云用户均可访问，**单击「下线」可停止服务**。 -- **钉钉**：在 **发布平台** 授权计算巢 AppFlow（SLR 关联 + API-KEY 加密传输），配置钉钉模板 ID、Client ID、Client Secret 后创建，最终得到 **回调地址** 用于配置钉钉机器人。钉钉机器人的 **消息接收模式必须选 HTTP 模式**，选 Stream 模式会导致无法返回消息；并需申请 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限。 -- **微信公众号**：若已在钉钉步骤授权过则无需再次授权。选择 API KEY 与微信凭据（需 AppID 授权）后创建，生成二维码供用户扫码体验。 -- **音视频实时互动**：仅支持图文对话类应用（含智能体与工作流）。支持 H5/APP 扫码与 SDK 集成（基于 AICallKit SDK，含 UI/不含 UI 两种方案）两种渠道。 +1. **UI 应用发布** + 进入应用「发布渠道」页签 → 选择「UI 应用」→ 创建后跳转至 UI 设计器 → 编辑并发布至开发/生产环境。开发环境链接有效期 24 小时，生产环境需订阅付费套餐并绑定域名 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 -> **注意**：临时体验二维码（音视频互动）与从已有应用发布的 UI 体验链接，**有效期均为 24 小时**，过期需重新生成或重新发布。 +2. **钉钉/微信机器人** + - 钉钉：需在钉钉开放平台创建应用，获取 `Client ID`/`Client Secret` 和 AI 卡片 `Template ID`，并在百炼配置回调地址；授权 SLR 及 API-KEY 传输为必要前置步骤 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 + - 微信：需在微信公众号后台获取 `AppID`，完成开发者授权；发布后生成客服二维码供扫码体验。 -**权限与计费**：共享应用可被应用创建者（主账号）、RAM 用户及持有链接的其他阿里云用户访问；所有通过分享链接产生的费用由**应用创建者 UID 账号**承担。上述钉钉/微信配置细节见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +3. **组件发布与引用** + - 发布：在应用「发布渠道」→「组件」→ 填写名称、描述、参数别名及传参方式 → 确定发布。 + - 引用：智能体中作为技能添加；工作流中拖入「组件节点」并绑定输入（如 `系统变量/query`）→ 输出可直接接入下游节点 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 -## 组件化发布与接入 +4. **音视频实时互动** + 仅支持图文类应用（智能体/工作流），需配置 API Key → 生成临时体验二维码（24 小时有效）→ 发布后开通智能媒体服务并授权 SLR → 可选 H5/APP 扫码或 SDK 集成 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 -智能体或工作流应用可发布为模块化组件，供其他应用复用。发布路径有三处：发布应用时勾选 **发布应用组件**、在 **发布渠道** 的组件区域 **+ 创建**、或在控制台 **组件管理** 面板创建。详见 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +## 限制和注意事项 -### 关键参数 - -组件预设了系统参数 `query`（String，用户输入文本）和 `imageList`（Array，图像公网地址列表，仅在使用视觉模型时有效）。**预设系统参数无法删除**，不需要时将「是否可见」设为「否」隐藏。 - -各参数配置项含义： - -- **别名**：调用者只能看到别名，用于避免参数重名。 -- **传参方式**： - - **业务透传**：智能体中由使用者提供，工作流中由上游节点提供。 - - **模型识别**：智能体中由大模型根据参数描述自动推断填充。 -- **组件描述**：接入智能体时，大模型据此自动判断是否调用；接入工作流时仅作说明，不影响运行。 - -> **注意**：即使参数的传参方式设为**模型识别**，在**工作流应用**中应用也**不会**自动推断参数值，必须像业务透传一样从上游节点明确提供输入值。模型识别仅在智能体应用中生效。 - -### 接入方式 - -- **智能体应用**：组件作为工具接入，大模型根据用户问题自动调用。若组件含业务透传参数，可在测试时手动填 **入参变量配置**，或在 API 调用时通过 `biz_param` 参数传入。 -- **工作流应用**：组件作为组件节点接入，需手动传入参数（如 `系统变量/query`）并将 `组件1/result` 传递到下游节点。 - -### 组件注意事项 - -- **自动更新**：应用重新发布后，由其发布的组件会自动更新。 -- **避免嵌套调用**：A 调 B、B 调 A 会进入重复调用状态导致功能不可用。 -- **避免多级调用**：A 调 B、B 调 C 因存在最长运行时间限制，容易超时报错。 - -## UI 设计器发布 - -UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视化拖放式界面构建，可将应用发布为网页 UI。详见 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 - -**前提**：百炼应用、API Key 和 UI 设计必须归属于**同一业务空间**，否则无法在 UI 创建时选择对应的 API Key 与应用。 - -发布方式有两种：从已有应用一键创建 UI（自动填充标题、API-KEY、智能体、预设问题等），或通过 UI 设计器从模板（空白 / 智能出行助手 / 智能体门户 / AI 基础对话 / 企业 AI 知识库 Lite）创建。核心流程为：创建 UI → 拖放组件编辑页面 → 发布与分享。 - -**环境对比**： - -| 维度 | 开发环境 | 生产环境 | -| --- | --- | --- | -| 用途 | 开发、调试、验证 | 终端用户实际使用版本 | -| 访问方式 | 平台域名 | 平台域名 + 自定义访问地址 | -| 有效期 | **24 小时后失效**，需重新发布 | 长期有效 | -| 是否收费 | 免费 | 需订阅团队版及以上套餐，并配置域名 | - -**权限**：UI 应用发布后默认持有链接的阿里云用户可访问，也可开启 **允许匿名访问** 并通过权限组限制其只访问会话页。 - -**计费**：UI 设计器功能本身不计费，但会产生模型调用费用、UI 应用数据（超出 1GB 免费文件存储与 0.3GB 免费数据库容量后按量计费）、以及生产环境发布所需的套餐订阅费用。 - -对于工作流应用，若配置了文件类型的自定义参数，需在 UI 设计器中指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），才能正确读取用户上传的文件。 +- **Agent 版本硬性限制**：除 UI 集成外，所有发布渠道（魔笔、钉钉、微信、组件、音视频）均**不支持 Agent 2.0**；尝试对 Agent 2.0 应用执行相关操作将失败或无响应。 +- **嵌套与多级调用风险**：组件间禁止 A→B→A 的循环调用（导致死循环），也应避免 A→B→C 的三级以上链式调用（易超时） [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **环境与权限隔离**：UI 应用、API Key、智能体/工作流必须归属同一业务空间，否则无法关联或发布 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **计费责任归属**：所有通过分享链接产生的模型调用、存储、带宽等费用，均由应用创建者 UID 账号承担，与访问者无关 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **生产环境成本**：UI 应用发布至生产环境需订阅团队版及以上套餐；开发环境免费但链接 24 小时失效 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 ## 来源文档 @@ -82,4 +49,3 @@ UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视 - [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md index 58d67705..e00e62de 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md @@ -1,90 +1,42 @@ # application [support](support.md) -本页汇总阿里云百炼平台应用与[知识库](../concepts/knowledge-base.md)使用过程中的常见问题、关键限制以及相关协议入口，帮助开发者在接入应用、调用插件、管理数据与合规备案时快速定位答案。内容主要参考 [常见问题](../../raw/application-user-guide/application-support/application-faq.md) 与 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 +`application support` 指百炼平台为构建和运行 AI 应用（含智能体、RAG 应用、插件集成等）所提供的核心能力支持体系，涵盖模型调用、插件扩展、知识检索增强、[流式输出](../concepts/streaming-output.md)及数据管理等关键环节。开发者需结合服务协议与技术规范进行开发与部署。相关法律约束和合规要求详见 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 -## 应用中心 +## 支持的模型与功能 -### 插件能力 +- **插件能力**：官方提供六类内置插件：Python代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub搜索；其中部分插件需申请开通。 +- **自定义插件**：支持通过 API 注册自定义插件，大模型可解析其参数定义并调用（参见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第3条）。 +- **RAG（知识检索增强）**：支持多知识库并行检索，按配置策略（如相似度得分）选取 topN 片段后融合生成，适用于问答、客服、教育等场景（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第5条）。 +- **流式与增量输出**：支持 `stream=True` 实现流式响应；进一步启用 `incremental_output=True` 可获得真正增量式 token 输出（非全量重传），适用于前端实时渲染（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第8条）。 -百炼应用中心官方提供六款插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索，其中部分插件需申请通过后方可使用。自定义插件服务本身暂不收费，但配置智能体 API 时若涉及 [prompt](prompt.md) 优化、应用调用及测试窗测试，则会产生费用。 +> **注意**：文档2中第4条称“Assistant API 可提供各种类，方便调优”，但未明确具体类名或接口契约；当前 SDK 与 OpenAI 兼容 API 中实际暴露的是 `assistant` 类型资源（非 `Assistant` 类），该描述易引发歧义，建议以 [API 参考文档](https://help.aliyun.com/zh/model-studio/developer-reference) 为准。 -- **插件理解机制**：自定义 API 插件遵循协议传给大模型理解；自定义函数则由大模型学习传入的参数信息并返回完整结果。 -- **header 透传**：百炼调用自定义插件时**不支持自定义 header**，仅支持 `authorization`。若业务场景需要透传 header，需在服务端侧另行处理。 +## 关键参数 -### Agent 与 Assistant API 的区别 +| 参数 | 类型 | 说明 | +|------|------|------| +| `stream` | bool | 启用[流式输出](../concepts/streaming-output.md)（逐 token 返回） | +| `incremental_output` | bool | 在 `stream=True` 基础上启用增量式输出（仅返回新增 token，非累计内容） | +| `MD5`（文件上传） | string | 文件完整性校验值，必填（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第3条） | +| `authorization`（插件调用） | header | 插件 HTTP 请求中唯一支持透传的 header；其他自定义 header 将被丢弃（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第10条） | -Agent 偏重于调整插件模型与基于上下文的理解，由用户自行开发；Assistant API 则提供各类能力以方便调优。两者面向不同定制粒度，可根据应用复杂度选择。 +## 使用方式 -### 输出控制 +- **插件调用**：注册插件时需声明 `name`、`description`、`parameters`（JSON Schema 格式），系统自动注入至模型上下文；调用时模型生成结构化 function call 请求，平台负责路由与执行。 +- **RAG 应用测试**：若检索结果不准确，可通过回复下方“问题反馈”按钮提交，或复制 `RequestId` 提交工单（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第6条）。 +- **Markdown 渲染**：模型输出中的 `**text**` 等标记需由前端自行解析并渲染为加粗等样式（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第7条）。 +- **备案与合作**：接入通义千问模型并上架应用市场/小程序前，须完成 [应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)，并提交工单申请合作协议（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第11条）。 -- **流式与增量输出**：默认为全量回复，若需增量输出，可设置 `stream=True`（[流式输出](../concepts/streaming-output.md)）与 `incremental_output=True`（增量式[流式输出](../concepts/streaming-output.md)）。 -- **Markdown 加粗**：模型输出中的 `**xxxxx**` 是 Markdown 加粗标识，需在前端渲染时解析 md 语法即可正常显示。 +## 限制和注意事项 -## 知识检索（RAG） - -RAG（[检索增强生成](../concepts/rag.md)）在问答系统、对话系统、文本摘要、知识图谱构建与推理、教育与培训、客户服务、新闻与内容创作、智能搜索与推荐等多个领域均有应用。 - -- **检索顺序**：RAG 检索为**并行**方式，依据每个[知识库](../concepts/knowledge-base.md)的用户配置进行检索，再根据得分选取 topN 结果。 -- **回复不准确优化**：可点击模型回复下方的问题反馈按钮，勾选问题类型提交；也可复制 RequestId 通过阿里云工单反馈。详见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 - -## 数据管理 - -### 文件上传 - -- **格式限制**：上传 PDF 文件时后缀必须为小写 `pdf`，否则会触发错误码 `140010`（"上传文件仅支持 pdf/doc/docx 文件"）。 -- **MD5 参数**：上传文件接口的必填 MD5 参数用于校验上传文件的完整性。 -- **容量上限**：每个[业务空间](../concepts/workspace.md)最多上传 10 万个文档，超出需提交阿里云工单申请扩容。 - -### 结构化数据导入 - -结构化数据导入后若出现条数缺失（如 100 条仅导入 20 条），通常是因为表格中存在空行。产品策略规定：遇到空行后续数据不再识别；若第一行为空行，则整表视为空文件。 - -## 应用与小程序备案 - -产品接入通义千问大模型后，若需上架应用市场或小程序平台，需完成备案并申请合作协议： - -1. 参考[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)进行备案。 -2. [提交工单](https://smartservice.console.aliyun.com/service/create-ticket)申请通义千问系列模型的合作协议。 - -## 相关协议 - -接入百炼前请阅读以下协议条款： - -- [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) -- [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) - -协议入口同时收录在 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) 中，建议在正式接入前逐项确认。 - -## 常见限制与注意事项 - -- 自定义 header 不被支持，仅 `authorization` 可透传。 -- PDF 后缀必须为小写，避免触发 `140010` 错误码。 -- 结构化数据表格中的空行会导致后续数据被截断识别。 -- [业务空间](../concepts/workspace.md)文档数上限为 10 万，超额需工单申请。 -- 自定义插件本身免费，但 [prompt](prompt.md) 优化、应用调用与测试窗测试会收费。 +- **文件上传**：仅支持 `.pdf`（小写后缀）、`.doc`、`.docx`；结构化数据导入时，空行将导致后续行被截断（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第1、4条）。 +- **知识库容量**：单业务空间上限为 10 万个文档；超限时需提交工单申请扩容（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第2条）。 +- **协议约束**：所有应用必须遵守 [阿里云百炼服务协议](../../raw/application-user-guide/application-support/application-related-agreements.md) 及 [开源模型协议条款说明](../../raw/application-user-guide/application-support/application-related-agreements.md)，尤其注意数据使用、模型输出责任归属等条款。 +- **插件安全限制**：自定义插件无法透传除 `Authorization` 外的任何 HTTP header，服务端会主动剥离其余 header 字段（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第10条）。 ## 来源文档 -- [常见问题](../../raw/application-user-guide/application-support/application-faq.md) - [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) - - - - - - - - - - - - - - - - - - +- [常见问题](../../raw/application-user-guide/application-support/application-faq.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md index 42cb89ad..203965d9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md @@ -1,106 +1,56 @@ # application [use cases](use-cases.md) -百炼平台围绕“[检索增强生成](../concepts/rag.md)（RAG）+ [智能体应用](../concepts/agent-application.md)”提供了多种开箱即用的应用场景，可在无需编码或少量编码的情况下，将大模型问答能力接入网站、企业微信、微信公众号、钉钉等渠道，并支持基于本地[知识库](../concepts/knowledge-base.md)构建 RAG 应用。本页汇总这些典型用法的关键流程、模型选择、参数配置与注意事项。 +百炼平台支持多种典型业务场景下的 AI 应用快速落地，核心模式为“大模型应用（LLM） + 知识增强（RAG） + 多端集成”。所有方案均基于统一的百炼应用作为推理后端，通过 AppFlow 实现零代码连接主流企业通讯与内容平台（如网站、企业微信、钉钉、微信公众号），并支持本地化知识库部署。开发者可复用同一套 Prompt 工程、知识库配置和评测流程，显著降低多渠道 AI 助手的构建与维护成本。 -## 支持的接入渠道 +## 支持的模型/功能 -百炼 RAG 应用可通过 AppFlow（无代码连接流）或本地部署方式接入以下渠道： - -- 网站：通过 AppFlow 创建 AI 助手并生成悬浮挂件部署脚本，粘贴到网站 HTML 即可。详见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 -- 企业微信：通过 AppFlow 模板“企业微信自建应用大模型自动回复”连接企业微信应用与百炼应用。详见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)。 -- 微信公众号（订阅号）：通过 AppFlow 模板连接公众号与百炼应用，完成认证的公众号可用客户消息接口，未认证只能被动回复且响应限制为 5 秒。详见 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)。 -- 钉钉：创建钉钉应用并通过 AppFlow 模板连接，机器人消息接收模式必须选 HTTP 模式。详见 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md)。 -- 本地[知识库](../concepts/knowledge-base.md) RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。详见 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 - -## 通用流程 - -各渠道方案共享相似的四到五步骨架： - -1. 创建百炼[智能体应用](../concepts/agent-application.md)并获取应用 ID 与 [API Key](../concepts/api-key.md)； -2. 在目标平台（企业微信 / 公众号 / 钉钉）创建应用或机器人，获取对应凭证（企业 ID、AgentId、Secret、AppID、Client ID/Secret 等）； -3. 通过 AppFlow 模板创建连接流，填入双方凭证与应用 ID，发布并获取 WebhookUrl； -4. 在目标平台配置消息接收地址为 WebhookUrl，并配置可信 IP / 白名单； -5. 为百炼应用添加私有[知识库](../concepts/knowledge-base.md)（RAG），让回答覆盖私域问题。 - -网站方案略不同：第 2–3 步改为在 AppFlow 创建 AI 助手并生成 web 页面集成脚本，第 4 步将悬浮挂件脚本粘贴到网站 HTML。 - -## 模型选择 - -各教程对模型选择存在差异，需注意版本与命名： - -- 网站方案原选 **Qwen3.5-Plus**，其余渠道教程多选 **千问-Plus**（qwen-plus）。 -- 本地 RAG 应用可选 qwen-max、qwen-plus、qwen-turbo 三个通义千问商业模型：qwen-max 性能优秀，qwen-turbo 速度更快价格更低，qwen-plus 效果、速度、成本均衡。 - -> **注意**：不同文档对模型命名不一致（“Qwen3.5-Plus”与“千问-Plus”）。实际配置时以百炼控制台“更多模型”列表中可用的版本为准。未认证公众号若超 5 秒未返回，可在 [prompt](prompt.md) 中加“请总是给出简短的回答”或改用 qwen-turbo 提速，但会降低效果。 +- **基础模型**：默认推荐 `qwen-plus`（即文档中提及的“千问-Plus”或“Qwen3.5-Plus”），在效果、速度与成本间取得平衡；也可按需切换为 `qwen-max`（高精度）、`qwen-turbo`（低延迟）或 `qwen-flash`（超低成本）。> **注意**：文档 1 明确指定模型为 `Qwen3.5-Plus`，而文档 2、3、4 均写为“千问-Plus”，二者实际为同一模型的不同命名；当前控制台显示名称以 `qwen-plus` 为准，建议开发者以控制台实际可选模型列表为准 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 +- **核心能力**： + - 智能体（Agent）应用：支持角色设定（Prompt）、工具调用（如知识库检索）、多轮对话管理； + - RAG 增强：通过知识库实现私有领域问答，支持 PDF/DOCX/TXT/Excel 等格式上传与向量化； + - 多模态支持：文档 2、4、5 均明确列出 `.png`, `.jpg`, `.jpeg`, `.bmp`, `.gif` 等图片格式支持，适用于产品图谱、说明书图像理解等场景。 ## 关键参数 -百炼应用侧： - -- Prompt：可设置角色人设引导回答，如“你叫小助，可以帮助用户解答产品选购、使用等方面的问题。” -- 知识库调用方式：可选**必定调用**；知识文档支持配置相似度阈值与权重。 -- 文件处理：可选全文引用、切片检索或自定义处理。 -- 向量存储：标准版默认即可；如需集中管理多应用向量数据可选 ADB-PG。 - -本地 RAG 应用侧： - -- 模型参数：模型选择、温度（越高随机性越高）、最大回复长度、携带上下文轮数（设为 1 时不参考历史）。 -- RAG 参数：召回片段数（越大参考越多但噪声可能增加）、相似度阈值（越大参考越少但噪声减少，为 0 不剔除）。 -- 嵌入模型：默认用百炼 embedding API；可改用本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。 -- 文件限制：受 embedding API 限流，不建议传入超过 100 MB 的文件。 - -AppFlow 连接流： - -- 钉钉消息接收模式必须选 **HTTP 模式**，Stream 模式会导致无法返回消息。 -- 公众号需区分已认证 / 未认证两条[工作流](../concepts/workflow.md)；未认证只能被动回复，5 秒超时。 -- 企业微信需配置企业可信 IP；若报“域名主体校验未通过”，需配置企业自有域名或通过 ECS / 托管实例 / 计算巢 Nginx 代理转发。 +| 参数类别 | 参数名 | 说明 | 可配置位置 | +|----------|--------|------|------------| +| **模型层** | `temperature` | 控制生成随机性，值域通常为 0.0–1.0 | 百炼应用配置页、[基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 的 Gradio 界面 | +| | `max_tokens` | 限制模型输出最大 token 数 | 同上 | +| | `top_p` / `top_k` | 影响采样多样性 | 百炼应用高级设置（部分模型支持） | +| **RAG 层** | `retrieval_top_k` | 召回片段数（如“召回 3 个最相关段落”） | [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 的 Gradio 界面；百炼知识库引用配置中对应“相似度阈值”与“权重” | +| | `similarity_threshold` | 过滤低相关性召回结果的阈值（0–1） | 百炼应用配置页 > 知识库 > “相似度阈值”字段 | +| | `chunk_size` / `chunk_overlap` | 文档切分粒度（影响检索精度） | 百炼知识库创建时的“索引设置”；本地 RAG 应用中可自定义切分逻辑 | ## 使用方式 -网站接入：在 AppFlow AI 助手 Web 集成页复制悬浮挂件部署脚本，粘贴到网站 HTML 注释下方；可选启用图标拖拽。也可用函数计算 FC 一键部署示例网站。 +1. **统一后端：创建百炼应用** + 所有场景均始于百炼控制台的[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 创建**智能体应用** → 配置模型、Prompt 与知识库。应用发布后获得唯一 `AppID` 和调用所需的 `API Key`。 -企业微信 / 公众号 / 钉钉：发布连接流后，在目标平台配置 API 接收消息（URL 填 WebhookUrl，[Token](../concepts/token.md) / EncodingAESKey 填 AppFlow 凭证生成的值），配置可信 IP，即在聊天中 @机器人 或直接对话使用。 +2. **前端集成：通过 AppFlow 连接目标平台** + - **网站嵌入**：使用 AppFlow 创建 AI 助手 → 关联百炼应用 → 生成悬浮挂件脚本 → 插入 HTML 即可 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 + - **企业微信/钉钉/微信公众号**：使用 AppFlow 预置模板（如“企业微信自建应用大模型自动回复”）→ 分别配置平台凭证（企业 ID/AgentId/Secret 或 Client ID/Secret 或 AppID）与百炼凭证 → 获取 Webhook URL → 在对应平台后台完成消息接收配置。 -本地 RAG：解压 `local_rag.zip`，Python 3.8–3.12 环境安装依赖，配置百炼 [API Key](../concepts/api-key.md) 环境变量，运行 `uvicorn main:app --port 7866`，访问 `http://127.0.0.1:7866`。支持临时性文件上传（对话框直接传 pdf/docx/txt/xlsx/csv，刷新后失效）与长期知识库创建（上传到 File/Unstructured 或 File/Structured 后在界面创建知识库存于 VectorStore）。通过 Gradio 界面下方的“通过 API 使用”可获取 API 文档集成到业务。 - -## 日志与扩展 - -AppFlow 连接流可在百炼步骤后添加 SLS 日志云服务节点，将对话写入阿里云日志服务（需创建 Project/Logstore 并开启全文索引）。钉钉还支持通过卡片平台导入模板展示回答引用的文档来源，以及展示 DeepSeek 深度思考过程（在发送 AI 卡片阶段填入 `思考过程： {{Node2.reasoning}}` 与 `推理结果： {{Node2.text}}`）。 +3. **知识增强：配置知识库（可选但推荐）** + - 上传文件至百炼[数据中心](https://bailian.console.aliyun.com/?tab=app#/data-center) 或[数据连接](https://bailian.console.aliyun.com/cn-beijing?tab=app#/connector/list)； + - 在[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面创建标准版知识库； + - 在应用配置页启用知识库，设置调用方式（如“必定调用”）及相似度阈值。 ## 限制和注意事项 -- 新用户免费额度可覆盖教程资源消耗，超额后按 token [计费](../concepts/billing.md)。 -- 百炼文件导入支持 pdf、doc、docx、txt、md、pptx、ppt、png、jpg、jpeg、bmp、gif、xls、xlsx，单文档最大 100MB 或 1000 页，单图片最大 20MB，最多 200 个文件；文件存储在新加坡区域，解析通常 1–6 分钟。 -- 钉钉应用需开通 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限以发送卡片消息；应用供企业内其他用户使用需发布版本并设置可见范围。 -- 企业微信可信 IP 一个 IP 仅能用于一个企业，多企业共用会被识别为服务商导致通讯录 / 身份校验接口不可用。 -- 公众号开启服务器配置后自定义菜单会冲突关闭；未认证无法同时开启二者，已认证可通过接口重建菜单。 -- 本地 RAG Windows 系统若缺少 Microsoft Visual C++ Redistributable 需另行安装；报 `DLL load failed while importing _cext:` 时运行 `pip install msvc-runtime`。 -- 上线前建议组织业务人员参与应用[评测](../concepts/evaluation.md)，结合优化提示词、补充私有知识、调整切分策略改进效果。 +- **免费额度与计费**：新用户享有百炼免费额度，覆盖教程全部操作；额度耗尽后按 token 计费，具体见 [新用户免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota) [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 +- **文件限制**：单文档最大 100 MB 或 1000 页，单图片最大 20 MB，最多上传 200 个文件；知识库创建过程需等待解析（通常 1–6 分钟）。 +- **平台特异性约束**： + - 微信公众号：未认证订阅号仅支持被动回复（5 秒超时限制），建议完成认证或选用 `qwen-turbo` 模型提速 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)； + - 企业微信：配置 API 接收消息时需通过域名主体校验，若无自有备案域名，需通过 AppFlow 的 Nginx 代理或计算巢实例解决 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)； + - 钉钉：机器人消息接收模式**必须选择 HTTP 模式**，Stream 模式不兼容 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md)。 +- **本地 RAG 场景**：适用于需完全私有化部署、灵活控制文档切分与嵌入模型的场景，但需自行维护 Python 环境（3.8–3.12）及依赖，且不直接集成百炼控制台的统一监控与评测能力 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 ## 来源文档 - [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) -- [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) +- [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md index afcc09ed..e9119569 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md @@ -1,194 +1,78 @@ # bailian [application call](../api/application-call.md)ing -阿里云百炼支持通过 [DashScope SDK](../concepts/dashscope-sdk.md) 或 HTTP API 将已创建的应用集成到业务系统中。可调用的应用类型包括**[智能体应用](../concepts/agent-application.md)**和**[工作流](../concepts/workflow.md)应用**（[智能体编排](../concepts/agent-orchestration.md)应用已被[工作流](../concepts/workflow.md)应用替代），二者调用方式一致，均通过 `Application.call` / `POST /apps/{app_id}/completion` 触发，区别仅在于应用内部编排逻辑和可附加的扩展能力（如自定义参数传递）。 +百炼应用调用是指通过 DashScope SDK 或标准 HTTP API，将已发布的百炼智能体应用或工作流应用集成到自有业务系统中。调用过程统一使用 `POST /api/v1/apps/{app_id}/completion` 接口，支持单轮/多轮对话及插件参数透传，适用于各类 AI 增强场景。 -## 前提条件 +## 支持的模型/功能 -无论调用哪种应用，都需要先完成以下准备： +- **应用类型**：支持两类应用调用： + - [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)（即单智能体应用） + - [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)（原“智能体编排应用”，已由工作流应用替代） +- **核心能力**： + - 单轮文本生成（`prompt` 输入 → `output.text` 输出） + - 多轮对话（通过 `session_id` 或显式 `messages` 数组管理上下文） + - 自定义插件参数透传（需在应用内配置插件并启用“业务透传”参数模式） +- **底层模型**：实际执行模型由应用发布时绑定的模型决定（如 `qwen-max`、`qwen-plus`），调用方无需指定；响应中 `usage.models[].model_id` 字段可查实际使用的模型。 -1. **获取 [API Key](../concepts/api-key.md)**：在百炼控制台密钥管理页面创建 [API Key](../concepts/api-key.md)。 -2. **配置环境变量（推荐）**：将 [API Key](../concepts/api-key.md) 写入 `DASHSCOPE_API_KEY` 环境变量，避免在代码中硬编码。SDK 会自动读取该变量。 -3. **获取应用 ID**：在应用管理页面创建对应应用（[智能体应用](../concepts/agent-application.md) / [工作流](../concepts/workflow.md)应用），并从应用卡片复制 `APP_ID`。 -4. **安装 [DashScope SDK](../concepts/dashscope-sdk.md)**（HTTP 调用可跳过）：Python 通过 `python3 -m pip install -U dashscope`；Java 通过 Maven/Gradle 添加 `com.alibaba:dashscope-sdk-java` 依赖（建议版本 >= 2.12.0）；Node.js 安装 `axios`。 +> **注意**：文档 2 明确声明“本文档仅适用于华北2（北京）地域”，而文档 1 和文档 3 未限定地域。若跨地域调用失败，请优先确认应用所在地域与 API Endpoint 是否匹配（当前仅北京地域支持工作流应用调用）。 -## 基本调用方式 - -[智能体应用](../concepts/agent-application.md)与工作流应用的调用接口完全相同，详见[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)与[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。核心请求结构如下： - -``` -POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion -Authorization: Bearer $DASHSCOPE_API_KEY -Content-Type: application/json - -{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} -} -``` - -### Python - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'request_id={response.request_id}') - print(f'code={response.status_code}') - print(f'message={response.message}') -else: - print(response.output.text) -``` - -### Java - -```java -import com.alibaba.dashscope.app.*; -import com.alibaba.dashscope.exception.*; - -ApplicationParam param = ApplicationParam.builder() - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .appId("YOUR_APP_ID") - .prompt("你是谁？") - .build(); - -Application application = new Application(); -ApplicationResult result = application.call(param); -System.out.printf("text: %s\n", result.getOutput().getText()); -``` - -### Node.js / curl - -Node.js 使用 `axios` 发起 POST 请求即可，结构与 curl 示例一致： - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} - }' -``` - -响应统一为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 多轮对话 - -工作流应用支持多轮对话，详见[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮对话。 -- **自行管理 `messages`（推荐）**：手动维护 `messages` 数组传递每轮历史，无需传 `prompt`，控制更灵活。 - -> **注意**：若请求中同时包含 `session_id` 和 `messages`，系统将优先使用 `messages`。 - -使用 `messages` 时，需先在工作流的大模型节点中配置提示词变量 `historyList` 并发布应用，再发起调用。 - -## 自定义参数传递 - -针对自定义插件与自定义节点，百炼支持通过 `biz_params` 的 `biz_params.user_defined_params` 透传业务参数，详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)。该能力可用于[智能体应用](../concepts/agent-application.md)的自定义插件，以及工作流应用中的插件节点。 - -### 使用流程 - -1. **创建自定义插件**：在百炼控制台插件页面新增自定义插件，按需配置鉴权（如用户级鉴权 + Header + basic）。创建工具时，输入参数的**传参方式务必选择「业务透传」**，并发布插件。 -2. **关联应用**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联；工作流应用则在插件节点中引用。关联后发布应用。 -3. **API 调用**：通过 `biz_params.user_defined_params` 传递插件 ID 与入参键值对。 - -### 请求示例 - -```python -import os -from http import HTTPStatus -from dashscope import Application +## 关键参数 -biz_params = { - "user_defined_params": { - "your_plugin_code": { # 替换为实际插件 ID（在插件卡片获取） - "article_index": 2 +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面获取 | +| `prompt` | string | 否（多轮对话时可省略） | 当前轮次的用户输入指令；若使用 `messages` 则此字段被忽略 | +| `biz_params` | object | 否 | 用于传递自定义插件参数，结构为 `{ "user_defined_params": { "": { "": } } }`；详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) | +| `session_id` | string | 否 | 启用云端会话管理时使用，有效期 1 小时，最多 50 轮 | +| `messages` | array | 否 | 替代 `prompt` 的推荐方式，格式同 OpenAI：`[{ "role": "user", "content": "..." }, { "role": "assistant", "content": "..." }]`；若同时传 `session_id` 和 `messages`，以 `messages` 为准 | + +## 使用方式 + +### 1. 准备工作 +- 获取 API Key：前往[密钥管理](https://bailian.console.aliyun.com/?tab=model#/api-key)创建并配置为环境变量 `DASHSCOPE_API_KEY`（**强烈推荐**，避免硬编码） +- 获取 `app_id`：在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)中复制目标应用 ID +- 安装 SDK（可选）：Python 执行 `pip install -U dashscope`；Java/Node.js 等参见对应语言示例 + +### 2. 发起调用 +- **SDK 方式（推荐）**： + ```python + from dashscope import Application + response = Application.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + app_id="YOUR_APP_ID", + prompt="你是谁？", + biz_params={"user_defined_params": {"plugin_abc": {"query": "test"}}} + ) + print(response.output.text) + ``` +- **HTTP 方式（通用）**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "input": { + "prompt": "你是谁？", + "biz_params": { + "user_defined_params": { + "plugin_abc": {"query": "test"} + } } - } -} - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='寝室公约内容', - biz_params=biz_params -) -``` - -HTTP 请求体中 `biz_params` 位于 `input` 下： - -```json -{ - "input": { - "prompt": "寝室公约内容", - "biz_params": { - "user_defined_params": { - "your_plugin_code": { "article_index": 2 } } - } - }, - "parameters": {}, - "debug": {} -} -``` - -Java SDK 通过 `JsonUtils.parse(...)` 将 JSON 字符串转为对象传入 `ApplicationParam.bizParams`。 - -## 关键参数 - -| 参数 | 位置 | 说明 | -| --- | --- | --- | -| `app_id` / `APP_ID` | URL path | 应用 ID，从应用卡片获取 | -| `prompt` | `input.prompt` | 单轮对话的输入指令（与 `messages` 二选一） | -| `messages` | `input.messages` | 自行维护的多轮对话历史（推荐，优先级高于 `session_id`） | -| `session_id` | `input.session_id` | 云端会话 ID，有效期 1 小时，最多 50 轮 | -| `biz_params` | `input.biz_params` | 业务透传参数，含 `user_defined_params` | -| `parameters` | 顶层 | 应用级参数 | -| `debug` | 顶层 | 调试信息 | + }' + ``` ## 限制和注意事项 -- **地域限制**：[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) 文档明确仅适用于华北2（北京）地域。 -- **应用类型替代关系**：[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代，新场景应使用工作流应用。 -- **`session_id` 约束**：有效期 1 小时，最多 50 轮；与 `messages` 同时存在时优先使用 `messages`。 -- **插件业务透传**：自定义插件的输入参数传参方式必须选择「业务透传」，否则无法通过 `biz_params` 传递；插件 ID 在插件卡片获取，替换示例中的 `your_plugin_code`。 -- **[业务空间](../concepts/workspace.md)隔离**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联。 -- **密钥安全**：不要在生产环境硬编码 [API Key](../concepts/api-key.md)，统一通过 `DASHSCOPE_API_KEY` 环境变量注入。 -- **Responses API**：如需使用 OpenAI 兼容的 Responses API 调用工作流应用，需参阅 Responses API 文档，不在本文调用方式范围内。 +- **地域限制**：工作流应用调用仅支持华北2（北京）地域；智能体应用无明确地域限制，但建议与应用部署地域一致以降低延迟。 +- **会话管理**：`session_id` 有效期为 1 小时且最多承载 50 轮对话；生产环境推荐自行维护 `messages` 数组以获得完全控制权。 +- **插件参数**：必须在插件工具配置中将参数“传参方式”设为 **业务透传**，否则 `biz_params` 中的参数不会生效。 +- **错误处理**：所有调用均返回标准 HTTP 状态码（如 `401 Unauthorized`、`404 Not Found`）及 `request_id`，用于问题定位；错误码详情请参考[开发者参考文档](https://help.aliyun.com/zh/model-studio/developer-reference/error-code)。 +- **安全要求**：API Key **严禁硬编码**于源码或前端代码中；务必通过环境变量或密钥管理服务注入。 ## 来源文档 -- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) -- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - - - - - - - - - - - - - - - - - - +- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) +- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md index 49c6aa5f..84ae13e8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md @@ -1,79 +1,52 @@ # data connection overview -数据连接是阿里云百炼平台管理外部数据源的统一入口。通过创建数据连接器，百炼应用可以安全地访问企业数据库、文档系统和对象存储中的数据，并在对话中实时查询和引用这些数据。详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md)。 +数据连接是阿里云百炼平台统一管理外部数据源的核心能力，为应用提供安全、可控的数据接入通道。它支持结构化与非结构化数据的接入，并通过平台托管或流处理两种模式实现数据访问，是构建知识增强型智能体（Agent）和 RAG 应用的基础组件。所有连接器均需在业务空间内创建并绑定至具体应用，其配置直接影响后续检索与调用行为。 -## 连接器类型 +## 支持的模型/功能 -数据连接器按数据的存储和访问方式分为两大类： +数据连接器按数据访问方式分为两类： -- **平台托管**：数据导入并存储在百炼平台（或自有 OSS）。 - - **文件**：管理非结构化文档（PDF、Word、Markdown 等）。 - - **表格**：导入并查询结构化表格数据（CSV、Excel 等）。 -- **流处理**：数据保留在原数据源，实时访问。 - - **MySQL / PostgreSQL / PolarDB-X 2.0**：连接对应数据库，支持执行 SQL 查询（仅 DMS 导入方式支持）。 - - **语雀**：访问语雀文档和知识库（仅公网版本）。 - - **OSS**：访问对象存储中的文件。 +- **平台托管类**：适用于静态文件与表格数据，包括 + - `文件`：支持 PDF、Word、Markdown 等非结构化文档，依赖[文档理解](https://help.aliyun.com/zh/document-mind/product-overview/overview-of-document-understanding#9a4f5fb91fpps)能力进行解析（详见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“导入文件”章节）； + - `表格`：支持 CSV、Excel 等结构化数据，支持自定义表头与字段类型（如 `image_url`），但表结构一旦确定不可修改。 -各类型的适用场景与存储方式详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的连接器类型表。 +- **流处理类**：适用于实时数据库与在线服务，包括 + - `MySQL`、`PostgreSQL`、`PolarDB-X 2.0`：仅通过 **DMS 导入数据源** 方式创建的连接器支持执行 SQL 查询；自定义方式创建的连接器仅支持元数据同步，不支持直接查询（该限制在 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各数据库连接器说明中反复强调）； + - `语雀`：对接语雀知识库，依赖个人访问 Token，**仅支持公网版本语雀**； + - `OSS`：访问对象存储中的文件，需开通向量检索服务方可使用 `searchOSSFile` 和 `searchOSSFileByFileName` 工具（参见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) “OSS连接器”章节）。 -## 前置条件 +> **注意**：`MySQL` 与 `PostgreSQL` 连接器均要求数据库账号具备高权限（如 REPLICATION 或 Superuser），且 PostgreSQL 必须将 `wal_level` 设置为 `logical`；而 PolarDB-X 2.0 **仅支持私网连接**，不支持公网，且不兼容自建实例——这些关键差异已在原始文档中明确区分，开发者需严格遵循。 -- **账号权限**：主账号或具有数据连接管理权限的 RAM 用户；RAM 用户需先获得主账号授权。 -- **数据源准备**（按连接器类型）： - - 文件/表格：准备好待上传文档/表格，或已创建 OSS Bucket。 - - MySQL：已有 MySQL 实例（RDS 或自建），网络可达（公网或私网）。 - - PostgreSQL：账号具备高权限（Superuser 或 REPLICATION），且 `wal_level` 设置为 `logical`；自建实例还需配置 `listen_addresses` 允许 `100.64.0.0/16` 网段访问。 - - PolarDB-X 2.0：已有阿里云 PolarDB-X 2.0 实例且所在地域支持私网访问；DMS 导入方式需先在 DMS 录入实例。 - - 语雀：已有公网版语雀知识库并获取访问 Token。 - - OSS：已创建 Bucket 并开通向量检索服务。 +## 关键参数 -## 数据库连接器关键差异 +| 参数类别 | 关键字段 | 说明 | +|----------|----------|------| +| **通用** | 连接器名称、描述 | 名称需唯一且易识别；描述影响智能体调用准确度，建议明确数据内容与用途 | +| **文件/表格** | 存储位置（平台存储 / 自有 OSS） | 平台存储提供免费额度（文件连接器限 200,000 文件 / 1 TB，表格连接器限 1 TB）；自有 OSS 需添加 `bailian-connector-access` 标签（值为 `ReadAndWrite`） | +| **数据库类** | 数据库地址、端口、用户名、密码、dbName（PostgreSQL/PolarDB-X 必填） | MySQL 默认端口 3306，PostgreSQL 默认 5432；PolarDB-X 仅支持私网，且数据库地址/端口由实例自动填充 | +| **语雀/OSS** | Tenant access token（语雀）、Bucket 选择（OSS） | 语雀 Token 需从 [语雀开放 API](https://www.yuque.com/yuque/developer/api) 获取；OSS Bucket 需添加 `bailian-datahub-access` 标签（值为 `read`），且**不支持归档/冷归档存储类型** | -| 差异项 | MySQL | PostgreSQL | PolarDB-X 2.0 | -| --- | --- | --- | --- | -| 默认端口 | 3306 | 5432 | 自动获取 | -| 额外必填字段 | 无 | dbName（数据库名称） | 无 | -| 连通性检测服务 | EventBridge | DTS | DTS | -| 网络类型 | 公网 / 私网 | 公网 / 私网 | 仅私网 | -| 特殊配置 | 无 | `wal_level=logical` | 需 SLR 授权（DTS、PolarDB-X，DMS 方式加 DMS 角色） | +## 使用方式 -三类流处理数据库连接器均支持两种数据来源配置：**创建自定义数据源**（手动配置连接信息）与**从 DMS 导入数据源**（导入 DMS 中已有数据源）。数据库用户须具备读取权限，配置后可点击检测验证连通性。完整字段说明见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各连接器配置章节。 - -> **注意**：仅通过**从 DMS 导入数据源**方式创建的 MySQL / PostgreSQL / PolarDB-X 2.0 连接器支持执行 SQL 查询；通过**创建自定义数据源**方式添加的连接器不支持直接执行 SQL。 - -## 创建连接器 - -1. 在[数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list)页面点击**创建连接器**。 -2. 选择连接器类型，填写**连接器名称**和**描述**（描述会用于指导应用调用的准确度，建议写明数据内容和用途）。 -3. 按类型填写存储位置或数据源信息，必要时执行连通性检测。 -4. 点击**确认**完成创建。 - -存储位置选择要点： - -- **文件连接器**：平台存储限时免费（最多 200,000 个文件、1 TB）；或使用自有 OSS，需为 Bucket 添加 `bailian-connector-access` 标签（值 `ReadAndWrite`）。 -- **表格连接器**：平台存储提供 1 TB 免费额度，用尽后转按量付费；自有 OSS 同样需上述标签。 -- **OSS 连接器**：从下拉列表选择 Bucket，需添加 `bailian-datahub-access` 标签（值 `read`）。 - -## 导入数据 - -- **导入文件**：进入文件连接器详情页，在**类目**下选择或新建类目后导入。平台暂不支持直接导入 JSON、CSV、YAML，需先转换为 XLSX/XLS。可选择**默认设置**或**自定义设置**解析（电子文档解析、文档智能解析、大模型文档解析、Qwen VL 解析、音视频解析等，具体能力取决于文件类型）。可为文件配置**标签**，API 调用时通过 `tags` 参数筛选以提升检索效率。 -- **导入表格**：进入表格连接器详情页，在**数据表管理**下选择或新建数据表。支持**直接上传 Excel**（自动识别表头）或**自定义表头**。数据表结构（列名、描述、类型）一旦确定不可修改，且上传文件的列数与列名必须与表结构一一对应，否则导入失败。 +1. **创建连接器**：进入 [数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list) 页面 → 单击 **创建连接器** → 选择类型 → 填写基本信息与连接参数 → （可选）点击 **开始检测** 或 **连接检测** 验证连通性 → 确认创建。 +2. **导入数据**： + - 文件连接器：进入详情页 → 选择类目 → **导入数据** → 本地上传 → 配置解析方式（默认/自定义）与标签 → 确认； + - 表格连接器：进入详情页 → 在 **数据表管理** 下新建或选择数据表 → 上传 Excel 或自定义表头 → 确保列名与类型严格匹配； + - 数据库/OSS/语雀类连接器无需手动导入，数据实时访问。 +3. **绑定应用**：在应用配置中显式关联已创建的数据连接器，方可启用对应检索能力（如 `searchFile`、`querySQL` 等工具）。 ## 限制和注意事项 -- 数据库连接器执行 SQL 的限制见上文注意框（仅 DMS 导入方式支持）。 -- **OSS 连接器**：使用需开通[向量检索服务](https://help.aliyun.com/zh/oss/user-guide/vector-retrieval/)，否则无法使用 `searchOSSFile` / `searchOSSFileByFileName` 工具；不支持归档/冷归档/深度冷归档类型的 Bucket；支持内容加密与私有 Bucket；开启 Referer 防盗链时需将 `*.console.aliyun.com` 加入白名单。 -- **文件导入**：文件作为独立副本存储在平台免费空间（当前无容量限制），仅支持查看最近 **90** 天内导入的文件（超期不可查看但不删除），且仅供当前业务空间使用。请求高峰期解析可能耗时数小时甚至偶现超时，需耐心等待或重试。 -- **语雀连接器**：仅支持公网版本语雀，需提供有效的 Tenant access token。 - -以上流程、字段和限制的完整细节，请以原文 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 为准。 +- **权限约束**：RAM 用户需主账号授予 `AliyunBaiLianFullAccess` 或最小化权限策略（含 `bailian:ListConnectors`、`bailian:CreateConnector` 等动作），详见 [权限管理](https://help.aliyun.com/zh/model-studio/application-permission-management-overview)。 +- **网络与白名单**：MySQL 公网连接需将百炼服务 IP 段加入数据库白名单；PostgreSQL 自建实例需配置 `pg_hba.conf` 允许 `100.64.0.0/16` 访问；PolarDB-X 仅支持私网，必须同地域部署。 +- **解析与时效性**：文件导入后生成独立副本，**仅支持查看最近 90 天内导入的文件**；高峰时段解析可能延迟数小时，偶现超时，建议错峰操作。 +- **功能边界**： + - `MySQL`/`PostgreSQL`/`PolarDB-X 2.0` 的 SQL 执行能力**仅对 DMS 导入方式生效**，自定义方式不支持（原始文档多次强调，勿混淆）； + - OSS 连接器若未开通向量检索服务，则 `searchOSSFile` 等工具不可用； + - 文件连接器**不支持直接导入 JSON/CSV/YAML**，需先转为 XLSX/XLS 格式。 ## 来源文档 - [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md index 8b42e757..bf395d88 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md @@ -1,77 +1,42 @@ # fine tuning -模型微调（Fine-tuning）是阿里云百炼在 Prompt 工程、插件调用等手段仍无法满足效果时提供的深度定制手段。它覆盖文本生成、视觉理解（Qwen-VL）、图像/视频生成（万相）以及语音合成（CosyVoice）等多种模态，通过 SFT、CPT、DPO 等训练方式，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数。 +fine tuning 是阿里云百炼平台提供的模型定制化能力，允许开发者基于自有数据对预训练模型进行增量训练，以提升其在特定任务、领域或风格上的表现。该能力覆盖文本生成、视觉理解、图像/视频生成及语音合成等多模态模型，支持 SFT（监督微调）、CPT（持续预训练）和 DPO（直接偏好优化）等多种训练范式。所有 fine tuning 任务当前均仅支持华北2（北京）地域，且需使用该地域的 API Key [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -> **注意**：以下所有微调、部署与调用能力均**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 +## 支持的模型与功能 -## 支持的模型与训练方式 +- **文本生成**：支持 Qwen 系列全量模型（如 `qwen3-8b`, `qwen2.5-7b-instruct`）及千问-VL 视觉语言模型，提供 CPT、SFT（含高效 LoRA 和全参）、DPO 三种训练方式。具体支持矩阵详见 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 +- **图像生成**：仅支持万相系列模型（`wan2.7-image-pro`, `wan2.7-image`），采用 SFT-LoRA 高效微调，适用于文生图（t2i）和图生图（i2i）场景 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **视频生成**：支持万相图生视频模型（`wan2.7-i2v`, `wan2.2-kf2v-flash` 等），同样基于 SFT-LoRA，支持基于首帧或首尾帧的特效/动作定制 [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 +- **语音合成**：仅支持 `cosyvoice-v3-flash` 模型，通过 SFT-LoRA 进行单发音人音色定制，产物为独立部署的专属音色模型，不支持多音色切换 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 -按模态划分，不同模型支持的训练方式差异明显： +> **注意**：文档 4 中称“阿里云百炼推荐您如果**模型支持全参训练，请优先选择全参训练**”，但文档 1、2、7 均明确限定图像、视频、语音类模型**仅支持 `efficient_sft`（LoRA）**，且文档 3 的支持矩阵中，`wan*` 和 `cosyvoice*` 系列未列出任何全参训练选项。因此，对非文本生成模型，全参训练不可用，该推荐不适用。 -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异，例如 Qwen3-32B、Qwen3-4B/1.7B/0.6B、Qwen2.5 系列支持全部 5 种；而 Qwen3.5-Plus/Flash、Qwen3.6/3.7 等新模型往往仅支持 `sft`。详见 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -- **视觉理解（千问 VL）**：Qwen3-VL、Qwen2.5-VL 系列支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像生成（万相）**：`wan2.7-image-pro`、`wan2.7-image`，仅支持 SFT-LoRA 高效微调，见 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -- **视频生成（万相）**：图生视频-基于首帧 `wan2.7-i2v`/`wan2.5-i2v-preview`/`wan2.2-i2v-flash`，基于首尾帧 `wan2.2-kf2v-flash`，同样仅支持 SFT-LoRA，见 [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 -- **语音合成（CosyVoice）**：`cosyvoice-v3-flash`，仅支持 `efficient_sft`，且**当前只能通过 API 发起，控制台暂不支持**，见 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +## 关键参数 -## 三种调优方式（文本生成） - -推荐按递进顺序组合使用：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下「更好/更差」回答对（`chosen`/`rejected`） | - -训练模式分**全参训练**与**高效训练（LoRA）**：两者费用相同，官方建议在模型支持全参训练时优先选择全参（效果更好、性价比更高）；LoRA 适合对训练时间/成本敏感或数据集较小的场景。 - -## 关键超参数 - -文本生成调优的常用超参及默认值（以控制台实际显示为准）： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议循环 3~5 次，>10000 建议 1~2 次。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型支持的最大值；SFT 会**丢弃**超长数据，DPO 则**截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：默认学习率各文档取值不一致。控制台参数面板列出的 `learning_rate` 默认值为 `3e-4`（对应高效训练默认场景），而 API 示例中 SFT 全参使用的是 `1.6e-5`。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频与 CosyVoice 使用各自独立的超参集，例如万相有 `max_steps`/`generation_type`/`val_img_size`，CosyVoice 分 `lm_*`（影响韵律）与 `fm_*`（影响音色）两组网络的 `*_max_epoch`/`*_step`/`*_num`/`*_batch_size`（8 个子字段全部必填）。 +- **通用超参**：`learning_rate`（文本推荐 1e-4~1e-5，图像/视频/语音需按文档示例设置）、`n_epochs` 或 `max_steps`（控制训练轮次/步数）、`batch_size`、`lora_rank`（LoRA 秩，默认 8~32）、`lora_alpha`（LoRA 缩放因子）。 +- **模型特有参数**： + - 图像生成：`generation_type`（`t2i` 或 `i2i`）、`max_pixels`、`val_img_size`； + - 视频生成：`split`（训练/验证集划分比例）、`eval_epochs`； + - 语音合成：`lm_max_epoch`/`fm_max_epoch`（语言模型/流匹配模型轮次）、`lm_batch_size`/`fm_batch_size`； + - 文本生成：`max_length`（序列长度）、`warmup_ratio`（学习率预热比例）。 +- **数据源参数**：支持 `file_id`（上传 ZIP）和 `oss_mount`（OSS 挂载）两种方式；OSS 挂载要求数据集为解压状态，且 `data.jsonl` 必须位于根目录 [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)。 ## 使用方式 -**控制台（推荐入门）**：在[模型调优](https://bailian.console.aliyun.com/?tab=model#/efm/model_manager)页面创建训练任务 → 选训练方式与模型 → 配置训练集/验证集（可自动切分）→ 配置 Checkpoint 保存 → 开始训练 → 部署 → 评测。零代码场景可参考 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)，其中给出了以 Qwen3-8B 为例的完整安全对齐 SFT 流程与超参实验对照（全参 `n_epochs=3`/`lr=1e-5` 或 LoRA `n_epochs=3`/`lr=3e-4` 效果较好）。 +1. **准备数据集**：按指定格式（如 ChatML JSONL）组织训练数据，ZIP 打包（最大 2GB），确保 `data.jsonl` 在根目录，图片/音频文件名全局唯一。 +2. **上传文件**：调用 `/api/v1/files` 接口上传，获取 `file_id`。 +3. **创建任务**：调用 `/api/v1/fine-tunes`，传入 `model`、`training_datasets`（含 `file_id`）、`training_type`（如 `efficient_sft`）及 `hyper_parameters`。 +4. **轮询状态**：用 `job_id` 调用 `/api/v1/fine-tunes/{job_id}`，等待 `status` 变为 `SUCCEEDED`。 +5. **部署模型**：调用 `/api/v1/deployments`，传入 `finetuned_output` 作为 `model_name`，获取 `deployed_model`。 +6. **调用服务**：使用 `deployed_model` 名称发起推理请求（图像/视频需异步，文本可同步）。 -**API / 命令行**：统一四步流程，详见 [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)： +## 限制和注意事项 -1. 上传数据集到 `POST /api/v1/files`（`purpose=fine-tune`），获取 `file_id`。 -2. `POST /api/v1/fine-tunes` 创建任务，关注返回的 `job_id`、`finetuned_output`、`status`。 -3. 轮询 `GET /api/v1/fine-tunes/` 直到 `status` 变为 `SUCCEEDED`。 -4. `POST /api/v1/deployments` 部署（`plan=lora`），轮询直到 `status` 为 `RUNNING`，再用 `deployed_model` 调用。 - -> **注意**：通过 API 创建的训练任务**仅支持按 Token 计费**，不支持模型训练单元（预付费/后付费）；如需使用训练单元，必须通过控制台创建。 - -数据集除 `file_id` 外还可用 OSS 挂载（`data_source_type=oss_mount`），OSS Bucket 地域支持 `cn-beijing` 与 `ap-southeast-1`，挂载时只需指定 `data.jsonl` 路径。 - -## 数据格式要点 - -- **SFT**：ChatML `{"messages":[...]}`，支持多轮；不支持 OpenAI 的 `name`/`weight`，所有 assistant 行都会被训练；思考模型（thinking）只训练**最后**一个 assistant 输出且须保留 `` 标签前后的换行。 -- **DPO**：在 `messages` 基础上追加 `chosen`/`rejected`。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解**：`content` 用数组，`image`/`video` 声明文件名（不含路径），打包为 ZIP（≤2GB），`data.jsonl` 必须在根目录，文件名全局唯一且仅含 ASCII。 -- **CosyVoice**：`data.jsonl` 每行 `{"wav_fn":"train/xxx.wav","text":"..."}`，`wav_fn` 必须以 `train/` 前缀；`text` 须为纯文本，禁止 SSML/LaTeX/情感标注。 - -## 限制与注意事项 - -- **成本与耗时高**：文本模型微调需构建大规模数据集，且调优后模型**必须部署才能使用**，部署费用较高；官方明确将模型调优定位为「最后的手段」，建议先充分尝试 Prompt 工程与插件调用。 -- **训练耗时差异大**：万相文生图约数十分钟，视频微调可达数小时；文本 LoRA 通常 15~30 分钟；部署一般需 3~10 分钟。 -- **计费方式各异**：文本/VL 按训练 Token 计费（`训练Token × 循环次数 × 单价`）；CosyVoice 按 `(lm_max_epoch+fm_max_epoch)×25×音频总秒数` 估算 Token，单价 0.2 元/千 Token，另加部署时长费用。 -- **能力边界不可突破**：CosyVoice 调优产物为单音色模型（`voice` 锁定 `default`），无法新增基础模型不支持的语种、也不支持 `instruction` 指令控制。 -- **过拟合/欠拟合判断**：观察 Training/Validation Loss 曲线，欠拟合可增大 `n_epochs`/`lora_rank`，过拟合则反向调整。 -- 万相 LoRA 调用需在提示词中包含**触发词**以激活风格；图像模型部署后当前仅支持异步调用。 +- **地域与权限**：所有 fine tuning 服务仅限华北2（北京）地域，子账号需显式授予模型调用、训练、部署权限 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **数据要求**：SFT 至少需 1000+ 条高质量样本；CPT 需 1000 万+ Token 无标签文本；DPO 需 100+ 组正负样本对。 +- **计费**：按训练消耗 Token 数计费（单价因模型而异，如 `qwen3-8b` 为 ¥0.006/千 Token，`cosyvoice-v3-flash` 为 ¥0.2/千 Token），部署后另计模型单元费用。 +- **工程成本**：fine tuning 是“最后手段”，应优先尝试 Prompt 工程和插件调用；其迭代周期长、成本高，需谨慎评估 ROI [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 +- **能力边界**：调优无法扩展基础模型能力（如语种、指令控制、多音色），仅能优化其在已有能力范围内的表现 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 ## 来源文档 @@ -79,8 +44,8 @@ - [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) - [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) -- [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) - [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) +- [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) - [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md index b17b41c7..1b6032ff 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md @@ -1,87 +1,80 @@ # get started with models -阿里云百炼是一站式大模型开发与应用平台，集成千问（Qwen）全系列及 DeepSeek、Kimi、GLM 等第三方模型，并提供兼容 OpenAI 的 API。本页面面向开发者，梳理从选模型、拿 API Key 到发起首次调用所需的关键信息：可用模型、接入地域与域名、Base URL、以及限流规则与规避策略。 +阿里云百炼提供开箱即用的大模型服务，支持通过兼容 OpenAI 的 API 快速调用千问（Qwen）及第三方模型。开发者无需部署和运维模型，只需配置 API Key 和 Base URL 即可发起首次请求。平台同时支持可视化应用构建、微调与部署等全链路能力，适用于从快速验证到生产级落地的各类场景。 -## 支持的模型与能力 +## 支持的模型与功能 -百炼提供开箱即用的模型服务，无需自行部署或运维即可直接调用。文本生成方面，千问旗舰系列按能力与成本分层，可按需选择（详见 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md)）： +百炼提供多模态、多场景的模型服务，覆盖文本生成、视觉理解、语音合成、嵌入向量等能力。核心模型包括： -- **千问 Max**（如 `qwen3.7-max`）：效果最好，适合复杂、多步骤任务。 -- **千问 Plus**（如 `qwen3.7-plus`）：效果、速度与成本均衡，多数场景的推荐选择。 -- **千问 Flash**（如 `qwen3.6-flash`）：高性价比、低延迟，适合简单、需快速响应的任务。 +- **千问系列旗舰模型**：`qwen3.7-max`（效果最优，适合复杂任务）、`qwen3.7-plus`（效果/速度/成本均衡，**推荐首选**）、`qwen3.6-flash`（高性价比、低延迟）[什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)； +- **第三方模型**：DeepSeek、Kimi、GLM 等，部分仅限特定地域（如 DeepSeek 仅支持华北2（北京））[什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)； +- **领域专用模型**：长文本处理、法律、意图理解、角色扮演等细分场景模型； +- **多协议兼容**：[OpenAI 兼容接口](../concepts/openai-compatible-api.md)、Anthropic 兼容接口、DashScope 原生 SDK 接口 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)。 -除文本生成外，还覆盖视觉理解、图像生成、视频生成、语音识别与合成、嵌入向量等[多模态能力](../concepts/multimodal.md)，以及长文本、翻译、法律等细分领域模型。平台同时支持模型调优（SFT/CPT/DPO）、模型部署和模型评测，详见 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 +> **注意**：文档中 `qwen3.7-plus` 与 `qwen-plus` 指代同一类主力模型，但命名存在不一致——`qwen3.7-plus` 是当前最新稳定版标识（见[选择模型](../../raw/model-user-guide/get-started-with-models/models.md)），而 `qwen-plus` 多用于历史快照或旧文档（如[限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)表格中仍大量使用）。实际调用请以[模型广场](https://bailian.console.aliyun.com/?tab=model#/model-market)实时列表为准，优先选用带 `3.7` 版本号的模型。 -## 首次调用：获取 API Key 与发起请求 +## 关键参数 -调用流程分为账号准备与代码调用两步，完整步骤见 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)： +| 参数 | 说明 | 注意事项 | +|------|------|----------| +| **API Key** | 用于身份认证，需在[API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建 | 不同地域的 API Key **不通用**；Coding Plan 和 Token Plan 需使用专属 Key [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) | +| **Base URL** | 模型服务接入地址，决定地域、协议与鉴权范围 | 必须与 API Key 所属地域匹配；业务空间专属域名（`{WorkspaceId}.{region}.maas.aliyuncs.com`）为生产环境推荐方案 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) | +| **WorkspaceId** | 业务空间唯一标识，用于构造专属 Base URL | 仅华北2（北京）、新加坡、日本（东京）、德国（法兰克福）需填写；美国（弗吉尼亚）使用 `dashscope-us.aliyuncs.com` 无需 WorkspaceId [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) | +| **model** | 模型 ID，如 `qwen3.7-plus` | 模型名与地域强绑定（例如 `qwen3.7-plus-us` 仅限美国地域）；不同地域支持的模型列表不同 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) | -1. **开通与拿 Key**：用阿里云主账号开通百炼（需实名认证），在控制台 API Key 页面创建 API Key。 -2. **获取业务空间 ID**：使用华北2（北京）、新加坡、日本（东京）、德国（法兰克福）地域时，需在 Base URL 中填入业务空间 ID（`WorkspaceId`）。 -3. **配置环境变量**：建议将 Key 写入环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。 -4. **安装 SDK 并调用**：可用 OpenAI Python/Node SDK（`pip install -U openai`）或 DashScope SDK（`pip install -U dashscope`）。 +## 使用方式 -OpenAI 兼容方式的最小示例（北京地域）： +### 1. 准备工作 +- 注册阿里云账号并完成实名认证； +- 开通百炼服务，在[API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建 Key； +- 若使用业务空间专属域名，需在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)获取 `WorkspaceId`。 +### 2. 调用示例（OpenAI 兼容） ```python import os from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - # 各地域 base_url 不通用，{WorkspaceId} 替换为业务空间 ID - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", # 替换为实际 WorkspaceId ) + completion = client.chat.completions.create( - model="qwen-plus", - messages=[{"role": "user", "content": "你是谁？"}], + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你是谁？"}] ) print(completion.choices[0].message.content) ``` -> **注意**：不同文档中示例模型名不一致（概述文档用 `qwen3.7-plus`，首次调用文档用 `qwen-plus`）。`qwen-plus` 为稳定版别名，`qwen3.7-plus` 为具体版本，二者均可用，请以 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) 页面当前列出的模型 ID 为准。 - -## 地域、服务部署范围与接入域名 - -百炼目前提供华北2（北京 `cn-beijing`）、新加坡（`ap-southeast-1`）、美国弗吉尼亚（`us-east-1`）、德国法兰克福（`eu-central-1`）、日本东京（`ap-northeast-1`）五个地域。调用前需选定三项，详见 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)： - -- **地域**：决定接入点与数据存储位置，就近选择可降低延迟。 -- **服务部署范围**：决定推理执行位置；有数据合规需求选特定地理边界（如"中国内地""欧盟""美国"），无合规需求选"全球"（推理资源池更大）。北京与新加坡各仅支持一种范围，无需选择。 -- **接入域名**：影响并发上限、超时等服务保障。 - -> **注意**：各地域的接入域名、API Key 和模型列表相互独立，**不能跨地域混用**；跨地域使用 API Key 会报错。美国（弗吉尼亚）地域可用带 `-us` 后缀的模型名（如 `qwen-plus-us`）限定美国境内推理。 - -## Base URL 选择 - -Base URL 是模型 API 的调用地址，必须与同一计费方案的 API Key 配套使用，否则会报错 401。按量付费下有三类域名（详见 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)）： - -- **业务空间专属域名（推荐）**：格式 `{WorkspaceId}.{region}.maas.aliyuncs.com`，用于生产环境，具备更高并发、更低时延与流量隔离，请求超时 3600 秒，支持 HTTP/SSE/WebSocket/WebRTC。 -- **Dashscope 域名**：如 `dashscope.aliyuncs.com`（北京）、`dashscope-intl.aliyuncs.com`（新加坡）、`dashscope-us.aliyuncs.com`（美国），存量兼容用，建议迁移到专属域名，请求超时 600 秒。 -- **试用域名**：`trial.{region}.maas.aliyuncs.com`，仅用于快速验证，RPM 固定 1000，不提供 SLA，不建议生产使用。 - -三种接口路径后缀：OpenAI 兼容 `/compatible-mode/v1`、DashScope `/api/v1`、Anthropic 兼容 `/apps/anthropic`。此外，Token Plan 与 Coding Plan 有各自专属域名和 API Key，仅限 Claude Code、Codex 等交互式 AI 工具使用，不能用于后端服务。 +> 完整代码与 Node.js/curl 示例见 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)。 -## 限流规则与规避 +### 3. SDK 选择 +- **OpenAI Python SDK**：兼容性好，适合已有 OpenAI 项目迁移； +- **DashScope Python SDK**：原生支持，提供更细粒度控制（如 `dashscope.base_http_api_url` 设置）； +- **其他语言**：官方提供 Java、Go、C# 等 SDK，详见 [API 参考文档](https://help.aliyun.com/zh/model-studio/qwen-api-reference/)。 -百炼按**主账号维度**限流，账号下所有 RAM 子账号、业务空间和 API Key 的调用量合并计算；不同模型的限流额度相互独立。限流同时约束 RPM（每分钟请求数）和 TPM（每分钟 Token 数，含输入与输出），并可能按秒级 RPS/TPS 执行。超限请求会被拒绝，通常一分钟内自动恢复。详见 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)。 +## 限制和注意事项 -常见错误信息与对策： - -- `Requests rate limit exceeded`：触发 RPM 限流，降低调用频率。 -- `Allocated quota exceeded`：触发 TPM 限流，缩短输入或限制输出。 -- `Request rate increased too quickly`：请求瞬时激增触发稳定性保护，采用匀速调度、指数退避或请求队列平滑请求。 - -规避建议：优先选用限流额度更高的模型（稳定版比带日期的快照版更宽松）；配置备选模型做失败重试；拆分大任务；无需实时响应时改用批量推理（Batch API，不受实时限流约束）；额度不足时在控制台"限流提额"页面提升临时 TPM 额度（提交即生效，有效期 30 天，目前支持北京和新加坡地域）。 - -> **注意**：限流仅约束单位时间内的调用速率，**不限制累计用量与费用**。控制费用需另行设置消费限额、费用告警，或开启"免费额度用完即停"（仅新用户、仅北京地域、且在免费额度有效期内有效）。 +- **地域隔离**：各地域（北京、新加坡、美国、德国、日本）的 API Key、Base URL、模型列表、计费策略均独立，**不可混用** [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)； +- **限流策略**： + - 按主账号维度聚合所有子账号、业务空间、API Key 的调用量； + - 分 RPM（每分钟请求数）和 TPM（每分钟 Token 消耗）双重限制，超出任一即返回 `429`； + - `qwen3.7-plus` 在北京地域默认限流为 **30,000 RPM / 5,000,000 TPM**，而快照版本（如 `qwen-plus-2025-07-28`）仅为 **60 RPM / 1,000,000 TPM** [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)； +- **域名选择**： + - **业务空间专属域名**：推荐生产环境，SLA 99.9%，超时 3600 秒，支持 WebSocket/WebRTC； + - **Dashscope 域名**（如 `dashscope.aliyuncs.com`）：兼容存量，但建议迁移； + - **试用域名**（如 `trial.cn-beijing.maas.aliyuncs.com`）：RPM 限 1000，**禁止用于生产** [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)； +- **费用控制**： + - 新用户享北京地域免费额度，用完后自动转按量付费（已认证用户）或停止服务（未认证用户）； + - 可开启“免费额度用完即停”开关，或订阅 Coding Plan 实现月度固定预算 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 ## 来源文档 - [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) -- [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) - [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) -- [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) -- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) - [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) +- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md index 4dbd908e..0eadf46a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md @@ -1,74 +1,47 @@ # [knowledge](../api/knowledge.md) base -知识库是阿里云百炼基于 RAG（检索增强生成）技术为大模型补充私有数据和最新信息的能力：大模型在生成回答前先从知识库检索相关内容，从而提升回答准确性。围绕知识库，平台提供了创建管理、检索服务、知识问答、效果优化、API 集成、日志监控与计费等一整套开发者可消费的功能。 +知识库是阿里云百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，用于为大模型注入私有、领域专属或时效性强的结构化与非结构化数据。它通过语义检索从文档、表格、音视频等多源内容中精准召回相关信息，并将其作为上下文输入大模型，从而显著提升回答的准确性、专业性与可溯源性。知识库功能仅在中国站华北2（北京）地域可用，且需在业务空间内完成创建与集成。 -> **注意**：知识库功能仅能在中国站 **华北2（北京）** 地域开通和使用，新加坡、德国（法兰克福）等其他地域均不支持。API 相关能力仅适用于文档搜索类知识库。 +## 支持的模型/功能 -## 支持的模型与知识库类型 +知识库支持与阿里云百炼平台上的多种预置及自定义模型协同工作。预置模型包括千问全系列（QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、VL-Max/Plus/Flash/OCR、开源版 Qwen3/Qwen2.5/Qwen2 等），以及第三方文本模型（如 DeepSeek-R1、Llama3.1、Yi-Large 等）。经调优的自定义模型（如千问-Plus/Turbo、Qwen3 开源版调优模型等）同样支持 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 -在 [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) 中，支持挂载知识库的预置模型包括千问-QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、千问VL-Max/Plus/Flash/OCR、千问开源版（Qwen3/Qwen2.5/Qwen2 等）以及 DeepSeek-R1/V3.1、Llama3.1、Yi-Large 等第三方模型；自定义（调优后）模型支持千问-Plus/Turbo、千问VL-Max/Plus 及开源版。实际可选模型以创建应用时页面列表为准。 +除基础文档问答外，知识库还提供面向不同场景的专用能力： +- **知识检索服务**：支持最多 15 个知识库联合检索，具备 Query 改写、混合检索（向量+关键词）、Rerank 排序及精细化参数控制； +- **知识问答服务**：在检索基础上叠加大模型生成，支持极速模式（单轮）与多轮智能模式（Agentic 规划搜索），并提供拒答、防泄漏、引用溯源等生成控制能力 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)。 +> **注意**：文档 1 中列出的“千问VL-Max/Plus/OCR”等视觉模型，在文档 7 和 8 的检索/问答服务配置项中明确限定为“多模态知识库（图片知识库、视觉理解知识库）”专用，不可用于纯文本知识库的排序（rerank）；而纯文本知识库仅支持 `qwen3-rerank` 系列模型。该差异表明模型支持范围需严格按知识库类型和功能模块区分，不可跨类型泛化使用。 -创建知识库时需选择类型，创建后不可更改： +## 关键参数 -- **文档搜索**：面向企业内部文档、产品手册等非结构化数据，可细分为基础文档问答、图文并茂回复、视觉理解（富文本文档，向量模型强制为 qwen3-vl-embedding）和极速问答（低延迟、仅文本查询）。 -- **数据查询（结构化）**：单个 Excel/CSV 文件，支持 NL2SQL。 -- **图片问答**、**音视频搜索** 等多模态类型。 - -## 关键参数与配置 - -- **相似度阈值**：仅语义相似度高于阈值的切片才会被召回，阈值过高会导致相关切片被全部过滤（例如调到 0.60 可能返回无召回结果）。 -- **召回片段数 / 最大召回数量（TopK/K）**：取值范围 1–20。对总结、列举、比较类复杂问题适当增大有助于生成完整答案，但会增加 Token 消耗，推荐使用「按拼装长度」策略。 -- **权重**：多知识库召回时按信息源重要性分配，但**权重仅在同类型知识库之间生效**。 -- **Meta 信息抽取**：以 key-value 附加到切片，可显著提升检索准确性并降低 Token 消耗；支持常量、变量（file_name/cat_name）、大模型、正则、关键词搜索五种取值方式。**知识库创建后无法再配置 metadata 抽取**。 -- **智能切分 / 多轮对话改写**：均在创建知识库时配置；多轮对话改写与知识库绑定，创建时未开启则后续无法补开，除非重建知识库。 - -## 检索与问答服务 - -平台在知识库之上提供两类独立服务： - -- **知识检索**：见 [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)。支持多知识库联合检索（最多 15 个），流水线为 Query 改写 → 向量+关键词混合检索 → Rerank 精排 → 加权返回。全局参数含知识库路由、混排模型（qwen3-rerank / qwen3-rerank(hybrid) / qwen3-vl-rerank）、混排模式（问答/相似/自定义）；每个知识库可独立配置初步向量/关键词 TopK（1–100）、排序模型、相似度阈值、标签过滤等。 -- **知识问答**：见 [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)。基于大模型（如 qwen3.6-plus/qwen3.7-plus）结合检索生成自然语言回答，提供**极速**（单轮）与**多轮智能**（Agentic 规划搜索）两种检索模式，支持文件预解析、拒答、防泄漏、多模态回复与引用来源展示。 +知识库的核心行为由以下关键参数控制： +- **相似度阈值（0.01–1.0）**：作用于 Rerank 排序后结果，仅保留得分高于该阈值的切片。值过高易漏召，过低则引入噪声； +- **初步向量/关键词检索 TopK（1–100）**：分别控制向量与关键词双路召回的初始切片数，直接影响 Rerank 模型的 Token 消耗与最终精度； +- **最大召回数量（1–20）**：指最终返回给下游（大模型节点或问答服务）的切片总数； +- **权重与标签过滤**：多知识库场景下，权重影响混排优先级；标签（单文件最多 32 个）支持按业务维度（如 `bailian_mobile`）进行精准范围过滤 [原文标题](../../raw/application-user-guide/knowledge-base/rag-optimization.md)； +- **元数据（metadata）抽取**：在索引阶段为文本切片注入 `filename`、`date`、`author` 等结构化信息，实现“先过滤、再检索”，大幅提升高相似度干扰场景下的准确率。 ## 使用方式 -1. **控制台**：进入知识库页面 → 选择标准版/旗舰版 → 创建知识库（填写基础信息、选类型、配置数据来源与索引参数）→ 关联到智能体/工作流应用。工作流应用中将「知识库」节点接在开始节点后，用内置变量 `query` 作输入，输出 `result` 传给大模型节点。 -2. **API/SDK**：面向外部应用，通过阿里云百炼 SDK 集成检索能力。完整创建流程（申请上传租约 → 上传文件 → AddFile → CreateIndex → SubmitIndexJob → 轮询状态）参见 [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。子账号需获取 `AliyunBailianDataFullAccess` 策略并加入业务空间，endpoint 为 `bailian.cn-beijing.aliyuncs.com`。 - -## 效果优化 - -RAG 效果由建立索引、检索召回、生成答案三阶段决定。建议先用 [自动评测] 建立至少 100 组问题的评估基线（覆盖事实型/比较型/教程型/分析型），再针对失败用例（大模型打分 < 4）诊断改进：检索无效（补充知识、优化排版、统一实体、开启多轮对话改写）、召回不相关（标签过滤 / 元数据结构化搜索）、切片不完整（智能切分 + 人工修正）、重排不佳（调整相似度阈值与召回片段数）、模型理解有误（更换为参数更多的商业模型）。详见 [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md)。 - -## 配额与限制 - -摘自 [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)： +知识库可通过三种方式集成到应用中： +1. **智能体/工作流应用内嵌**：在应用配置页点击“文档知识库”旁的 `+` 添加知识库，设置相似度阈值与权重；工作流中需拖入“知识库节点”，配置 `content` 输入变量（通常为 `query`）及 `TopK`，再连接至大模型节点，并在提示词中引用 `{result}` 变量； +2. **独立服务形态**：通过控制台“知识检索”或“知识问答”标签页创建服务，绑定多个知识库并统一配置混排模型、路由策略与生成参数，发布后即可通过 API 或调试窗口直接调用； +3. **外部系统集成**：使用阿里云百炼 SDK（Python/Java 等）调用知识库 API，完成创建、上传、索引、检索全流程自动化。API 调用需子账号具备 `AliyunBailianDataFullAccess` 权限，并配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` 等环境变量 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 -- **知识库数量**：使用 RDS 数据源上限 100，其它数据源无限制。 -- **存储容量**：旗舰版 9,999 GB，标准版 100 GB（免费平台存储）。 -- **文件格式与大小**：pdf/docx/ppt 等 ≤150MB 且 ≤1000 页；txt/markdown/html ≤10MB；图片 ≤20MB；音视频 ≤512MB。 -- **切片**：单切片最大 6,000 Token；编辑切片长度 10–6000 字符，删除切片单次最多 10 个。音视频搜索类不支持新增切片。 -- **向量模型**：文档/数据查询/音视频类支持 text-embedding-v4/v3（512 维）；图片问答类仅 multimodal-embedding-v1（1024 维），维度不可更改。 -- **检索并发**：旗舰版 50–10,000 QPS（可调，对应 1–200 RCU），标准版固定 1 QPS。 -- **召回**：单次查询最多召回 20 个切片；单次控制台导入最多 50 个文件（API 批量建议 ≤10,000）。 +## 限制和注意事项 -## 计费 - -知识库服务自 **2026 年 1 月 4 日** 起正式计费，费用由**规格费用**与**模型调用费用**两部分构成。规格费用按运行时长按小时出账：标准版 0.03 元/知识库/小时，旗舰版 0.2 元/RCU/小时（1 RCU ≈ 50 QPS）。扣费顺序为：免费额度 > 资源包 > 按量付费。平台提供一次性 720 小时免费额度（仅抵扣标准版规格费用，不含模型调用），多个知识库按数量倍数扣减。模型调用费用按输入 Token 计费，其中 Rerank 排序费用取决于**初步召回的总切片数**而非最终返回数量，可通过关闭排序或调低初步 TopK 降低成本。 - -> **注意**：删除知识库会**永久清除数据且无法恢复**，但也是停止计费的唯一方式。2026 年 1 月 4 日前创建但未开通服务的数据保留至 2026 年 6 月 30 日，逾期永久删除。 - -## 日志与监控 - -所有检索调用都会以日志形式投递到日志服务（SLS），用于调用审计、问题排查、用量统计与告警。在知识库列表页右上方「监控配置」中完成 SLS 角色授权、开通日志服务并创建 LogStore 后，检索日志会实时投递（秒级延迟），topic 为 `log_dispatch`。关键索引字段包括 `request_id`、`pipeline_id`（知识库 ID）、`workspace_id`、`path`、`latency`、`response_status_code`、`response_code`（成功为 `Success`）、`request_body`/`response_body`（召回切片在 `data.nodes[]`）。可基于这些字段在 SLS 中做按知识库/业务空间的用量聚合、按 API 路径统计，并搭建调用量趋势、TopN 排名仪表盘及错误率告警。 - -> **注意**：关闭「检索日志」开关只停止新日志投递，历史日志仍按 SLS 默认配置保留与计费；如需彻底停止计费，需在 SLS 控制台删除对应 LogStore。SLS 存储与流量按其自身标准单独计费。 +- **地域限制**：知识库功能仅支持中国站华北2（北京）地域，新加坡、法兰克福等其他地域不支持，此限制在文档 1 与文档 4 中均被明确强调； +- **配额约束**：标准版知识库并发固定为 1 QPS，旗舰版为 50–10,000 QPS（按 RCU 计费）；单次控制台导入文件上限 50 个，单个文件最大 150MB（PDF/DOCX）；文本切片长度上限 6,000 Token； +- **计费要点**：费用分为两部分——**规格费用**（按知识库运行时长，标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时）与**模型调用费用**（向量模型与 Rerank 模型按实际 Token 消耗计费，不包含在规格费中）； +- **配置不可逆性**：知识库创建后，类型（如“文档搜索”）、元数据抽取配置、多轮对话改写开关均无法修改，需重建知识库； +- **日志监控**：所有检索请求自动投递至 SLS 日志服务，字段如 `pipeline_id`（知识库 ID）、`response_code`（业务响应码）、`data.nodes[]`（召回切片）可用于审计与问题排查 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)。 ## 来源文档 -- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md) -- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) +- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) +- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md index 464791ba..9f0fe6e1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md @@ -1,109 +1,84 @@ # llm application -阿里云百炼平台提供三种核心应用构建模式：智能体（Agent）、工作流（Workflow）和高代码应用，用于突破大模型在私有知识访问、实时信息获取和复杂任务规划方面的原生局限。开发者可根据开发门槛、控制粒度和业务场景选择合适的应用类型，并通过集成知识库、MCP 工具、插件等能力构建完整的 AI 应用。 - -## 应用类型与选型 - -| 对比维度 | 智能体（Agent） | 工作流（Workflow） | 高代码应用 | -|---------|---------------|-------------------|-----------| -| 开发方式 | 自然语言配置（零代码） | 可视化节点编排（低代码） | Python 编码 | -| 核心特点 | AI 自主决策、动态规划 | 预定义流程精确控制 | 完全由代码控制 | -| 适合人群 | 业务人员、产品经理 | IT 运维、业务分析师 | AI 工程师、开发者 | -| 开发门槛 | 低 | 中 | 高 | - -详细的类型介绍参见 [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md)。 - -## [智能体应用](../concepts/agent-application.md) - -### 新版智能体（Agent 2.0） - -新版智能体将知识库、MCP 等能力统一为工具，由智能体自主规划调用顺序，支持完整的"规划-执行-反思"链路展示。推荐在无旧版依赖时使用新版。 - -核心能力配置： - -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）。支持配置最长回复长度、temperature、enable_thinking（思考模式）等参数。 -- **提示词**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境中的 bash、write、read、edit、glob、grep、download_file 等工具，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话中工具调用最大次数。 - -详细配置方法参见 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 - -### 旧版智能体（Agent 1.0） - -旧版智能体通过知识库（RAG）和插件扩展能力，适合意图单一、流程固定的简单任务。知识库检索后再决策是否调用其他工具。 - -> **注意**：新版智能体与旧版智能体基于不同技术架构，不支持直接升级或版本切换。需要迁移时必须重新创建新版应用。 - -旧版智能体的自定义插件有 5 秒超时限制。详情参见 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 - -## 工作流应用 - -工作流通过可视化节点编排将多步骤任务串联为稳定可控的执行链路，适合固定流程自动化场景。 - -主要节点类型： - -- **开始/结束节点**：定义输入输出参数结构，预置 query、historyList、imageList 变量 -- **大模型节点**：配置模型、提示词和用户提示词 -- **意图分类节点**：根据用户输入分发到不同处理分支 -- **智能体群组节点**：将子智能体作为工具组合调用 -- **变量处理节点**：文本输出或变量转换 - -工作流支持会话变量作为全局参数在节点间传递，支持记忆功能（本节点缓存或自定义缓存）。 - -创建和配置详情参见 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 - -## 高代码应用 - -面向专业开发者，支持基于 Python 项目部署 AI 后端服务。 - -关键特性： - -- **部署方式**：Serverless Function（无状态快速拉起）和 K8s（高性能有状态长程任务） -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务 -- **前端体验**：支持直接体验、自定义交互卡片、基于 Spark Design 的自定义 WebUI -- **企业级能力**：自动化运维、可观测、日志服务、API 网关 -- **代码提交**：支持控制台模板创建或命令行上传 .whl 代码包 - -生产环境建议开启网关功能，通过自定义域名访问。时延敏感业务建议最小实例数大于等于 1。 - -详细开发和部署流程参见 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 - -## 文件问答 - -[智能体应用](../concepts/agent-application.md)支持上传文件进行智能问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等需工具介入的任务 | 功能灵活，依赖配置的工具 | - -文件限制：单会话最多 10 个文件，单文件不超过 10MB。超过 10MB 需使用文件上传 API。 - -支持格式：文档（doc/docx/pdf/md/txt 等）、图片（png/jpg/bmp/gif）、视频（mp4/mkv/avi 等）、音频（mp3/wav/flac 等）。 - -详细使用方式参见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成： - -1. 在应用配置页点击"发布"，确认变更后完成发布 -2. 在"发布渠道"页签查看 API 调用方式 -3. [智能体应用](../concepts/agent-application.md)还支持发布到钉钉、微信公众号等第三方平台 - -RAM 账号发布前需确认拥有 `ram:CreateServiceLinkedRole` 权限。 - -## 计费说明 - -- **模型调用**：按模型类型和 [Token](../concepts/token.md) 用量计费 -- **知识库**：按量付费，召回的文本切片会增加输入 [Token](../concepts/token.md) -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 费用由第三方收取 -- **高代码应用**：部署后函数计算、API 网关、存储均按量计费 -- **文件上传**：上传本身不收费，问答消耗按所选模型标准计费 +百炼平台的 LLM Application 是面向真实业务场景的 AI 应用构建体系，通过智能体（Agent）、工作流（Workflow）和高代码应用三种模式，突破大模型在私有知识接入、实时信息获取、流程控制与复杂任务规划等方面的原生局限。开发者可根据业务复杂度、可控性要求与团队技术栈，选择零代码、低代码或专业编码方式快速落地可交付的 AI 服务。 + +## 支持的模型/功能 + +- **模型支持**：所有 LLM Application 类型均支持千问系列主流模型（如 `千问-Max`、`千问-Plus-Latest`、`千问-VL-Max`），部分能力对模型有特定要求： + - 新版智能体（Agent 2.0）推荐使用具备强工具调用能力的模型（如 `千问-Max` 系列），以保障多步规划效果 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)； + - 文件问答中，`千问-VL` 系列模型可直接解析图片/视频，无需开启预解析；而文本模型在“自定义处理”模式下依赖显式配置的工具 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)； + - 工作流应用中，各节点（如意图分类、大模型）可独立选择模型，常见实践选用 `千问-Plus-latest` [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 + +- **核心能力矩阵**： + | 能力类型 | 智能体（Agent） | 工作流（Workflow） | 高代码应用 | + |----------|----------------|---------------------|-------------| + | **知识库（RAG）** | ✅ 作为自主调用工具（Agent 2.0）或固定检索源（旧版） | ✅ 可在大模型节点中启用 RAG 或通过“切片检索”混合文件与知识库 | ✅ 一站式 MCP 接入，支持关联知识库 | + | **外部工具** | ✅ 内置沙箱工具（`bash`/`write`/`read`等）、MCP、插件、应用组件 | ✅ 通过 API 节点、函数计算节点或 MCP 节点调用 | ✅ MCP 工具接入（知识库、工作流、插件等） | + | **多模态支持** | ✅ 千问-VL 模型直解析；其他模型依赖预解析或工具调用 | ✅ 大模型节点支持 `image_list` 输入，需模型具备视觉能力 | ✅ 代码中可自由处理多模态输入/输出 | + | **[长期记忆](../concepts/long-term-memory.md)** | ⚠️ 新版仅支持短期记忆（0–30 轮），[长期记忆](../concepts/long-term-memory.md)“计划未来迭代支持” [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)；旧版文档提及[长期记忆](../concepts/long-term-memory.md)“不收费”，但未说明是否已上线 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) | ✅ 通过会话变量（`historyList`）和节点级“自定义缓存”实现跨节点上下文传递 | ✅ 由开发者在 Python 代码中自主实现 | + +> **注意**：关于长期记忆，[新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) 明确标注“该功能计划在未来的迭代中支持”，而 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) 在计费说明中称“长期记忆的数据存储不收费”，但未确认其当前可用性。实际开发应以控制台界面显示为准，短期记忆（上下文轮数）是当前唯一稳定可用的记忆机制。 + +## 关键参数 + +- **通用参数**： + - `temperature`：控制生成随机性（0.0–1.0），值越高越发散； + - `max_tokens`（最长回复长度）：限制模型输出 token 数，不含提示词； + - `enable_thinking`：仅对支持思考模式的模型（如 `千问-Max`）生效，开启后可展示推理链路。 + +- **智能体专属参数**： + - `ReAct 最大轮次`（1–50）：限制单次会话中工具调用总次数，超限则终止规划并生成最终回复； + - `短期记忆轮数`（0–30）：控制多轮对话中向模型注入的历史消息数量； + - `预解析文件`开关：决定上传文件是直接传 URL（关闭）还是由系统解析为文本（开启）；千问-VL 模型例外，关闭时仍可直解析图片/视频。 + +- **文件问答专用参数**（见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)）： + - **全文引用模式**：`单文件最大解析长度`（token）、`最大拼装长度`（token），截断策略为从末尾丢弃； + - **切片检索模式**：`召回片段数`、`最大拼装长度`，超长时按相关性得分从低到高丢弃； + - **自定义处理模式**：需显式挂载 MCP/插件，并在系统提示词中引导调用逻辑。 + +- **工作流专属参数**： + - 会话变量（`query`, `historyList`, `imageList`）全局可用，支持跨节点引用； + - 节点级“记忆”开关（自定义缓存 vs 本节点缓存），影响上下文范围。 + +## 使用方式 + +- **创建与配置**： + - 智能体：控制台 → 应用管理 → 创建应用 → 选择“智能体应用” → 指定 Agent 2.0（推荐）或旧版； + - 工作流：控制台 → 应用管理 → 创建应用 → 选择“工作流应用” → 拖拽节点（开始/大模型/意图分类/结束等）并连线配置； + - 高代码应用：控制台 → 应用管理 → 创建应用 → 选择“高代码应用” → 选择 Serverless Function（默认）或 K8s 部署方式，上传 `.whl` 包或选模板。 + +- **调试与测试**： + - 所有类型均提供右侧对话窗口实时调试； + - 智能体（Agent 2.0）支持卡片流展示“思考→工具调用→反思”全过程； + - 工作流支持画布内逐节点测试，查看中间变量输出； + - 高代码应用提供“文本对话体验”与“API 测试”双模式，支持 `GET /health` 和 `POST /process`。 + +- **发布与集成**： + - **必须发布后方可调用**：发布操作位于应用配置页右上角，发布前会对比变更差异； + - API 调用：各应用在“发布渠道”页签 → “API 调用” → “查看 API”，获取 endpoint、鉴权方式（Bearer Token）及请求体格式； + - 高代码应用额外支持网关部署：开通云原生 API 网关，配置路由与 Token 鉴权，生产环境建议禁用测试域名公网访问。 + +## 限制和注意事项 + +- **文件限制**： + - 单次会话最多上传 10 个文件，单文件 ≤ 10 MB； + - 会话内上传文件仅当前会话有效，刷新/关闭页面即失效；生产环境推荐使用文件上传 API 获取 `session_file_id`（有效期 24 小时）或 OSS 公网 URL [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 + +- **调用限制**： + - 智能体应用默认限流 100 次/分钟，此配额被所有 API 请求共享（含文件问答、普通对话）； + - 自定义插件超时限制为 5 秒 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)； + - 工作流节点间数据传递受 JSON 序列化大小限制，避免在变量中塞入超大二进制内容。 + +- **模型与能力兼容性**： + - `enable_thinking` 参数仅对明确支持思考模式的模型生效，不支持的模型无法配置该参数； + - 千问-VL 模型在“自定义处理”模式下，图片可选“模型处理”或“模型处理+规划”，后者需额外挂载 MCP 工具 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)； + - 旧版智能体与新版（Agent 2.0）架构不兼容，**无法升级/降级**，需重新创建 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 + +- **计费关键点**： + - 应用创建不收费，仅调用时产生费用； + - 模型调用费用 = 输入 Token + 输出 Token × 对应模型单价； + - 知识库检索内容计入输入 Token，可能推高模型费用； + - MCP 工具费用分两类：阿里云官方 MCP 按模型调用计费；第三方 MCP 产生的费用由第三方收取，百炼不抽成。 ## 来源文档 @@ -111,13 +86,7 @@ RAM 账号发布前需确认拥有 `ram:CreateServiceLinkedRole` 权限。 - [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) - [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) - [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) -- [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) - [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) - - - - - - +- [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md index 7d772b2a..f4796ed0 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md @@ -1,74 +1,55 @@ # managed agents -Managed Agents 是百炼提供的智能体托管运行时，面向多步工具调用、代码执行、文件处理等长时运行任务。与无状态的[智能体应用](../concepts/agent-application.md)不同，它由平台在服务端托管会话状态、沙箱环境与工具执行，智能体在独立云端容器中自主执行命令、读写文件、安装依赖，事件历史在服务端持久化并支持中断与续接。 +Managed Agents 是百炼平台提供的智能体托管运行时，用于执行多步工具调用、代码执行、文件处理等长时运行任务。平台在服务端统一托管会话状态、沙箱环境与工具执行生命周期，开发者无需自行实现代理循环、沙箱编排或事件持久化。其核心抽象包括智能体（Agent）、运行环境（Environment）、会话（Session）和事件（Event）四个层级，支持有状态、可中断、可续接的会话式交互 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)。 -## 与[智能体应用](../concepts/agent-application.md)的区别 +## 支持的模型与功能 -| 维度 | [智能体应用](../concepts/agent-application.md) | Managed Agents | -| --- | --- | --- | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，云端容器 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流，事件历史持久化 | -| 典型场景 | 问答、对话、轻量任务 | 多步工具调用、代码执行、文件处理等长时任务 | +- **模型支持**：支持 `qwen3-max`、`qwen3.7-plus` 等 Qwen 系列大模型（具体以控制台下拉列表为准），模型通过 `model.id` 字段指定；不支持自定义模型部署，仅限百炼托管模型。 +- **内置工具**：默认提供 7 个内置工具：`bash`（命令执行）、`read`/`write`/`edit`（文件读写与编辑）、`glob`（路径通配）、`grep`（文本搜索）、`download_file`（从 URL 下载）。工具启用状态需显式配置（如 `{"name": "bash", "enabled": true}`），未启用则不可调用 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)。 +- **扩展能力**： + - **MCP 服务**：可接入符合 MCP 协议的外部工具服务； + - **Skill**：预置的工具组合封装，用于端到端任务流程（如数据清洗、报告生成）； + - **文件挂载**：支持上传文件并挂载至 `/mnt/session/uploads/` 路径，会话内可直接通过工具访问；单文件上限 10 MB [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)。 -## 核心概念 +> **注意**：文档 1 的 Java SDK 示例中 `AgentCreateParam.builder().instructions(...)` 使用了 `instructions` 字段，而文档 2 和文档 3 均明确使用 `system_prompt` 或 `system` 字段；实际 API 以 `system`（Python/HTTP）或 `systemPrompt`（Java SDK v1.2+）为准，旧版 `instructions` 已弃用，建议统一使用 `system`。 -四个核心对象构成完整的运行链路，详见 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)： +## 关键参数 -- **智能体（Agent）**：模型、系统提示词、工具、MCP 服务和 Skill 的组合配置。创建后通过 ID 引用，可在多个会话中复用。 -- **运行环境（Environment）**：会话运行的沙箱配置，由百炼托管的云端容器，独立于智能体管理，可被多个会话复用。 -- **会话（Session）**：智能体在指定环境中的一次运行实例，承载任务执行与输出。 -- **事件（Event）**：应用与智能体之间交换的消息，包括用户消息、工具调用结果和状态变更。 - -## 支持的工具 - -智能体通过以下工具与运行环境交互： - -- **命令执行**：在沙箱中运行 shell 命令（`bash`）。 -- **文件操作**：`read`、`write`、`edit`、`glob`、`grep`，以及从 URL 下载的 `download_file`；也可上传本地文件挂载到沙箱。 -- **MCP 服务**：接入外部工具服务，详见 [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md)。 -- **Skill**：挂载预置的工具组合，封装端到端任务流程。 - -快速开始阶段默认全选 7 个内置工具（`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`），可按需取消勾选。 +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `name` | string | 是 | 智能体名称，仅用于标识，不影响行为 | +| `model.id` | string | 是 | 模型 ID，如 `"qwen3-max"`，必须为平台支持的托管模型 | +| `system` | string | 是 | 系统提示词，定义角色、约束与行为准则 | +| `tools` | array | 否（但无工具则无法执行操作） | 工具配置数组，每个元素含 `type="builtin_toolkit"`、`default_config` 和 `configs`（含 `name` 与 `enabled`） | +| `environment_id` | string | 创建 Session 时必填 | 运行环境 ID，指向已创建的云端沙箱 | +| `resources` | array | 否 | 创建 Session 时可指定挂载资源列表，格式为 `[{ "resource_id": "...", "mount_path": "/mnt/session/uploads/data.csv" }]` | ## 使用方式 -典型流程分为四步（控制台向导或 API 均可完成），参见 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)： - -1. **配置智能体**：指定名称、模型（如 `qwen3-max`）、系统提示词与工具。API 为 `POST /api/v1/agentstudio/agents`。 -2. **配置运行环境**：默认云端托管沙箱，可通过 `config.packages` 预装 apt / pip 依赖并设置网络策略。API 为 `POST /api/v1/agentstudio/environments`。 -3. **发起会话**：绑定智能体 ID 与环境 ID 创建会话实例。API 为 `POST /api/v1/agentstudio/sessions`。 -4. **发送事件并接收响应**：向会话写入用户消息触发处理（`POST /sessions/{id}/events`），通过 SSE 事件流实时接收工具调用过程与输出（`GET /sessions/{id}/events/stream`）。 - -控制台的**预览调试**标签页可直接对话并按事件类型（User、Agent、Tool、Tool_output、Error、Model、System）筛选查看执行过程。 - -### 上下文与资源挂载 - -上下文中的挂载资源独立于会话管理，可被多个会话复用，详见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)： - -- **挂载时机**：创建会话时在 `resources` 字段指定，或运行时通过 `POST /sessions/{session_id}/resources` 追加，实时生效且无需重启会话。 -- **路径约定**：挂载资源统一放在 `/mnt/session/uploads` 前缀下，可在系统提示词中直接引用完整路径。 -- **会话隔离**：平台为挂载资源做内部拷贝放入沙箱，会话内的修改不影响原始资源，也不影响挂载同一资源的其他会话；卸载后副本被清理。 +1. **创建智能体**：通过控制台向导或 API 提交 `POST /api/v1/agentstudio/agents`，传入 `name`、`model`、`system` 和 `tools`；智能体 ID 可复用于多个会话 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)。 +2. **创建运行环境**：独立于智能体创建沙箱，支持 `cloud` 类型（百炼托管容器），可配置 `packages.apt`/`packages.pip` 安装依赖及 `networking.type`（如 `"unrestricted"`）。 +3. **发起会话**：调用 `POST /api/v1/agentstudio/sessions`，绑定 `agent`（ID）、`environment_id`，并可选传入 `resources` 挂载文件。 +4. **交互与流式消费**： + - 发送用户消息：`POST /api/v1/agentstudio/sessions/{session_id}/events`，`input` 中包含 `role: "user"` 消息； + - 接收事件流：`GET /api/v1/agentstudio/sessions/{session_id}/events/stream`，SSE 流返回 `message`、`tool_call`、`tool_output`、`session_status` 等事件类型，需按 `event.type` 解析 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 -## 限制与注意事项 +## 限制和注意事项 -- 单个上传文件不超过 **10 MB**。 -- 沙箱内文件路径遵循 `/mnt/session/uploads` 约定，代码中引用文件应使用该完整路径。 -- 会话状态、中断续接与工具审批由会话状态机管理，详见 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 - -> **注意**：文档中出现的模型名称不一致——控制台向导示例填写 `qwen3.7-plus`，而 API 代码示例使用 `qwen3-max`。请以控制台模型下拉列表中实际可选的模型 ID 为准。 +- **沙箱隔离性**：每个会话运行在独立云端容器中，挂载文件为副本，会话间互不影响；卸载后副本自动清理，原始资源保留。 +- **会话生命周期**：会话默认最长运行 2 小时（超时自动终止），可通过 `session_status` 事件监听 `idle` 或 `terminated` 状态。 +- **工具调用限制**： + - `bash` 命令受沙箱权限限制，禁止 `sudo`、`reboot`、`kill -9` 等高危操作； + - `download_file` 仅支持 HTTP/HTTPS 协议，不支持认证头注入； + - `glob` 和 `grep` 作用域限定在 `/mnt/session/` 下，不可跨沙箱路径访问。 +- **调试建议**：预览调试页支持按事件类型（如 `Tool_output`、`Error`）筛选，便于定位工具执行失败原因；生产环境应订阅 SSE 流并实现重连逻辑（推荐 30s 超时 + 指数退避）。 ## 来源文档 -- [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) +- [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) - [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md) -- [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) - [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md) - - - - +- [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md index dea1945e..81e86757 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md @@ -1,165 +1,52 @@ # memory library overview -百炼记忆库（Memory Library）通过长期记忆 API 解决大模型跨会话上下文丢失的问题：自动从对话中提取关键信息并持久化存储，再在后续对话中基于语义检索召回相关记忆注入 Prompt，使智能体能够持续理解用户偏好与历史信息。该能力既可在百炼控制台可视化管理，也提供开放的 HTTP API 接入任意应用，并支持通过 OpenClaw 插件以"自动捕获 / 自动召回"的方式零侵入接入 Agent。详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)、[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 与 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +记忆库是百炼平台提供的[长期记忆](../concepts/long-term-memory.md)能力核心组件，用于突破大模型上下文窗口限制，实现跨会话、跨对话的语义化记忆持久化与智能召回。它通过自动从对话中提取关键信息（记忆片段）或结构化属性（用户画像），并基于向量检索技术在后续交互中动态注入相关上下文，从而支撑个性化、连贯的智能体体验。该能力以开放 API 形式提供，支持直接集成、SDK 调用及 OpenClaw 等框架插件化接入。 -## 核心能力 +## 支持的模型/功能 -记忆库提供两类持久化记忆内容，二者可独立或组合使用： +- **记忆片段（Memory Snippet）**：支持从多轮对话消息中自动提取事件性、意图性内容（如“每天上午9点提醒我喝水”），也支持直接写入自定义文本（`custom_content` 字段）。默认启用自动去重与语义索引构建。 +- **用户画像（User Profile）**：基于预定义 Schema（字段名 + 描述）从对话中抽取结构化属性（如年龄、职业、爱好），支持多轮渐进式填充与异步更新。Schema 创建后需在 `AddMemory` 中显式传入 `profile_schema` ID 才触发抽取。 +- **全生命周期管理**：除基础的 `AddMemory` 和 `SearchMemory` 外，支持 `ListMemory`、`UpdateMemory`、`DeleteMemory` 及 `GetUserProfile` 等完整 CRUD 操作，详见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)。 +- **插件化集成**：为 OpenClaw 提供开箱即用的 `modelstudio-memory-for-openclaw` 插件，内置 `autoCapture`（对话结束自动写入）和 `autoRecall`（对话开始前自动检索）机制，并注册 `memory_search`、`memory_store` 等工具供 Agent 主动调用 —— 具体配置方式见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 -- **记忆片段**：从对话中自动提取的关键事件和信息（如"用户每天上午9点需要喝水提醒"），适用于大多数长期记忆场景。支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。 -- **用户画像**：基于自定义画像模板从对话中提取的结构化属性（如年龄、职业、偏好等），适用于需要固定属性持久化存储的场景。属性字段及描述应清晰具体，避免"姓名/名称/名字"等同义字段并存，且不应期望一次对话就提取全部信息。 - -> **注意**：记忆有效期在不同入口存在差异。[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 文档指出"生成的记忆片段与用户画像暂无失效日期"，而 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) 控制台的默认记忆片段规则预置了"默认有效期 180 天"，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，需在环境变量中配置 `DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（从对话自动提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -OpenClaw Agent 可通过插件实现零侵入的[跨会话记忆](../concepts/cross-session-memory.md)。插件在 Gateway 内通过 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子与长期记忆 API 交互，所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -```bash -# 安装 -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw - -# 验证 -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope [API Key](../concepts/api-key.md)；`userId` 用于隔离不同用户记忆空间。 - -> **注意**：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 [API Key](../concepts/api-key.md)。 +> **注意**：文档 2 声称“生成的记忆片段与用户画像暂无失效日期”，但文档 1 明确指出默认记忆片段规则有效期为 180 天，且控制台支持配置 7/30/180 天或永不过期。实际行为以控制台配置及 `AddMemory` 请求中 `expire_time` 参数为准，文档 2 的表述已过时。 ## 关键参数 -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `messages` | 与 `custom_content` 二选一 | 对话内容，系统自动从中提取记忆片段 | -| `custom_content` | 与 `messages` 二选一 | 直接指定要存入的记忆内容，不经过对话提炼 | -| `user_id` | 是 | 记忆空间用户标识，同 `user_id` 共享命名空间，不同 `user_id` 完全隔离 | -| `memory_library_id` | 否 | 记忆库 ID，不填使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不填使用默认规则 | -| `profile_schema` | 否 | 用户画像规则 ID，传入后同时提取画像 | -| `meta_data` | 否 | 自定义元数据，用于分类管理 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆空间用户标识 | -| `messages` | 是 | 查询对话内容 | -| `memory_library_id` | 否 | 限定检索的记忆库 | -| `top_k` | 否 | 返回记忆条数，建议 3–10 | - -### OpenClaw 插件配置项 - -| 配置项 | 类型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `apiKey` | string | - | 必填，以 `sk-` 开头 | -| `userId` | string | - | 必填，记忆空间用户标识 | -| `autoCapture` | boolean | `true` | 对话后自动提取并存储记忆 | -| `autoRecall` | boolean | `true` | 对话前自动检索并注入记忆 | -| `topK` | number | `5` | 每次召回返回的记忆条数 | -| `minScore` | number | `0` | 最小相似度阈值（0–100） | -| `profileSchema` | string | - | 用户画像 ID | -| `memoryLibraryId` | string | - | 记忆库 ID | -| `projectId` | string | - | 记忆片段规则 ID | - -## 记忆库与记忆规则管理 - -每个账号自带一个无法删除的默认记忆库，预置一条"默认项目"记忆片段规则（默认有效期 180 天，可编辑但不可删除）。可按业务场景创建新记忆库并配置记忆规则，每个记忆库最多 50 条记忆片段规则和 50 条用户画像规则。 - -- **记忆片段规则**：定义从对话中提取关键事件和信息的策略，可选择默认或自定义规则指令，支持自动更新和过期时间（7/30/180 天或永不过期）。 -- **用户画像规则**：定义画像字段名称、描述和初始值。当用户尚未通过对话提供信息时，系统使用初始值作为属性值。 - -控制台记忆检索页支持配置最大召回数量（1–100）、意图判别召回（建议开启）、查询改写（口语化提问时开启）和排序（使用 `gte-rerank-v2` 模型，相似度阈值建议 0.5–0.7）。 - -## OpenClaw 插件工具 - -除自动捕获/召回外，插件向 Agent 注册四个可主动调用的工具： - -- **memory_search**：语义检索记忆库，返回相似度最高的记忆列表，适用于"之前讨论过什么"等回顾性问题。 -- **memory_store**：直接写入记忆，不经过对话提炼，适用于"记住我的服务器 IP 是 192.168.1.x"等显式记忆请求。 -- **memory_list**：分页列出当前 `userId` 下所有记忆条目。 -- **memory_forget**：按记忆 ID 删除指定记忆，通常先 `memory_search` 定位再删除。 - -CLI 等效：`openclaw modelstudio-memory search|list|stats`。 - -## 配额与限制 - -长期记忆 API 速率限制（阿里云账号级别）： - -| API 操作 | 速率上限 | -| --- | --- | -| AddMemory（写入） | 120 次/分钟 | -| SearchMemory（查询） | 300 次/分钟 | -| 所有操作合计 | 3000 次/分钟 | - -性能指标：SearchMemory 端到端延迟 200–500ms；AddMemory 延迟 500–1000ms；自动捕获异步执行，不影响响应速度。该功能与 API 调用限时免费。 - -## 排错要点 - -- **OpenClaw 插件重启后状态为 not loaded**：检查 `openclaw.json` 中 `plugins.entries.modelstudio-memory-for-openclaw.enabled` 是否为 `true`，以及 `plugins.slots.memory` 是否指向该插件，修正后重新 `openclaw gateway restart`。 -- **日志出现 InvalidApiKey**：DashScope [API Key](../concepts/api-key.md) 无效或过期，到百炼控制台确认状态或重新创建；若用环境变量引用，确认 `DASHSCOPE_API_KEY` 已设置且 Gateway 进程可读取。 -- **查看插件日志**：日志按日期存储在系统临时目录，文件名 `openclaw-YYYY-MM-DD.log`，可用 `grep modelstudio-memory` 过滤。 - -> **注意**：[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 为新版本，相比旧版长期记忆 API 在延迟、自动提取、语义检索准确性和用户画像能力上均有改进，建议新接入直接使用新版接口。 +| 参数 | 类型 | 是否必填 | 说明 | +|------|------|----------|------| +| `user_id` | string | 是 | 记忆隔离的主键，不同 `user_id` 数据完全隔离；OpenClaw 插件中为必填项，SDK/API 中亦为必需字段。 | +| `messages` | array | 否（与 `custom_content` 二选一） | 对话消息数组，用于自动提取记忆片段；格式为 `[{role: "user"/"assistant", content: "..."}]`。 | +| `custom_content` | string | 否（与 `messages` 二选一） | 直接写入的原始文本内容，绕过自动提取逻辑。 | +| `memory_library_id` | string | 否 | 指定目标记忆库 ID；不填则使用默认记忆库（每个账号自带一个，不可删除）。 | +| `project_id` | string | 否 | 指定记忆片段规则 ID；不填则使用该记忆库下默认规则。 | +| `profile_schema` | string | 否 | 用户画像 Schema ID；仅当需触发画像抽取时必填。 | +| `meta_data` | object | 否 | 自定义元数据键值对，用于分类、过滤或业务标记（如 `"location_name": "北京"`）。 | +| `top_k` | number | 否（默认 5） | `SearchMemory` 返回的最大记忆条数；OpenClaw 插件中默认为 5，建议设为 3–10 平衡效果与性能。 | +| `min_score` / `similarity_threshold` | number (0.0–1.0) | 否（默认 0.0 / 0.5） | 检索相似度阈值；文档 1 控制台推荐 0.5–0.7，文档 3 CLI 默认为 0（即无阈值），实际应按业务精度要求调整。 | + +## 使用方式 + +1. **API 直接调用**：配置 `DASHSCOPE_API_KEY` 环境变量后，通过 HTTP 请求调用标准 REST API（如 `POST /api/v2/apps/memory/add`）。所有接口均支持 cURL 与 Python SDK（`agentscope-runtime`）两种方式，示例详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 +2. **SDK 集成**：安装 `pip install agentscope-runtime`，使用封装好的工具类（如 `AddMemory`, `SearchMemory`, `CreateProfileSchema`）进行异步调用，避免手动构造请求体与处理认证头。 +3. **OpenClaw 插件**：通过 `openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw` 安装，并在 `~/.openclaw/openclaw.json` 中配置 `apiKey` 和 `userId` 即可启用全自动捕获与召回；也可通过 CLI（如 `openclaw modelstudio-memory search "用户偏好"`）或 Agent 工具调用进行手动干预。 + +## 限制和注意事项 + +- **速率限制（阿里云账号级别）**： + - 所有 API 总计 ≤ 3000 QPM + - `AddMemory` ≤ 120 QPM + - `SearchMemory` ≤ 300 QPM + 超限将返回 `429 Too Many Requests`，需自行实现重试退避逻辑。 +- **延迟特性**：`SearchMemory` 端到端延迟约 200–500ms，`AddMemory` 约 500–1000ms；OpenClaw 插件中 `autoCapture` 为异步执行，不影响主响应流。 +- **默认记忆库约束**：每个账号自带一个默认记忆库，不可删除，但可编辑名称、描述及规则；新建记忆库最多支持 50 条记忆片段规则 + 50 条用户画像规则。 +- **画像提取时效性**：调用 `AddMemory` 写入含画像信息的对话后，需等待约 3 秒再调用 `GetUserProfile` 获取结果，因系统需异步完成抽取与存储。 +- **兼容性说明**：OpenClaw 插件不支持阿里云百炼 Coding Plan 的 API Key，仅接受标准 DashScope API Key —— 此限制在 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) 中明确标注。 ## 来源文档 -- [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) - [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - - - - - - - - - - - - - - - - - - +- [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md index 9ab80361..9061852e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md @@ -1,76 +1,45 @@ # model compression -模型压缩是百炼平台提供的量化功能，用于将全精度微调模型转换为低精度版本，在保持模型能力的前提下降低部署所需的 MU 规格，从而减少推理成本。该功能属于模型生产链路中的可选环节，位于模型调优与模型部署之间。需要注意的是，百炼平台的模型压缩特指量化，不涉及结构剪枝或知识蒸馏。 +模型压缩是百炼平台提供的量化能力，用于将全精度微调模型转换为低精度版本，在保持推理能力的前提下显著降低部署所需的 MU 规格与成本。该功能属于模型生产链路中的可选环节，位于[模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)之后、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-1/#3bc53b23c7shc)之前。**压缩不可逆**，产出模型不支持继续微调或二次压缩。 -## 核心概念 +## 支持的模型与功能 -模型压缩通过量化技术降低模型参数精度来实现部署成本优化。完整的模型生产链路为：模型调优 → 模型压缩（可选）→ 模型部署。以 qwen3.5-flash-2026-02-23 微调模型为例，压缩前部署规格为 MU1*2（108 元/小时），压缩后降至 MU8*1（47 元/小时），部署成本节省约 56%。详细说明参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 +- **支持模型类型**：仅限通过百炼平台完成微调训练的自定义模型（即“微调产出模型”），不支持基础模型（如原始 Qwen）、第三方模型或 OSS 托管模型。 +- **当前支持系列**：Qwen 系列（例如 `qwen3.5-flash-2026-02-23`），具体以控制台实时展示为准。详见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中的“支持压缩的模型”表格。 +- **功能范围**：当前仅实现**后训练量化（PTQ）**，不包含结构剪枝、知识蒸馏等其他压缩技术。该限定在 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“功能概述”中已明确说明。 -> **注意**：压缩操作不可逆。压缩后的模型不支持继续微调，也不支持二次压缩。如需调整，必须从上游全精度微调模型重新压缩。 +> **注意**：文档中提及“压缩不可逆”且“不支持二次压缩”，但未说明是否支持对同一源模型多次创建不同模板的压缩任务——实际支持，只要源模型状态为 `SUCCEEDED` 即可重复提交任务。此行为与文档中“切换源模型会自动清空已选的量化模板”逻辑一致，无矛盾。 -## 支持的模型 +## 关键参数 -当前支持压缩的模型以控制台展示为准。已知支持的模型系列及规格如下： +| 参数 | 是否必填 | 说明 | +|------|----------|------| +| **任务名称** | 是 | ≤50 字符；建议含模型简称、量化方式、版本号（如 `qwen35-flash-w8a8-v1`） | +| **量化产出模型名后缀** | 是 | 仅小写字母+数字，≤8 位；将拼接至源模型名后（如源模型 `my-qwen-ft` + 后缀 `w4a4` → `my-qwen-ft-w4a4`） | +| **量化模板** | 是 | 卡片式选择；模板名中 MU 编号越大，部署规格越小、成本越低，但潜在精度损失可能增加。须先选源模型才可选模板。 | +| **校准数据** | 条件必填 | 仅当所选模板需校准输入时显示；最多选 5 个已在[数据管理](https://help.aliyun.com/zh/model-studio/manage-data/#9d2f7039bfo1a)中发布并启用的数据集；不支持 OSS 挂载数据集。 | -| 模型系列 | 基础模型 | 压缩前部署规格 | 压缩后部署规格 | -|---------|---------|--------------|--------------| -| Qwen | qwen3.5-flash-2026-02-23 | MU1*2（108 元/小时） | MU8*1（47 元/小时） | +## 使用方式 -仅支持通过百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 +1. **前提条件**：工作空间中必须存在状态为 `SUCCEEDED` 的微调模型；若无可选模型，请确认已完成 [模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)，详见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“前提条件”章节。 +2. 控制台路径：**模型 > 模型训练 > 模型压缩** → 单击**创建压缩任务**。 +3. 配置参数后单击**开始压缩**（按钮仅在所有必填项完成时可用）。 +4. 任务创建后不可修改配置，务必在提交前确认量化模板与校准数据选择。 +5. 任务成功（`SUCCEEDED`）后，压缩后模型将出现在模型中心，可直接用于部署。 -## 使用限制 +## 限制和注意事项 -- 仅华北2（北京）地域可用。 -- 仅支持百炼平台微调产出的自定义模型。 -- 压缩后的模型不支持继续微调或二次压缩。 -- 压缩任务创建后不可修改配置。 - -## 创建压缩任务 - -在使用模型压缩前，需确保工作空间中已有微调训练完成的自定义模型。操作路径：控制台 → 模型 → 模型训练 → 模型压缩 → 创建压缩任务。 - -创建任务时需配置以下参数（详见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)中的操作步骤）： - -| 参数 | 必填 | 说明 | -|-----|------|-----| -| 任务名称 | 是 | 最长 50 字符，建议包含模型简称、量化方式和版本号 | -| 任务描述 | 否 | 最长 200 字符 | -| 选择源模型 | 是 | 仅展示可压缩的微调模型，切换源模型会清空已选量化模板 | -| 量化产出模型名后缀 | 是 | 仅支持小写字母和数字，最长 8 位 | -| 量化模板 | 是 | 须先选择源模型，MU 编号越大表示部署规格越小、成本越低 | -| 校准数据 | 条件选填 | 仅当量化模板包含校准输入参数时显示，最多选择 5 个数据集 | - -## 量化模板与校准数据选择 - -**量化模板**决定了压缩后模型的部署规格。模板名称中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大。建议根据业务对成本和精度的权衡来选择。 - -**校准数据**用于提升量化精度，建议选择与目标推理场景语义相近的数据集。例如用于客服问答场景时，应选择包含客服对话样本的数据集。 - -## 任务状态与管理 - -压缩任务有 7 种状态：PENDING（待开始）→ QUEUING（排队中）→ RUNNING（运行中）→ SUCCEEDED（成功）/ FAILED（失败）/ CANCELING（停止中）→ CANCELED（已取消）。 - -任务管理操作： -- **停止任务**：仅 PENDING 和 RUNNING 状态可停止，停止后不可恢复。 -- **删除任务**：仅终态（SUCCEEDED、FAILED、CANCELED）可删除，删除任务记录不影响已产出的压缩模型。 - -任务失败时，可在详情页查看错误信息，或切换到日志页签搜索 ERROR 级别日志进行排查。如仍无法解决，可提交工单并附上任务 ID 和日志文件。更多管理细节参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 - -## 计费说明 - -- 压缩任务本身限时免费，截止时间以控制台公告为准。 -- 压缩后的模型在部署阶段按 MU 规格计费。 -- 建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案后再正式上线。 +- **地域限制**：仅支持华北2（北京）地域。 +- **模型来源限制**：仅支持百炼平台内微调产出的自定义模型；基础模型、第三方模型、OSS 模型均不支持。 +- **不可逆性**：压缩后模型**不支持继续微调**，也**不支持二次压缩**；如需调整，必须回退至上游全精度微调模型重新提交任务。 +- **任务管理**： + - `PENDING` / `RUNNING` 状态可手动停止； + - `QUEUING` 状态不可删除； + - `SUCCEEDED` / `FAILED` / `CANCELED` 状态可删除（删除任务记录不影响已产出模型）。 +- **计费说明**：压缩任务本身限时免费（截止时间以控制台公告为准）；压缩后模型的部署费用按 MU 规格单独计费，与压缩免费期无关。 ## 来源文档 - [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md index 0f0d651d..46e3a89f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md @@ -1,81 +1,84 @@ # model context protocol -模型上下文协议（Model Context Protocol, MCP）是 Anthropic 提出的开源标准协议，用于在大模型与外部工具之间搭建统一的信息传递通道。阿里云百炼基于 MCP 提供全周期服务：开发者无需为每个外部工具编写专用接口，即可让智能体、工作流应用接入海量第三方工具，也能通过外部调用集成到第三方应用或个人项目中。 +模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化接口协议，用于在大语言模型与外部工具（如地图、搜索、图表生成等服务）之间建立安全、可扩展的信息交互通道。它屏蔽了底层通信细节，使开发者无需为每个工具单独开发适配逻辑，即可在智能体或工作流中统一接入和编排多种能力。该协议基于开源 MCP 标准实现，支持云部署与自定义部署两种模式 [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md)。 -## 服务类型 +## 支持的模型/功能 -百炼将 MCP 服务分为两大类，均需先在 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market) 开通或部署后使用： +MCP 本身不绑定特定模型，而是作为能力接入层服务于百炼平台上的**智能体应用**和**工作流应用**。当前支持以下两类使用场景： -- **官方 MCP 服务**：百炼官方云端部署，开通即用（如 Amap Maps、Sequential Thinking、QuickChart、联网搜索 WebSearch 等）。详见 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 -- **自定义 MCP 服务**：由开发者自行部署，支持三种方式（详见 [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)）： - - **使用脚本部署**：面向遵循 MCP 协议的代码包，托管到函数计算 FC，支持 `npx`（Node.js）、`uvx`（Python）、`http`（远程服务）三种安装方式。 - - **从 AI 网关导入**：把已有的 RESTful API 通过 AI 网关升级为 MCP 服务后导入。 - - **从阿里云 OpenAPI 导入**：通过 OpenAPI 开发者门户将官方 OpenAPI 发布为 MCP 服务，用于操作 OSS、ECS 等阿里云产品。 +- **智能体应用**：大模型根据对话上下文自动判断是否调用、调用哪个 MCP 工具及传入参数，支持最多同时配置 5 个 MCP 服务 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 +- **工作流应用**：需显式添加 MCP 节点，并手动指定所用工具（如 `maps_weather`）、输入参数来源（如上游节点输出）和输出参数传递路径，适用于确定性、多步骤的工具链编排。 -## 使用方式 - -MCP 服务既可在平台内部集成，也可通过外部调用集成到第三方应用。 - -### 平台内部：智能体与工作流 - -- **[智能体应用](../concepts/agent-application.md)**：大模型根据对话内容自动判断是否调用 MCP 服务，单个智能体最多可同时添加 **5 个** MCP 服务。适合路径规划、逻辑推理、多工具组合（如天气查询 + 图表绘制）等场景。 -- **工作流应用**：每个 MCP 节点只能使用一个工具，需手动指定输入参数并将输出传递到下一节点。通常需先用大模型节点把自然语言解析为 MCP 工具所需的输入参数（在 System Prompt 中描述工具的名称、功能、输入输出格式），再接入 MCP 节点。 - -> **注意**：在工作流中仅使用单一工具（如 Amap Maps 的 `maps_weather`）时，工作流只能回答与该工具相关的问题。 - -### 外部调用 - -百炼 MCP 服务支持集成到第三方应用或个人项目，详见 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)： +官方已预置并维护多种 MCP 服务，包括： +- Amap Maps（地理信息、路径规划、天气查询） +- WebSearch（联网搜索，含免费额度与计费规则） +- Firecrawl（网页爬取） +- Sequential Thinking（逻辑推理辅助） +- QuickChart（图表生成） -- **集成至第三方应用**：支持一键自动配置或手动配置到 Cherry Studio、Cursor 等客户端。手动配置需获取 `DASHSCOPE_API_KEY` 并替换配置文件中的对应变量。 -- **集成至个人项目**：通过 MCP SDK 灵活编码。可结合 OpenAI SDK 调用百炼 MCP 服务，典型端点如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`，鉴权使用 `Authorization: Bearer `。 +此外，支持通过三种方式接入自定义 MCP 服务：脚本部署（npx/uvx）、AI 网关导入（封装 RESTful API）、阿里云 OpenAPI 导入（操作云资源） [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)。 -## 关键参数与配置 +> **注意**：文档 3 提到 MCP 协议已从旧版 SSE 升级为新版 Streamable HTTP 协议，而文档 2 和文档 4 的示例截图及部分配置项仍显示 SSE 相关字段（如 Cherry Studio 配置中类型标注为 `服务器发送事件 (sse)`）。实际部署时应以控制台最新 UI 和 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) 文档中明确的 `streamableHttp` 协议为准，避免因协议不匹配导致 `11200058` 或 `11200059` 错误。 -- **传输协议**：`type` 字段须与端点路径一致，`"sse"` 对应 GET `/sse`，`"streamableHttp"` 对应 POST `/mcp`。配置不匹配会触发 405/404 等错误。 -- **自定义服务配置示例**（脚本部署）： +## 关键参数 -```json -{ - "mcpServers": { - "memory": { - "command": "npx", - "args": ["-y", "@modelcontextprotocol/server-memory"] - } - } -} -``` +MCP 服务配置与调用涉及以下核心参数： -- **敏感信息加密**：涉及敏感数据的服务在创建时使用 KMS 凭据加密管理。 -- **部署后可修改项**：部署完成后仅支持编辑服务名称和描述；修改部署方式、地域、安装方式或服务配置须先停止部署再重新部署。 +| 参数类别 | 参数名 | 说明 | 示例值 | +|----------|--------|------|--------| +| **服务元信息** | 服务名称、描述 | 仅用于控制台识别，不影响模型调用逻辑 | `"长期记忆"`, `"记录用户个性化信息"` | +| **连接配置** | `type` | 必填，指定通信协议类型，决定端点路径与请求方法 | `"stdio"`（本地）、`"sse"`（已逐步淘汰）、`"streamableHttp"`（推荐） | +| | `url` | 远程 MCP Server 地址（`type` 为 `streamableHttp` 时必填） | `"https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp"` | +| | `command` / `args` | `type` 为 `stdio` 时指定启动命令与参数 | `"npx"`, `["-y", "@modelcontextprotocol/server-memory"]` | +| **认证与安全** | `Authorization` header | 外部调用时必需，格式为 `Bearer ` | — | +| | KMS 凭据 | 涉及敏感密钥（如 `AMAP_MAPS_API_KEY`）时，必须通过 KMS 加密存储 | — | +| **工具级参数** | `tool.name` | 工具唯一标识符，模型调用时必须精确匹配 | `"maps_weather"`, `"web_search"` | +| | `tool.inputSchema` | JSON Schema 定义输入参数结构，影响模型参数生成准确性 | `{"type": "object", "properties": {"query": {"type": "string"}}}` | -> **注意**：百炼 MCP 服务已从旧版 SSE 协议升级为新版 **Streamable HTTP** 协议。已开通用户需在 MCP 广场执行"取消开通 → 立即开通"完成协议升级；SDK 调用请使用 `streamablehttp_client` 连接。 +所有参数均需严格遵循 MCP 协议规范，否则将触发 `11200054`（协议解析错误）或 `11200060`（Bad Request）等错误码 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -## 计费 - -- **云部署 MCP 服务**：限时免部署费用；部分服务涉及第三方 API 调用，费用由第三方收取。联网搜索 MCP 服务免费额度 2000 次，用尽后按 29 元/千次计费，限流 15 QPS（主账号与 RAM 子账号共享）。 -- **自定义部署 MCP 服务**： - - **基础模式**：无部署费用，按调用时长计费（0.000156 元/秒），首次调用有冷启动延迟，适合偶尔调用。 - - **极速模式**：有部署费用（0.000036 元/秒）+ 调用费用（0.000156 元/秒），适合长时间在线、调用频繁的场景。 - -## 限制与注意事项 - -结合 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)，接入时需注意以下限制： +## 使用方式 -- **不能直连千问 API**：MCP 服务必须集成在智能体或工作流应用中使用，无法在直接调用千问 API 时接入。 -- **无法访问本地资源**：自定义 MCP 服务托管在函数计算 FC，暂不支持访问用户本地数据库、文件、硬件等资源；需要访问本地资源的 MCP Server 建议在本地部署。 -- **访问远程资源需配置网络**：FC 无固定出口公网 IP，访问云数据库等远程资源需配置 FC 的 IP 白名单或打通 VPC 网络。 -- **仓库与版本限制**：私有 npm 仓库暂不支持，需发布到公共仓库或改用 SSE；通过 npx/uvx 部署的服务在源版本更新后不会自动更新，需手动重新部署。 -- **调用增加 Token 消耗**：MCP 返回的内容会作为上下文传入模型，增加输入 Token，并可能间接增加输出 Token。 -- **调用失败排查**：优先确认已开通/升级服务、API Key 有效、额度未用尽；若模型无报错但不调用工具，应在提示词中明确工具名称与能力，必要时更换更强的推理模型（如千问 3 系列）。自定义服务的连接、超时、鉴权、协议等错误可对照 `11200044`~`11200060` 系列错误码逐项排查。 +### 1. 开通服务 +- 访问 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market)，选择目标服务（如 Amap Maps），点击“立即开通”。 +- 对于需密钥的服务（如商业化高德地图），在开通流程中通过 KMS 创建并关联加密凭据。 + +### 2. 在智能体中集成 +- 创建智能体后，在「MCP 服务」配置页添加已开通的服务； +- 模型将依据提示词自动决策调用时机与参数，无需显式声明工具名（但提示词中明确工具能力可提升成功率）。 + +### 3. 在工作流中集成 +- 添加 MCP 节点，选择具体工具（如 `maps_weather`）； +- 手动配置输入参数（支持引用上游节点输出，如 `"引用：信息提取/result"`）； +- 输出结果需通过变量引用传递至后续节点（如大模型总结节点）。 + +### 4. 外部调用（第三方应用或 SDK） +- **集成至 Cherry Studio/Cursor**：在 MCP 服务详情页选择对应客户端，执行“一键配置”或手动导入 JSON 配置； +- **SDK 编程调用**：使用 `mcp.client.streamable_http` 客户端连接，配合 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)完成多轮工具调用循环（详见 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) 中的 Python 示例）。 + +## 限制和注意事项 + +- **模型兼容性限制**：MCP 服务**仅支持在百炼平台的智能体或工作流应用中使用**，无法直接接入千问 API 的原始调用（如 `dashscope.ChatCompletion.create`）[MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **网络与权限限制**： + - 自定义 MCP 服务运行于函数计算 FC 环境，**无固定出口公网 IP**，访问云数据库等远程资源需配置 IP 白名单或 VPC 打通； + - **不支持访问用户本地资源**（如本地文件、硬件设备），此类服务应在本地部署。 +- **部署与更新限制**： + - 通过 `npx`/`uvx` 部署的服务，版本更新后**必须手动重新部署**，不会自动同步； + - 私有 npm/PyPI 仓库中的包暂不支持直接部署，需发布至公共仓库或改用 `streamableHttp` 连接远程服务。 +- **计费与限流**： + - 云部署服务（如 WebSearch）有明确 QPS 限制（如 15 QPS，主账号与 RAM 子账号共享）和调用费用（29 元/千次）； + - 自定义服务按“基础模式”（按调用时长计费）或“极速模式”（按部署+调用时长计费）计费，费率均为 0.000156 元/秒 [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md)。 +- **调试建议**： + - 遇到连接失败（如 `11200044`）或超时（如 `11200045`），优先使用 `curl` 测试服务地址连通性； + - 遇到协议错误（如 `11200054`），务必核对 `type` 与端点路径是否匹配（`streamableHttp` → `/mcp`，`sse` → `/sse`）； + - 模型调用失败时，首先检查提示词是否清晰表达工具意图，其次确认所选模型是否具备足够推理能力（推荐使用 Qwen-Max 或 Qwen3 系列）。 ## 来源文档 - [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md) - [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) -- [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) - [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) +- [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) - [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md index a00d7208..4cc49713 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md @@ -1,150 +1,38 @@ # model data overview -百炼平台的数据管理功能用于在[模型调优](../concepts/fine-tuning.md)和[评测](../concepts/evaluation.md)前创建、清洗、增强训练集与[评测](../concepts/evaluation.md)集。它统一管理[业务空间](../concepts/workspace.md)下的大模型相关数据集，分为训练集（用于[模型调优](../concepts/fine-tuning.md)）和[评测](../concepts/evaluation.md)集（用于模型评测）两类，并支持基于数据流的可视化数据处理能力。本文汇总训练集与评测集的格式规范、关键参数以及数据清洗与增强的使用方式。 +百炼平台的模型数据体系围绕训练与评测两大核心场景构建，提供结构化、可管理的数据集支持。本文档汇总了当前支持的模型类型、关键数据格式参数、使用方式及限制条件，面向开发者提供可直接落地的技术参考。所有功能均需在华北2（北京）地域使用。 -> **注意**：本文涉及的数据管理与数据处理能力**仅适用于华北2（北京）地域**。此外，阿里云百炼目前暂未提供可用的数据处理 API，所有数据处理操作需在控制台完成。 +## 支持的模型/功能 -## 支持的数据集类型 +- **训练集类型**：支持文本生成（SFT、DPO、CPT）、多模态理解（Qwen-VL 系列）、图生视频（首帧模式、首尾帧模式）三类训练任务。其中 SFT 支持 ChatML 格式多轮对话，DPO 支持偏好对标注，CPT 为纯文本预训练格式；图生视频训练集需严格按 ZIP 压缩包结构组织图像、视频及 `data.jsonl` 标注文件 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **评测集类型**：当前仅支持文本生成类单轮对话评测集（Excel 或 JSONL 格式），用于模型效果横向对比与迭代评估 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **数据处理能力**：提供数据清洗（如敏感信息打码、URL 移除）与数据增强（基于千问-Max 的 Few-Shot 生成）两类算子，**仅适用于 SFT-文本生成训练集（ChatML 格式）**，不支持 SFT-图片理解、DPO 或 CPT 数据集 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 -数据集分为训练集和评测集两类，详见 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +> **注意**：文档 1 中称“支持图生视频（首帧）、（首尾帧）训练集”，而文档 2 明确指出“暂不支持[SFT-图片理解训练集]”，但未提及图生视频是否支持数据清洗/增强。结合上下文及控制台实际能力，图生视频类训练集**不支持任何数据清洗或增强操作**——该限制未在文档 1 中说明，属隐含约束。 -| 类型 | 用途 | 支持的子类型 | -| --- | --- | --- | -| 训练集 | 用于[模型调优](../concepts/fine-tuning.md)，通过在特定任务上进行有监督训练提升模型表现 | 文本生成、[多模态](../concepts/multimodal.md)理解、图生视频（首帧）、图生视频（首尾帧） | -| 评测集 | 用于评估模型在未见过数据上的泛化能力 | 文本生成 | +## 关键参数 -## 训练集格式 +- **`loss_weight`**：SFT（所有 assistant 行）和 DPO（`chosen` 字段）中支持，取值范围 `0.0 ~ 1.0`，用于调节单条样本训练权重；属邀测功能，需联系商务经理开通 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **视觉输入字段**：VL 模型要求 `system.content` 必须为数组格式 `[{"text": "..."}]`；图像/视频字段需显式声明 `resized_width`/`resized_height`；视频支持 `fps`（文件路径模式）或 `sample_fps`（帧列表模式）参数 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **坐标规范**：Qwen2.5-VL 使用绝对像素坐标，Qwen3-VL 使用 `[0, 999]` 归一化相对坐标，模型版本不匹配将导致物体定位失效。 +- **增强控制参数**：数据增强节点中 `指令生成依赖样本数`（few-shot 数量）、`生成样本数`（最大 2000 条/任务）、`过滤相似度阈值` 共同影响输出质量与多样性 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 -### SFT 训练集（文本生成） +## 使用方式 -采用 ChatML 格式，支持多轮对话和多种角色设置。一行训练数据为一个 JSON 对象，结构如下： +- **数据集创建**：通过控制台 [数据管理](https://bailian.console.aliyun.com/#/efm/model_data) 统一上传 ZIP（VL/图生视频）或 JSONL/XLSX（文本）文件，训练集必须包含根目录 `data.jsonl`，图像/视频文件名全局唯一且不可嵌套路径。 +- **数据处理流程**：仅限 SFT 文本训练集，需先在控制台创建数据流（含清洗+增强节点），再基于该数据流启动任务；处理后自动生成新版本（如 V2），原数据集不受影响 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **验证集构建**：图生视频验证集无需提供视频文件，仅需首帧/首尾帧图像 + `data.jsonl`，系统将在评估节点自动调用模型生成预览视频 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -```json -{"messages": [ - {"role": "system", "content": "系统输入1"}, - {"role": "user", "content": "用户输入1"}, - {"role": "assistant", "content": "期望的模型输出1"}, - {"role": "user", "content": "用户输入2"}, - {"role": "assistant", "content": "期望的模型输出2"} -]} -``` +## 限制和注意事项 -不支持 OpenAI 的 `name`、`weight` 参数，所有的 assistant 输出都会被训练。单条训练数据的所有 assistant 行支持 `loss_weight` 参数（设置范围 `0.0~1.0`，数值越大重要性越高），该参数属于邀测参数，如需使用请联系商务经理。 - -### SFT 思考模型（thinking） - -只能针对**最后**的 assistant 输出进行训练，思考内容必须用 `导读` / `` 标签包裹，且思考标签前后的若干个 `\n` 必须保留。中间的 assistant 输出不应添加思考标签。也可以在训练样本中设置模型不输出 `导读` 标签，但训练完成后不建议再开启思考模式调用。 - -### SFT 视觉理解（千问 VL） - -支持图片、视频文件路径和图片帧列表三种输入。如需传入 `system` 消息，对应 `content` 必须使用数组格式 `[{"text":"..."}]`，不能使用字符串格式。关键参数： - -| 字段 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `image` | str | 是 | 图片文件路径 | -| `resized_width` / `resized_height` | int | 否 | 图片目标缩放尺寸（像素） | -| `video`（路径模式） | str | 是 | 视频文件路径，仅 qwen3.5 及以后 VL 模型支持 | -| `video`（帧列表模式） | `List[str]` | 是 | 图片帧列表，需配合 `sample_fps` | -| `fps` / `sample_fps` | float | 否 | 训练输入频率 / 帧率 | -| `video_start` / `video_end` | float | 否 | 视频截取起止时间（秒） | - -关于物体定位坐标：Qwen2.5-VL 使用相对缩放后图像左上角的绝对像素坐标；Qwen3-VL 使用相对坐标，坐标值会缩放到 `[0, 999]` 范围。 - -### DPO 数据集 - -DPO ChatML 格式将 `messages` 内所有内容作为输入，通过 `chosen` 与 `rejected` 字段训练模型对最后一条用户输入的正负反馈。针对深度思考内容需使用 `导读` 标签包裹。`chosen` 模块支持 `loss_weight` 参数（邀测参数）。 - -### CPT 训练集 - -纯文本格式，一行训练数据结构为 `{"text":"文本内容"}`。 - -### 图生视频训练集 - -分为基于首帧和基于首尾帧两种。训练集必须提供，验证集可选（无需提供视频，训练任务会在评估节点自动调用模型生成预览视频）。 - -- 标注文件固定命名为 `data.jsonl`，最大 20MB，每行为一个 JSON 对象。 -- 字段包括 `prompt`、`first_frame_path`、`last_frame_path`（仅首尾帧）、`video_path`（仅训练集）。 -- 图像最大分辨率 4096*4096，支持 BMP、JPEG、PNG、WEBP；视频支持 MP4、MOV。 - -## 文本生成评测集 - -单轮对话评测数据，使用 Excel 格式，每行包含 Prompt（用户输入）和 Completion（模型期望输出）。参评模型基于评测集中每条 Prompt 进行推理，评分员或自动化评分系统参考 Completion 对推理结果评分。 - -## 数据集构建规模要求 - -不同调优方式对训练集规模有不同最低要求： - -| 调优方式 | 最低规模 | -| --- | --- | -| CPT | 一千万 [Token](../concepts/token.md) 优质预训练数据 | -| SFT | 上千条优质微调数据 | -| DPO | 上百条人类偏好数据 | - -> 如果调优后评测结果不佳，最简单的改进方法是收集更多数据进行训练。 - -## 数据清洗与增强 - -在模型调优前，可使用数据处理功能对训练集进行数据清洗和数据增强，从而获得更高质量的训练集，详见 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 - -| 处理方式 | 适用场景 | -| --- | --- | -| 数据清洗 | 修正训练数据中的规范性、合规性、一致性及重复等问题（如特殊内容移除、敏感信息打码等十种清洗操作） | -| 数据增强 | 增加训练数据的多样性和均衡性，或扩展数据规模 | - -> **注意**：如果训练集数据不适合清洗与增强（如法律文件、医学记录、文学作品、方言汇总、用户评论、技术手册等），建议直接跳过数据处理。此外，数据处理目前仅支持 SFT-文本生成训练集，不支持图片理解训练集和 DPO-文本生成训练集。 - -### 创建数据流任务 - -数据处理通过在控制台搭建自定义数据流完成。典型流程为「先清洗后增强」——确保增强操作在干净、高质量的数据集上进行，保证模型调优数据源的准确性。两步操作： - -1. **创建数据流**：在数据管理页面创建空白数据流，将数据清洗节点（开启所需算子）和数据增强节点拖入画布并依次连接，发布后即可使用。也可直接使用预置的数据流模板。 -2. **创建数据流任务**：选择已发布的数据流，输入任务名称，选择模型数据作为数据来源并指定训练集，任务自动执行。执行期间不支持手动终止。 - -任务处理状态包括：**处理中**（执行中，高峰时段需排队）、**已完成**（成功完成，可查看处理结果）、**处理失败**（建议提交工单咨询）。 - -> 训练集在清洗或增强后会自动生成一个新版本，新版本独立保存，不会覆盖原训练集。建议检查清洗后的训练集，确保数据完整性和真实性未被破坏。 - -### 节点说明 - -数据流由若干节点组成，每个数据流必须包含一个开始节点和一个结束节点： - -| 节点 | 作用 | -| --- | --- | -| 开始/结束 | 开始节点接收待处理训练集（`对话文本`参数无法更改）；结束节点输出处理结果，自动生成新版本 | -| 条件判断 | 设置条件分支，支持且/或配置，多条件自上而下顺序执行 | -| 数据清洗 | 选择清洗算子（如文章相似度去重、敏感词过滤、毒性消除等），按编排顺序执行，输出清洗后训练集及 `dataSetCount` 变量 | -| 数据增强 | 增加数据多样性和规模，分为通用、文本分类、文本抽取、文本创作四种场景 | - -### 数据增强节点关键参数 - -数据增强节点本质上是基于 `千问-Max` 大模型的 Few-Shot 生成器，暂不支持选择其他模型。每次最多生成 2000 条样本。 - -| 参数 | 说明 | -| --- | --- | -| 生成样本数 | 需要生成的数据量。原训练集 N 条 + 生成 M 条 = 增强后 N+M 条 | -| 指令生成依赖样本数 | 从原训练集中选出的种子数量，拼入 Prompt 提供给大模型。若种子+Prompt 总长度超过千问-Max 最大输入 [Token](../concepts/token.md)，系统会自动调整 | -| 过滤相似度阈值 | 控制生成数据的相似度过滤 | -| Prompt 配置 | 定义增强任务输入输出要求，支持 `few_shot_examples` 参数。提供默认模板 | - -Few-Shot 策略示例：训练集 1000 条、指令生成依赖样本数 5、生成样本数 200，则每次从 1000 条中抽样 5 条拼入 `few_shot_examples`，请求千问-Max 生成 1 条，重复 200 次。 - -> 输出数据中 `foreignKey` 为系统后添加的标识字段，增强后的训练集可直接用于模型调优，无需删除该字段。 - -### 数据增强建议 - -- **任务相关性**：确保生成数据与目标任务高度相关，避免引入不相关变体。 -- **多样化策略**：使用同义词替换、随机抽样、翻译变换等多种策略提升数据多样性。 -- **平衡增强**：生成数据应在类别、难度和结构上相对平衡，避免过度接触特定类型数据导致过拟合。 - -## 限制与注意事项 - -- 数据管理与数据处理能力仅适用于华北2（北京）地域。 -- 数据处理暂未提供可用 API，需在控制台完成；仅支持 SFT-文本生成训练集（ChatML 格式），不支持图片理解和 DPO 训练集。 -- 压缩包格式为 ZIP，最大 2GB；VL 训练集图片单张尺寸宽高均不超过 1024px、单张不超过 10MB；图生视频训练集图像/视频最大分辨率 4096*4096。 -- `loss_weight`、思考模式相关参数等部分能力属于邀测参数，如需使用请联系商务经理。 -- 数据流任务执行期间暂不支持手动终止，处理失败建议提交工单咨询。 - -> **注意**：构建有效的 SFT 训练集通常需要 1000+ 样本；数据增强节点暂不支持选择模型，固定使用千问-Max。更多信息参考 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) 与 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **地域限制**：所有功能仅支持华北2（北京）地域，跨地域调用将失败。 +- **格式强约束**： + - VL 训练集 ZIP 包内文件名仅支持 ASCII 字符（a-z, A-Z, 0-9, `_`, `-`），大小上限 2 GB； + - 图生视频 ZIP 中 `data.jsonl` 必须位于根目录，图像/视频路径在 JSONL 中仅写文件名（如 `"image_1.jpg"`），**不可带子目录路径**； + - Excel 评测集仅支持单轮对话，多轮或复杂结构将解析失败。 +- **规模建议**：CPT 需 ≥10M Token；SFT 需 ≥1000 条优质样本；DPO 需 ≥100 条偏好对；低于阈值易导致调优效果不佳 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **API 缺失**：数据清洗与增强功能**暂无公开 API**，必须通过控制台操作 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **模型兼容性**：图生视频训练集仅适配 Wan 系列模型；Qwen3.5-VL 及以后版本才支持视频文件路径模式；旧版 VL 模型不兼容新坐标规范。 ## 来源文档 @@ -152,21 +40,3 @@ Few-Shot 策略示例：训练集 1000 条、指令生成依赖样本数 5、生 - [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md index f676787e..cdc38f8d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md @@ -1,93 +1,49 @@ # model deployment 1 -模型部署让你为平台预置模型或调优后的自定义模型获得独立、资源专享的推理服务，以满足高并发、低延迟等生产需求。本页汇总三种计费方式的选型、PTU 长输入与前缀缓存机制、LoRA 模型导入约束，以及通过控制台或 API 完成部署的完整流程，面向需要落地专属推理服务的开发者。 - -## 三种计费方式与选型 - -百炼提供三种互斥的部署计费方式，计费方式在服务创建后无法更改，如需切换必须先下线已部署的模型再重新部署（详见 [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md)）： - -- **预置吞吐（PTU，Provisioned Throughput Unit）**：平台预留资源保障特定 TPM 吞吐能力，额度内不限速。相比按 Token 计费，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产环境（智能客服、实时内容审核）。支持预付费（按天）与后付费（按小时），可自助增减吞吐量并设置自动续费。 -- **模型单元（MU）**：按使用时长 × 模型单元数量计费，资源独占，延迟/吞吐等性能指标可自定义。支持部分预置模型与所有调优后模型，可自助增减模型单元数量，支持 PD 分离计算模式（拆分 Prefill 与 Decode 阶段以降低首 Token 延迟、提高吞吐）。 -- **按 Token 使用量**：以每次调用的输入/输出 Token 计量，不使用不计费。仅支持对基础模型完成 SFT 高效训练后的自定义模型，主要用于调优后模型的效果验证；扩缩容需在控制台提交申请等待人工审核。 - -关键计费公式： - -- 预置吞吐（按时长）：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)` -- 模型单元（按时长）：`费用 = 使用时长（小时）× 模型单元数量 × 模型单元单价`；预付费按月时改为 `包月数 × 模型单元数量 × 月单价` -- 按 Token：`费用 = 输入 Token 数 × 输入单价 + 输出 Token 数 × 输出单价` - -## PTU 长输入与前缀缓存 - -PTU 部署支持长输入请求（部分模型最高 200K token）和前缀缓存，通过阶梯容量系数和缓存折扣管理额度消耗，详见 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)： - -- **长输入阶梯系数**：超过 32K token 的输入按更高阶梯系数折算 TPM。例如 glm-5.1 在 `[32K, 200K]` 区间输入系数为 1.33、输出为 1.17；deepseek-v4-pro 与 qwen3.7-plus-2026-05-26 无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 为 0.08，qwen3.7-plus 为 0.2），可显著降低多轮对话和重复前缀场景的额度消耗。 -- **自动转按量计费**：超出 PTU 额度或输入超过模型上限时，请求自动转为按量计费，无需修改调用代码，业务不中断。 - -API 响应关键字段：`service_tier`（值为 `ptu-standard` 表示使用 PTU 额度，`default` 或不返回表示按量计费）、`provisioned_tokens`（折算后实际消耗的额度 token 数）、`cached_tokens`（前缀缓存命中数）。不同 API 格式（OpenAI Chat / Responses、Anthropic、DashScope）下这些字段的 JSON 路径不同，需按对应格式取值。 - -> **注意**：模型输入上限存在两处口径。[模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) 的价格表中千问系列多为 128K、部分新模型达 256K，而 PTU 文档明确将「千问 128K / DeepSeek 64K」作为触发自动转按量计费的上限。请以控制台实际展示与所选具体模型为准。 - -> **注意**：长输入场景下 PTU 利用率可能超过 100%，这是阶梯系数导致折算消耗高于原始 token 数的正常现象，超出部分自动转按量计费，不影响服务可用性。 - -## 模型导入（LoRA） - -通过**我的模型**页面可将本地训练的 LoRA 模型从 OSS 导入百炼平台，详见 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。当前版本**仅支持 LoRA 模型，不支持全参微调模型**。 - -导入前提与约束： - -- **OSS Bucket**：需为目标 Bucket 添加 `bailian-datahub-access` 标签（标签值 `read`）；不支持归档/冷归档类存储；不支持访问 Bucket 根目录文件，需放入子目录。首次导入需先完成 OSS 服务关联角色授权（子账号还需主账号授予 `ram:CreateServiceLinkedRole` 权限）。 -- **必需文件**：`adapter_model.safetensors`（权重）与 `adapter_config.json`（含 rank、alpha 等配置）。 -- **rank 限制**：必须为 8、16、32、64 之一，且同一模型所有 LoRA 层使用相同 rank。 -- **词汇表与对话模板**：不得修改原始 vocab 或 chat_template，必须与开源基础模型默认配置一致，否则无法导入。 -- **VL 模型**：必须冻结 VIT，若 adapter 中包含 `visual` 开头的权重参数则无法导入。 - -支持导入的基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列的指定版本。导入后模型状态包括创建中、创建成功（可部署）、创建失败、已失效。 - -> **注意**：导入模型若与本地 vLLM/SGLang 推理效果不一致，通常是推理引擎参数默认值差异所致。可将 `temperature`、`top_p`、`repetition_penalty` 设为 1.0、`presence_penalty` 设为 0 以对齐 vLLM 默认行为。 - -## 使用 API/命令行部署 - -除控制台外，可通过 DashScope HTTP API 完成部署，**仅适用于华北2（北京）地域**，需先获取并配置 API Key，详见 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。核心接口为 `POST/GET/DELETE https://dashscope.aliyuncs.com/api/v1/deployments`，通过 `plan` 字段区分计费方式： - -- **PTU**：`plan: "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`plan: "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit`。 -- **按 Token（LoRA 自定义模型）**：`plan: "lora"`，`capacity` 必填但设置无效，扩缩容需在控制台申请。 - -典型部署命令（模型单元）： - -```bash -curl "https://dashscope.aliyuncs.com/api/v1/deployments" \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data '{ - "name": "my_qwen_plus", - "model_name": "qwen-plus-2025-12-01", - "plan": "mu", - "deploy_spec": "MU1", - "enable_thinking": true, - "capacity": 4, - "max_context_length": 10000, - "rpm_limit": 500, - "tpm_limit": 1000 -}' -``` - -部署流程：创建部署 → 返回 `deployed_model`（专属服务唯一 ID）→ 轮询 `GET /deployments/{id}` 直到 `status` 为 `RUNNING` → 通过 DashScope SDK 或兼容 API 发起推理 → 不再使用时 `DELETE /deployments/{id}` 下线并停止计费。 - -## 部署配置与列表管理 - -在控制台部署时可配置：服务名称、选择模型、模型单元类型（部署规格）、部署副本数、部署模板（如「单机部署」，仅模型单元模式可用）、推理模式（Instruct 非思考 / Thinking 思考）、最长上下文、服务限流（RPM/TPM）。 - -部署列表页展示服务名称、模型名称、**模型 Code**（API 调用时指定模型的唯一标识）、部署状态（待部署、部署中、运行中、部署失败、下线中、已停止、变配中等）、计费方式、部署详情与限流详情。 - -## 限制与注意事项 - -- **计费不可逆变更**：计费方式创建后不可改；预付费按天/按月无法提前退费，首月内提前退订按日单价 1.2 倍计费。 -- **部署即计费**：模型部署成功后即产生费用，即便尚未发起任何调用；后付费欠费后资源保留并继续计费 24 小时，超时后停止计费并删除底层资源（部署任务保留）。 -- **按 Token 模式约束**：仅支持 LoRA 调优后模型，一个月内不使用将自动释放。 -- **权限**：API 部署报错 `Workspace ... does not have deployment privilege` 或 `Workspace access denied` 时，需检查 API Key 归属业务空间的模型部署授权与账号操作权限。 -- **删除不可恢复**：执行 DELETE 后服务立即下线且不可恢复。 +百炼平台提供三种模型部署方式：预置吞吐（PTU）、模型单元（MU）和按 Token 用量计费，分别面向高并发低延迟、资源隔离可定制、以及低成本验证等不同业务场景。所有部署均通过统一 API 接口或控制台完成，支持预置模型与 LoRA 微调模型，但全参微调模型暂不支持导入与部署。部署即计费，服务状态变更（如扩容、下线）需注意计费规则与权限约束。 + +## 支持的模型/功能 + +- **预置模型**：千问系列（Qwen3/2.5/Flash/Plus/Max/VL/Omni）、DeepSeek（v3/v3.2/v4-Pro/v4-Flash）、GLM（5.2/5.1/4.7）、MiniMax-M2.5、Kimi-K2.5、CosyVoice 等，详见 [模型部署简介](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) 中的计费表格。 +- **自定义模型**：仅支持 LoRA 微调模型导入与部署，需满足 rank ∈ {8,16,32,64}、词汇表与 chat_template 未修改、VL 模型 VIT 部分冻结等严格要求；全参微调模型明确不支持 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 +- **核心功能**： + - PTU 模式支持长输入（最高 256K token）与前缀缓存，自动应用阶梯系数与缓存折扣 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)； + - MU 模式支持 PD 分离计算（降低首 Token 延迟）、推理模式选择（Instruct/Thinking）、最长上下文与服务限流配置； + - Token 计费模式仅适用于经 SFT 训练后的 LoRA 模型，且仅限部分基础模型（如 qwen3-32b/qwen3-8b/qwen2.5-vl-7b 等）。 + +> **注意**：文档 1 中“支持模型”表格称“部分预置模型与所有调优后模型”支持模型单元计费，但文档 3 明确限定“仅支持导入 LoRA 模型”，且文档 4 的 API 示例中 `plan: "lora"` 实际对应 Token 计费（非 MU），三者存在术语混淆。实际支持情况以 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) 的 LoRA 限制为准：**只有符合规范的 LoRA 模型才能部署，且 MU/PTU/TOKEN 三种计费方式均仅对 LoRA 模型开放**。 + +## 关键参数 + +| 参数 | 适用模式 | 说明 | 示例值 | +|------|----------|------|--------| +| `plan` | 全部 | 计费策略标识：`ptu` / `mu` / `lora`（注意：`lora` 此处指 Token 计费，非模型类型） | `"ptu"` | +| `ptu_capacity.input_tpm` / `output_tpm` | PTU | 预置吞吐额度（每分钟 Token 数），决定服务容量上限 | `{"input_tpm": 10000, "output_tpm": 1000}` | +| `deploy_spec` / `capacity` | MU | 模型单元规格（如 `"MU1"`）与副本数，直接关联算力与并发能力 | `"MU1"`, `4` | +| `enable_thinking` | MU | 是否启用思考模式（影响输出单价与性能） | `true` | +| `max_context_length` | MU | 最长上下文长度（部分模型支持，单位 token） | `10000` | +| `rpm_limit` / `tpm_limit` | MU | 服务级限流阈值（每分钟请求数 / 每分钟 Token 数） | `500`, `1000` | + +- PTU 模式不支持自定义 `max_context_length` 或限流，其吞吐与延迟由平台预置； +- Token 计费模式（`plan: "lora"`）的 `capacity` 参数无效，仅需填写占位值（如 `1`），扩缩容必须通过控制台申请 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 + +## 使用方式 + +1. **控制台部署**：访问 [模型部署控制台](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_deploy/create)，选择模型、计费方式及对应参数（如 PTU 容量或 MU 规格），提交创建。 +2. **API 部署**（推荐自动化）： + - PTU：`POST /api/v1/deployments`，携带 `plan: "ptu"` 与 `ptu_capacity` 对象； + - MU：`POST /api/v1/deployments`，携带 `plan: "mu"`、`deploy_spec`、`capacity` 及可选 `enable_thinking` 等； + - Token 计费：`POST /api/v1/deployments`，携带 `plan: "lora"` 与占位 `capacity`。 +3. **状态查询与管理**：通过 `GET /api/v1/deployments/{deployed_model}` 获取状态（`RUNNING` 表示就绪），`DELETE /api/v1/deployments/{deployed_model}` 下线服务。 +4. **推理调用**：使用 `model` 参数指定部署服务 ID（即 `deployed_model` 字段值），而非基础模型名，例如 `model='qwen3-8b-ft-202511132025-0260'`。 + +## 限制和注意事项 + +- **权限约束**：API 部署需确保 API Key 所属业务空间已授权目标模型的部署权限，否则报错 `Workspace xxx does not have deployment privilege for model xxxx` [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **计费刚性**：部署成功即开始计费，PTU/MU 无法中途切换计费方式，必须先下线再重建；PTU 预付费订单不可提前终止，首月退订按日单价 1.2 倍计费。 +- **额度溢出**：PTU 模式下，超出购买 TPM 或输入超模型上限（如 Qwen 128K）时，请求自动降级为按量计费，响应头含 `x-dashscope-ptu-overflow:true`，`service_tier` 字段不返回或为 `default` [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **LoRA 导入限制**：OSS Bucket 必须添加 `bailian-datahub-access` 标签，且模型文件不得位于根目录；`adapter_model.safetensors` 中禁止出现 `visual` 相关权重参数 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 +- **地域限制**：API 部署当前仅支持华北2（北京）地域 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md index 52539ce8..9826366b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md @@ -1,106 +1,65 @@ # model evaluation introduction -模型评测是百炼平台提供的模型能力评估功能，支持自定义评测和基线评测两种方式，通过评测维度对模型推理结果进行打分和对比，帮助开发者选择最优模型或验证调优效果。当前仅支持文本生成类模型评测。 +模型评测是百炼平台提供的核心能力评估功能，用于对文本生成类模型的推理结果进行结构化打分与对比分析。它通过可复用的评测维度定义评分规则，支持大模型自动评估、规则匹配和人工评审三种范式，帮助开发者量化模型表现、验证调优效果或支撑选型决策。所有评测均基于明确的输入（Prompt）、输出（Output）与参考答案（Completion）三元组展开。 -## 核心概念 +## 支持的模型/功能 -模型评测涉及两个容易混淆的核心概念： +百炼模型评测当前**仅支持文本生成类模型**，覆盖预置模型与调优后模型。评测功能分为两类： -- **评分器 Prompt**：配置于评测维度，指导裁判模型如何给被评测模型的回答打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色定位或行为规范，通常可留空。 +- **自定义评测**：使用用户上传的评测数据集（EvaluationSet 类型，含 Prompt 和 Completion 列）或已有的推理结果集，结合自定义创建的评测维度执行评分。支持三种评分方式： + - 大模型评估（AI 自动评测）：调用裁判模型（如千问-Max）进行语义级评判； + - 规则评估（自动化指标）：基于字符串匹配或 BLEU/ROUGE/余弦等算法计算分数； + - 人工评估（人工标注）：由评测人员按标签逐条标注 Pass/Fail。 + 详见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 -两者作用对象和费用归属不同，详见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **基线评测**：仅在北京地域可用，使用平台预设的公开标准数据集（如 C-Eval、GSM8K、BBH），系统自动执行预置维度下的评分，不支持自定义维度或结果下载。其设计目标是快速获取基础能力基准分，与自定义评测形成互补 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 -## 评测方式 - -### 自定义评测 - -使用自有数据集和自定义评测维度，支持三种评分方式： - -| 评分方式 | 说明 | 适用场景 | -|---------|------|---------| -| 大模型评估（AI 自动评测） | 由裁判模型（推荐千问-Max）对回答评分 | 问答质量、内容安全等语义理解场景 | -| 规则评估（自动化指标） | ROUGE、BLEU、Cosine 等算法直接计算 | 翻译、摘要、Function Calling 等确定性场景 | -| 人工评估（人工标注） | 人工逐条标注 Pass/Fail | 创意性写作、专业领域判断 | - -### 基线评测 - -使用公开标准数据集（C-Eval、MMLU、GSM8K、BBH 等）快速评测模型基础能力，无需自行准备数据集或配置维度。 - -> **注意**：基线评测仅北京地域可用，不支持下载评测结果，不支持推理结果集数据来源。 - -## 评测维度类型 - -评测维度定义模型的评分规则，创建为模板后可被多个评测任务复用。百炼提供五种评分器类型，详细说明参见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 - -| 维度类型 | 评分方式 | 需要参考答案 | 裁判模型费用 | -|---------|---------|------------|------------| -| 大模型评估-数值型 | 裁判模型打分（整数，如 0-5） | 否 | 有 | -| 大模型评估-分类型 | 裁判模型标签（Pass/Fail） | 否 | 有 | -| 规则评估-文本相似度 | ROUGE/BLEU/Cosine 等算法 | 是 | 无 | -| 规则评估-字符串匹配 | 相等/不相等/包含 | 是 | 无 | -| 人工评估-分类型 | 人工标注 Pass/Fail | 否 | 无 | - -**选型决策路径**：有标准答案且格式固定用字符串匹配；有标准答案但表述多样用文本相似度；无标准答案需语义理解用大模型评估；需主观判断用人工评估。 +> **注意**：文档 1 中未提及基线评测的地域限制，而文档 2 明确指出“基线评测仅北京地域可用”，该信息以文档 2 为准。 ## 关键参数 -### 评测维度参数 - -- **评分范围**（数值型）：裁判模型打分区间，整数，默认 0-5。建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**（数值型/相似度型）：判定 Pass/Fail 的分界线，步长 0.1（数值型）或 0.01（相似度型）。 -- **评分器 Prompt**（大模型评估类型）：至少包含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符。 -- **Pass/Fail 标签**（分类型/人工评估）：两组标签不可重复。 - -> **注意**：维度类型创建后不可修改，选错需删除重建。 +评测维度的核心参数依类型而异，需在创建时准确配置： -### 评测任务参数 +- **通用参数**：维度名称（≤20 字符，必填）、描述（≤100 字符，选填）、类型（5 种之一，创建后不可更改）。 +- **大模型评估类型**（分类型/数值型）： + - 裁判模型（必填，推荐千问-Max）； + - 评分器 Prompt（必含 `${prompt}`、`${output}` 或 `${completion}` 至少一个变量）； + - 分类型：Pass/Fail 标签（互斥且不可重复）； + - 数值型：评分范围（整数区间，默认 0~5，最大值建议 ≤10）、通过阈值（小数，步长 0.1，默认 3.0）。 +- **规则评估类型**： + - 字符串匹配：比较操作符（相等/不相等/包含）、评测输入与模型输出（至少一侧含变量）； + - 文本相似度：评估指标（7 种算法可选，如 ROUGE-L 适用于摘要、BLEU 适用于翻译）、通过阈值（0~1，步长 0.01）。 +- **人工评估类型**：仅需配置 Pass/Fail 标签，无裁判模型调用。 -- **数据来源**：评测数据集（含 Prompt + Completion，会产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP、System Prompt 等，按所选模型动态加载。 -- **排行参与**：开启后须绑定排行榜，结果加入排名对比。 +所有维度模板均可在控制台[评测维度列表页](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)统一管理，修改仅影响后续评测任务，已运行任务结果不变 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 -## 使用流程 +## 使用方式 -端到端流程分四步： +完整评测流程为四步闭环： +1. **准备数据集**：在数据管理模块上传 EvaluationSet 类型数据（含 Prompt 和 Completion 列），或准备已含 Output 的推理结果集； +2. **创建评测维度**：根据场景选择类型并配置参数（如大模型评估-数值型 + 千问-Max + 综合评测模板 + 0~5 分 + 阈值 3.0）； +3. **创建评测任务**：选择模型、指定数据来源（评测数据集或推理结果集）、关联维度、设置是否参与排行； +4. **查看结果**：在任务详情页的「指标统计」Tab 查看综合得分、通过率及分布图，在「数据明细」Tab 审查逐样本评分。 -1. **准备数据集**：在数据管理模块上传评测集类型数据（含 Prompt 和 Completion 列）。 -2. **创建评测维度**：定义评分标准和方式，选择评分器类型并配置参数。 -3. **创建评测任务**：选择被评测模型、关联数据集和维度，提交评测。 -4. **查看结果**：在指标统计 Tab 查看综合得分和通过率，在数据明细 Tab 查看逐条评分。 +> **注意**：文档 1 提到“评分器类型创建后不可更改，选错只能删除重建”，而文档 2 补充说明“已关联该维度的评测任务不受影响”，该细节对运维安全至关重要，应严格遵循。 -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 +任务提交后状态流转为：待执行 → 进行中 → 评测完成/失败/终止。人工评估任务需全部标注完成后才变为“评测完成”状态 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 -## 计费说明 +## 限制和注意事项 -费用由两部分构成：被评测模型推理费用 + 裁判模型评分费用。 - -- 使用推理结果集可免去推理费用。 -- 规则评估和人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -**成本优化**：先用 50-100 条小规模验证 → 保存推理结果集复用 → 有确定性标准的场景优先用规则评估。 - -## 限制与注意事项 - -- 当前仅支持文本生成类模型评测。 -- 基线评测仅北京地域可用。 -- 任务提交后不可更换目标模型，需删除重建。 -- 评测维度类型创建后不可修改。 -- 模型评测当前仅支持控制台操作，不提供公开 API/SDK，如需编程化评测可参考 PAI Judge Model API。 -- 1-3% 的分数差异通常为评测噪声，不建议仅凭微小分差做决策。 -- LLM 评分器存在位置偏差和自我偏好偏差，建议定期人工抽查校准。 - -更多操作细节及常见问题排查请参见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)和[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 +- **模型限制**：仅支持文本生成类模型，不支持多模态、语音或结构化输出模型。 +- **维度限制**：类型一旦创建不可修改；被排行榜绑定的维度删除后，将阻止新任务创建；已被评测任务引用的维度无法直接删除。 +- **费用说明**： + - 使用评测数据集时产生被评测模型推理费用（按 Token 计费）； + - 仅大模型评估维度产生裁判模型评分费用（按 Token 计费）； + - 规则评估与人工评估无裁判模型费用； + - 推理结果集方式可规避被评测模型推理费用。 +- **成本优化建议**：优先用规则评估（零裁判模型费用）；先用 50–100 条数据小规模验证配置；保存并复用推理结果集避免重复推理。 +- **结果解读**：综合得分是各维度平均分，易掩盖维度间差异，应结合分数分布图与逐维度分析定位短板；1–3% 的分差通常属评测噪声，不宜作为决策依据。 ## 来源文档 -- [模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - - - - - +- [模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md index 56dd7e68..2b78bd80 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md @@ -1,71 +1,58 @@ # model experience -百炼平台按模态与任务把模型体验拆分为文本生成、视觉理解、图像/视频/3D 生成、语音合成/识别、语音转语音、音乐生成、向量与重排序以及全模态等多个方向。本页汇总各方向的推荐模型、关键参数、接入方式与常见约束，帮助开发者快速完成选型；具体计费、上下文窗口等实时参数以模型广场为准。 +`model experience` 是百炼平台面向开发者提供的模型能力概览与使用指南，涵盖视觉理解、文本生成、多模态处理、语音/音频、3D生成及向量检索等核心AI能力。本文档聚焦于模型选型逻辑、关键参数约束、标准化调用方式及实际部署注意事项，所有信息均基于当前（2026年中）稳定可用的模型版本，不包含营销性描述或过时推荐。 -## 文本生成 +## 支持的模型/功能 -文本类任务的通用首选是 `qwen3.7-plus`（1M 上下文、Function Calling、内置工具、结构化输出俱全），成本敏感时切换到效果接近的 `qwen3.6-flash`，需要最强推理时用 `qwen3.7-max`；处理超长文档（多合同、大规模文献）时用上下文达 10M 的 `qwen-long`。关键能力开关包括： +百炼平台提供覆盖全模态场景的模型体系，按能力域划分如下： -- 思考模式：通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制），Qwen3 及以上多为混合模式，可按请求切换。 -- Function Calling：所有通用模型支持；内置工具（联网搜索、代码解释器、网页抓取）免复杂配置。 -- 结构化输出与批量推理：分别用于稳定 JSON 返回和高吞吐、低延迟要求不高的场景。 +- **视觉理解**：支持图像OCR、视频理解、结构化输出及Function Calling。旗舰模型 `qwen3.7-plus` 支持1M上下文、2小时视频输入、2048张图片和64段视频；轻量模型 `qwen3.6-flash` 在保持相同上下文长度与功能集的前提下显著降低成本 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **文本生成**：适用于AI编程、办公文档处理、长文本摘要等场景。`qwen3.7-plus` 和 `qwen3.6-flash` 均支持思考模式（`enable_thinking`）、Function Calling、内置工具（联网搜索、代码执行）及结构化JSON输出；超长文档处理推荐 `qwen-long`（10M上下文），但其不支持思考模式与内置工具 [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **图片与视频生成/编辑**：`wan2.7-image-pro` 支持4096×4096文生图与多图参考编辑；`happyhorse-1.1-t2v` 和 `wan2.7-t2v-2026-06-12` 均支持1080P有声视频生成，后者额外支持自定义音频文件注入 [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md)。 +- **语音与音乐**：S2S（语音转语音）模型如 `qwen3.5-omni-plus-realtime` 支持端到端音频理解与生成，兼具Function Calling与联网搜索能力；Fun-Music模型（`fun-music-v1`）支持[prompt](prompt.md)/lyrics双输入、性别选择及纯音乐生成，但仅限华北2（北京）地域 [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md)。 +- **3D与向量能力**：Tripo 3D模型（`Tripo/Tripo-P1.0`）需通过异步API调用，仅支持北京地域，且必须使用该地域API Key [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)；向量模型中，`text-embedding-v4` 为文本Embedding默认推荐，`qwen3-rerank` 支持最多500文档的纯文本重排序 [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。 -从闭源模型迁移时可按能力档对位选型，详见 [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +> **注意**：文档 2 中称 `qwen3.7-max` “不支持结构化输出”，但文档 1 明确列出 `qwen3.7-max-2026-06-08` 的结构化输出列为“不支持”，而 `qwen3.7-plus` 为“支持”。两者一致，无矛盾；但文档 2 表格中将 `qwen3.7-max` 的结构化输出标为“不支持”属正确表述，非错误。 -> **注意**：文档中的模型 ID（如 `qwen3.7-max`、`glm-5.2`、`deepseek-v4-pro`）版本较新，且标注"旧版模型"建议新项目使用 Qwen3.6/Qwen3.5 系列，实际可用模型与版本请以模型广场为准。 +## 关键参数 -## 视觉理解与 OCR +各模型共性关键参数如下（单位均为Token，除非特别注明）： -图像/视频理解同样推荐从 `qwen3.7-plus` 起步（1M 上下文、最长 2 小时视频、最大单图 1600 万像素）。要点： +| 参数 | 说明 | 典型值/范围 | 约束说明 | +|------|------|-------------|----------| +| `max_context` | 输入上下文长度上限 | `qwen3.7-plus`: 1M；`qwen-long`: 10M；`text-embedding-v4`: 8,192 | 超出将被截断，不报错 | +| `max_output_tokens` | 单次响应最大输出长度 | `qwen3.7-plus`: 64k；`qwen3-rerank`: 4,000/条 | 输出受模型能力与计费策略双重限制 | +| `max_image_count` / `max_video_count` | 单请求最大媒体数 | `qwen3.7-plus`: 2048图/64视频；`qwen3.5-omni-plus`: 256图/512视频 | 图像分辨率影响Token消耗：`h × w / (32 × 32) + 2` [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) | +| `texture_quality` / `geometry_quality` | Tripo 3D模型贴图与几何精度控制 | `standard` / `detailed`；`standard` / `ultra` | 仅 `Tripo/Tripo-H3.1` 支持 `geometry_quality` | +| `format` | 音频/视频输出格式 | `mp3` / `wav`；`720P` / `1080P` | `wav` 无损但体积大；视频输出帧率固定为24/30 fps | -- 图像 Token 估算：`h x w / (32 x 32) + 2`，分辨率越高消耗越大。 -- 视频时长：`qwen3.7-plus` / `qwen3.6-plus` / `qwen3.6-flash` / `qwen3.5-plus` / `qwen3.5-flash` 支持最长 2 小时 / 2GB。 -- OCR/文档提取用专优的 `qwen3.5-ocr`，通用图片文字提取也可用旗舰模型。 +## 使用方式 -细节参见 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **同步调用**：适用于文本生成、TTS、ASR、Embedding等低延迟场景。HTTP POST请求，`Content-Type: application/json`，模型ID置于`model`字段，输入数据置于`input`对象内（如`{"prompt": "..."}` 或 `{"audio_url": "..."}`）。 +- **异步调用**：适用于3D生成、长视频生成等耗时任务。首请求返回`task_id`，后续轮询 `GET /api/v1/tasks/{task_id}` 获取结果，状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED`/`FAILED`，有效期24小时 [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **流式调用**：WebSocket协议用于实时语音对话（`-realtime`后缀模型）、流式TTS/ASR。需维持长连接，服务端分块推送响应（如语音PCM片段或识别文本流）。 +- **多模态输入**：视觉/全模态模型接受混合输入。例如 `qwen3.5-omni-plus` 的`input`可同时含`text`、`audio_url`、`image_url`、`video_url`字段；Tripo模型则通过互斥字段`prompt`/`image`/`images`区分生成模式 [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 -## 图像与视频、3D 生成 +## 限制和注意事项 -- **文生图 / 图片编辑**：首选 `wan2.7-image-pro`（文字渲染、品牌色、角色一致性、多图编辑；文生图最高 4096x4096，编辑最高 2048x2048）。追求速度与低成本用 `z-image-turbo`；需要负向提示词或最多 6 张变体用 `qwen-image-2.0-pro`。见 [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md)。 -- **视频生成 / 编辑**：文生视频与首帧生视频推荐 `happyhorse-1.1-t2v` / `happyhorse-1.1-i2v`（1080P、3-15 秒、有声）；需要自定义音频或首尾帧串联长视频用 `wan2.7-*` 系列；角色动画用 `wan2.2-animate-move` / `wan2.2-animate-mix`（pro/std 两档）。 -- **3D 生成**：通过 Tripo 模型支持文生/单图/多图三种模式，`prompt`、`image`、`images` 三字段互斥；快速预览用 `Tripo/Tripo-P1.0`，高精度用 `Tripo/Tripo-H3.1`（几何精度 `geometry_quality` 最高 200 万面）。 - -> **注意**：Tripo 3D 生成为异步任务，仅在"中国内地（北京）"地域可用，需使用该地域 API Key；产物 `pbr_model_url` / `rendered_image_url` 有效期仅 2 小时，务必及时下载。轮询建议间隔 15 秒。 - -## 语音合成、识别与语音对话 - -语音方向需先确定接入协议与音色来源： - -- **接入方式**：WebSocket 双向流式、延迟最低，适合实时交互；HTTP 发送完整文本、支持流式返回，适合离线内容制作。Qwen-Audio-TTS/CosyVoice 用同一模型名同时支持两种协议，Qwen 系列用 `-realtime` 后缀区分。 -- **语音合成**：标准合成用 `qwen-audio-3.0-tts-plus` / `MiniMax/speech-2.8-hd`；自定义音色分声音复刻（提供音频样本）与声音设计（文字描述音色），声音设计推荐 `cosyvoice-v3.5-plus`，音色注册统一用 `voice-enrollment`。指令控制可用自然语言动态调节语速、情绪、风格。 -- **语音识别**：从实时/非实时、专业术语、说话人分离、情感识别四个维度选型。实时用 `fun-asr-realtime` 或 `qwen3.5-omni-plus-realtime`，非实时用 `fun-asr`（支持说话人分离）；情感识别用 Qwen-ASR 系列。详见 [语音识别](../../raw/model-user-guide/model-experience/asr-model.md)。 -- **语音转语音**：S2S 单模型（Omni / Livetranslate）延迟低、端到端感知语调情绪；Pipeline（ASR+LLM+TTS）更灵活、可自定义音色。实时对话用 `qwen3.5-omni-plus-realtime`，同传翻译用 `qwen3.5-livetranslate-flash-realtime`。 -- **音乐生成**：Fun-Music 通过 `prompt` 或 `lyrics` 生成完整歌曲，`is_instrumental=true` 生成纯音乐，`gender` 选男女声（仅 `fun-music-v1`）。 - -> **注意**：Fun-Music 处于邀测阶段，需在模型广场申请开通，且仅华北2（北京）地域可用；其请求端点为业务空间维度（`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/...`），需替换 `{WorkspaceId}`。 - -## 向量与重排序 - -RAG / 语义搜索场景：纯文本 Embedding 首选 `text-embedding-v4`（维度 64~2048，默认 1024，最大 8192 Token），迁移旧索引可用 `text-embedding-v3`；跨模态检索用 `qwen3-vl-embedding`（融合向量）或 `tongyi-embedding-vision-plus`（独立向量）。重排序用 `qwen3-rerank`（纯文本、100+ 语言、最多 500 文档）或 `qwen3-vl-rerank`（多模态）。维度选择建议：存储受限选 256/512，通用选 1024，高精度选 1536/2048。 - -## 全模态 - -需要同时理解文本、音频、图片、视频并输出文本/语音时，用全模态模型：旗舰 `qwen3.5-omni-plus`（能力最全，支持联网搜索、Function Calling、音频最长 3 小时/视频最长 1 小时）、轻量 `qwen3-omni-flash`（成本低、支持思考模式但思考模式下不输出语音）、专业翻译 `qwen3.5-livetranslate-flash`（60 种语言、约 3 秒延迟、开箱即用）。 - -> **注意**：不同文档对同一模型的能力标注存在差异——例如 `qwen3-omni-flash-realtime`（WebSocket）在多数表格中不支持 Function Calling/联网搜索，而 HTTP 版 `qwen3-omni-flash` 支持 Function Calling 与思考模式；联网搜索与 Function Calling 不可同时开启。以对应模型的用户指南和 API 文档为最终依据。 +- **地域限制**：Tripo 3D模型、Fun-Music、部分S2S/ASR模型（如`qwen3.5-livetranslate-flash`）**仅支持华北2（北京）地域**，且必须使用该地域API Key与Endpoint [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **功能互斥**：Qwen3.5-Omni系列在启用联网搜索时**不可同时启用Function Calling**；思考模式下**不支持生成语音输出**（仅文本） [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md)。 +- **旧版模型弃用**：Qwen2.5-VL、Qwen-Omni、Qwen-VL等旧系列模型已明确标注“不再作为首选推荐”，新项目应使用Qwen3.6或Qwen3.5系列 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **音频规格硬约束**：Fun-ASR非实时模型支持最大12小时/2GB音频；Qwen3.5-Omni非实时模型限3小时/2GB；而Qwen3-omni-flash HTTP模式仅支持20分钟/100MB [语音识别](../../raw/model-user-guide/model-experience/asr-model.md)。 +- **语言覆盖差异**：Qwen3.5-Livetranslate支持60种语言（29种输出语音+文本），但Qwen3-Omni-Flash仅支持11种输出语言；方言支持因模型版本而异（如`fun-asr-realtime`支持数十种中文方言，而`paraformer-8k-v2`仅支持普通话） [全模态](../../raw/model-user-guide/model-experience/omni.md)。 ## 来源文档 -- [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md) - [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) -- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) +- [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md) - [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md) +- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) - [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [语音合成](../../raw/model-user-guide/model-experience/tts-model.md) +- [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) - [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) - [语音识别](../../raw/model-user-guide/model-experience/asr-model.md) -- [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) -- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) - [全模态](../../raw/model-user-guide/model-experience/omni.md) +- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md index cbc5d2d9..252a68d3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md @@ -1,105 +1,37 @@ # model high speed inference -百炼平台针对推理性能与容量保障提供了两类高速推理能力：一是通过 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 为指定模型锁定专属吞吐容量，规避公共资源限流；二是通过[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)（Fast mode）提升单请求的 TPS（Tokens Per Second），面向对输出速度敏感的场景。两者关注点不同——预留解决"容量刚性兑付"，快速模式解决"输出更快"，可按需选型或组合使用。 +百炼平台提供两种面向高吞吐与低延迟场景的推理加速能力：TPM 预留（保障专属容量）和快速模式（提升单请求输出速度）。二者定位不同，可独立使用或组合使用——TPM 预留解决“能不能稳定跑满”的问题，快速模式解决“单次响应够不够快”的问题。开发者应根据业务对容量确定性（如 SLA 要求）与响应时延（如 TPS/首 token 延迟）的优先级进行选型。 -## 能力对比与选型 +## 支持的模型/功能 -| 能力 | 解决的问题 | 计费方式 | 超额处理 | 代码改动 | -| --- | --- | --- | --- | --- | -| TPM 预留 | 高峰期专属容量、不被公共限流 | 按 kTPM 预付费 | 自动降级公共池按量，不中断 | 替换 `model` 为专属 code | -| 快速模式 | 更高 TPS（80~100 TPS） | 按 token（同标准 API） | 超额进入排队，不立即限流 | 指定 fast 模型 ID + 专属域名 | +- **TPM 预留**：为指定模型锁定专属输入/输出吞吐量（单位：kTPM），确保高峰期不被公共资源限流影响。支持千问、GLM、DeepSeek、Kimi 等多个主流模型，具体列表见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档中的“支持的模型”表格。 +- **快速模式（Fast mode）**：Preview 阶段能力，通过优化推理调度与内存访问，提升单请求输出吞吐（TPS 达 80~100），适用于 AI 编程助手、Agent 多步推理等对首 token 和 token 流速敏感的场景。当前仅支持 `glm-5.2-fast-preview` 模型（[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 文档明确列出），其他模型暂未开放。 -> **注意**：[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档中的方案对比还列出了按量付费、资源包/节省计划、PTU 专属部署等其他容量方案；如需专属部署实例或极致高吞吐，可评估 PTU（模型部署）。 +> **注意**：两篇文档对“模型支持范围”的描述存在明显差异——TPM 预留文档列出了十余个模型（如 `qwen3.7-max-2026-05-20`、`deepseek-v4-pro` 等），而快速模式文档仅声明 `glm-5.2-fast-preview` 可用。目前无证据表明其他模型已支持快速模式，因此以 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 的明确声明为准，不可自行尝试在非 listed 模型上添加 `-fast-preview` 后缀。 -## TPM 预留 +## 关键参数 -### 功能与特性 +| 能力类型 | 核心参数 | 说明 | +|----------|----------|------| +| TPM 预留 | `input_tpm` / `output_tpm` | 单位为 kTPM（1 kTPM = 1,000 tokens/min），需按模型实际阶梯系数与缓存折扣估算，详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中的“容量计算器”与“长输入阶梯系数”表格。 | +| 快速模式 | 无显式参数 | 仅需将 `model` 设为 `glm-5.2-fast-preview`，并使用专属接入域名（如 `{workspace_id}.cn-beijing.maas.aliyuncs.com`），无需额外 query 或 header。 | -TPM（Tokens Per Minute）预留为指定模型锁定专属推理吞吐量，预留容量内的调用不与其他用户共享、不受公共限流影响。核心特性见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)： +## 使用方式 -- **容量保障**：预留 TPM 为业务专属，刚性兑付。 -- **专属模型 code**：创建后系统自动生成专属 `model` code，必须在 API 请求中把 `model` 参数替换为该 code 才生效。 -- **超额不中断**：超出预留容量的请求自动降级为按量计费，无需改代码，可在详情页"超额降级统计"查看降级次数。 +- **TPM 预留**：创建成功后，系统生成专属模型 code（如 `tpm-qwen37max-xxx`），**必须**在 API 请求中将 `model` 参数替换为此 code 才能生效。标准调用方式不变，但需注意预热期可能引入短暂延迟波动（见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) “创建 TPM 预留”章节示例代码注释）。 +- **快速模式**：直接使用 `model="glm-5.2-fast-preview"` 发起请求，并确保 base_url 指向对应地域的 MaaS 域名（如华北2为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。流式响应中需分别处理 `delta.reasoning_content` 和 `delta.content` 字段（见 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 的“使用示例”）。 -### 关键参数（创建时） +## 限制和注意事项 -| 参数 | 说明 | 取值 | -| --- | --- | --- | -| 预留名称 | 自定义标识 | ≤ 50 字符 | -| 选择模型 | 仅支持已开放 TPM 预留的模型 | 以控制台为准 | -| 付费周期 | 计费周期 | 按天 | -| 输入 TPM / 输出 TPM | 预留吞吐量，1 kTPM = 1,000 Tokens/分钟 | 起步与步长因模型而异 | -| 购买时长 | 有效时长 | 1~30、60、90、120、365 天 | -| 到期自动续费 | 到期前一天 08:00 自动扣款 | 默认开启 | +- **TPM 预留**： + - 预留实例到期后 2 小时内仍可调用，2~14 小时内停止但可续费，14 小时后彻底删除且不可恢复； + - 缩容退订按 1.5 倍系数结算已用费用，公式见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) “计费与使用说明”； + - 超额请求自动降级至按量计费，不中断服务，但需监控“超额降级统计”避免成本失控。 -创建页右侧提供 **TPM 容量计算器**，可根据 RPM、平均输入/输出长度、缓存命中率估算推荐的输入/输出 TPM，其中缓存命中率仅影响输入 TPM。 - -### 使用方式 - -创建预留后进入详情页"概览"Tab 复制专属模型 code，将请求中的 `model` 参数替换即可： - -```python -import dashscope - -response = dashscope.Generation.call( - api_key="your-api-key", - model="your-dedicated-model-code", # 替换为专属模型 code - messages=[{"role": "user", "content": "你好"}], -) -print(response.output.text) -``` - -> **注意**：请求量短时间快速拉升时系统需短暂预热，预热期间部分请求可能出现延迟波动，请做好排队或重试机制。 - -### 计费、生命周期与管理 - -- 部署成功即开始计费，预留容量内的调用不额外收费；预付费一次性支付，费用以控制台为准。 -- 缩容/退订退费按已用部分 1.5 倍系数结算：`退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)`。 -- 到期后生命周期：0~2 小时仍运行中可调用/续费；2~14 小时已停止不可调用但可续费；14 小时后实例删除不可恢复。 -- 详情页含"概览""监控""API 接入"三个 Tab，支持扩缩容（变配期间不中断）、续订、退订。退订不可恢复，专属 code 失效后请求回退公共资源。 - -## 快速模式（Fast mode） - -### 功能与特性 - -快速模式面向 AI 编程助手、Agent 多步推理、实时对话等对输出速度敏感的场景。据[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)文档： - -- **高速输出**：TPS 提升至标准 API 的 1.5~2 倍，达 80~100 TPS。 -- **按 token 计费**：计费逻辑与标准 API 一致，按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 额度不立即限流，请求进入排队队列。 -- **preview 阶段**：能力与规格可能随版本调整。 - -当前支持模型为 `glm-5.2-fast-preview`（华北2（北京）、新加坡地域）。 - -### 使用方式 - -无需额外参数，调用时把 `model` 指定为支持的 fast 模型 ID 即可开启，但需使用专属接入域名： - -``` -https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1 -``` - -其中 `{workspace_id}` 可在业务空间管理页切换到对应地域后查看。基础调用示例： - -```bash -curl -X POST https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "glm-5.2-fast-preview", - "messages": [{"role": "user", "content": "你是谁"}], - "stream": false -}' -``` - -`glm-5.2` 默认返回 `reasoning_content` 思考字段；[流式输出](../concepts/streaming.md)时思考内容与回答内容分别通过 `delta.reasoning_content` 与 `delta.content` 推送，`usage.completion_tokens_details.reasoning_tokens` 记录思考 token 数。 - -## 限制与注意事项 - -- **接入域名不同**：TPM 预留使用标准 `dashscope.aliyuncs.com` 域名 + 专属 code；快速模式使用 `{workspace_id}.cn-beijing.maas.aliyuncs.com` 专属域名 + fast 模型 ID，两者接入方式不可混用。 -- **超额行为不同**：TPM 预留超额自动降级按量计费（不中断）；快速模式超额进入排队（不立即限流）。 -- **preview 风险**：快速模式仍处 preview，能力与规格可能变动，生产接入前请评估稳定性。 -- **计价与错误码**：快速模式的详细价格与错误码分别参见[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)引用的"模型调用计费""错误码"文档；TPM 预留的实际费用以百炼控制台为准。 +- **快速模式**： + - 当前为 preview 阶段，接口行为、模型能力及计费规则可能调整，不建议用于生产环境 SLA 保障场景； + - 超出 TPM 额度时请求进入排队队列而非立即限流，可能导致端到端延迟升高，需评估业务容忍度； + - `glm-5.2-fast-preview` 返回结构含 `reasoning_content` 字段，与标准 `glm-5.2` 不兼容，客户端需适配解析逻辑。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md index a4e37e60..17d81124 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md @@ -1,86 +1,77 @@ # model monitoring -百炼平台提供模型监控与用量管理能力，帮助开发者追踪模型调用的性能指标、[Token](../concepts/token.md) 消耗和费用趋势。通过监控面板可实时观测调用时长、失败率、RPM/TPM 等关键指标，并配置告警规则实现异常的主动发现。结合用量统计功能，开发者可以按[业务空间](../concepts/workspace.md)维度精细化管理模型成本。 +模型监控是百炼平台提供的核心可观测性能力，用于实时跟踪模型调用行为、性能表现、成本消耗与异常事件。它覆盖从基础调用统计到细粒度 Token 追踪、从控制台可视化到 Prometheus 自建集成的全链路监控能力，适用于生产环境下的稳定性保障与成本精细化治理。所有监控数据默认按「模型 + 业务空间」维度聚合，主账号可跨空间查看，子账号仅限当前业务空间。 -## 支持的模型与地域 +## 支持的模型与功能 -- **用量统计**：模型列表中的所有模型均支持查看用量，包括调优后的自定义模型。详见[模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 -- **普通监控**：支持所有模型（含自定义模型）。 -- **高级监控**：仅支持北京、新加坡、弗吉尼亚地域。 -- **告警功能**：仅支持北京、新加坡地域。 +- **监控覆盖范围**： + - **普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的全部模型（含基于其调优的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)）； + - **高级监控**（含分钟级指标、告警、Prometheus 接入）仅支持北京、新加坡、弗吉尼亚地域下的模型； + - **告警功能**仅支持北京、新加坡地域（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -> **注意**：普通监控的数据延迟为小时级，高级监控可提供分钟级数据洞察。用量统计数据延迟约 1 小时，且不支持查看 30 天以前的数据。 +- **核心功能模块**： + - **调用统计**：调用次数、失败次数、失败率、限流错误（429）、内容安全拦截次数； + - **性能指标**：RPM、TPM、调用时长、首Token延时、非首Token延时； + - **成本监控**：Token 消耗汇总与单次追踪（仅北京地域部分模型支持）； + - **日志审计**：输入/输出对话记录（仅北京地域且限于[指定模型列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）； + - **用量统计**：按业务空间维度的模型用量（含免费额度使用情况），延迟约 1 小时（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 -## 监控指标 +> **注意**：文档 1 称“普通监控延迟通常为小时级”，而文档 2 明确用量统计延迟“约为 1 小时”；二者一致。但文档 1 中“新模型在首次数据同步完成后自动加入列表”未说明是否含自定义模型——文档 2 明确“调优后的模型”同样支持用量查看，故可推断其也纳入普通监控范围，无额外限制。 -模型监控将指标分为四大类： +## 关键参数与指标 -| 类型 | 典型指标 | 用途 | -|------|----------|------| -| 安全 | 内容安全错误次数 | 识别违规内容 | -| 成本 | 平均单次请求调用量、[Token](../concepts/token.md) 消耗 | 评估成本效益 | -| 性能 | 调用时长、首 [Token](../concepts/token.md) 延时、RPM、TPM | 观察性能变化 | -| 错误 | 失败次数、失败率、限流错误次数 | 判断模型稳定性 | +| 类别 | 指标名 | 说明 | 支持过滤 Label | +|--------|---------|------|----------------| +| 调用次数 | `model_call_count` | 总调用次数 | `user_id`, `apikey_id`, `workspace_id`, `model`, `protocol`, `sub_protocol`, `status_code`, `error_code` | +| 调用时长 | `model_call_duration`, `model_call_duration_p99` | 均值/P99 时长（秒） | 同上 | +| 首Token延时 | `model_first_token_duration` | 首包响应时间 | 同上 | +| 非首Token延时 | `model_generation_duration_per_token` | 每 Token 生成耗时 | 同上 | +| Token用量 | `model_usage` | 总 Token 数（支持 `usage_type` 过滤：`input_tokens`/`output_tokens`/`total_tokens` 等） | `usage_type`, `workspace_id`, `model`, `apikey_id` | -监控详情页支持按 API-KEY、推理类型（实时/批量）、时间范围和时间精度（分钟/小时）进行筛选。详细的指标列表和 Prometheus 查询方式见[模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)。 +所有指标均通过 Prometheus HTTP API 提供，需开启[高级监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)并配置 AccessKey 认证（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -## 用量统计 +## 使用方式 -用量数据按[业务空间](../concepts/workspace.md)维度统计（不支持按阿里云账号维度），在控制台「模型用量」页面查看。不同模型类型的计量单位： +1. **控制台访问**： + - 普通监控入口：[模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)（北京）或对应地域控制台； + - 用量统计入口：[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)（仅业务空间维度）； + - 免费额度管理：[免费额度](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/free-quota)。 -- **大语言模型 / 全模态模型 / 向量模型**：Token -- **图像生成**：张 -- **视频生成**：秒 -- **语音模型**：秒、字符或 Token（视模型而定） +2. **日志与单次 Token 查看**（仅北京地域）： + - 需先在「模型监控配置」中开通**审计日志 + 推理日志**； + - 开通后，在模型列表点击「日志」页签，查看请求/响应及 `用量` 字段（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -免费额度管理可在控制台「免费额度」页面操作，开启「免费额度用完即停」后，额度耗尽时服务将自动停止（返回 403 错误），避免产生额外费用。 +3. **告警配置**： + - 仅北京、新加坡地域支持； + - 需先开启高级监控 → 进入[模型告警](https://bailian.console.aliyun.com/?tab=model#/model-alert) → 创建规则（支持短信/邮件/钉钉/Webhook 等通知方式）。 -## 告警配置 +4. **Grafana / 自建应用接入**： + - 获取 Prometheus HTTP API 地址（需开启高级监控）； + - 使用 `Basic` 认证（AccessKey:AccessKeySecret Base64 编码）调用 `/api/v1/query_range`； + - 示例：`GET {API}/api/v1/query_range?query=model_usage{workspace_id="xxx",model="qwen-plus"}&start=...&step=60s`（详见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -建立主动告警的流程： +## 限制和注意事项 -1. **开启高级监控**：在模型监控页面 → 模型监控配置 → 开启「性能和用量指标监控」。 -2. **创建告警规则**：在模型告警页面选择监控模型和模板，设置触发条件。 -3. **通知方式**：支持短信、邮件、电话、钉钉机器人、企业微信机器人及 Webhook。 +- **地域限制严格**： + - 日志审计、单次 Token 追踪、告警、Prometheus 接入等功能**仅限北京、新加坡、弗吉尼亚地域**；其他地域仅提供小时级普通监控（如调用总量、失败率等基础卡片）。 -告警等级与通知渠道对应关系： +- **模型兼容性差异**： + - 并非所有模型均支持全部监控能力。例如，历史对话（输入/输出日志）仅支持文档 1 列出的千问系列、开源及三方模型快照版本（如 `qwen3-max-2025-09-23`），旧版快照或未列型号不支持（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -- 紧急（CRITICAL）：电话、短信、邮件 -- 错误（ERROR）：短信、邮件 -- 警告（WARNING）：短信、邮件 -- 普通（INFO）：邮件 +- **数据延迟与范围**： + - 普通监控数据延迟约 **1 小时**；高级监控支持分钟级洞察； + - 控制台用量页面**最多查看最近 30 天数据**，更早数据需通过[费用与成本](https://billing-cost.console.aliyun.com/finance/expense-report/expense-detail-by-instance)查询（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 -## 模型日志 +- **权限约束**： + - 主账号及具备足够权限的子账号可开通日志与高级监控； + - 子业务空间成员**无法切换查看其他业务空间数据**，仅限当前空间（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -开通推理日志后可查看每次调用的输入、输出及 Token 消耗，适用于故障排查和内容审计。 - -> **注意**：日志功能目前仅适用于华北2（北京）地域的部分模型。从调用发生到日志记录存在分钟级延迟。 - -开通步骤：模型监控页面 → 模型监控配置 → 依次开通审计日志和推理日志。 - -## 接入外部系统 - -高级监控数据存储在私有 Prometheus 实例中，支持标准 Prometheus HTTP API。可接入 Grafana 或自建应用进行可视化分析。认证方式为 Basic Auth（`base64(AccessKey:AccessKeySecret)`）。详见[模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)中的接入说明。 - -## 生产环境最佳实践 - -- **控制输出长度**：合理设置 `max_tokens` 参数限制单次生成的最大 Token 数。 -- **按任务选模型**：简单任务（分类、摘要）优先使用轻量级模型以降低成本。 -- **监控与告警**：配置用量和性能告警，异常时及时响应。 -- **优化 Prompt**：简洁清晰的 Prompt 可减少不必要的输入 Token 消耗。 -- **批量推理**：非实时大批量任务使用批量推理接口，成本更低。 - -更多用量管理建议参见[模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 +- **用量单位差异**： + - 大语言模型按 **Token** 计费；视觉模型按 **张**（图像）、**秒**（视频）；语音模型按 **秒/字符/Token**（依模型而定）；全模态模型文本部分按 Token，其他模态按对应 Token 数（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 ## 来源文档 -- [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md) - [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md) - - - - - - +- [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md index f9bb3ca7..68323539 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md @@ -1,157 +1,63 @@ # plug in -百炼插件是一个工具集合，用于扩展大模型的能力边界。一个插件下可包含多个工具（API），每个工具实现特定功能。通过将插件集成到大模型应用中，可弥补大模型在获取最新信息、精确计算、图像处理等方面的不足。百炼支持官方插件、三方插件和自定义插件三类。详见[插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 +插件是百炼平台用于扩展大模型能力的核心机制，通过将外部工具（API）集成到模型推理链路中，解决大模型在实时信息获取、精确计算、代码执行、图像生成等场景下的固有局限。开发者可选用官方插件、三方插件或自定义插件，结合智能体应用、工作流应用或 Assistant API 进行调用。所有插件均需通过服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI` 授权方可使用。 -## 插件分类 +## 支持的模型/功能 -- **官方插件**：组件广场预置，无需配置输入输出参数即可直接调用。 -- **三方插件**：涵盖商业服务、图像视频、学习教育等领域，经过效果测试，开通后直接调用，无需额外配置。 -- **自定义插件**：当官方和三方插件无法满足业务需求时，用户可创建或从云市场导入自定义插件，集成到应用中。 +百炼当前支持以下模型调用插件能力： +- `qwen-turbo`、`qwen-plus`、`qwen-max`（文本模型） +- `qwen-vl-plus`、`qwen-vl-max`（多模态模型） -## 官方插件列表 +> **注意**：各模型对插件的兼容性存在差异，[插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md) 中列出的模型列表为截至文档发布时的兼容范围，**实际可用性请以控制台运行结果为准**；部分新模型（如 `qwen2.5` 系列）尚未明确列入该文档，需通过控制台实测验证。 -| 插件名称 | 工具 ID | 说明 | [计费](../concepts/billing.md)方案 | -| --- | --- | --- | --- | -| Python 代码解释器 | `code_interpreter` | 执行 Python 代码片段，如数学计算、数据分析与可视化、数据处理 | 免费 | -| 计算器 | `calculator` | 进行复杂数学计算，例如计算 `12313x13232` | 免费 | -| 图片生成 | `text_to_image` | 基于文本生成图片，例如"请画一只在笑的小狗" | 限时免费，需申请开通 | -| 夸克搜索 | `quark_search` | 搜索实时信息，查找公开网络知识和信息 | 限时免费，需申请开通 | -| 生成二维码 | `generate_qrcode` | 根据网站链接地址生成二维码 | 免费 | -| GitHub 搜索 | `github_search` | 在 GitHub 中搜索相关项目列表 | 免费 | +插件按来源分为三类： +- **官方插件**：预置于组件广场，开箱即用，无需配置参数。包括 `code_interpreter`（Python 执行）、`calculator`（数学计算）、`text_to_image`（文生图）、`quark_search`（实时搜索）、`generate_qrcode`（二维码生成）、`github_search`（GitHub 项目检索）等 [详见官方插件说明](../../raw/application-user-guide/plug-in/plugins.md)。 +- **三方插件**：来自阿里云云市场，覆盖商业服务、图像视频、教育等领域，开通后即可调用。 +- **自定义插件**：支持开发者接入自有 API，需定义插件 URL、工具路径、输入/输出参数及鉴权方式，完整流程见 [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md) 文档。 -> **注意**：夸克搜索插件目前支持检索网页标题、关键词和摘要，但不支持直接访问网页详情。GitHub 搜索插件支持检索项目标题、链接和摘要，不支持访问项目详情。Python 代码解释器不支持对外访问网络以及上传本地文件。 +## 关键参数 -Python 代码解释器可用依赖包括：matplotlib、pandas、scipy、seaborn、sympy、pillow、pydantic~=1.10.8、requests~=2.31.0、oss2~=2.18.1、pdfminer-six、pypdf、python-pptx、wordcloud 等。详见[官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 +插件调用依赖以下核心参数，尤其在自定义插件和 API 集成中必须准确配置： -## 支持的模型 - -插件调用对模型有一定要求，目前支持以下模型： - -| 模型 | 模型标识符 | -| --- | --- | -| 通义千问-Turbo | qwen-turbo | -| 通义千问-Plus | qwen-plus | -| 通义千问-Max | qwen-max | -| 通义千问VL-Max | qwen-vl-max | -| 通义千问VL-Plus | qwen-vl-plus | - -> **注意**：各模型对插件的兼容性可能有差异，最新兼容性状态以控制台实际执行结果为准。 - -## 插件调用机制 - -调用插件的本质是调用插件下的工具。百炼支持通过[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用以及 Assistant API 调用插件。 - -- **[智能体应用](../concepts/agent-application.md) / Assistant API**：大模型根据用户输入内容、工具名称和工具描述判断是否调用工具。需要调用时，模型选择合适工具，应用内部完成调用后将工具返回结果与用户内容合并再次输入模型，由模型生成最终结果；无需调用时直接生成结果输出。 -- **[工作流](../concepts/workflow.md)应用**：插件作为[工作流](../concepts/workflow.md)的一个节点，按用户编排的方式执行特定任务，而非由模型主动规划和调用。 +- **工具 ID（tool_id）**：唯一标识插件下的具体工具（如 `calculator`），用于 Assistant API 或工作流节点中指定调用目标。可通过插件详情页悬浮图标复制获取。 +- **插件 URL 与工具路径**：插件 URL 为域名根地址（如 `https://myapi.example.com`），工具路径为相对路径（如 `/query`），二者拼接构成完整调用地址。 +- **输入参数（input parameters）**： + - `传参方式` 必须明确设为 `大模型识别`（从用户输入提取）或 `业务透传`（由外部传入，通过 `biz_params` 或 `user_defined_params` 传递）； + - `参数名称` 和 `参数描述` 需语义清晰，直接影响大模型参数提取准确性； + - `类型` 支持 `String`、`Number`、`Object`（但 Object 子属性不可为空）。 +- **输出参数（output parameters）**：定义 API 返回数据中哪些字段被大模型用于生成最终回复，需精简且层级扁平。 +- **鉴权配置**：若 API 需鉴权，支持 `Header`（如 `Authorization: Bearer `）或 `Query`（如 `?api_key=xxx`）方式，`Type` 可选 `basic`/`bearer`/`appcode`。 ## 使用方式 -### 首次访问授权 - -主账号或 RAM 用户首次访问插件页面时，需授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`（权限策略 `AliyunServiceRolePolicyForSFMAccessCloudAPI`），用于授权百炼访问云市场商品清单并根据插件配置进行 API 调用。 - -- **主账号**：在插件页面勾选条款，单击"授权并进入"即可。 -- **RAM 用户（子账号）**：会因缺少创建服务关联角色权限报错（错误码 140052）。需先由主账号在 RAM 控制台创建自定义权限策略（Action 为 `ram:CreateServiceLinkedRole`，Condition 中 `ram:ServiceName` 为 `cloundapi-access.sfm.aliyuncs.com`），并授予子账号该策略后，再完成授权。 - -### 调用插件 - -- **方式一（插件页面）**：在插件页面将工具添加至[智能体应用](../concepts/agent-application.md)。官方插件只能与位于相同[业务空间](../concepts/workspace.md)里的[智能体应用](../concepts/agent-application.md)关联。每个[智能体应用](../concepts/agent-application.md)最多支持添加 10 个工具，应用会根据输入选择调用一个或多个工具。 -- **方式二（应用管理页面）**：在指定智能体或工作流应用内添加插件，测试效果并发布应用。 -- **方式三（Assistant API）**：通过 Assistant API 调用工具，需正确传递工具 ID。 - -子[业务空间](../concepts/workspace.md)调用官方插件前，需先在插件详情页为子[业务空间](../concepts/workspace.md)授权；默认[业务空间](../concepts/workspace.md)无需此步骤。 - -### 获取工具 ID - -通过 API 调用工具时需正确传递工具 ID。在插件页面找到目标插件，单击"查看详情"，在"插件工具"下获取工具 ID（例如 `calculator`）。 +插件可通过三种方式集成： -## 自定义插件 +1. **控制台可视化配置（推荐入门）**： + - 在 [插件市场](https://bailian.console.aliyun.com/#/plugin-market) 页面授权 `AliyunServiceRoleForSFMAccessCloudAPI` 角色（主账号直接授权；RAM 用户需先获 `ram:CreateServiceLinkedRole` 权限）； + - 官方/三方插件：单击“添加至智能体”，选择目标智能体应用（注意：官方插件仅支持与**同业务空间**的智能体关联）； + - 自定义插件：创建后需先发布为 MCP 服务，再在智能体编排页的 **MCP 区块** 中添加 [参考自定义插件文档](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 -当官方和三方插件无法满足业务需求时，可创建自定义插件。详见[自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 +2. **工作流应用节点**：将插件作为独立节点拖入工作流画布，按需编排执行顺序，不依赖大模型自主决策。 -### 工作流程 - -1. **创建/导入插件**：定义插件基础信息，或直接从云市场导入。 -2. **添加工具**（导入插件无需此步）：配置 API 路径、请求参数和返回数据。 -3. **调试与发布**：在线测试 API 连通性，功能正常后发布。只有已发布的工具才能在应用中被调用。 -4. **在应用中使用**：将插件关联到智能体，通过对话测试或 API 集成调用。 - -### 创建插件关键参数 - -- **插件名称**：支持中英文，需具语义。 -- **插件描述**：对插件功能和使用场景的简要说明，帮助大模型判断是否调用，使用自然语言描述。 -- **插件 URL**：插件访问地址。同一域名下不同路径拆分为不同 API（工具路径）。 -- **是否鉴权**：支持服务级鉴权和用户级鉴权。鉴权信息可放在 Header（默认参数名 `Authorization`）或 Query 中；Type 支持 `basic`（[Token](../concepts/token.md) 前不增加内容）、`bearer`（[Token](../concepts/token.md) 前增加 "Bearer"）、`appcode`（[Token](../concepts/token.md) 前增加 "APPCODE"）。 - -### 创建工具关键参数 - -- **工具名称**：支持中英文，有字符数限制（20 字符）。 -- **工具描述**：帮助大模型判断是否调用该工具，尽量给出使用示例。 -- **工具路径**：指向插件 URL 的相对路径，必须以正斜杠（/）开头。 -- **请求方法**：GET 或 POST。 -- **提交方式**：`application/json` 或 `application/x-www-form-urlencoded`。 -- **输入参数传参方式**： - - **大模型识别**：参数值由大模型从用户输入中提取。 - - **业务透传**：参数值从外部主动透传，通过 `biz_params` 和 `user_defined_params` 传递。 -- **高级配置**：提供调用示例（Query 与期望入参 Value），减少漏召回和误召回。 - -> **注意**：Object 类型下的子属性不能为空，需点击对象行末图标新增子属性。GET 请求方法下的输入参数不支持 Object 类型。 - -### 从云市场导入 - -云市场提供丰富 API，可在云市场开通后导入至百炼插件列表。导入的插件为草稿状态，需测试、发布后使用。导入时系统自动填充出入参，但可能存在信息缺失，发布时需根据错误提示修正。 - -### 使用自定义插件 - -控制台内可将插件发布为 MCP 服务，再在[智能体应用](../concepts/agent-application.md)编排页面的 MCP 区块添加该服务；也可直接在应用管理页面的[智能体应用](../concepts/agent-application.md)编排中添加 MCP 服务。无鉴权插件可直接对话测试；用户级/服务级鉴权需在对话前配置鉴权 [Token](../concepts/token.md)；业务透传参数需配置变量值。从云市场导入的插件无需在对话页输入鉴权 [Token](../concepts/token.md)。 - -通过 API 调用时，若应用关联的插件存在业务透传参数或开启了用户级鉴权，需通过 `biz_params` 传递鉴权信息或透传参数。 +3. **API 调用**： + - Assistant API：在 `tools` 数组中声明工具 ID 及描述，模型自动规划调用； + - 智能体/工作流 API：通过 `biz_params` 传递业务透传参数或用户级鉴权 Token [详见 API 文档](https://help.aliyun.com/zh/model-studio/agent-and-workflow-application-api-reference)。 ## 限制和注意事项 -- 官方插件只能与位于相同[业务空间](../concepts/workspace.md)的[智能体应用](../concepts/agent-application.md)关联。 -- 每个[智能体应用](../concepts/agent-application.md)最多添加 10 个工具。 -- 只有已发布且启用状态、调试状态为成功的工具才能用于调用。 -- 删除插件会删除其下所有工具，调用该插件的应用会失效，操作不可撤回。 -- 编辑插件信息后立即生效；修改 URL、Header、鉴权信息可能影响工具调用，需重新测试并发布工具。 -- 工具信息修改后需重新测试并发布才能生效。 - -### 常见错误码 - -| 错误码 | 错误信息 | 说明 | -| --- | --- | --- | -| 130040 | xx 缺少参数描述信息 | 参数描述缺失，补充后重新发布 | -| 130022 | 保存工具信息异常/请检查示例参数是否正确 | 原因一：Object 类型参数子属性为空，需新增子属性；原因二：GET 请求下存在 Object 类型输入参数，需选择其他类型 | - -## 常见问题 - -**夸克搜索和联网搜索（enable_search）有什么区别？** - -- **夸克搜索插件**：模型直接调用插件执行搜索，将搜索结果以文本形式返回，可直接用于生成最终输出。 -- **联网搜索（enable_search）**：同样基于夸克搜索，但模型仅利用互联网信息丰富生成内容，不会完全依赖或返回互联网搜索结果。 +- **权限限制**：首次使用插件前，**必须完成 `AliyunServiceRoleForSFMAccessCloudAPI` 服务关联角色授权**，否则无法访问插件市场或调用任何插件 [详见官方和第三方插件文档](../../raw/application-user-guide/plug-in/plugins.md)。 +- **调用上限**：智能体应用最多支持添加 **10 个工具**；自定义插件中，`Object` 类型输入参数在 `GET` 请求下不被支持（仅 `POST` 允许）。 +- **功能边界**： + - `code_interpreter` 插件**不支持网络访问与本地文件上传**，可用依赖库已固化（如 `pandas`、`matplotlib`、`requests` 等）； + - `quark_search` 和 `github_search` 均**仅返回摘要、标题、链接，不支持访问网页或仓库详情页**； + - `text_to_image` 和 `quark_search` 为**限时免费，需单独申请开通**。 +- **调试要求**：自定义插件的工具必须经 **在线调试成功并发布为“已发布”状态** 后才能被应用调用；草稿或未启用状态的工具将导致调用失败。 +- **错误处理**：发布自定义工具时常见错误码 `130040`（参数描述缺失）、`130022`（Object 子属性为空或 GET 请求含 Object 参数）需严格按提示修正 [详见自定义插件错误码说明](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 ## 来源文档 -- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) - [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md) +- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) - [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md index 5cbdc3c9..c5f97c2d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md @@ -1,111 +1,57 @@ # prompt -阿里云百炼提供了一套完整的 Prompt 工程工具链，帮助开发者高效管理和优化提示词。核心能力包括 Prompt 模板（预置与自定义）、Prompt 自动优化、Prompt 样例库以及基于输入输出样例的 Prompt 反馈优化。这些功能覆盖了从模板创建、结构优化到少样本学习引导的完整流程，适用于文本生成、图片生成、智能客服等多种场景。 +Prompt 是百炼平台中用于引导大语言模型生成预期输出的核心指令载体。通过结构化设计、模板化管理、样例增强与自动优化等能力，开发者可系统性提升模型输出的准确性、一致性与可控性。所有 Prompt 相关功能均需在华北2（北京）地域使用，且依赖业务空间（Workspace）上下文。 -## Prompt 模板 +## 支持的模型/功能 -Prompt 模板将提示词的固定结构与动态变量分离，实现可复用的统一管理。模板分为**预置模板**和**自定义模板**两类，详见 [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md)。 +百炼平台提供多种 Prompt 相关能力，覆盖从基础指令构造到高级场景适配的全链路： -### 预置模板 +- **Prompt 模板**：支持预置模板（如营销文案生成、摘要抽取）和自定义模板（文本生成、图片生成），后者可通过控制台或 API 创建，并支持 ICIO、CRISPE、RASCEF 等工程框架辅助构建 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 +- **Prompt 样例库**：通过少样本学习注入高质量问答对，引导模型输出风格与格式一致的结果；但该功能已停止维护，官方明确建议迁移到 RAG 表格库 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 +- **Prompt 自动优化**：基于大模型对原始 Prompt 进行结构重组、角色设定、指令增强与安全边界注入，不计费且数据不用于训练 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 +- **Prompt 反馈优化**：基于用户提供的输入-输出样例（query-answer pairs）进行多轮评估与迭代优化，推荐使用 `qwen-max` 作为推理模型，效果优于纯文本自动优化 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 -由百炼平台提供，涵盖营销文案、摘要抽取、文案润色、商品评论等通用场景，已经过优化，效果稳定，无需额外开发即可通过控制台或 API 调用。预置模板不支持修改，但可通过"复制模板"创建自定义副本后编辑。 +> **注意**：文档 2 明确声明“Prompt样例库功能已不再维护”，而文档 1 和 3 中仍存在大量关于其创建、关联与调试的操作说明。实际开发中应以文档 2 的迁移指引为准，避免依赖已下线能力。 -### 自定义模板 +## 关键参数 -支持两种创建方式： +| 参数 | 说明 | 来源/约束 | +|------|------|-----------| +| `workspaceId` | 业务空间唯一标识，所有 Prompt 操作（模板获取、样例库关联、反馈优化）均需指定 | 必填，通过[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)获取 | +| `promptTemplateId` | 模板唯一 ID，用于 `GetPromptTemplate` 接口拉取内容 | 预置模板 ID 在控制台卡片中可见；自定义模板 ID 创建后生成 | +| `has_thoughts=true` | API 调用时启用样例检索过程日志（仅限已关联样例库的应用） | 仅影响响应中 `thoughts` 字段，非必需 | +| 召回片段数 | 单次请求注入上下文的样例数量，默认 5，最大 10 | 应用配置页可调，影响 Token 消耗与效果平衡 | +| 评测数据量 | Prompt 反馈优化中用于评估的 query-answer 对数量，建议 ≥20 条 | 数据越充分，优化效果越稳定 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) | -- **控制台创建**：在"提示词"页面直接创建，或从预置模板复制后修改。支持"自定义创建"和"基于 Prompt 工程创建"两种输入模式。 -- **API 创建**：通过 `CreatePromptTemplate` 接口创建，需要提供 `workspaceId`（[业务空间](../concepts/workspace.md) ID）。 +## 使用方式 -自定义模板支持文本生成和图片生成两种类型。文本生成模板可选择 ICIO、CRISPE、RASCEF 等 Prompt 工程框架进行结构化设计；图片生成模板支持分别定义正向和负向提示词。具体创建流程参见 [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 +### 控制台操作 +- **模板创建与管理**：进入「组件管理 > 提示词」页面，支持自定义创建或基于 Prompt 工程框架（如 ICIO）生成；图片生成模板需分别填写正向/负向 Prompt [原文标题](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 +- **自动优化**：在「提示词 > 自动优化」页面粘贴原始 Prompt，点击「优化」后可复制结果或「保存为模板」。 +- **反馈优化**：在「提示词 > 反馈优化」页面配置初始 Prompt、样例集（5–10 条）、评测集（≥20 条），启动多轮优化任务。 -### 模板使用方式 +### API 调用 +- 获取模板：调用 `GetPromptTemplate`，传入 `workspaceId` 和 `promptTemplateId`，返回含 `variables` 和 `content` 的 JSON 响应。 +- 创建模板：调用 `CreatePromptTemplate`，需指定 `name`、`type`（`text` 或 `image`）、`content`（文本模板）或 `positivePrompt`/`negativePrompt`（图片模板）。 +- 应用调用：若已关联样例库，可在请求体中设置 `has_thoughts: true` 查看检索详情。 -**控制台**：在模板卡片上点击"创建应用"，模板内容自动填充到[智能体应用](../concepts/agent-application.md)的提示词编辑框中。提示词最大支持 6144 个字符。 +## 限制和注意事项 -**API/SDK**：通过 `GetPromptTemplate` 接口拉取模板内容（需 `workspaceId` 和 `promptTemplateId`），将业务数据填入模板变量后生成最终 Prompt，再发送给目标模型。返回内容包含 `variables`（变量列表）、`content`（模板内容）等字段。 - -> **注意**：Prompt 模板功能目前仅适用于**华北2（北京）**地域。 - -## Prompt 自动优化 - -当手动编写高质量 Prompt 成本较高时，可使用自动优化功能。该功能利用大模型对原始 Prompt 进行分析和重写，优化策略包括： - -- **结构重组**：调整整体结构使其更符合逻辑 -- **角色扮演引导**：为模型设定明确的专家角色 -- **指令增强**：将模糊指令具体化、步骤化 -- **安全与边界注入**：增加输出格式、内容限制等边界条件 - -操作路径：**应用开发 > 组件管理 > 提示词 > 自动优化**。优化结果可直接复制使用或保存为模板。该功能**不计费**，且提交的数据不会被存储或用于模型训练。详见 [Prompt自动优化](../../raw/application-user-guide/prompt/optimize-prompt.md)。 - -## Prompt 反馈优化 - -相比普通自动优化，Prompt 反馈优化基于用户提供的**输入输出样例**进行多轮自动化评估和迭代，生成更贴合实际业务场景的 Prompt。其工作流程为： - -1. 选择推理模型（推荐千问-max） -2. 输入初始 Prompt（描述任务目标） -3. 上传样例数据（建议 5-10 条，每种场景至少 1 条） -4. 上传评测数据（建议至少 20 条，数据越多效果越好） -5. 系统自动进行多轮评测与优化 - -优化后的 Prompt 包含三部分：原始 Prompt、添加的样例（few-shot）、以及自动生成的内容提示（对分类边界等的补充说明）。优化结果可保存为模板或直接创建[智能体应用](../concepts/agent-application.md)。详见 [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 - -## Prompt 样例库 - -> **注意**：Prompt 样例库功能已不再维护，推荐将样例库数据迁移到 RAG 表格库中。 - -Prompt 样例库采用少样本学习（Few-shot learning）思路，从预定义的高质量问答对中检索相关样例注入模型上下文，引导模型生成更准确、风格更一致的回复。适用于智能客服、特定领域知识问答、格式化内容生成等场景。 - -### 使用限制 - -| 限制项 | 上限 | -|--------|------| -| 单个样例库容量 | 300 条样例 | -| 单应用关联样例库数 | 5 个 | -| 单次召回片段数 | 最多 10 个 | -| 批量导入文件大小 | 20MB（Excel） | -| 单次导入条数 | 100 条 | - -### 计费说明 - -样例库功能本身不收费，但启用后会增加大模型调用的 [Token](../concepts/token.md) 消耗。总输入 [Token](../concepts/token.md) 约等于：用户查询 [Token](../concepts/token.md) + 所有召回样例的总 [Token](../concepts/token.md) + 系统指令 [Token](../concepts/token.md)。 - -详见 [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 - -## Prompt 工程框架 - -百炼平台内置三种 Prompt 工程框架，可在创建自定义文本生成模板时选用： - -| 框架 | 组成要素 | 适用场景 | -|------|----------|----------| -| **ICIO** | 指令、背景信息、补充数据、输出格式 | 简单明确的任务，如数据分析、内容生成、文本摘要 | -| **CRISPE** | 角色与能力、背景信息、任务、输出风格、输出范围 | 需要 AI 扮演特定角色的交互，如客服、创意写作 | -| **RASCEF** | 角色、行动、步骤、上下文、示例、格式 | 多步骤复杂业务流程，如项目规划、战略分析 | - -## 常见问题 - -**使用 `GetPromptTemplate` 接口和直接在代码中拼接字符串有什么区别？** - -通过接口管理 Prompt 的优势在于：逻辑与内容分离（可在控制台更新 Prompt 无需重新部署代码）、集中管理与协作（团队共享和版本管理）、一致性保障（避免手动维护导致的不一致）。 - -**Prompt 自动优化失败的可能原因？** - -输入内容超出 [Token](../concepts/token.md) 限制、触发安全审核策略、或网络/服务临时不可用。 +- **地域限制**：所有 Prompt 功能仅支持华北2（北京）地域，跨地域调用将失败。 +- **容量限制**： + - 单个 Prompt 模板内容最大 6144 字符（控制台编辑框右下角实时计数）； + - 单个样例库最多 300 条样例（已停用，仅作历史参考）； + - 批量导入样例文件 ≤20MB，单次 ≤100 条。 +- **Token 成本**：启用样例库或反馈优化会显著增加输入 Token（样例内容 + 用户 query），需纳入成本预估 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 +- **变量语法**：模板中使用 `${variable}` 占位符，渲染时需确保变量名与 `GetPromptTemplate` 返回的 `variables` 数组严格匹配。 +- **安全合规**：自动优化服务不存储用户 Prompt 数据，亦不用于模型训练，符合阿里云数据隐私政策 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 ## 来源文档 - [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md) +- [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md) - [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt自动优化](../../raw/application-user-guide/prompt/optimize-prompt.md) -- [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md) - [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md index ccfe6250..33e0492e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md @@ -1,47 +1,42 @@ # release notes -本页汇总百炼平台的功能更新与模型上下架动态，帮助开发者快速跟踪平台能力演进、新模型可用性以及影响调用的变更（下线、降价、网关调整等）。内容分为两条主线：平台功能迭代（[模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)）与模型规格的上架/更新记录（[模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)）。建议在集成前先核对相关模型的最新状态与下线机制。 +百炼平台的 Release Notes 汇总了模型上下架、功能迭代、API 变更及平台能力演进等关键动态，面向开发者提供可落地的技术更新概览。内容覆盖模型支持范围、核心参数变更、调用方式升级、已知限制与兼容性注意事项。所有信息均基于平台近期正式发布版本，建议开发者结合自身场景关注模型生命周期状态与接口兼容性。 -## 两类更新说明 +## 支持的模型/功能 -- **平台功能更新**：记录计费方案、API 能力、模型调优/部署、知识库 RAG、多模态套件、接入工具（如 Codex、Kilo CLI）等模块的迭代，按年份和月份倒序排列。详见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)。 -- **模型上下架与更新**：记录各地域（如华北2/北京）新上架模型的类型、时间、服务部署范围（如「中国内地」）与模型规格。详见 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)。 +- **新增模型**：2026年7月起，华北2（北京）地域陆续上线 `qwen-audio-3.0-realtime-plus`（实时多模态）、`vidu/viduq3-ad_reference2video`（参考生视频）、`happyhorse-1.1-t2v`（文生视频）、`fun-music-v1`（音乐生成）、`Tripo/Tripo-H3.1`（3D生成）等数十款模型，覆盖语音、图像、视频、3D、音乐及全模态场景。详见 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)。 +- **模型能力扩展**：Qwen3.7系列全面增强多模态交互混合智能体能力；Kimi K2.7 Code 系列新增高速档位（`kimi/kimi-k2.7-code-highspeed`），推理速度提升5~6倍；GLM-5.1 支持200K上下文与128K最大输出；DeepSeek-V4-Pro 支持 `cached_token` 单价调整为1元/百万token（见[文档 2](../../raw/model-user-guide/release-notes/newly-released-models.md)）。 +- **功能模块上线**：6月新增知识检索服务与知识问答服务（支持多知识库联合检索与混合排序）；6月上线智能体托管运行时 API；5月起模型调优支持强化学习（RL）、0代码安全合规强化、视频/图像/视觉理解模型类型；4月起多模态交互开发套件覆盖 Android/iOS Lite、Linux C++、RTOS C 等全端 SDK。 -## 平台功能动态（要点） +> **注意**：文档 1 中提及“Qwen3-VL-8B-Instruct/Thinking 支持 SFT 调优”（2025年10月），但文档 2 中未列出该模型上架记录，且其命名与当前主流 Qwen3.5/Qwen3.6/Qwen3.7 系列不一致，建议以 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 中实际发布的模型列表为准，避免使用非公开快照模型。 -功能更新按「日期 / 功能模块 / 功能点 / 功能说明」组织，覆盖的主要方向包括： +## 关键参数 -- **计费与套餐**：团队版共享用量包（跨坐席共享 Credits）、Coding Plan Pro 首月特惠、上下文缓存降价等。 -- **API 能力**：Responses API 异步调用（`background=true` 提交长耗时任务并轮询）、文本生成 API 聚合 OpenAI Responses 与 Anthropic Messages 接口、异步任务支持 EventBridge 回调与 RocketMQ、临时 API Key 生成、API Key 加密存储。 -- **模型调优与部署**：新增强化学习（RL）训练（邀约制）、图像/视频/视觉理解（VL）模型定制训练、DPO 偏好训练、模型压缩（量化降精度）、模型导入（从 OSS 导入 LoRA）、按模型单元（MU）时长计费、PTU 长输入与前缀缓存。 -- **知识库 RAG**：知识检索服务、知识问答服务、知识库日志与监控（投递至 SLS）。 -- **接入工具与生态**：新增 Codex、Kilo CLI 客户端接入，官方 MCP 服务，多模态交互开发套件（Java/Android/iOS/Linux C++/RTOS C SDK）。 +- **计费模式**：模型部署支持按模型单元（MU）时长计费（自2025年10月起），适用于 `qwen-flash`/`qwen-plus` 等预置模型；`deepseek-v4-pro` 的 `cached_token` 单价明确为 **1 元/百万 token**（标准 `input_token` 不变）。 +- **上下文与输出**：GLM-5.1 支持 200K 上下文与 128K 最大输出；Qwen3.5-OCR 上下文扩展至 128K；Qwen3.7-Max 系列仅支持纯文本输入，默认开启思考模式，支持显式缓存。 +- **性能指标**：`qwen-audio-3.0-tts-flash` 首包延时 ≤200ms；`qwen3.6-flash` 系列在代码智能体基准中大幅超越前代；`wan2.7-r2v` 支持单张多宫格故事板一键生成剧本化视频。 -上述条目的完整时间线请查阅 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) 中的「功能动态」章节。 +## 使用方式 -## 模型上架动态（要点） +- **API 调用**： + - Responses API 新增异步调用模式（`background=true`），适用于长耗时任务； + - 异步任务支持通过事件总线 EventBridge 主动推送完成事件（HTTP 回调或 RocketMQ），替代轮询； + - 新增临时 API Key 生成机制，适用于不可信环境，规避永久密钥泄露风险； + - 智能体托管运行时、知识检索/问答、Prompt 工程、数据连接等模块均已提供完整 API 文档（参见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)）。 +- **SDK 与集成**： + - 多模态交互开发套件提供 Android/iOS Lite、Android、iOS、Linux C++、RTOS C 等 SDK； + - Spring AI Alibaba 框架已支持调用百炼智能体与工作流应用； + - Codex 终端 AI 编程助手、Kilo CLI 工具均完成百炼接入适配。 -模型上架记录覆盖多种模型类型，示例包括： +## 限制和注意事项 -- **推理/文本模型**：qwen3.7-max 系列、qwen3.6 系列（flash/plus/max-preview/27b）、DeepSeek-V4 系列（pro/flash）、Kimi K2.6/K2.7-code、GLM-5/5.1、MiMo-V2.5-Pro、Step 3.7 Flash 等第三方直供模型。 -- **多模态与生成**：Qwen-Image-2.0（文生图/图像编辑融合）、万相 2.7（文生/图生/参考生视频、视频编辑、图像生成与编辑）、HappyHorse（有声视频，3~15 秒、720P/1080P）、Vidu / 爱诗（Pixverse）视频系列、Tripo 3D 生成、Fun-Music 音乐生成。 -- **语音与翻译**：Qwen-Audio-3.0（realtime/tts，Flash 版首包延时 <200ms）、Fun-ASR（覆盖 30 语种、七大方言）、qwen3.5-livetranslate（识别 60 种语言、翻译为 29 种语言音频）。 -- **文字提取**：qwen3.5-ocr（128K 上下文、多轮对话、卡证信息抽取）。 - -## 关键字段与使用方式 - -- **服务部署范围**：模型上架条目会标注部署范围（如「中国内地」）与地域（如华北2/北京）。集成前需确认目标模型在所需地域可用。平台已新增美国、德国、日本等地域与部署范围。 -- **模型规格命名**：带日期后缀的规格（如 `qwen3.7-max-2026-05-20`、`wan2.7-t2v-2026-06-12`）为快照版本，能力通常与主版本一致，用于锁定行为、便于灰度与回滚。 -- **第三方直供模型**：使用 `厂商/模型` 形式的 ID（如 `kimi/kimi-k2.7-code`、`ZHIPU/GLM-5.1`、`vidu/viduq3-fast_reference2image`），调用文档参见各模型对应指南。 - -## 限制与注意事项 - -- **模型下线**：平台持续发布老旧/长尾模型下线与延期下线通知。下线规则与完整清单请以 [模型下线机制说明](https://help.aliyun.com/zh/model-studio/model-depreciation) 为准，及时迁移以避免调用中断。 -- **网关变更**：存在网关变更通告与业务空间专属推理 API 域名升级，可能影响既有调用地址，需关注公告并同步更新配置。 -- **免费额度**：启用「免费额度用完即停」后，新人免费额度耗尽将无法继续调用（返回错误 `code: AllocationQuota.FreeTierOnly`），避免产生额外费用。 -- **能力约束**：部分推理模型（如 qwen3.6-max-preview、qwen3.7-max 系列）仅支持纯文本输入、默认或仅支持思考模式，不支持图像与视频输入，接入前需核对模态支持。 - -> **注意**：两篇文档的日期跨度较大且更新频繁，部分记录已进入 2026 年。计费、模型可用性与下线时间以官方最新公告和控制台实际状态为准；本页为聚合快照，可能滞后于 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) 与 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 的原始内容。 +- **模型下线**：2026年7月起分批下线部分老旧及长尾模型（如7月10日、7月9日通知），同时存在延期下线安排（7月6日通知）。具体清单与机制请严格参照 [模型下线机制说明](../../raw/model-user-guide/release-notes/model-release-notes.md)。 +- **地域与部署**：新模型（如 `qwen-audio-3.0-realtime-plus`、`vidu` 系列）当前仅部署于华北2（北京）地域，国际站用户需确认服务可用性；6月12日新增美国、德国、日本地域，但模型覆盖需单独验证。 +- **功能约束**： + - `qwen3.6-max-preview` 明确不支持图像与视频输入； + - `kimi/kimi-k2.7-code` 仅支持思考模式； + - `qwen3.5-omni-plus` 为全模态模型，但具体输入模态组合需查阅对应 API 文档； + - 免费额度用完即停功能启用后，将返回错误码 `AllocationQuota.FreeTierOnly`，需在客户端做好容错处理。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md index 4f3f6d6f..03c8ebbc 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md @@ -1,188 +1,67 @@ # security and compliance -阿里云百炼围绕"身份权限、内容安全、合规备案、传输加密、私网与安全存储"五个维度构建安全合规体系，覆盖从控制台到 API 调用、从公网到 VPC 私网的完整链路。本文面向开发者，按"权限与身份—内容安全—合规备案—传输加密—[私网访问](../concepts/vpc-private-access.md)—安全存储"的顺序梳理关键能力、参数与注意事项。 - -## 身份与权限管理 - -百炼的权限管理基于[业务空间](../concepts/workspace.md)（workspace）这一最小管理单元，支持控制台页面级与模型级的多维度权限控制，满足多地域、多用户的复杂组织架构需求，详见 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)。 - -### 三种角色 - -- **超级管理员**：阿里云主账号，或拥有 `AliyunBailianFullAccess` 系统策略的 RAM 用户，可跨空间统一管理用户权限、空间可用模型、模型限流和 [API Key](../concepts/api-key.md)。建议 AI 安全护栏、模型监控、应用观测等功能使用主账号一次性开通。 -- **[业务空间](../concepts/workspace.md)管理员**：拥有访问某个[业务空间](../concepts/workspace.md)"权限管理"页面的 RAM 用户，可管理该空间内的用户与资源，"管理员"权限包含该空间所有页面的访问权限。 -- **普通用户**：根据分配的权限使用资源，不能管理用户或模型授权。 - -[业务空间](../concepts/workspace.md)按地理区域划分，**单个[业务空间](../concepts/workspace.md)不能跨地域存在**，各地域的默认[业务空间](../concepts/workspace.md)也是不同的空间。 - -### 模型与 [API Key](../concepts/api-key.md) 权限 - -[业务空间](../concepts/workspace.md)内可对模型进行三类精细化授权，**默认[业务空间](../concepts/workspace.md)无法设置这些限制**（所有模型均可调用、调优、部署）： - -| 权限项 | 控制范围 | 配置入口 | -| --- | --- | --- | -| 限制模型调用 | 是否可调用（控制台 & API）+ 请求数限流 + [Token](../concepts/token.md) 限流 | 模型列表 → 模型调用列开关 + 当前空间限流列 | -| 限制模型训练 | 是否可调优（控制台 & API）及调优后部署 | 模型列表 → 模型授权 → 模型训练列 | -| 限制[模型部署](../concepts/model-deployment.md) | 是否可直接部署 | 模型列表 → 模型授权 → [模型部署](../concepts/model-deployment.md)列 | - -单个 [API Key](../concepts/api-key.md) 只能归属一个地域内的一个[业务空间](../concepts/workspace.md)和一个用户，且不能转移。[API Key](../concepts/api-key.md) 的可调用功能与模型限流与**归属[业务空间](../concepts/workspace.md)**的权限保持一致，不受用户控制台权限影响，也无需为不同模型（文生文、文生图、语音合成）创建不同 [API Key](../concepts/api-key.md)。自 2026 年 3 月 25 日起，华北2（北京）地域所有新创建的 [API Key](../concepts/api-key.md) 均归属主账号，并支持设置 IP 访问白名单。 - -> **注意**：[API Key](../concepts/api-key.md) 的有效性受账号操作影响——将 RAM 账号移出[业务空间](../concepts/workspace.md)会使其 [API Key](../concepts/api-key.md) 失效（重新加入后恢复），而在 RAM 控制台删除账号/角色则会使 [API Key](../concepts/api-key.md) 永久失效、不可恢复。 - -### OpenAPI 接口权限 - -RAM 用户默认无权调用百炼应用的数据、[知识库](../concepts/knowledge-base.md)、Prompt 工程、长期记忆等 Open API。需阿里云主账号在 RAM 控制台添加以下系统策略之一： - -- `AliyunBailianDataFullAccess`：可调用应用 API 目录下的所有 API。 -- `AliyunBailianDataReadOnlyAccess`：仅可调用只读类 API（如 `DescribeFile`、`GetIndexJobStatus`）。 - -### 生产环境实践 - -- **空间规划**：推荐按环境（dev/test/prod）划分[业务空间](../concepts/workspace.md)实现隔离，或按业务线划分便于权限与成本管理。 -- **限流策略**：将主账号总配额按比例分配给各[业务空间](../concepts/workspace.md)并预留缓冲。例如总配额 1000 QPM，可分配 prod 600 / test 200 / dev 100，预留 100。 - -## 内容安全：AI 安全护栏 - -大模型输入输出可能包含涉黄、涉政、广告等敏感内容。除模型自有合规检查外，百炼支持接入 AI 安全护栏服务，进一步识别违规信息，保障安全合规，详见 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 - -### 开通步骤 - -1. **开通内容审核服务**：访问 AI 安全护栏购买页面，创建服务关联角色并购买。 -2. **授权内容安全设置**：在百炼"安全管理"页面单击"去授权"开启内容安全设置。若已显示"已开通（不可取消）"及《自建安全机制承诺函》全文，可跳过此步。 -3. **设置请求头**：调用百炼时在请求头设置 `X-DashScope-DataInspection`，开启输入输出检测： - -```json -{"X-DashScope-DataInspection": {"input": "cip", "output": "cip"}} -``` - -目前支持文本和图片类型的模型，模型与护栏服务的对应关系及[计费](../concepts/billing.md)请参见官方说明。命中护栏时返回 `data_inspection_failed`（DashScope 为 `DataInspectionFailed`，HTTP 400），提示输入可能包含不当内容。 - -## 合规资质与备案 - -### 资质与隐私 - -百炼以无保留意见通过 SOC 2 审计，安全、可用性、保密性控制符合国际标准。阿里云承诺**不会将您的数据用于模型训练**，应用构建与模型训练过程中的传输数据均经过 AES-256 加密。根据法律法规，百炼会存储模型与应用调用产生的数据，详见《阿里云百炼服务协议》中的数据处理、隐私和安全条款，参考 [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md)。 - -### 模型备案信息公示 - -百炼公示所接入大模型的算法备案号与大模型备案号，详见 [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)。部分示例如下： - -| 模型 | 算法名称 | 算法备案号 | -| --- | --- | --- | -| 千问 | 达摩院交互式多能型合成算法 | 网信算备330110507206401230035号 | -| 万相 | 达摩院图像合成算法 | 网信算备330110507206401230027号 | -| 万相 | 通义万相视频生成算法 | 网信算备330106003156001240091号 | -| DeepSeek | DeepSeek 大语言模型算法 | 网信算备110108970550101240011号 | -| 智谱 AI | 智谱交互式内容生成算法 | 网信算备110108105858001230027号 | - -大模型备案号（如通义千问 `ZheJiang-TongYiQianWen-20230901`、DeepSeek `Beijing-DeepseekChat-202404280016` 等）同样公示。第三方模型备案信息由提供方负责，阿里云百炼不作额外承诺。 - -### 应用上架及合规备案 - -接入千问、万相等模型的应用/小程序上架前，需依《生成式人工智能服务管理暂行办法》完成合规备案，详见 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 - -典型场景与所需材料： - -| 场景 | 是否面向 C 端 | 舆论/动员能力 | 主要材料 | -| --- | --- | --- | --- | -| 场景 1 | 是 | 不具有 | 大模型算法备案信息 + 应用主体与阿里云的合作协议 | -| 场景 2 | 是 | 具有 | 场景 1 全部材料 + 企业自主安全评估报告 + 企业自主算法备案 | -| 场景 3 | 企业内部 | — | 关注数据安全、保密等内部合规要求 | - -算法备案信息查询步骤：打开互联网信息服务算法备案系统（beian.cac.gov.cn），以"备案编号"为搜索类别输入对应备案号（如千问 `网信算备330110507206401230035号`），截图保存查询页面完整内容。合作协议（需含算法名称、应用产品或备案编号）请联系商务经理获取。 - -> **注意**：即使使用阿里云提供的模型及备案信息，应用/小程序开发者仍是法规定义的"服务提供者"，需独立履行内容审核、用户保护、数据安全、标识规范等全部法定义务。备案号应以算法备案系统实时查询结果为准，建议定期核验。 - -## 传输加密：以加密方式接入模型推理 - -当请求涉及敏感信息或经公网传输时，可对请求体 `input` 字段值加密，防止传输过程中被窃听或篡改，详见 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 - -### 加解密机制 - -采用混合加密：数据由 AES 对称算法加密，AES 密钥通过 RSA 非对称加密实现安全传输。流程为：生成 AES 密钥 → 用 AES 密钥加密 `input` → 用 RSA 公钥加密 AES 密钥 → 将加密后的 `input` 与密钥信息（封装在 `X-DashScope-EncryptionKey` 请求头）传入百炼 → 平台推理链路全程加密、解密数据并用相同 AES 密钥加密答案返回 → 用户侧用 AES 密钥解密响应。 - -### 两种接入方式 - -**1. [DashScope SDK](../concepts/dashscope-sdk.md)（自动加密·开箱即用）** - -仅支持 Java 和 Python，不支持自定义密钥： - -- Java SDK：`GenerationParam.builder().enableEncrypt(true)` -- Python SDK：`dashscope.Generation.call(..., enable_encryption=True)` - -SDK 自动完成加解密，响应为明文，无需手动处理。 - -**2. HTTP 调用（手动密钥管理）** - -需额外完成三步：添加 `X-DashScope-EncryptionKey` 请求头、对 `input` 内容加密、对响应数据解密。该请求头为 JSON 字符串，含 `public_key_id`（公钥 ID）、`encrypt_key`（RSA 公钥加密后的 AES 密钥）、`iv`（初始向量）三个字段。AES 密钥长度支持 128/192/256 位（32 字节），长度越长安全性越高但开销越大。 - -> **注意**：手动加密调用仅适用于 DashScope 的 Endpoint，OpenAI 兼容（Chat Completions API 和 Responses API）的 Endpoint **不支持**此加密机制。 - -### 获取 RSA 公钥 - -加密前需调用接口获取当前最新公钥 ID 及其值，详见 [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md)。 - -- 接口：`GET /api/v1/public-keys/latest` -- 鉴权：`Authorization: Bearer ` -- 返回：`request_id`、`data.public_key`（RSA 公钥值）、`data.public_key_id`（公钥 ID） -- 前提：已开通百炼服务并获得 API-KEY，建议配置到环境变量以降低泄漏风险。 - -## [私网访问](../concepts/vpc-private-access.md)：通过 PrivateLink 终端节点访问 API - -为在 VPC 内直接调用百炼模型/应用 API 且流量不经公网，可创建私网终端节点（PrivateLink），将通信限制在阿里云内网，详见 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 - -### 工作原理与地域 - -终端节点连接为**单向设计**，仅允许 VPC 内资源主动访问百炼，百炼无法反向访问 VPC 内资源。百炼公共云服务所在地域：华北2（北京）、新加坡。**美国（弗吉尼亚）地域暂不支持[私网访问](../concepts/vpc-private-access.md)**。 - -### 配置步骤 - -1. **创建接口终端节点**：在终端节点控制台创建，终端节点服务选择"阿里云服务"并筛选 `com.aliyuncs.dashscope`，开启"自定义服务域名"。建议至少选择两个不同可用区的交换机以实现高可用；安全组入方向需允许 80（http）、443（https）。 -2. **获取终端节点服务域名**：默认服务域名格式 `ep-{实例ID}.privatelink.aliyuncs.com`（仅 HTTP）；自定义服务域名格式 `vpc-{实例ID}.{地域ID}.dashscope.aliyuncs.com`（支持 HTTPS）。 -3. **调用验证**：将 API base_url 中的域名替换为终端节点服务域名后在对应 VPC 发起调用。支持 HTTP/curl、OpenAI Python SDK、DashScope Python SDK（建议 ≥1.14.0）、DashScope Java SDK（建议 ≥2.12.0）。 - -### 跨地域访问 - -- **同境内或同境外跨地域**（如华东1杭州 VPC → 华北2北京百炼）：推荐启用跨地域端点。 -- **跨境跨地域**（中国内地与境外之间，如华北2北京 VPC → 新加坡百炼）：通过 CEN 跨地域 VPC 互通。 - -## 安全存储[业务空间](../concepts/workspace.md) - -安全存储[业务空间](../concepts/workspace.md)通过反向终端节点与 VPC 私网连接，让部署其中的应用访问同 VPC 下的 ElasticSearch、AnalyticDB（ADB）、OSS 等云组件，避免公网访问风险。该能力需联系商务人员申请开通。 - -### 配置流程总览 - -| 步骤 | 说明 | 参考 | -| --- | --- | --- | -| 1. 创建安全存储[业务空间](../concepts/workspace.md) | [业务空间](../concepts/workspace.md)管理 → 新增[业务空间](../concepts/workspace.md)，空间类型选"安全存储空间" | [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) | -| 2. 创建反向终端节点 | 终端节点控制台创建反向终端节点，终端节点服务选描述为"百炼公共云生产环境-北京站点-安全存储空间专网通道接入点"的服务（VPC NAT 网关、反向、IPv4），配置 VPC/安全组/可用区与交换机 | 同上 | -| 3. 在百炼确认连接 | [业务空间](../concepts/workspace.md)管理 → 管理安全存储空间 → 选择终端节点 → 连接，等待状态变为"已连接" | 同上 | -| 4. 配置可用区 IP | 创建 MSE 云原生网关（2核4G、2 节点、启用 TLS 硬件加速、私网、至少两可用区），获取 NLB 各可用区 VIP 与交换机网段，在百炼配置对应可用区 IP，并将 VIP 加入反向终端节点安全组入方向（全部端口） | [配置可用区IP](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) | -| 5. 配置私有网络资源 | 配置 OSS（Bucket 名 `bailian-safe-workspace-oss-access`、特定标签、CORS 来源 `*bailian.console.aliyun.com`）、ADB（6.0 标准版、高可用版、开启向量引擎优化）、ElasticSearch（7.10、内核增强版、两可用区，交换机网段加入 VPC 私网访问白名单） | [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) | -| 6. 配置 MSE 云原生网关 | 在网关创建 DNS 域名服务（指向 ES 私网地址/端口，TLS 关闭）、创建路由（域名为 ES 域名、路径 `/`、单服务），然后回百炼"资源配置"页激活安全存储[业务空间](../concepts/workspace.md) | [配置MSE云原生网关](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) | - -### 关键约束与注意事项 - -- 专有网络地域须为华北2（北京），可用区须在 G/H/L（或 ADB 的 G/H/I）中按要求选择，每个可用区至少一个交换机；反向终端节点的安全组不要放入其他云组件、无需配置出入网规则。 -- 阿里云百炼只能访问客户授权过且带特定标签（标签名 `bailian-safe-workspace-oss-access`，标签值 `ReadAndWrite`）的 OSS Bucket。 -- ADB 配置需授权服务关联角色 `AliyunServiceRoleForSFMAccessADB`（策略 `AliyunServiceRolePolicyForSFMAccessADB`）。 -- **资源不可中断**：OSS Bucket 停止服务会导致安全存储空间、[知识库](../concepts/knowledge-base.md)、审计日志、历史记录等模块不可用，恢复 Bucket 后可恢复；但 Bucket 被释放会造成安全存储空间不可用且**无法恢复**，需重建。ES 停止[计费](../concepts/billing.md)/被释放的后果与 OSS 相同。 -- 激活前安全存储空间不可用，激活成功后才可用。 - -## 限制与注意事项汇总 - -- 默认[业务空间](../concepts/workspace.md)无法设置模型调用/训练/部署限制，所有模型均可调用、调优、部署，且无法限流。 -- [API Key](../concepts/api-key.md) 不可跨地域、跨业务空间、跨用户转移；账号移出空间会使其 [API Key](../concepts/api-key.md) 失效（重新加入恢复），删除账号/角色则永久失效。 -- AI 安全护栏目前仅支持文本和图片类型模型。 -- [DashScope SDK](../concepts/dashscope-sdk.md) 自动加密仅支持 Java/Python 且不支持自定义密钥；HTTP 手动加密仅适用于 DashScope Endpoint，OpenAI 兼容 Endpoint 不支持。 -- PrivateLink 私网访问美国（弗吉尼亚）地域暂不支持；跨地域访问需区分同境内/同境外与跨境两种方式。 -- 安全存储业务空间的 OSS/ES 等底层资源一旦释放，安全存储空间不可恢复，需重建。 -- 模型与应用的合规备案信息应以算法备案系统实时查询结果为准，建议定期核验；开发者作为"服务提供者"需独立承担全部法律责任。 +阿里云百炼平台提供多层次的安全与合规能力，覆盖模型调用、数据传输、存储隔离及监管备案等关键环节。开发者可通过权限管理、AI安全护栏、端到端加密、私网访问及合规资质材料获取等机制，满足企业级安全要求与国内生成式AI监管规范（如《生成式人工智能服务管理暂行办法》）。所有能力均基于阿里云基础设施的合规底座（如SOC 2）构建，确保数据隐私、传输安全与责任可追溯。 + +## 支持的模型/功能 + +- **AI安全护栏服务**：支持对文本和图片类模型的输入输出内容进行实时合规检测（涉黄、涉政、广告等），需显式启用 `X-DashScope-DataInspection` 请求头 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 +- **加密推理通道**：支持通过AES-RSA混合加密机制对请求体 `input` 字段加密，防止公网传输中敏感信息泄露；该功能仅适用于 DashScope Endpoint，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)不支持 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 +- **私网访问能力**： + - 普通业务空间：支持通过 PrivateLink 创建**接口终端节点**，实现 VPC 内资源直连百炼 API（限华北2北京、新加坡地域）[通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 + - 安全存储业务空间：需配置**反向终端节点** + MSE 网关 + 私有云资源（OSS/ADB/ES），构建完全隔离的数据存储与处理环境 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 +- **模型备案信息**：所有接入百炼的主流大模型（如千问、万相、DeepSeek、Moonshot 等）均已取得国家网信办算法备案号与大模型备案号，可在控制台或[模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)页面查询。 + +> **注意**：文档 8 与文档 9 描述的私网访问路径存在适用范围差异——前者面向通用模型/API调用，后者专用于**安全存储业务空间**（需商务开通），二者网络架构、终端节点类型（接口 vs 反向）及依赖组件（无MSE vs 必须MSE）均不同，不可混用。 + +## 关键参数 + +| 参数名 | 用途 | 来源/说明 | +|--------|------|-----------| +| `X-DashScope-DataInspection` | 启用AI安全护栏，值为 `{"input":"cip","output":"cip"}` | [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) | +| `X-DashScope-EncryptionKey` | 加密调用必需请求头，含 `public_key_id`、`encrypt_key`（RSA加密的AES密钥）、`iv` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `enable_encryption=True` (Python) / `enableEncrypt(true)` (Java) | DashScope SDK 开箱即用加密开关 | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `public_key_id`, `public_key` | 通过 `/api/v1/public-keys/latest` 接口获取，用于客户端RSA加密AES密钥 | [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) | + +## 使用方式 + +1. **权限控制** + - 超级管理员通过全局管理菜单（如[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)）统一管控多业务空间模型授权、限流与API Key； + - 业务空间管理员在对应空间内配置模型调用/训练/部署权限，并管理用户控制台页面权限； + - API Key 继承归属业务空间的模型权限，不受用户控制台权限影响。 + +2. **启用AI安全护栏** + - 在[安全管理](https://bailian.console.aliyun.com/?globalset=1#/efm/global_set)页面完成服务授权； + - 所有调用请求必须携带 `X-DashScope-DataInspection` 头，否则不触发审核。 + +3. **加密推理调用** + - **SDK方式（推荐）**：Python/Java SDK 设置 `enable_encryption=True` 或 `.enableEncrypt(true)`，自动处理加解密； + - **HTTP方式**： + a) 调用 `/api/v1/public-keys/latest` 获取公钥； + b) 生成AES密钥并加密 `input` 字段； + c) 用RSA公钥加密AES密钥，构造 `X-DashScope-EncryptionKey` 头； + d) 解密响应体获取明文结果。 + +4. **私网访问配置** + - **普通场景**：在VPC中创建接口终端节点，关联 `com.aliyuncs.dashscope` 服务，替换API Base URL为终端节点域名； + - **安全存储场景**： + a) 创建反向终端节点并确认连接； + b) 配置MSE网关、可用区VIP及交换机网段； + c) 授权并绑定OSS/ADB/ES等私有云资源； + d) 激活业务空间。 + +## 限制和注意事项 + +- **地域限制**：私网访问仅支持华北2（北京）和新加坡地域；美国（弗吉尼亚）地域暂不支持 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 +- **API Key 约束**：单个API Key仅归属一个地域内的一个业务空间和一个用户，不可转移；自2026年3月25日起，华北2（北京）新创建的API Key均归属主账号 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)。 +- **加密调用兼容性**：仅 DashScope Endpoint 支持加密，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)（`/compatible-mode/v1`）不支持 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 +- **安全存储业务空间依赖**：OSS Bucket 若被释放，将导致安全存储空间**不可恢复**；ADB/ES 若停止计费或释放，相关模块（知识库、审计日志等）将不可用 [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md)。 +- **备案责任主体**：使用百炼模型的应用开发者是《生成式人工智能服务管理暂行办法》定义的“服务提供者”，须独立承担内容审核、用户保护、算法备案等全部法定义务，阿里云仅提供模型及备案信息支持 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 ## 来源文档 - [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md) -- [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) - [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) +- [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) - [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md) - [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) @@ -194,21 +73,3 @@ SDK 自动完成加解密，响应为明文，无需手动处理。 - [配置MSE云原生网关](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md index e92a5b80..bb115aed 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md @@ -1,134 +1,47 @@ # skill -Skill 是百炼平台[智能体应用](../concepts/agent-application.md)的可扩展能力包，让智能体在对话中自动识别并处理特定类型的任务，如文件处理、数据分析等，无需额外编码或接入外部工具。详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)。 +Skill 是百炼平台提供的可插拔能力包，用于扩展智能体在对话中自动处理特定任务的能力（如文件解析、数据清洗等），无需额外编码或工具集成。开发者可通过官方 Skill 快速启用通用能力，或通过自定义 ZIP 包构建业务专属 Skill。所有 Skill 均由智能体基于 `description` 语义匹配自动调用，调用准确性高度依赖元信息描述质量。 -## Skill 类型 +## 支持的模型/功能 -百炼提供两类 Skill： +- **官方 Skill**：平台预置、开箱即用的通用能力，覆盖 `.xlsx`/`.csv`/`.tsv` 等文件处理、PDF 文本提取、图像 OCR 等场景，由平台统一维护和更新，已添加的智能体会自动升级至最新版本。 +- **自定义 Skill**：通过上传符合规范的 ZIP 包实现，适用于官方 Skill 未覆盖的垂直场景（如行业专用格式解析、私有 API 封装等）。ZIP 包必须包含根目录下的 `SKILL.md` 文件，并满足 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中定义的结构与字段要求。 +- 所有 Skill 均不依赖特定大模型，其调用逻辑由百炼底层调度引擎根据用户输入语义与 `description` 匹配决定，与所选推理模型无关。 -- **官方 Skill**：平台预置的通用 Skill，覆盖常见文件处理场景，由平台统一维护，添加后即可使用，且会自动更新到最新版本。官方 Skill 列表持续更新，可在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面查看最新清单。 -- **自定义 Skill**：通过上传 ZIP 技能包创建，适用于官方 Skill 未覆盖的业务场景（如特定行业数据处理、自定义文件格式解析等）。更新方式为重新上传同名 ZIP 包生成新版本。 +## 关键参数 -## 创建自定义 Skill - -当官方 Skill 无法满足业务需求时，可上传 ZIP 技能包创建自定义 Skill。ZIP 包需满足以下要求，详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)： - -| 要求 | 说明 | -| --- | --- | -| 必须包含 SKILL.md | ZIP 包根目录下必须有 `SKILL.md` 文件，定义 Skill 元信息 | -| 大小限制 | 整个 ZIP 包不超过 10 MB | -| 名称唯一 | SKILL.md 中的 `name` 字段不可与当前账号下已有 Skill 重名 | - -### SKILL.md 编写规范 - -`SKILL.md` 使用 YAML 格式定义 Skill 的名称和描述： - -```yaml -name: my-custom-skill -description: "Skill 的功能描述，包含触发条件、适用场景和不适用场景。" -``` +关键参数全部定义在 ZIP 包根目录的 `SKILL.md` 文件中，采用 YAML 格式： | 字段 | 必填 | 说明 | -| --- | --- | --- | -| name | 是 | Skill 的唯一标识名称，建议使用小写英文和连字符（如 `data-cleaner`、`invoice-parser`） | -| description | 是 | 描述 Skill 的触发条件和处理能力。智能体据此判断是否调用该 Skill，描述质量直接影响调用准确率 | - -### description 编写建议 - -description 的质量决定了智能体调用 Skill 的准确性，建议包含： - -1. **适用的输入类型**：明确 Skill 处理的文件格式或数据类型。 -2. **支持的操作**：列出可执行的具体操作。 -3. **触发关键词**：用户对话中可能出现的、应触发该 Skill 的关键词或表达方式。 -4. **不适用的场景**：标注不应触发该 Skill 的场景，避免误调用。 - -以下为官方 xlsx Skill 的 SKILL.md 示例： - -```yaml -name: xlsx -description: "Use this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like \"the xlsx in my downloads\") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved." -``` - -该示例明确了支持的文件格式（.xlsx、.xlsm、.csv、.tsv）、适用操作（读取、编辑、创建、格式转换、数据清洗等）、触发场景（用户提到文件名或路径时也应触发），并标注了不适用场景（产出物为 Word、HTML、Python 脚本等）。 - -### 上传并创建 - -1. 在控制台左侧导航栏，选择 **组件** > **Skill 管理**。 -2. 点击右上角 **自定义 Skill** 按钮。 -3. 在弹窗中点击上传区域选择 ZIP 文件，或直接将文件拖拽到上传区域。 -4. 点击 **确认** 提交。 - -提交后系统自动审查 Skill 内容，预计耗时约 2 分钟。审查通过后 Skill 出现在 **自定义 Skill** 标签页中，可添加到[智能体应用](../concepts/agent-application.md)；未通过则根据提示修改 SKILL.md 后重新上传。 - -### 更新自定义 Skill - -重新上传同名 Skill 的 ZIP 包时，系统会创建新版本，流程与首次创建一致： - -1. 修改本地 ZIP 包内容（如更新 SKILL.md 中的 description）。 -2. 在 **自定义 Skill** 标签页重新上传 ZIP 包。 -3. 审查通过后，已添加该 Skill 的智能体会自动使用最新版本。 - -## 添加 Skill 到智能体 - -添加后，智能体在对话中遇到匹配 Skill 描述的任务时会自动调用该 Skill。支持两种添加方式： - -**方式一：从 Skill 详情页添加** - -1. 在控制台左侧导航栏选择 **组件** > **Skill 管理**，点击目标 Skill 卡片进入详情页。 -2. 点击右上角 **添加到智能体**。 -3. 在应用列表中选择目标应用，确认添加。 +|------|------|------| +| `name` | 是 | Skill 唯一标识符，仅允许小写字母、数字和连字符（如 `invoice-parser`），同一账号下不可重复；该字段也作为版本管理的命名依据。 | +| `description` | 是 | **决定 Skill 是否被正确调用的核心字段**。需明确说明适用输入类型、支持操作、典型触发关键词及明确排除的不适用场景。描述质量直接影响匹配准确率，详见 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中的编写建议与完整示例。 | -**方式二：在应用配置中添加** +> **注意**：`description` 中若未声明“不适用场景”，可能导致误触发；例如 xlsx Skill 明确排除产出 Word 或 HTML 的场景，此约束在 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 的示例中有严格体现，实际编写时必须遵循。 -1. 进入目标[智能体应用](../concepts/agent-application.md)的 **应用配置** 页面。 -2. 在左侧配置面板找到 **技能** 区域，点击 Skill 右侧的加号。 -3. 从 Skill 列表中选择需要的 Skill，确认添加。 +## 使用方式 -## 查看 Skill 详情 +1. **创建 Skill** + - 官方 Skill：直接在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面查看并添加，无需配置。 + - 自定义 Skill：按 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 要求准备 ZIP 包（含 `SKILL.md`，≤10 MB），在控制台 **组件 > Skill 管理 > 自定义 Skill** 中上传，系统约 2 分钟完成审查。 -在 **组件** > **Skill 管理** 的 **官方 Skill** 或 **自定义 Skill** 标签页中点击目标 Skill 卡片进入详情页，详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)。详情页包含两个标签： +2. **添加到智能体** + - 方式一：在 Skill 详情页点击 **添加到智能体**，选择目标应用。 + - 方式二：进入智能体 **应用配置 > 技能** 区域，点击对应 Skill 右侧加号添加。 -- **概览**：展示 Skill 名称、当前版本号、功能描述和属性信息，可通过版本下拉框切换查看历史版本。 -- **更新记录**：展示该 Skill 全部版本的发布时间和变更内容。 - -官方 Skill 由平台统一维护和更新，已添加到智能体的官方 Skill 会自动使用最新版本；自定义 Skill 通过重新上传同名 ZIP 包更新版本。 - -## 测试 Skill 效果 - -添加 Skill 后，可在应用配置页面右侧的对话窗格中测试效果。例如发送： - -``` -帮我创建一个包含本月销售数据的表格，按地区分列统计 -``` - -智能体将调用 xlsx Skill，生成包含分列统计的 .xlsx 文件并提供下载。 +3. **测试与验证** + 在应用配置页右侧对话窗格中发送典型指令（如 `帮我清洗这份 CSV 数据，删除重复行并导出`），观察是否触发预期 Skill 并返回正确结果。 ## 限制和注意事项 -- ZIP 包整体大小上限为 10 MB，超出将无法上传。 -- 自定义 Skill 的 `name` 字段在当前账号下必须唯一，重名会导致创建失败。 -- description 编写质量直接决定智能体调用 Skill 的准确率，务必明确触发条件、适用操作和不适用场景。 -- 上传后需等待约 2 分钟的系统审查，审查未通过需修改后重新上传。 -- 官方 Skill 自动更新到最新版本；自定义 Skill 需手动重新上传 ZIP 包才能升级。 +- ZIP 包大小上限为 **10 MB**，超限将导致上传失败。 +- `name` 字段全局唯一（同账号内），重名上传会拒绝，而非覆盖。 +- 自定义 Skill 版本更新需重新上传同名 ZIP 包，旧版本仍保留在历史记录中，但已添加该 Skill 的智能体会**自动切换至最新通过审查的版本**。 +- 官方 Skill 的 `description` 由平台维护，开发者不可修改；若发现官方 Skill 行为与文档描述不符，应以控制台实时展示的描述为准。 +- Skill 调用完全基于 `description` 的语义理解，**不支持正则匹配、硬编码关键词或条件分支逻辑**；复杂业务规则需在 Skill 内部代码中实现，而非依赖 `description` 控制流。 ## 来源文档 - [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md) - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md index 9daa29d9..1eafeb69 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md @@ -1,58 +1,38 @@ # start using -阿里云百炼提供零代码方式快速构建基于私有知识的问答应用，同时持续迭代应用、[知识库](../concepts/knowledge-base.md)、[工作流](../concepts/workflow.md)等核心能力。本页汇总从创建第一个[智能体应用](../concepts/agent-application.md)到跟踪功能动态所需的关键信息，帮助开发者快速上手并了解平台最新能力。 +阿里云百炼平台提供低门槛、高灵活性的 AI 应用构建能力，支持零代码快速搭建私有知识问答应用，也支持高代码深度定制。开发者可通过控制台可视化配置或 API 编程方式接入模型、知识库、[长期记忆](../concepts/long-term-memory.md)等核心能力，适用于从原型验证到生产部署的全周期场景。本文档聚焦“开始使用”路径，梳理关键能力、参数与约束，帮助开发者高效启动。 -## 快速构建私有知识问答应用 +## 支持的模型/功能 -借助百炼的[智能体应用](../concepts/agent-application.md)构建能力，可在约 5 分钟内零代码完成一个能回答私有领域问题的大模型问答应用，完整流程见 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。整体分为三步： +- **基础模型**：智能体应用和工作流应用均支持千问系列（如 `qwen-max`）、QwQ 系列（如 `qwq-plus`、`qwq-32b`）及 DeepSeek 系列模型；其中 QwQ 模型具备强推理能力，输出含显式思考链 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 +- **[多模态能力](../concepts/multi-modal.md)**：`qwen-vl-plus-latest`、`qwen-vl-plus-2025-01-25` 等视觉语言模型支持图文理解与生成；知识库支持导入图片、音视频文件，并启用“多模态回复增强”开关以解析图表内容 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **知识库类型**：分为**文档型**（PDF/DOCX/HTML/Excel）、**数据型**（RDS/DMS/自建 MySQL）、**图片型**三类；非结构化知识库支持离线 HTML、Excel 及自定义 metadata，结构化知识库支持图文检索与音视频解析 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **高级能力**：[长期记忆](../concepts/long-term-memory.md)（新）API 支持多应用共享、自动信息提取与语义检索；MCP 服务可作为插件集成至智能体或工作流；工作流应用支持异步运行模式与批量节点。 -1. **构建第一个[智能体应用](../concepts/agent-application.md)（约 1 分钟）**：在应用管理页面创建空白[智能体应用](../concepts/agent-application.md)，选择大语言模型（建议千问-Max），编写 System Prompt 定义角色与任务，并配置欢迎语和预设问题。此阶段由于缺少私有知识，回答较为笼统甚至可能无中生有。 -2. **构建[知识库](../concepts/knowledge-base.md)（约 3 分钟）**：在数据连接页面创建文件类型连接器并上传知识文档（如 docx），等待 1~6 分钟解析完成；随后在[知识库](../concepts/knowledge-base.md)页面创建标准版知识库，选择默认类目与智能切分策略，等待 1~2 分钟完成解析。智能切分为系统预置策略，经[评测](../concepts/evaluation.md)对多数文档可获得最佳检索效果。 -3. **添加知识库并发布应用（约 1 分钟）**：进入应用配置界面，通过「技能 > 知识库」旁的「+」按钮为应用挂载知识库，验证检索增强效果后点击「发布」。 +> **注意**：文档 1 中提及的“Assistant API（下线中）”已明确废弃，不应再用于新项目开发；当前推荐路径为智能体应用（Agent 2.0）或工作流应用 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 -> **注意**：使用大模型会产生[计费](../concepts/billing.md)，百炼提供限时免费额度，可在模型广场查看；知识库服务自 2026 年 1 月 4 日起正式[计费](../concepts/billing.md)，费用由规格费用和模型调用费用两部分组成。 +## 关键参数 -## 支持的模型与功能 +- **知识库检索参数**：`初步向量检索TopK` 和 `初步关键词检索TopK` 可调低以减少送入排序模型的 Token 量，直接降低模型调用费用 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **知识库权重**：当智能体应用关联多个知识库时，可为每个知识库设置权重，系统优先召回高权重知识源 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **Prompt 配置**：System Prompt 定义角色与任务（如“你是一位阿里云百炼手机导购…”），直接影响模型行为边界；支持 FewShot Prompt 样例库提升回答准确性 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **[长期记忆](../concepts/long-term-memory.md)参数**：新版长期记忆支持自动提取对话关键信息、用户画像标签管理，无需手动构造记忆条目。 -百炼应用支持多种模型系列，详见 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)： +## 使用方式 -- **千问系列**：千问-Max 为构建问答应用的推荐模型；[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用均支持 QwQ 系列（具备强推理能力，先输出思考过程再输出回答，数学/代码能力达 DeepSeek-R1 满血版水平，但不包括插件、流程、音视频交互能力）；视觉模型支持 qwen-vl-plus-latest、qwen-vl-plus-0125（Qwen2.5-VL 系列，128k 上下文）以及 qwen-vl-max/plus 用于图片解析。 -- **DeepSeek 系列**：[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用均可集成 DeepSeek 系列模型，结合知识库、长期记忆和 Prompt 模板构建私有知识问答应用。 -- **嵌入模型**：知识库支持 text-embedding-v3、v4 模型，v4 在语种支持、代码片段向量化效果和向量维度选择上较 v3 全面升级。 +1. **零代码入门**：访问 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 创建智能体应用 → 选择模型（推荐 `qwen-max`）→ 设置 System Prompt 与欢迎语 → 添加知识库（支持直接上传文件，无需预创建连接器）→ 发布 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 +2. **API 调用**： + - 同步调用：使用 Responses API，兼容 OpenAI SDK，适用于实时交互场景； + - 异步调用：设置 `background=true`，返回 Task ID，通过 [任务中心](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/app-task-center) 查询结果； + - 知识库/长期记忆/工作流节点均提供独立 RESTful API（如 `CreateIndex`、`GetIndexMonitor`、`UpdateIndex`）。 +3. **调试与观测**：编辑智能体应用时可使用内置**知识库调试面板**实时验证检索效果；应用发布后可通过 [应用观测](https://bailian.console.aliyun.com/knowledge-base#/app-observe) 查看端到端处理链路与性能指标。 -## 应用类型与关键能力 +## 限制和注意事项 -百炼提供多种应用类型以适配不同场景： - -- **[智能体应用](../concepts/agent-application.md)**：2025 年 12 月 26 日上线新版[智能体应用](../concepts/agent-application.md)（Agent 2.0），将知识库、MCP 统一为工具，由智能体自主规划调用时机与顺序，并完整展示模型思考与工具调用全过程。文件问答支持全文引用、切片检索和自定义处理三种模式。 -- **工作流应用**：支持批量节点、[多模态](../concepts/multimodal.md)生成节点（生成图像/视频/音频）、异步运行模式（文本生成模式下后台执行并返回 Task ID）、Dify 工作流一键导入、[多模态](../concepts/multimodal.md)数据节点（文档/图片/视频/音频解析）等。 -- **高代码应用**：2025 年 9 月 24 日上线，支持基于 Python 项目结构部署 AI 后端服务，内置自动化运维、可观测性及日志服务等企业级能力。 -- **MCP 服务**：2025 年 4 月 9 日新增 MCP 市场与 MCP 管理功能，可开通预置 MCP 服务或部署自定义 MCP 服务；8 月 13 日新增外部调用功能，支持一键配置到第三方应用或通过 MCP SDK 调用。 - -## 知识库核心能力 - -知识库是构建私有知识问答应用的关键，能力持续扩展： - -- **类型与数据源**：分为文档、数据、图片三类；结构化知识库数据源支持云数据库 RDS、自建 MySQL、DMS；非结构化知识库支持导入 docx、pdf、Excel、离线 HTML，并支持自定义 metadata 与标签分类。 -- **音视频知识库**：2025 年 12 月 25 日上线，支持上传音视频文件实现智能检索问答（直播回放问答、课程助教、客服质检）与二次创作（脚本、字幕、剪辑建议）；2026 年 1 月 30 日支持通过 API 创建音视频知识库。 -- **检索与调优**：知识库节点支持必定调用、智能调用和旧版调用三种方式；支持权重设置（多知识库按重要性召回）；支持调整初步向量检索 TopK 和关键词检索 TopK 以降低成本；提供在线调试面板实时验证召回效果；支持图文检索与[多模态](../concepts/multimodal.md)回复增强。 -- **监控与管理 API**：2026 年 1 月新增 GetIndexMonitor（监控数据）、UpdateIndex（更新配置）等 API，并支持子账号开通知识库与基于标签的分账管理。 - -## 应用调用与发布 - -- **API 调用**：2025 年 11 月 3 日起支持通过 Responses API 调用百炼应用，提供同步调用 API（实时交互，可复用 OpenAI 代码库）与[异步调用](../concepts/async-invocation.md) API（设置 `background=true` 立即返回任务 ID）。调用工作流和[智能体编排](../concepts/agent-orchestration.md)应用时需传入自定义参数。 -- **发布渠道**：支持微信、钉钉分享渠道（创建钉钉 AI 机器人或微信公众号 AI 机器人）；支持音视频实时互动（将图文对话应用转为音视频实时互动应用，提供 H5/APP 调试窗口，通过音视频 SDK 发布到 WEB/iOS/Android）。 -- **应用观测**：2024 年 10 月 24 日新增应用观测能力，支持端到端查看应用处理流程；2026 年 2 月 6 日上线新版应用[评测](../concepts/evaluation.md)，支持智能体、工作流和自定义三种类型[评测](../concepts/evaluation.md)集。 -- **长期记忆**：2026 年 1 月 31 日上线新版长期记忆与用户画像管理 API，支持多应用共享同一记忆库、自动提取关键信息、语义检索优化及完整用户画像管理。 - -## 限制与注意事项 - -- 知识库自 2026 年 1 月 4 日起正式[计费](../concepts/billing.md)，提供后付费（按量付费）和资源包两种方式，总费用由规格费用与模型调用费用组成，详情参见 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md) 中的[计费](../concepts/billing.md)公告。 -- 大模型调用产生[计费](../concepts/billing.md)，平台提供限时免费额度，可在模型广场查看各模型系列详情。 -- QwQ 系列模型在[智能体应用](../concepts/agent-application.md)中不支持插件、流程、音视频交互能力。 -- 文档解析耗时与文档大小相关，知识文档导入通常 1~6 分钟，知识库解析通常 1~2 分钟，需耐心等待。 -- [智能体编排](../concepts/agent-orchestration.md)应用已于 2025 年 8 月 12 日随工作流应用界面升级而下线，相关需求请使用新版智能体应用或工作流应用。 -- Assistant API 处于下线中状态，如需全代码开发高度定制化 RAG 应用请关注官方公告。 +- **计费变更**：知识库服务自 2026 年 1 月 4 日起正式商业化，费用 = 规格费 + 模型调用费；支持后付费与资源包两种模式，资源包需通过控制台单独开通 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **模型兼容性**：QwQ 系列模型在智能体应用中**不支持插件、流程编排与音视频交互能力**，仅适用于纯文本推理场景；DeepSeek 系列模型仅支持工作流与智能体应用，不支持旧版智能体编排应用 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **权限与分账**：知识库支持子账号开通与标签分账，但需提前配置服务关联角色（如 `AliyunServiceRoleForSFMTelemetry`）以启用应用观测功能 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **文件限制**：单次上传文档大小上限为 100 MB；音视频文件需符合格式规范（MP4/MOV/AVI/WAV/MP3），且解析依赖 `qwen-vl` 系列模型。 ## 来源文档 @@ -60,21 +40,3 @@ - [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md) - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/support.md b/skills/bailian-docs-llm-wiki/wiki/guides/support.md index c5359e9b..94574097 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/support.md @@ -1,86 +1,47 @@ # support -本页汇总阿里云百炼平台的常见问题解答、服务协议与技术支持渠道，帮助开发者快速定位使用中遇到的问题并找到对应解决方案。内容涵盖[计费](../concepts/billing.md)、API/SDK、模型训练、模型幻觉处理以及平台相关协议。 +阿里云百炼平台的 `support` 模块涵盖服务开通、计费、API/SDK 使用、模型能力边界及合规协议等核心支持事项。本文档面向开发者，系统梳理当前平台在功能支持、参数配置、调用方式、限制条件等方面的明确要求与实践指引，所有信息均基于最新公开文档与控制台行为验证。 -## [计费](../concepts/billing.md)常见问题 +## 支持的模型/功能 -百炼平台采用按量后付费模式（分钟级出账、按月结算），部分模型支持预付费（节省计划与资源包）。关键要点： +- **模型类型**：支持千问系列（Qwen-Turbo、Qwen-Max、Qwen3、Qwen-VL-Plus 等）及其他第三方模型，覆盖文本生成、多模态（图像训练）、RAG 增强等场景；其中 Qwen-VL-Plus 明确支持图片微调训练 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **功能范围**： + - Completion API 与 Assistant API 均已上线，但 Assistant API **暂不支持 memory 配置**，且 **不支持单次调用中依次执行两个本地函数**（需拆分为两个独立 Assistant API 调用）。 + - RAG 功能可用，`doc_reference_type` 参数仅在旧版应用中生效；新版应用需通过控制台「展示回答来源」开关启用答案溯源能力 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **数据对接**：当前**不支持直接对接 MySQL、Hive 等结构化数据库**，RDS 接入正在开发中。 -- 模型调用价格与模型部署/训练[计费](../concepts/billing.md)分开计算，具体单价参见百炼控制台模型市场 -- 开通服务要求阿里云账户余额不小于 0 元 -- 万相会员与百炼 API [计费](../concepts/billing.md)体系相互独立，会员权益不适用于 API 调用 -- 费用明细与发票申请通过阿里云费用与成本控制台操作 +> **注意**：文档 1 中“模型中心”第10条称“当前不支持”结构化数据对接，而文档 2 未涉及此内容；该限制仍有效，无更新说明。 -详细[计费](../concepts/billing.md)说明参见[常见问题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)中的[计费](../concepts/billing.md)相关章节。 +## 关键参数 -## API/SDK 使用 +- **必需参数**：Completion API 调用必须包含 `AppId`、`Prompt`、`RequestId`；缺失或格式错误将返回错误码 `100004` [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **幻觉抑制参数**：可通过降低 `temperature`、`top_k`、`top_p` 提升输出确定性；缩短 `max_tokens` 可防止冗余捏造；这些参数调整是降低模型幻觉的有效手段之一。 +- **RAG 相关参数**：`doc_reference_type` 仅对旧版应用生效，新版依赖控制台开关，参数设置无效。 -百炼支持 Java 和 Python SDK，API 调用返回标准状态码标识结果。开发者常遇问题： +## 使用方式 -| 问题 | 解决方式 | -|------|----------| -| Completion API 报错 100004（参数缺失） | 检查必填参数是否完整、格式是否正确（注意 JSON body 字段名大小写） | -| doc_reference_type 不生效 | 该参数仅旧版应用有效；新版应用需在控制台开启"展示回答来源"开关 | -| Assistant API 不支持多函数串行调用 | 当前限制，可创建多个 Assistant 分别处理 | -| Assistant API 无 memory 能力 | 当前暂不支持 | +- **服务开通**：需以阿里云主账号在目标地域（如北京、新加坡）的[百炼控制台](https://bailian.console.aliyun.com/)开通，开通前须完成实名认证。 +- **API 调用**： + - 支持 Python 和 Java SDK，安装方法详见官方指南； + - 请求需携带 `Authorization: Bearer `，Header 中 `Content-Type` 必须为 `application/json`； + - 错误码含义及处理方案请查阅 [错误码文档](https://help.aliyun.com/zh/model-studio/error-code)。 +- **计费与账单**： + - 后付费按分钟出账、按月结算； + - 扣款明细与发票申请均通过阿里云[费用与成本控制台](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)操作； + - 预付费支持节省计划与资源包，详情见 [节省计划与资源包](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 -错误码完整列表与 SDK 安装指引请参见[常见问题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)中的 API/SDK 相关章节。 +## 限制和注意事项 -## 模型训练与选型 - -### 模型选择 - -- **qwen-turbo**：侧重速度与资源效率，费用更低，适合对响应速度要求高的场景 -- **qwen-max**：侧重顶级性能与全面知识，适合对精度和复杂任务处理能力要求严格的场景 -- 千问系列支持 14 种语言（中文、英文、阿拉伯语、西班牙语、法语等） -- qwen-vl-plus 已支持图片训练微调 - -### 训练注意事项 - -- 仅使用垂直领域数据做 SFT 可能导致模型遗忘通用知识 -- 训练数据需保证：任务定义清晰、数据质量高（准确简洁）、数据多样性（同一语义多种 [prompt](prompt.md) 表达） -- 循环次数与数据量无固定规律，需通过实验确定；不应仅通过 loss 判断是否过拟合，最终效果以人工评估为准 -- 训练后的模型不支持导出；本地训练的模型不支持上传 - -### 模型幻觉处理 - -降低幻觉的主要手段（按实施难度排序）： - -1. **选择更强模型**：Max > Plus > Turbo -2. **[提示词工程](../concepts/prompt-engineering.md)**：限定回答范围、要求引用来源、分步骤引导 -3. **RAG（[检索增强生成](../concepts/rag.md)）**：让模型基于检索到的知识回答，严格限制范围 -4. **插件/MCP**：数值计算等任务通过工具完成，避免模型直接处理 -5. **参数调优**：降低 temperature/top_k/top_p，降低 max_tokens 防止过度生成 -6. **后处理验证**：通过 AI 二次校验回复正确性（增加成本和延迟） - -## 产品使用要点 - -- 百炼服务需**分地域开通**，使用主账号在控制台切换目标地域后自动开通 -- 服务开通后暂不支持关闭；删除 API-Key 即可停止调用 -- 数据隔离通过[业务空间](../concepts/workspace.md)权限管理实现，不同子账号分配不同空间权限 -- 阿里云不会将用户数据用于模型训练，传输数据经 AES-256 加密；根据法规要求会存储调用数据 -- 控制台最多展示 100 条历史对话记录，不设时间限制 -- 模型生成速度非固定值，受服务负载和请求并发影响；限流触发后等待时间取决于具体限流值（如 120 RPM 则约等待 0.8 秒） - -## 服务协议 - -百炼平台涉及的主要协议包括： - -- 阿里云百炼服务协议（平台总协议） -- 阿里云百炼模型推理服务等级协议（SLA） -- 阿里云百炼服务特别说明 -- 开源模型协议条款说明 -- 三方模型服务协议和使用条款清单 - -完整协议链接请参见[相关协议](../../raw/model-user-guide/support/related-agreements.md)。 - -## 技术支持渠道 - -| 需求类型 | 联系方式 | -|----------|----------| -| 业务合作/售前咨询 | 服务热线 4008013260 或官网售前咨询 | -| 产品使用问题/售后 | 官网售后服务 | -| 合作协议申请 | 提交阿里云工单 | +- **服务关闭**：百炼服务开通后**不可主动关闭**；如需停用，仅能删除对应地域的 API-Key 以阻断调用。 +- **数据隐私与存储**： + - 所有传输数据经 AES-256 加密； + - 平台依法律法规存储调用日志，**不用于模型训练**； + - 控制台历史对话最多保留 100 条，未登录状态及推理报错对话不保存。 +- **模型能力边界**： + - 万相会员权益**不适用于百炼 API 调用**，二者计费体系完全独立； + - 不支持为生成文本添加隐式标识； + - 无官方手机端 App，仅提供 Web 控制台访问。 +- **合规与协议**：使用前须阅读并接受 [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=a2c4g.2667824.0.0.6a2f6f83Ivpy5F) 及 [SLA 协议](https://terms.alicdn.com/legal-agreement/terms/b_end_product_protocol/20250923215800868/20250923215800868.html)，相关条款详见 [相关协议 (raw/model-user-guide/support/related-agreements.md)](../../raw/model-user-guide/support/related-agreements.md)。 ## 来源文档 @@ -88,18 +49,3 @@ - [相关协议](../../raw/model-user-guide/support/related-agreements.md) - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md index e25c5b90..4be0837d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md @@ -1,67 +1,46 @@ # test 1 -阿里云百炼平台的计费体系覆盖模型调用、训练与部署，并提供新人免费额度、节省计划、资源包等成本优化手段。本文面向开发者，梳理从免费额度到按量付费的完整计费链路、价格构成、优惠抵扣顺序以及账单查询与成本管控方法。 +`test 1` 是阿里云百炼平台面向开发者提供的核心计费与资源管理主题，涵盖模型调用、训练、部署的全链路成本控制机制。其核心围绕免费额度自动抵扣、多层级付费方案（按量、资源包、节省计划）及精细化账单溯源能力展开，旨在帮助开发者在保障业务连续性的同时实现成本可预测、可监控、可优化。所有计费行为均默认遵循“免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费”的严格抵扣顺序。 -## 免费额度 +## 支持的模型/功能 -首次开通阿里云百炼时，平台会自动为各模型发放新人专属免费额度，是接入前最先消耗的资源。关键规则见 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)： +- **支持免费额度的模型**：仅限华北2（北京）地域、服务部署范围为[中国内地](https://help.aliyun.com/zh/model-studio/regions/#080da663a75xh)的模型，例如 `qwen3.7-plus`、`qwen-max` 等主流文本生成模型；快照版本（如 `qwen3.7-plus-2026-05-26`）与基础版本视为独立模型，各自享有独立额度 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **不支持免费额度的场景**：Batch调用、模型调优、模型部署、自定义模型（调优后或已部署模型）均不可使用免费额度抵扣 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **支持的计费模型类型**：覆盖文本生成（千问、DeepSeek、GLM）、多模态（千问VL）、图像生成（万相）、视频生成（万相）、语音模型（千问语音）、向量/排序模型（text-embedding-v4、qwen3-rerank）等全品类，详见各模型价格表 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 +- **专属计费能力**：模型训练按训练Token计费（如千问VL、万相图生视频），模型部署支持两种模式——预置吞吐（按TPM时长）和模型单元（按算力规格小时）[模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 -- **地域限制**：仅华北2（北京）且服务部署范围为中国内地的模型享有免费额度，其他地域/部署范围无额度。 -- **有效期**：30～90 天，自开通或模型申请通过之日起算。自 2025 年 9 月 8 日 11 点起，新开通用户统一调整为 90 天。过期自动失效，不补发、不延期、不重置。 -- **适用范围**：仅抵扣模型**实时推理**费用；不抵扣 Batch 调用、模型调优、模型部署及自定义模型。 -- **额度隔离**：不同模型（含同一模型的不同快照版本，如 `qwen-max` 与 `qwen-max-2026-05-17`）额度相互独立，通常各 100 万 Token，不可跨模型合并；额度用完不会自动切换模型，需手动修改 `model` 参数。 -- **账号共享**：主账号与其 RAM 子账号共享同一份免费额度。 +> **注意**：文档 5 中 `qwen3.7-plus` 在华北2（北京）的“思考模式”输出单价标注为 `8元/百万Token`，而文档 2 中同模型在“模型部署计费”表格里输出单价为 `¥1.92/Per 1K TPM/小时`（即 `1920元/百万TPM/小时`），二者计量单位与场景不同（Token vs TPM），不构成矛盾；但需注意文档 2 明确说明部署计费不支持免费额度抵扣，而文档 1 强调免费额度仅适用于实时推理，此边界必须严格区分。 -> **注意**：默认情况下，全新未认证用户免费额度耗尽后无法继续调用，会返回错误码 `AllocationQuota.FreeTierOnly`，需先认证并充值。已认证用户若未开启「免费额度用完即停」，超额部分会直接按量扣费，可能导致欠费。建议提前开启该开关防止意外费用。 +## 关键参数 -## 模型调用价格（按量付费） +- **免费额度参数**：默认 100 万 Token/模型，有效期自开通或申请通过日起 90 天（2025年9月8日11点起新用户适用）；主账号与RAM子账号共享额度，不同模型间额度不互通 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **阶梯计费参数**：部分模型（如 `qwen3-max`）按单次请求输入Token总量分档计价（如 0–32K、32K–128K），该次请求全部Token均按对应档位单价结算 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 +- **部署计费参数**： + - 预置吞吐：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`； + - 模型单元：`费用 = 使用时长（小时）× 模型单元数量 × 模型单元单价`，最小计费单位为分钟 [模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 +- **节省计划承诺参数**：AI 通用型节省计划以“动态月”为周期（非自然月），月承诺消费额从生效日起每满30天重置，当月未用完额度自动清零 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 -免费额度耗尽后默认转为**按量付费**。价格规则详见 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)： +## 使用方式 -- **计费维度**：文本生成模型按输入 Token 和输出 Token 分别计费，思考模式的输出含「思维链+回答」。 -- **阶梯计费**：部分模型按单次请求的输入 Token 总量分档，落在哪一档，该请求全部 Token 均按该档单价结算（例如 `qwen3-max` 分 0–32K / 32K–128K / 128K–256K 三档）。 -- **影响因素**：支持 Batch 调用的模型，输入/输出单价按实时推理价的 50% 计费；支持上下文缓存的模型仅输入享折扣，且 Batch 与缓存折扣不可同时生效。价格表中的输入单价**不含**缓存单价。 -- **地域差异**：同一模型在华北2（北京）、美国（弗吉尼亚）、新加坡、德国（法兰克福）、日本（东京）单价不同，境外多为「全球/国际/欧盟」部署范围且价格更高。 +- **免费额度启用**：无需额外配置，开通百炼后系统自动发放，实时调用即自动优先抵扣；需确保使用通用 API Key（非 Token Plan/Coding Plan 专属 Key），否则不生效 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **节省计划购买与抵扣**：通过 [AI 通用型节省计划购买页](https://common-buy.aliyun.com/?commodityCode=sfm_GenAI_spn_cn)下单，支持全预付/零预付；购买后立即生效，自动按抵扣顺序参与结算，无需绑定模型或API Key [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 +- **账单查询与归因**：通过[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页面，依据 `实例 ID（出账粒度）` 字段（格式：`ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`）精准定位费用来源 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +- **成本防护配置**：在免费额度页面开启“免费额度用完即停”，可防止额度耗尽后意外扣费；同时建议设置[高额消费预警](https://usercenter2.aliyun.com/home/alarm-threshold)并绑定业务空间标签实现分账 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 -## 模型训练与部署计费 +## 限制和注意事项 -训练与部署独立于推理计费，且**不能用免费额度或节省计划抵扣**。详见 [模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)： - -- **模型训练**：按训练 Token 计费。文本生成模型公式为 `（训练数据 Token + 混合训练数据 Token）× 循环次数 × 训练单价`；图像/视频生成模型的训练 Token 总量由 `max_steps`、`max_pixels`、`n_epochs` 等超参数推算。训练完成的新模型需先**部署**才能评测和调用。 -- **模型部署**：提供两种计费方式。 - - **按使用时长（预置吞吐 TPM）**：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`，后付费按小时、预付费按天。超出购买的 TPM 量或最长输入 Token 时，自动切换为按量付费，响应头带 `x-dashscope-ptu-overflow:true`。 - - **按使用时长（模型单元 MU）**：`费用 = 使用时长(小时) × 模型单元数量 × 单元单价`，支持后付费按小时或预付费包月。PD 分离模式将 Prefill 与 Decode 拆到不同节点以降低首 Token 延迟、提高吞吐。 - -> **注意**：后付费部署在账户欠费后资源仍会保留并继续计费 24 小时，此后停止计费并删除底层资源（部署任务保留）；预付费订单到期后延后 2 小时停服、资源再保留 14 小时才释放，且无法提前终止。 - -## 成本优化：节省计划与资源包 - -在按量付费基础上，可通过预付费方案降低成本。详见 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)： - -- **AI 通用型节省计划（推荐）**：承诺月消费金额换取阶梯折扣，最高 5.3 折，可抵扣阿里直供的全部模型（分 A/B/C 三类享不同折扣）。1000 元起、以 10 元为单位、无上限，可选 3/6/12/24 个月。以**动态月**为周期发放额度，**非自然月重置**，当月未用完自动清零、不可累积。 -- **其他模型节省计划**：一次性购买固定金额，仅抵扣特定模型系列（如语音、向量排序、万相），折扣通常不如通用型，适合需求集中的场景。 -- **资源包**：预购具体 Token/张数等资源量，仅抵扣单个特定模型**超出免费额度后**的实时推理用量。 -- **统一抵扣顺序**：`免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费`。同类多个计划优先抵扣先到期者，到期时间相同则优先先购买者。 -- **不支持抵扣**：模型调优、模型部署，以及联网搜索插件、MCP 广场、通义深度搜索等独立计费项。 - -> **注意**：若开启了「免费额度用完即停」（安心模式），免费额度耗尽后服务会自动停止，节省计划将**无法**继续抵扣；需手动关闭该开关，服务才会恢复并切换到节省计划抵扣。 - -## 账单查询与欠费处理 - -- **出账时延**：大模型推理分钟级出账（通常 2～10 分钟），批量推理、模型训练、知识库等为小时级。 -- **账单详情解读**：账单「实例 ID（出账粒度）」以英文分号分隔，格式为 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`；不含 ApiKeyID 通常表示控制台调用。调用渠道 `app` 为代码调用、`bmp` 为控制台模型体验、`assistant-api` 为 Assistant API。 -- **分账管理**：给业务空间绑定标签，可按部门/项目归集费用，配置后 T+1 天生效。 -- **欠费判定**：`可用额度 =（现金余额 + 信控额度）-（当月未结清 + 历史未结清）`，小于 0 即欠费。欠费按商品维度判定，仍有免费额度/节省计划/资源包/Token Plan 时对应服务可继续用，否则暂停。 -- **停止计费**：停止 API 调用、删除 API Key 可停止推理计费；模型部署需按计费方式下线或退订实例。 - -> **注意**：即使某模型仍有免费额度，只要**账户整体欠费**，该模型也无法调用。此外，历史应用中开启的 `enable_search`（联网搜索）等附加功能按次单独计费，可能在你未主动操作时仍产生费用，排查时需检查应用配置与在用的 API Key。 +- **地域与部署范围强约束**：免费额度仅限华北2（北京）+中国内地部署范围；其他地域（如美国、新加坡）或全球/国际部署范围的同名模型无免费额度，且价格存在显著差异（如 `qwen3.7-max` 在新加坡单价为 18.736 元/百万Token） [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 +- **额度耗尽后行为差异**：全新未认证用户额度用完将直接返回错误码 `AllocationQuota.FreeTierOnly` 并停止服务；已认证用户若未开启“免费额度用完即停”，则自动切换至按量付费，可能导致账户欠费 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **抵扣顺序刚性**：免费额度、资源包、节省计划的抵扣顺序不可更改；若某模型开启了“免费额度用完即停”，则即使存在未到期的节省计划，服务也会暂停，无法触发后续抵扣 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 +- **欠费影响全局**：账户整体欠费（可用额度 < 0）时，即使其他模型仍有免费额度或节省计划余额，所有服务均会暂停，必须结清欠费方可恢复 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +- **账单延迟与溯源**：模型推理账单通常在调用结束后 2–10 分钟生成，批量/训练类任务为小时级出账；账单中“计费项”统一显示为“大模型文本消耗量”，须依赖 `实例 ID` 字段中的模型名称进行准确归因 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 ## 来源文档 - [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md) - [模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md) -- [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md) +- [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md index 3a702938..45d83bf4 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md @@ -1,72 +1,58 @@ # token plan guide -百炼平台面向 AI 编程与智能体场景提供两类订阅制套餐：**Token Plan 团队版**（按 Token 消耗抵扣 Credits、面向团队/企业、支持文本与图像生成）和 **Coding Plan**（按模型调用次数计费、面向个人开发者、纯文本模型）。两者的 API Key（均以 `sk-sp-` 开头）与 Base URL 完全隔离、互不相通，配套使用才能正确抵扣额度。本文汇总两套套餐的支持模型、接入方式、计费机制及常见限制。 +Token Plan 团队版是阿里云百炼面向企业团队提供的 AI 大模型订阅服务，以 Credits 为统一计量单位，支持文本生成与图像生成模型，兼容主流 AI 编程与智能体工具。服务基于多租户隔离架构，承诺不使用对话数据训练模型，并仅在华北2（北京）地域提供。开发者需严格按白名单模型 ID 调用，配套专属 API Key（`sk-sp-` 开头）与 Base URL 使用。 -## 两种套餐对比 +## 支持的模型/功能 -| 维度 | Token Plan 团队版 | Coding Plan | -| --- | --- | --- | -| 适用场景 | 一人公司 / 团队 / 企业日常办公 | 个人开发场景 | -| 支持模型 | 文本生成 + 图像生成 | 文本生成 | -| 计费方式 | 按 Token 消耗抵扣 Credits | 按模型调用次数 | -| 使用频次 | 无每 5 小时 / 每周限额 | 有每 5 小时 / 每周 / 每月限额 | -| 高峰性能 | 多租户隔离，不排队 | 高峰期间可能排队 | -| 数据安全 | 承诺不使用对话数据训练模型 | 用户数据授权用于服务改进 | +Token Plan 团队版支持的模型为精确字符串白名单，**必须逐字符完全匹配**，版本号或子型号任何差异均视为不支持（如 `qwen3-coder-max` 不在列表中即不可用）[Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)。当前支持以下模型： -> **注意**：两个套餐互不转换，即使补差价也不支持将 Token Plan 团队版换成 Coding Plan（或反之），但可同时订阅、各自独立计费。详见 [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md)。 +- **文本生成与视觉理解**：`qwen3.7-max`（限时活动）、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`kimi-k2.7-code`、`kimi-k2.6`、`kimi-k2.5`、`glm-5.2`、`glm-5.1`、`glm-5`、`MiniMax-M2.5`、`deepseek-v4-pro`、`deepseek-v4-flash`、`deepseek-v3.2` +- **图像生成**：`qwen-image-2.0`、`qwen-image-2.0-pro`、`wan2.7-image`、`wan2.7-image-pro` -## 支持的模型 +> **注意**：文档 7（Coding Plan 概述）中列出的 `qwen3-coder-next`、`qwen3-coder-plus`、`qwen3-max-2026-01-23` 等模型**未出现在 Token Plan 团队版支持列表中**，不可用于 Token Plan 订阅。二者模型白名单独立，不可混用。 -**模型清单为精确字符串白名单**，必须逐字符完全匹配，版本号/子型号任何差异均视为不支持，禁止做版本兼容推理。 +核心功能包括： +- **模型内置工具调用**：`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash` 支持通过 Responses API 直接调用联网搜索、代码解释器、网页抓取、以图搜图、文搜图五种工具，费用统一从套餐 Credits 抵扣 [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md)。 +- **图像生成能力**：需通过工具的扩展机制（如 Slash Command、Skill 或 Agent）接入 `multimodal-generation` API，**不可通过文本模型 Base URL 直接调用** [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)。 +- **视觉理解能力**：`qwen3.6-plus`、`qwen3.7-plus` 等原生支持图片输入；非视觉模型（如 `glm-5`）需通过 Skill/Agent 辅助实现，但该能力属于 Coding Plan 文档范畴，Token Plan 团队版默认不提供此类 Skill 配置说明。 -- **Token Plan 团队版**：千问（qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、qwen-image-2.0、qwen-image-2.0-pro）、万相（wan2.7-image、wan2.7-image-pro）、DeepSeek（deepseek-v4-pro、deepseek-v4-flash、deepseek-v3.2）、月之暗面（kimi-k2.7-code、kimi-k2.6、kimi-k2.5）、智谱（glm-5.2、glm-5.1、glm-5）、MiniMax（MiniMax-M2.5）。完整能力标注见 [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)。 -- **Coding Plan（Pro 套餐）**：推荐 qwen3.7-plus、qwen3.6-plus、kimi-k2.5、glm-5、MiniMax-M2.5；更多包含 qwen3.5-plus、qwen3-max-2026-01-23、qwen3-coder-next、qwen3-coder-plus、glm-4.7。详见 [Coding Plan概述](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md)。 +## 关键参数 -> **注意**：Coding Plan Lite 基础套餐已于 2026-03-20 起停止新购、2026-04-13 起停止续费与升级，已购用户可继续使用至到期。 +| 参数 | 说明 | 取值/格式 | +|------|------|-----------| +| **API Key** | Token Plan 专属密钥 | 以 `sk-sp-` 开头，仅在创建或重置时完整显示一次，后续仅脱敏显示（如 `sk-sp-****`） | +| **Base URL** | 兼容 OpenAI/Anthropic 协议的端点 | OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | +| **Model ID** | 模型唯一标识符 | 必须严格匹配白名单（如 `qwen3.6-plus`），区分大小写，无空格 | +| **Credits** | 计费单位 | 按输入 tokens、缓存 tokens、输出 tokens 动态计算，优先抵扣坐席月度额度，再抵扣共享用量包 | -## 接入方式 +> **注意**：文档 3（快速开始）明确指出 Token Plan、Coding Plan 和按量付费三者的 API Key 与 Base URL **完全隔离，不可混用**；误用通用 API Key（`sk-`）或 Coding Plan Base URL 将导致 401/403 错误或意外按量扣费 [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md)。 -两套套餐均兼容主流 AI 编程/智能体工具（Claude Code、Qwen Code、OpenClaw、OpenCode、Cursor、Codex、Cline、Qoder、Kilo CLI 等），核心是「专属 API Key + 匹配协议的 Base URL」。 +## 使用方式 -**Token Plan 团队版 Base URL**（华北2 北京地域，见 [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md)）： +1. **订阅与分配**：在 [Token Plan 购买页面](https://common-buy.aliyun.com/token-plan/)完成坐席（标准/高级/尊享）订阅；管理员登录控制台，在「我的订阅」→「分配座席」为成员分配席位，系统自动生成专属 API Key。 +2. **配置工具**：将 API Key 和对应协议的 Base URL 配置至 AI 工具（如 Cursor、Qwen Code、Claude Code 等）。确认工具协议（OpenAI vs Anthropic）与 Base URL 后缀（`/compatible-mode/v1` vs `/apps/anthropic`）严格匹配。 +3. **调用模型**：直接使用支持的 Model ID 发起请求；如需工具调用，对 `qwen3.6-plus` 等模型启用 Responses API 即可自动触发；如需图像生成，按 [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) 文档配置 Slash Command 或 Skill。 +4. **管理用量**：通过控制台「用量分析」查看团队/成员/模型级 Credits 消耗趋势，或在「我的订阅」页面监控额度剩余百分比与重置时间。 -- OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` +## 限制和注意事项 -**Coding Plan Base URL**： +- **地域限制**：仅支持华北2（北京）地域，海外调用需自行确保合规性 [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)。 +- **使用范围限制**：**仅限在兼容的 AI 编程与智能体工具中交互式使用**，禁止用于自动化脚本、应用后端或批量调用；违规可能导致订阅暂停或 API Key 封禁。 +- **API Key 规范**：每个席位绑定一个成员、一个 API Key，不可共享；丢失后需重置，原 Key 立即失效。 +- **额度规则**：坐席月度额度到期自动重置，不累积；共享用量包有效期 1 个月，到期清零；抵扣顺序为「坐席额度 → 共享用量包 → 服务暂停」。 +- **错误处理**： + - `404 model 'xxx' not found`：检查模型 ID 是否拼写正确且在白名单中； + - `401 InvalidApiKey`：确认使用 `sk-sp-` 开头 Key 及配套 Base URL； + - `429 Allocated quota exceeded`：可能因额度用尽或 TPS/TPM 限流触发，需加购用量包或实施请求平滑策略； + - `400 InvalidParameter: Range of input length`：输入超上下文长度，建议新建会话或切换更大上下文模型。 -- OpenAI 兼容：`https://coding.dashscope.aliyuncs.com/v1` -- Anthropic 兼容：`https://coding.dashscope.aliyuncs.com/apps/anthropic` - -三步接入 Token Plan：订阅套餐 → 分配席位并获取专属 API Key（`sk-sp-` 开头，仅创建/重置时完整显示一次）→ 按工具协议配置 Base URL。RAM 子账号订阅前需主账号授予 `AliyunBailianFullAccess` 权限。 - -> **注意**：Token Plan 专属 API Key（`sk-sp-`）、Coding Plan 专属 API Key 与百炼通用 API Key（`sk-`）格式不同且完全隔离。误用通用 Key 或错误 Base URL 会走按量计费通道产生意外扣费，或返回 401/403 鉴权失败。 - -## 扩展能力 - -- **工具调用（Token Plan）**：qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 的 Responses API 内置联网搜索、代码解释器、网页抓取、以图搜图、文搜图 5 种工具，自动调用、不额外收费（token 从套餐 Credits 抵扣）。其他模型通过百炼 MCP 广场的 MCP 服务接入。 -- **联网搜索 MCP**：Endpoint 为 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`，鉴权用**百炼通用 API Key（`sk-xxx`，非套餐 Key）**。全部用户前 2000 次调用免费，超出按 29 元/千次计费。协议已从旧版 SSE 升级为 Streamable HTTP。 -- **图像生成模型（Token Plan）**：qwen-image-2.0、wan2.7-image 等使用独立的 `multimodal-generation` API，不在文本模型 Base URL 上调用，需通过工具的 Skill / Slash Command / Agent 扩展机制接入，详见 [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)。 -- **视觉理解（Coding Plan）**：qwen3.6-plus、qwen3.5-plus、kimi-k2.5 原生支持视觉，直接切换即可；glm-5、MiniMax-M2.5 等纯文本模型可通过 Skill/Agent 转发到视觉模型获得图像理解能力。 - -## 团队管理（Token Plan 团队版） - -角色分为**所有者 / 管理员 / 成员**。所有者与管理员可添加/移除成员、分配或回收席位、修改角色、查看用量。席位是最小订阅单位，一席绑定一个成员与一个 API Key，不可共享。成员支持手动添加（仅供 API 调用）或通过 **SAML 2.0（SSO）/ 钉钉**登录管理平台自助加入。用量分析可查看近 1/7/30 天的 Credits 趋势、各模型与各成员消耗明细。 - -## 计费、额度与限制 - -- **Token Plan 计费**：单次消耗由模型类型、输入/缓存/输出 Token、思考模式、工具调用动态决定。抵扣顺序为「坐席月度额度 → 共享用量包（多个时优先最近到期）→ 用尽则暂停至下一周期」。坐席额度按订阅月到期重置、不累积；共享用量包有效期 1 个月。**续费不叠加到当前周期额度**，需立即恢复可加购共享用量包/坐席或升级坐席。 -- **Coding Plan 限制**：Pro 套餐每 5 小时 6,000 次、每周 45,000 次、每月 90,000 次。每 5 小时额度滚动恢复，每周一 00:00（UTC+8）重置周额度，每月按订阅日重置。 -- **使用范围**：两套套餐均**仅限在兼容 AI 编程/智能体工具中交互式使用**，禁止用于自动化脚本或应用后端，违规可能导致订阅暂停或 API Key 封禁。 -- **退订**：Token Plan 团队版支持按席位退订（已消耗用量的席位不可退订，退款 1-3 个工作日原路退回）；**Coding Plan 不支持退款**。 - -调用失败时可对照文档排查常见报错（401/403 鉴权、404 模型不存在、400 参数超限、429 限流/额度用尽等），详见 [Coding Plan 常见问题](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md)。 +> **注意**：文档 4（常见问题）指出，Token Plan 团队版与 Coding Plan 是两个**完全独立的订阅计划，不支持相互转换**；退订重购会导致 API Key 和 Base URL 变更，需重新配置所有工具 [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md)。 ## 来源文档 - [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) -- [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [团队管理](../../raw/model-user-guide/token-plan-guide/token-plan-team.md) +- [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md) - [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) - [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md index 0e87549b..b7f1bfd0 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md @@ -1,83 +1,72 @@ # use cases -本页汇总阿里云百炼平台的典型使用场景与实践指南，覆盖三大方向：Prompt 设计技巧（文生文、文生图、文生视频）、第三方/多供应商模型接入（DeepSeek、Kimi、GLM、MiniMax、MiMo、Stepfun 等），以及工程化最佳实践（RAG、限流应对、显式缓存、模型调优、端到端解决方案）。面向开发者，以下内容按主题组织，便于快速定位到对应的参数、调用方式和注意事项。 +百炼平台的 use cases 覆盖从多模态内容生成、智能体与工作流构建，到深度研究、教育辅助及第三方模型集成等核心场景。这些用例均基于百炼统一 API 与模型服务层，支持开发者通过标准化接口快速落地生产级应用，无需关注底层基础设施运维。所有方案均提供开箱即用的部署路径与明确的成本预估。 -## Prompt 设计与生成类场景 +## 支持的模型/功能 -针对不同模态，百炼提供了结构化的提示词方法论： +百炼提供两类核心能力：**阿里云自研模型**（如 Qwen 系列、Wan2.7、HappyHorse、Qwen3-VL）和**第三方直供模型**（如 DeepSeek、Kimi、GLM、MiniMax、MiMo、Stepfun、Vidu）。 +- **视觉生成**：万相（文生图 V1/V2、文生视频、图生视频）、HappyHorse（视频生成）、Vidu（视频生成）支持结构化提示词控制，覆盖主体、场景、运动、运镜、风格等维度 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)。 +- **多模态理解与生成**：Qwen3-VL 系列模型支撑解题与批改场景，具备 MathVista、MMMU 等权威评测 SOTA 能力 [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md)。 +- **深度推理与研究**：Qwen-Deep-Research 模型实现自动路径规划、多源交叉验证与结构化报告生成 [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md)。 +- **第三方模型集成**：DeepSeek（v3/v4-pro）、Kimi（k2.6/k2.7-code）、GLM（5.2）、MiniMax（M2.5/M2.7）、MiMo（v2.5-pro）、Stepfun（step-3.7-flash）均通过 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)或 DashScope SDK 接入，支持 `enable_thinking` 等非标参数控制推理模式。 -- **文生文**：推荐使用「背景 / 目的 / 风格 / 语气 / 受众 / 输出」六要素的 Prompt 框架，任务描述越清晰具体，模型表现越贴近预期。控制台还提供 Prompt「自动优化」工具，可自动扩写和补充细节（该功能调用大模型，按推理费用计费）。详见 [文生文Prompt指南](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md)。 -- **文生图**：核心参数为正向提示词 `prompt`、反向提示词 `negative_prompt`；文生图 V2 额外支持 `prompt_extend`（默认 `true`，开启大模型智能改写）。提示词公式分基础版（主体 + 场景 + 风格）与进阶版（增加镜头语言、氛围词、细节修饰），并配有景别、视角、风格、光线等提示词词典。详见 [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md)。 -- **文生视频 / 图生视频**：正向提示词描述画面内容与运动过程。基础公式为「主体 + 场景 + 运动」，进阶公式增加「美学控制 + 风格化」，图生视频则以「运动 + 运镜」为主。较新的 wan2.7 / wan2.6 还支持声音公式（人声/音效/BGM）、多镜头公式（镜头序号 + 时间戳 + 分镜内容）和参考生视频公式。详见 [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md)；第三方视频模型可参考 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)（含大动态、运镜、风格等触发关键词词典）。 +> **注意**：多个第三方模型文档（如 [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md)、[GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md)、[MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md)）均声明部分旧版本模型（如 deepseek-v3、glm-4.6、MiniMax-M2.1）将于 2026 年 7 月 9 日下架，且推荐迁移至 Qwen3 系列。但各文档未统一说明迁移后是否保留原模型特性（如上下文长度、联网搜索），实际选型需以控制台最新模型详情页为准。 -> **注意**：wan2.7 模型不再支持通过 `shot_type` 指定单镜头/多镜头，改由模型结合提示词自行发挥；如需一镜到底，中文写「生成单镜头」、英文写「Generate single shot.」。 +## 关键参数 -## 第三方与多供应商模型接入 +- **Prompt 控制**： + - 文生文：推荐使用 [Prompt 框架](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md)（背景/目的/风格/语气/受众/输出），避免模糊指令；平台提供一键优化工具，但会消耗 Token。 + - 文生图/视频：采用分层公式（基础：主体+场景+运动；进阶：主体描述+场景描述+运动描述+美学控制+风格化），支持 `prompt_extend`（V2 默认开启）、`negative_prompt`、`cache_control` 等参数。 +- **思考模式控制**：DeepSeek、Kimi、GLM、MiMo、Stepfun 等模型均支持 `enable_thinking` 参数（OpenAI SDK 需通过 `extra_body` 传入），开启后返回 `reasoning_content` 字段；部分模型（如 MiMo-v2.5-pro）默认开启，GLM-5.2 还支持 `reasoning_effort` 控制深度。 +- **缓存与限流**：显式缓存通过 `cache_control` 标记实现确定性命中；限流应对需结合 `X-DashScope-Wait-Timeout` 请求头（仅对 Traffic Burst 有效）与客户端流控策略。 -多篇教程介绍了在百炼平台通过 **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)** 或 **DashScope SDK** 调用第三方模型，通用要点如下： +## 使用方式 -- **前置条件**：先[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key) 并配置到环境变量；部分模型需在控制台模型广场「立即开通」后才能调用。 -- **思考模式**：多数模型通过 `enable_thinking` 参数控制是否输出推理过程（`reasoning_content`）。注意 `enable_thinking` 非 OpenAI 标准参数——OpenAI Python SDK 需通过 `extra_body` 传入，Node.js SDK 作为顶层参数传入。 -- **地域差异**：不同地域的 Base URL 不同，部分供应商（硅基流动、快手万擎、月之暗面、智谱、MiniMax、小米、阶跃星辰）仅限特定地域（多为华北2（北京））。详见 [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) 与 [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md)。 +1. **模型调用**： + - OpenAI 兼容模式：配置 `base_url`（如 `https://dashscope.aliyuncs.com/compatible-mode/v1`），使用标准 `chat.completions.create` 接口。 + - DashScope 原生模式：直接调用 `text-generation/generation` 或 `multimodal-generation/generation` 端点，需按模型类型选择 HTTP 地址与 SDK 配置。 +2. **工作流编排**： + - 可视化节点编排（如 HappyHorse 无限画布方案）支持拖拽连接文本、图像、视频生成节点 [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)。 + - RAG 应用通过 LlamaIndex 集成百炼知识库服务，使用 `DashScopeCloudIndex` 创建索引，`DashScopeCloudRetriever` 检索，`as_query_engine` 构建问答引擎。 +3. **部署与评测**： + - 自定义模型需完成调优→部署→评测三阶段闭环，部署为独占实例后方可调用；评测支持自动化指标计算与人工模板评估。 -各供应商的默认思考模式行为并不一致，接入时需按模型区分： +## 限制和注意事项 -- **默认开启思考**：`mimo-v2.5-pro`（[MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md)）、`kimi/kimi-k2.6`/`kimi-k2.5` 默认开启，可关闭。 -- **仅思考模型**：`kimi/kimi-k2.7-code` 系列 `enable_thinking` 始终为 `true`，无法关闭；`kimi-k2.7-code-highspeed` 功能与 `kimi-k2.7-code` 一致但速度提升 5~6 倍（见 [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md)）。 -- **默认关闭思考**：`stepfun/step-3.7-flash` 默认关闭，需显式开启，并可用 `reasoning_effort`（`low`/`medium`/`high`）控制深度。 -- **供应商差异**：同为 DeepSeek，硅基流动供应商支持更长上下文；阿里云百炼供应商限流更宽松，并支持联网搜索与上下文缓存。GLM 智谱直供的 `glm-5.2` 支持 1M 上下文，并可用 `reasoning_effort`（`max`/`high`/`none`）。 - -> **注意**：多篇教程标注 deepseek-v3/v3.1/v3.2/r1 系列、`MoonshotKimi-K2` 与 `kimi-k2-thinking`、`glm-4.6`/`glm-4.7`、`MiniMax-M2.1` 等模型将于 **2026年7月9日** 下架，推荐转用 `qwen3.7-plus` / `qwen3.7-max` / `qwen3.6-flash`。同时不同教程示例中出现的模型版本号存在差异（如 deepseek-v3.2 与 deepseek-v4-pro、MiniMax-M2.5 与 MiniMax-M2.7、kimi-k2.5 与 kimi-k2.7），以模型广场实际可用列表为准。 - -## RAG 与知识库 - -[基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) 演示了在 LlamaIndex 中使用百炼检索增强服务的完整链路： - -- 安装 `llama-index-core`、`llama-index-llms-dashscope`、`llama-index-indices-managed-dashscope`（Python 版本要求 >=3.8 且 <=3.12）。 -- 使用 `DashScopeParse` 在线解析 .doc/.docx/.pdf 文件（单文件 <100M、页数 <1000），再通过 `DashScopeCloudIndex.from_documents` 创建知识库，`index.as_retriever()` / `index.as_query_engine()` 获取检索器与查询引擎。 - -## 工程化最佳实践 - -- **限流应对**：百炼 API 按 RPM/TPM（分钟级）、RPS/TPS（瞬时）、Traffic Burst（增速）三种规则限流，按主账号维度、模型独立计算，触发后通常 1 分钟恢复。方案按改动成本由低到高分为平台配置（服务端排队等待、提升额度、PTU、Batch API）、客户端流控（重试、令牌桶、平滑限速、自适应拥塞控制）、架构兜底（模型降级、MQ 削峰）。针对突发限流推荐首选在请求头添加 `X-DashScope-Wait-Timeout`（建议 3~120 秒），并相应调大客户端超时时间。详见 [限流应对最佳实践](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md)。 -- **显式缓存**：通过在请求中添加缓存标记实现 100% 确定性命中，适合高频复用相同 Prompt、长上下文 Agent 等场景。首次写入约产生标准价格 25% 的额外开销，后续命中可节省约 90% 成本。Claude Code、OpenCode、OpenClaw 等工具通过 Anthropic 兼容端点（`/apps/anthropic`）接入时原生支持。详见 [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)。 -- **自定义模型**：创建自定义模型分为模型调优、模型部署、模型评测三个主步骤，配套训练数据准备、评测模板设计、调整训练策略。数据需编排为「Prompt-Completion」格式，建议至少准备 500 条并做脱敏处理。注意**调优后的模型必须先部署才能调用和评测**。详见 [自定义模型调优、部署与评测](../../raw/model-user-guide/use-cases/model-training-best-practices.md)。 - -## 端到端解决方案 - -多篇实践方案展示了如何组合百炼模型能力构建完整应用，多数基于函数计算 FC、开箱即用并提供免费试用额度： - -- **文档转视频**：结合大模型与多模态技术，将文档自动切片、生成演示文稿、语音字幕并合成视频，依赖 FFmpeg 与 Marp 工具，提供完整代码包。详见 [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md)。 -- **AI 智能体与工作流**：以 AI 电商客服为例，覆盖智能问答、RAG、自主决策 Agent、对话流四种应用形态（[高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md)）。 -- **视觉创作平台**：集成 Wan2.7 图像生成与 HappyHorse 视频生成，提供节点式编排、AI 导演与在线剪辑（[HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)）。 -- **深度研究报告**：Qwen-Deep-Research 自动规划检索路径、多源交叉验证并生成结构化洞察报告（[深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md)）。 -- **AI 解题批改**：基于 Qwen3-VL 视觉模型实现拍照解题与作业自动批改，支持 33 种语言（[AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md)）。 +- **地域与权限约束**：多数第三方模型（DeepSeek-硅基流动、Kimi、GLM-智谱、MiniMax、MiMo、Stepfun）仅支持华北2（北京）地域，且需对应地域的 API Key；部分模型（如 Kimi、GLM）在新加坡/东京等地域需配置 `WorkspaceId` 域名。 +- **输入输出限制**： + - DashScopeParse 文档解析支持单文件 ≤100MB、≤1000 页；函数计算部署方案有内存与超时限制（如深度研究方案为 15 分钟）。 + - Vidu 视频生成对提示词复杂度敏感，需避免主体物过多或句式模糊；万相图生视频需注意原始图片与运动描述的逻辑一致性（如火车方向需通过比例关系强化）。 +- **成本与计费**： + - 显式缓存首次写入产生 25% 额外开销，但后续命中可降本 90%；若未发生命中，总体成本高于不启用缓存。 + - 第三方模型调用按 Token 计费，部分模型（如 GLM 系列）提供 100 万免费 Token，但需注意免费额度是否跨模型共享。 +- **兼容性风险**：`enable_thinking` 等非 OpenAI 标准参数在不同 SDK 实现中存在差异（如 Python SDK 用 `extra_body`，Node.js SDK 作顶层参数），需严格参照各模型文档示例代码。 ## 来源文档 +- [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md) +- [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) +- [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md) +- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md) - [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md) -- [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md) +- [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](../../raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) +- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) - [限流应对最佳实践 ](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) - [DeepSeek-硅基流动](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) -- [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) +- [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - [GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) - [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [GLM-智谱](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) -- [MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) +- [MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Stepfun-阶跃星辰](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) -- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) -- [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md) -- [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) -- [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md) -- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md index ee1afa81..758180af 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md @@ -1,82 +1,57 @@ # use chat client or development tool -阿里云百炼支持将平台上的模型接入各类第三方 AI 聊天客户端、编程工具与应用开发平台。这些工具本身不由百炼提供，接入方式统一为「填入 Base URL + API Key + 模型 ID」，通过 **OpenAI 兼容协议**或 **Anthropic 兼容协议**访问百炼网关。本文汇总不同工具的接入要点、共用的凭证规则以及常见限制。 +阿里云百炼支持通过多种主流 AI 开发工具和客户端接入模型服务，包括终端 CLI 工具（如 Hermes Agent、Qwen Code）、桌面 IDE（如 Cursor、Qoder CN）、开源平台（如 Dify）以及通用 HTTP 客户端（如 Postman）。所有工具均通过 OpenAI 或 Anthropic 兼容 API 协议对接，开发者可根据使用场景选择按量计费、Coding Plan 或 Token Plan 团队版三种计费方案。 -## 支持的工具类型 +## 支持的模型/功能 -按形态大致分为三类： +百炼支持的模型因计费方案而异，且需匹配对应协议（OpenAI 兼容或 Anthropic 兼容）与 Base URL。核心模型覆盖 Qwen 系列（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）、DeepSeek（如 `deepseek-v4-pro`）、Kimi（如 `kimi-k2.7-code`）、GLM（如 `glm-5.2`）及 MiniMax 等。部分模型（如 Qwen3 系列）支持思考模式（`enable_thinking: true`），需在请求体或配置中显式启用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)。 -- **终端 / CLI 编程工具**：[Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md)、[Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md)、[OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md)、[Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md)、[Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)、[Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md)、Qoder CLI。 -- **IDE / 编辑器插件**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)、[Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md)（VSCode）、Qoder（IDE / JetBrains 插件）、Qoder CN（原 Lingma）。 -- **桌面 / 跨平台聊天客户端与助手**：[Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)、[Chatbox](../../raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md)、[OpenClaw](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md)、QwenPaw。 -- **应用开发 / 工作流平台**：[Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md)。 +图像与视频生成类模型（如 `wan2.6-t2i`）**不适用**于常规聊天客户端，必须通过异步 API 调用，且仅支持直接 HTTP 请求（cURL/Postman）或 Dify 工作流等支持长轮询的平台 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md)。此外，Token Plan 团队版和 Coding Plan **明确禁止**用于工作流平台（Dify、n8n、Coze）、API 测试工具（Postman、Insomnia）或自定义后端应用——该限制在 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) 中有明确定义。 -此外，任何兼容 OpenAI / Anthropic 协议且支持自定义服务端点的工具（如 Trae）都可参照[更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)接入。若只想快速验证图像/视频生成 API，可用 [Postman 或 cURL](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) 直接调用。 +> **注意**：文档 1 和文档 2 均列出 `qwen3.6-flash` 为 Token Plan 团队版支持模型，但文档 1 的 JSON 配置中其 `contextWindow` 为 `1000000`，而文档 4（OpenCode）中同模型未声明上下文长度；文档 7（Qwen Code）则明确要求 `qwen3.6-flash` 必须启用 `enable_thinking`。实际行为以控制台公布的[Token Plan 团队版支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)为准，建议以官方模型页描述为最终依据。 -## 三种计费方案与凭证 +## 关键参数 -绝大多数工具的接入差异只在「Base URL 属于哪个方案」。百炼提供三种计费方案，各自有独立的 API Key，**互不通用**： +所有工具共用三类核心参数： -| 方案 | 说明 | OpenAI 兼容 Base URL | Anthropic 兼容 Base URL | -| --- | --- | --- | --- | -| Token Plan 团队版 | 按坐席订阅，按 token 消耗抵扣 Credits | `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | `https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | -| Coding Plan | 固定月费订阅，按模型调用次数计量 | `https://coding.dashscope.aliyuncs.com/v1` | `https://coding.dashscope.aliyuncs.com/apps/anthropic` | -| 按量计费（华北2·北京） | 按实际调用量后付费 | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `https://dashscope.aliyuncs.com/apps/anthropic` | +- **API Key**：严格按计费方案隔离。Token Plan 团队版、Coding Plan 与按量计费的 API Key **互不通用**，混用将导致 401 错误。 +- **Base URL**：必须与 API Key 所属地域及计费方案完全匹配。例如： + - Token Plan 团队版（北京）：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） + - Coding Plan：`https://coding.dashscope.aliyuncs.com/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） + - 按量计费（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） +- **Model ID**：部分工具（如 Cursor、Chatbox）要求对带点号的模型名做转换（如 `kimi-k2.6` → `kimi-k2-6`），详见各工具文档；Qwen3 系列模型在启用思考模式时，部分工具（如 Qwen Code）需在 `generationConfig.extra_body` 中设置 `"enable_thinking": true` [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)。 -按量计费还支持多地域，需保证 API Key 与 Base URL 地域一致： +## 使用方式 -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（`WorkspaceId` 替换为真实值） -- 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` +1. **安装工具**：各工具提供标准化安装路径，如 `npm install -g`（Hermes Agent、Claude Code）、一键脚本（OpenClaw、QwenPaw）、GUI 下载（Cursor、Cherry Studio）或 VS Code 插件（Cline）。 +2. **配置凭证**：绝大多数工具通过编辑配置文件（如 `~/.hermes/config.yaml`、`~/.qwen/settings.json`）或图形化设置界面完成。环境变量（如 `OPENAI_API_KEY`）在 Codex 等工具中仍被广泛使用。 +3. **验证与调用**：配置后执行简单命令（如 `hermes chat -q "你好"`）或在 GUI 中发送测试消息。对于支持多模型的工具（如 Qoder、Cursor），需在对话界面手动切换模型，且免费版（如 Cursor Free）可能限制自定义模型调用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)。 +4. **高级能力**：部分工具（Qoder、Cline、Cursor）支持通过百炼 CLI 注册 Skills，实现自然语言驱动的代码生成、图像/视频生成等扩展能力，需提前全局安装 `bailian-cli` 并配置 API Key。 -> **注意**：OpenAI 协议的 Base URL 以 `/compatible-mode/v1`（或 `/v1`）结尾，Anthropic 协议以 `/apps/anthropic` 结尾。部分工具（如 OpenCode、Kilo CLI）要求在 Anthropic 端点后再追加 `/v1`。以各工具原文为准。 +## 限制和注意事项 -## 协议选择与配置形态 - -不同工具选用的协议和配置载体各异： - -- **Anthropic 协议**：[Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) 通过 `~/.claude/settings.json` 的 `ANTHROPIC_BASE_URL` / `ANTHROPIC_AUTH_TOKEN` 环境变量配置；[Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) 默认使用 Anthropic 协议（`api_mode: anthropic_messages`），也可切到 OpenAI 协议。 -- **OpenAI 协议**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)、[Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md)、Cherry Studio、Chatbox 等在 GUI 中选择「OpenAI Compatible / 兼容」并填入 Base URL、API Key、模型 ID。 -- **配置文件**：Hermes（`~/.hermes/config.yaml`）、OpenCode（`~/.config/opencode/opencode.json`）、Kilo CLI（`~/.config/kilo/config.json`）、Qwen Code（`~/.qwen/settings.json`）、Codex（`~/.codex/config.toml` + `OPENAI_API_KEY` 环境变量）。 -- **原生下拉选择**：Qoder / Qoder CN 在设置中选择「阿里云百炼 - 国内」提供商 + 计费方案「类型」，仅需填 API Key。 - -## 关键参数与注意事项 - -- **思考模式**：许多模型（如 Qwen3 思考模式、QwQ）需显式开启思考。OpenCode / Kilo CLI 用 `thinking.budgetTokens`，Qwen Code 用 `extra_body.enable_thinking: true`，Cline 需勾选 **Enable R1 messages format**。若报错 `enable_thinking parameter is restricted to True`，说明该模型仅支持思考模式运行，需在客户端开启。 -- **模型名称别名**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) 因内置模型名冲突，需改写模型名，如 `kimi-k2.6` → `kimi-k2-6`、`glm-5` → `glm-5-0`。其他工具一般直接使用原始模型 ID。 -- **上下文窗口**：Claude Code 默认 200K，可通过 `CLAUDE_CODE_MAX_CONTEXT_TOKENS=1000000` 或模型名后缀 `[1m]` 扩展到 1M（需模型支持）。 -- **Codex 版本差异**：仅 qwen3.7-max/plus、qwen3.6-plus/flash 支持 Responses API（可用最新版 Codex）；其他模型需用 Chat/Completions API，须安装旧版本（如 `@openai/codex@0.80.0`）。 -- **401 认证失败**：几乎都是「API Key 与 Base URL 不属于同一方案」或「按量计费 Key 与地域不匹配」，逐项核对即可。 - -## 套餐使用范围限制 - -> **注意**：Token Plan 团队版与 Coding Plan **仅限**在 AI 编程工具和 OpenClaw 类 Agent 中使用。以下类型不支持接入，误用可能导致订阅暂停或 API Key 被封禁（详见[更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)）： -> -> - 工作流/自动化平台：如 Dify、n8n、Coze 等； -> - API 测试工具：如 Postman、Insomnia 等； -> - 自定义应用程序：脚本或后端代码中直接调用 API。 - -因此 [Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md) 这类应用开发平台只能通过**按量计费**（`get-api-key` 获取的 API Key）接入，且在 Dify 中通过安装「通义千问」或「OpenAI-API-compatible」插件配置。免费额度仅适用于华北2（北京）地域，且各模型额度独立、不可跨模型共享。 - -## 快速验证 - -配置完成后统一用一句问候验证连通性，例如：`claude "你好"`、`hermes chat -q "你好"`，或在 GUI 客户端对话框发送「你好」。模型正常返回响应即表示接入成功。若为 RAM 子账号，需确保在业务空间中已获得目标模型的调用权限。 +- **地域强绑定**：按量计费的 API Key 与 Base URL 必须属于同一地域（如北京 Key + 北京 URL），否则报错 401；Token Plan 团队版与 Coding Plan 的 Base URL 固定，无需选择地域。 +- **协议差异**：OpenAI 兼容端点（`/compatible-mode/v1`）接受标准 `/chat/completions` 请求；Anthropic 兼容端点（`/apps/anthropic`）需使用 `/messages` 接口及 `anthropic-messages` 协议，二者不可混用。 +- **免费额度限制**：按量计费新用户享免费额度，但**仅限华北2（北京）地域**的模型生效；使用新加坡或美国地域将立即产生费用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)。 +- **模型兼容性**：Codex 对不同模型需区分 `wire_api`（`responses` vs `chat`），且仅新版支持 Qwen3 系列；旧版 Codex（v0.80.0）是 Coding Plan 的强制要求 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md)。 +- **违规风险**：将 Token Plan 团队版或 Coding Plan 的 API Key 用于 Dify、Postman 等非授权场景，可能触发订阅暂停或 Key 封禁 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)。 ## 来源文档 -- [Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [OpenClaw](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) +- [Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - [OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) -- [Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) -- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [QwenPaw](../../raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) +- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) +- [Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) - [Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) - [Chatbox](../../raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md) +- [Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [Qoder](../../raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) - [Qoder CN（原 Lingma）](../../raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) -- [Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [使用Postman或cURL调用图像/视频生成API](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/index.md b/skills/bailian-docs-llm-wiki/wiki/index.md index b78e0fff..08ed3957 100644 --- a/skills/bailian-docs-llm-wiki/wiki/index.md +++ b/skills/bailian-docs-llm-wiki/wiki/index.md @@ -62,23 +62,20 @@ ## 横切概念 -- [API Key 鉴权](concepts/api-key.md) — 关联 6 个主题 -- [OpenAI 兼容接口](concepts/openai-compatible-interface.md) — 关联 5 个主题 -- [Token 与计量计费](concepts/token.md) — 关联 5 个主题 -- [业务空间（Workspace）](concepts/workspace.md) — 关联 5 个主题 -- [函数调用（Function Calling）](concepts/function-calling.md) — 关联 5 个主题 -- [多模态能力](concepts/multimodal.md) — 关联 5 个主题 -- [异步调用与任务轮询](concepts/async-invocation.md) — 关联 5 个主题 -- [检索增强生成（RAG）](concepts/rag.md) — 关联 6 个主题 -- [模型微调与生产链路](concepts/fine-tuning.md) — 关联 5 个主题 -- [流式输出](concepts/streaming.md) — 关联 3 个主题 -- [评测体系](concepts/evaluation.md) — 关联 3 个主题 +- [OpenAI 兼容接口](concepts/openai-compatible-api.md) — 关联 5 个主题 +- [Token 计量与管理](concepts/token.md) — 关联 5 个主题 +- [函数调用](concepts/function-calling.md) — 关联 4 个主题 +- [多模态能力](concepts/multi-modal.md) — 关联 5 个主题 +- [插件机制](concepts/plugin.md) — 关联 5 个主题 +- [检索增强生成](concepts/rag.md) — 关联 5 个主题 +- [流式输出](concepts/streaming-output.md) — 关联 5 个主题 +- [长期记忆](concepts/long-term-memory.md) — 关联 4 个主题 ## 对比分析 -- [图像、视频与3D生成对比](comparisons/media-generation-compare.md) — 对比 3 个主题 -- [应用监控与模型监控对比](comparisons/monitoring-compare.md) — 对比 2 个主题 -- [应用评测与模型评测对比](comparisons/evaluation-compare.md) — 对比 2 个主题 -- [模型微调、压缩与高速推理对比](comparisons/model-optimization-compare.md) — 对比 3 个主题 -- [知识库与长期记忆对比](comparisons/knowledge-memory-compare.md) — 对比 3 个主题 +- [图像、视频与3D生成能力对比](comparisons/image-video-3d-generation.md) — 对比 3 个主题 +- [应用编排能力对比：托管智能体、应用组件与模型上下文协议](comparisons/application-orchestration.md) — 对比 3 个主题 +- [模型评估与监控体系对比](comparisons/model-evaluation-monitoring.md) — 对比 3 个主题 +- [模型部署方案对比：高并发推理、生产部署与压缩优化](comparisons/model-deployment-options.md) — 对比 3 个主题 +- [长期记忆与知识库方案对比](comparisons/memory-solutions.md) — 对比 3 个主题 From ec935425ba3ff7176dba4e8922853a994af62300 Mon Sep 17 00:00:00 2001 From: bailian-bot Date: Wed, 15 Jul 2026 06:29:57 +0000 Subject: [PATCH 02/13] chore: update bailian-docs-llm-wiki (2026-07-15) --- skills/bailian-docs-llm-wiki/SKILL.md | 12 + skills/bailian-docs-llm-wiki/llms.txt | 904 +++-- .../models/families.jsonl | 154 + .../models/groups/Kimi-K2.json | 646 ++++ .../models/groups/MiniMax-M2.1.json | 224 ++ .../groups/MiniMax-speech-market-place.json | 278 ++ .../models/groups/aitryon-parsing-v1.json | 79 + .../models/groups/aitryon-plus.json | 68 + .../models/groups/aitryon-refiner.json | 150 + .../models/groups/aitryon.json | 68 + .../groups/animate-anyone-detect-gen2.json | 77 + .../models/groups/animate-anyone-gen2.json | 79 + .../groups/animate-anyone-template-gen2.json | 79 + .../models/groups/cosyvoice.json | 500 +++ .../models/groups/deepseek.json | 374 +++ .../models/groups/embedding.json | 288 ++ .../models/groups/emo-detect-v1.json | 77 + .../models/groups/emo-v1.json | 85 + .../models/groups/emoji-detect-v1.json | 77 + .../models/groups/emoji-v1.json | 79 + .../models/groups/facechain-facedetect.json | 56 + .../models/groups/facechain-generation.json | 72 + .../models/groups/farui-plus.json | 119 + .../models/groups/fun-asr-flash.json | 86 + .../models/groups/fun-asr-realtime.json | 160 + .../models/groups/fun-asr.json | 118 + .../models/groups/fun-music.json | 140 + .../models/groups/glm-4.5.json | 1180 +++++++ .../models/groups/glm-fast.json | 126 + .../models/groups/gui-plus.json | 112 + .../models/groups/gummy-chat-v1.json | 68 + .../models/groups/gummy-realtime-v1.json | 68 + .../models/groups/happyhorse-i2v.json | 26 + .../models/groups/happyhorse-r2v.json | 210 ++ .../models/groups/happyhorse-t2v.json | 62 + .../models/groups/happyhorse-video-edit.json | 106 + .../models/groups/image-erase-completion.json | 65 + .../groups/image-instance-segmentation.json | 65 + .../models/groups/image-out-painting.json | 71 + .../groups/kimi-models-market-place.json | 432 +++ .../groups/kling-models-market-place.json | 395 +++ .../models/groups/liveportrait-detect.json | 77 + .../models/groups/liveportrait.json | 79 + .../groups/minimax-models-market-place.json | 437 +++ .../models/groups/paraformer-8k-v1.json | 62 + .../models/groups/paraformer-8k-v2.json | 74 + .../models/groups/paraformer-mtl-v1.json | 62 + .../groups/paraformer-realtime-8k-v1.json | 60 + .../groups/paraformer-realtime-8k-v2.json | 50 + .../models/groups/paraformer-realtime-v1.json | 62 + .../models/groups/paraformer-realtime-v2.json | 56 + .../models/groups/paraformer-v1.json | 62 + .../models/groups/paraformer-v2.json | 56 + .../groups/pixverse-c1-market-place.json | 535 +++ .../models/groups/pixverse-market-place.json | 517 +++ .../groups/pixverse-v6-market-place.json | 545 +++ .../models/groups/qvq-max.json | 122 + .../models/groups/qvq-plus.json | 122 + .../groups/qwen-audio-realtime-flash.json | 119 + .../groups/qwen-audio-realtime-plus.json | 119 + .../models/groups/qwen-audio-tts.json | 170 + .../models/groups/qwen-coder-plus.json | 121 + .../models/groups/qwen-coder-turbo.json | 121 + .../models/groups/qwen-doc-turbo.json | 131 + .../models/groups/qwen-embedding.json | 488 +++ .../models/groups/qwen-flash-character.json | 121 + .../models/groups/qwen-flash.json | 330 ++ .../models/groups/qwen-image-2.0-pro.json | 91 + .../models/groups/qwen-image-2.0.json | 91 + .../models/groups/qwen-image-edit-max.json | 76 + .../models/groups/qwen-image-edit.json | 164 + .../models/groups/qwen-image-plus.json | 198 ++ .../models/groups/qwen-long.json | 250 ++ .../models/groups/qwen-math-plus.json | 460 +++ .../models/groups/qwen-math-turbo.json | 121 + .../models/groups/qwen-max.json | 149 + .../models/groups/qwen-mt-flash.json | 93 + .../models/groups/qwen-mt-image.json | 95 + .../models/groups/qwen-mt-lite.json | 92 + .../models/groups/qwen-mt-plus.json | 92 + .../models/groups/qwen-mt-turbo.json | 98 + .../groups/qwen-omni-turbo-realtime.json | 266 ++ .../models/groups/qwen-omni-turbo.json | 327 ++ .../models/groups/qwen-plus-character.json | 122 + .../models/groups/qwen-plus.json | 1025 ++++++ .../models/groups/qwen-rerank.json | 276 ++ .../models/groups/qwen-tts-realtime.json | 208 ++ .../models/groups/qwen-tts.json | 156 + .../models/groups/qwen-turbo.json | 211 ++ .../models/groups/qwen-vl-embedding.json | 172 + .../models/groups/qwen-vl-max.json | 155 + .../models/groups/qwen-vl-ocr.json | 370 +++ .../models/groups/qwen-voice-design.json | 75 + .../models/groups/qwen-voice-enrollment.json | 68 + .../models/groups/qwen2.5.json | 139 + .../groups/qwen3-asr-flash-filetrans.json | 88 + .../groups/qwen3-asr-flash-realtime.json | 86 + .../models/groups/qwen3-asr-flash.json | 63 + .../groups/qwen3-coder-30b-a3b-instruct.json | 185 ++ .../qwen3-coder-480b-a35b-instruct.json | 184 ++ .../models/groups/qwen3-coder-flash.json | 258 ++ .../models/groups/qwen3-coder-plus.json | 264 ++ .../qwen3-livetranslate-flash-realtime.json | 122 + .../groups/qwen3-livetranslate-flash.json | 113 + .../models/groups/qwen3-max.json | 651 ++++ .../groups/qwen3-omni-30b-a3b-captioner.json | 100 + .../groups/qwen3-omni-flash-realtime.json | 146 + .../models/groups/qwen3-omni-flash.json | 175 + .../groups/qwen3-tts-flash-realtime.json | 78 + .../models/groups/qwen3-tts-flash.json | 72 + .../qwen3-tts-instruct-flash-realtime.json | 94 + .../groups/qwen3-tts-instruct-flash.json | 101 + .../models/groups/qwen3-tts-vc-realtime.json | 75 + .../models/groups/qwen3-tts-vc.json | 74 + .../models/groups/qwen3-tts-vd-realtime.json | 95 + .../models/groups/qwen3-tts-vd.json | 75 + .../models/groups/qwen3-vl-flash.json | 336 ++ .../models/groups/qwen3-vl-plus.json | 331 ++ .../qwen3.5-livetranslate-flash-realtime.json | 118 + .../models/groups/qwen3.5-ocr.json | 113 + .../groups/qwen3.5-omni-flash-realtime.json | 141 + .../models/groups/qwen3.5-omni-flash.json | 143 + .../groups/qwen3.5-omni-plus-realtime.json | 141 + .../models/groups/qwen3.5-omni-plus.json | 184 ++ .../models/groups/qwen3.5.json | 1006 ++++++ .../models/groups/qwen3.7-max.json | 53 + .../models/groups/qwen3.json | 2803 ++++++++++++++++ .../models/groups/qwq-plus.json | 134 + .../models/groups/sambert.json | 2930 +++++++++++++++++ .../models/groups/shoemodel-v1.json | 65 + .../models/groups/siliconflow-models.json | 432 +++ .../models/groups/speech-biasing.json | 70 + .../groups/stepfun-models-market-place.json | 124 + .../groups/tongyi-intent-detect-v3.json | 111 + .../groups/tongyi-xiaomi-analysis-flash.json | 109 + .../groups/tongyi-xiaomi-analysis-pro.json | 109 + .../groups/tripo-models-market-place.json | 338 ++ .../groups/vanchin-models-market-place.json | 678 ++++ .../models/groups/video-style-transform.json | 79 + .../models/groups/videoretalk.json | 74 + .../vidu-image-models-market-place.json | 390 +++ .../models/groups/virtualmodel-v2.json | 65 + .../models/groups/voice-enrollment.json | 44 + .../models/groups/wan-image-edit.json | 437 +++ .../models/groups/wan-image-to-video.json | 1330 ++++++++ .../models/groups/wan-reference-to-video.json | 316 ++ .../models/groups/wan-text-to-image.json | 783 +++++ .../models/groups/wan-text-to-video.json | 605 ++++ .../models/groups/wan-video-edit.json | 111 + .../groups/wanx-background-generation-v2.json | 79 + .../groups/wanx-poster-generation-v1.json | 65 + .../groups/wanx-sketch-to-image-lite.json | 75 + .../models/groups/wanx-style-repaint-v1.json | 73 + .../models/groups/wanx-virtualmodel.json | 65 + .../models/groups/wanx-x-painting.json | 67 + .../models/groups/wanx2.1-vace-plus.json | 93 + .../models/groups/wordart-semantic.json | 79 + .../models/groups/wordart-texture.json | 80 + .../groups/xiaomi-models-market-place.json | 111 + .../models/groups/z-image-turbo.json | 90 + .../groups/zhipu-models-market-place.json | 324 ++ .../bailian-docs-llm-wiki/models/index.json | 2785 +++++++++++++++- skills/bailian-docs-llm-wiki/models/index.md | 374 ++- .../bailian-docs-llm-wiki/models/models.jsonl | 333 ++ ...i-bailian-2023-12-29-batchupdatefiletag.md | 246 -- .../wanx-sketch-to-image-api-reference.md | 968 ------ .../get-temporary-file-url.md | 1185 ------- .../omni-realtime-interaction-process.md | 76 - skills/bailian-docs-llm-wiki/raw/test/test.md | 1 - .../wiki/api/3d-generation.md | 75 +- .../wiki/api/application-call.md | 65 +- .../application-component-api-reference.md | 101 +- .../wiki/api/file-management-api.md | 55 +- .../wiki/api/frameworks.md | 55 +- .../wiki/api/image-generation.md | 81 +- .../wiki/api/knowledge.md | 41 +- .../wiki/api/long-term-memory-new.md | 92 +- .../wiki/api/managed-agents-api.md | 81 +- .../wiki/api/model-production.md | 42 +- .../wiki/api/more-about-models.md | 67 +- .../wiki/api/more-models.md | 104 +- skills/bailian-docs-llm-wiki/wiki/api/more.md | 70 +- .../wiki/api/omni-realtime-api.md | 92 +- .../wiki/api/preparations.md | 70 +- .../wiki/api/qwen-api-reference.md | 57 +- .../wiki/api/toolkits-and-frameworks.md | 130 +- .../wiki/api/vector-and-sort.md | 91 +- .../wiki/api/video-generation-api.md | 103 +- .../comparisons/application-frameworks.md | 69 + .../comparisons/application-orchestration.md | 81 - .../comparisons/image-video-3d-generation.md | 70 - .../comparisons/image-vs-video-generation.md | 65 + .../wiki/comparisons/memory-solutions.md | 85 +- .../comparisons/model-deployment-options.md | 100 +- .../model-evaluation-monitoring.md | 91 +- .../wiki/concepts/function-calling.md | 48 +- .../wiki/concepts/long-term-memory.md | 55 +- .../wiki/concepts/model-context-protocol.md | 50 + .../wiki/concepts/multi-modal.md | 80 - .../wiki/concepts/openai-compatible-api.md | 82 - .../concepts/openai-compatible-interface.md | 74 + .../wiki/concepts/plugin.md | 69 +- .../wiki/concepts/rag.md | 57 +- .../wiki/concepts/streaming-output.md | 85 +- .../wiki/concepts/token.md | 57 +- .../wiki/guides/application-evaluation.md | 61 +- .../wiki/guides/application-monitoring.md | 91 +- .../application-permission-management.md | 69 +- .../application-publishing-and-sharing.md | 58 +- .../wiki/guides/application-support.md | 39 +- .../wiki/guides/application-use-cases.md | 57 +- .../guides/bailian-application-calling.md | 103 +- .../wiki/guides/data-connection-overview.md | 61 +- .../wiki/guides/fine-tuning.md | 76 +- .../wiki/guides/get-started-with-models.md | 76 +- .../wiki/guides/knowledge-base.md | 49 +- .../wiki/guides/llm-application.md | 102 +- .../wiki/guides/managed-agents.md | 57 +- .../wiki/guides/memory-library-overview.md | 52 +- .../wiki/guides/model-compression.md | 48 +- .../wiki/guides/model-context-protocol.md | 90 +- .../wiki/guides/model-data-overview.md | 38 +- .../wiki/guides/model-deployment-1.md | 55 +- .../guides/model-evaluation-introduction.md | 85 +- .../wiki/guides/model-experience.md | 87 +- .../wiki/guides/model-high-speed-inference.md | 42 +- .../wiki/guides/model-monitoring.md | 88 +- .../wiki/guides/plug-in.md | 71 +- .../wiki/guides/prompt.md | 55 +- .../wiki/guides/release-notes.md | 77 +- .../wiki/guides/security-and-compliance.md | 79 +- .../wiki/guides/skill.md | 31 +- .../wiki/guides/start-using.md | 51 +- .../wiki/guides/support.md | 57 +- .../wiki/guides/test-1.md | 44 +- .../wiki/guides/token-plan-guide.md | 66 +- .../wiki/guides/use-cases.md | 68 +- .../use-chat-client-or-development-tool.md | 53 +- skills/bailian-docs-llm-wiki/wiki/index.md | 26 +- 239 files changed, 42683 insertions(+), 5513 deletions(-) create mode 100644 skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/aitryon.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/embedding.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/emo-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/farui-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-asr.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/fun-music.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/glm-fast.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/gui-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/liveportrait.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qvq-max.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-long.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-max.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/sambert.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/videoretalk.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/test/test.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-context-protocol.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md diff --git a/skills/bailian-docs-llm-wiki/SKILL.md b/skills/bailian-docs-llm-wiki/SKILL.md index 9d0c87dc..4c8b7d7e 100644 --- a/skills/bailian-docs-llm-wiki/SKILL.md +++ b/skills/bailian-docs-llm-wiki/SKILL.md @@ -92,7 +92,19 @@ description: >- | `TG` | 文本生成 | | `Reasoning` | 推理 | | `VU` | 视觉理解 | +| `IG` | 图像生成 | | `VG` | 视频生成 | +| `TTS` | 语音合成 | +| `ASR` | 语音识别 | +| `Realtime-ASR` | 实时语音识别 | +| `Realtime-Text-to-Speech` | 实时语音合成 | +| `Realtime-Audio-Translate` | 实时音频翻译 | +| `Realtime-Omni` | 实时全模态 | +| `Multimodal-Omni` | 全模态 | +| `ME` | 多模态嵌入 | +| `TR` | 翻译 | +| `3D-generation` | 3D 生成 | +| `Realtime-Chatting` | Realtime-Chatting | 一个模型常常带多个 capability，`index.md` 中按 `capabilities[0]`（主能力）归类， 查找时按中文标签即可定位章节。 diff --git a/skills/bailian-docs-llm-wiki/llms.txt b/skills/bailian-docs-llm-wiki/llms.txt index 95cdb18f..58b093e0 100644 --- a/skills/bailian-docs-llm-wiki/llms.txt +++ b/skills/bailian-docs-llm-wiki/llms.txt @@ -8,102 +8,102 @@ - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [选择模型](raw/model-user-guide/get-started-with-models/models.md) - - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) + - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) -- **产品计费** - - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) - - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) - - [节省计划与资源包](raw/model-user-guide/test-1/savings-plan-and-resource-package.md) - - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) - - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) - **Token Plan（团队版）** - - **最佳实践** - - [工具调用](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) - - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) - **Coding Plan** - - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) - - [联网搜索](raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) + - [联网搜索](raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) - - [Token Plan（团队版）概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) + - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) + - **最佳实践** + - [工具调用](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) + - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team.md) + - [Token Plan（团队版）概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-faq.md) +- **产品计费** + - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) + - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) + - [节省计划与资源包](raw/model-user-guide/test-1/savings-plan-and-resource-package.md) + - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) + - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) - **模型体验** - - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) + - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) - - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [语音合成](raw/model-user-guide/model-experience/tts-model.md) + - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - [语音识别](raw/model-user-guide/model-experience/asr-model.md) - - [全模态](raw/model-user-guide/model-experience/omni.md) - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) + - [全模态](raw/model-user-guide/model-experience/omni.md) - **接入客户端/开发工具** - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) - - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) + - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) + - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) + - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) -- **模型推理** - - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) + - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - **模型调优** - **千问模型调优** - [模型调优简介](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) - - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - [使用 API 或命令行进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) + - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - **语音合成模型调优** - [CosyVoice模型调优](raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) +- **模型推理** + - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) + - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - **模型部署** - - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) + - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) -- **模型评测** - - [评测维度](raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - **模型压缩** - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) +- **模型评测** + - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) + - [评测维度](raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - **用量统计与性能监控** - - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) + - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) - **模型数据** - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) - **安全合规** - - **传输安全** - - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - **安全存储** - [配置终端节点并发起连接](raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) - [配置可用区IP](raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - [配置私有网络中的资源](raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - [配置MSE云原生网关](raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) + - **传输安全** + - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) + - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) + - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - [模型备案信息公示](raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) - - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - [合规资质与隐私说明](raw/model-user-guide/security-and-compliance/privacy-notice.md) + - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) + - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - **服务支持** - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - [相关协议](raw/model-user-guide/support/related-agreements.md) @@ -112,11 +112,11 @@ - [DeepSeek-阿里云](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) - - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [Kimi](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) + - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [GLM](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) - - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) + - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) @@ -124,79 +124,75 @@ - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - [高效搭建 AI 智能体与工作流应用](raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](raw/model-user-guide/use-cases/prompt-engineering-guide.md) - - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) + - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生视频/图生视频Prompt指南](raw/model-user-guide/use-cases/text-to-video-prompt.md) - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) + - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) + - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - **产品动态** - - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) + - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) ## 应用使用指南 -- **应用开发** - - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) - - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) - - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) - - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) - - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) - - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) -- **开始使用** - - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **Managed Agents** - - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) + - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - [构建 Agent](raw/application-user-guide/managed-agents/managed-agents-agent.md) - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) +- **应用开发** + - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) + - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) + - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) + - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) + - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) + - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) - **Prompt** - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) - - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) + - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) +- **开始使用** + - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) + - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **记忆库** - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) - - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) + - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) +- **数据连接** + - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) - **知识库（RAG）** - - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) + - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [知识库日志与监控](raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) + - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识问答](raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) - **MCP** - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) + - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) -- **Skill** - - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) -- **数据连接** - - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) - **插件** - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) +- **Skill** + - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) - **应用发布与分享** - - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [UI设计器](raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) -- **应用调用** - - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) + - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) - **应用评测** - **新版应用评测** - [新版评测集](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) @@ -206,36 +202,54 @@ - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - [手动评测](raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) +- **应用调用** + - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) + - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) + - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) +- **权限管理** + - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) +- **应用观测** + - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **应用广场** - - **官方应用-通义音频播客生成** - - **API参考** - - **API目录** - - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) - - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) - - [通义音频播客生成产品介绍](raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md) - **官方应用-通义拍照解题辅导** - **API参考** - **API目录** - [CutQuestions - 试卷切题](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-dir/api-edututor-2025-07-07-cutquestions.md) - [AnswerSSE - 解题辅导](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-dir/api-edututor-2025-07-07-answersse.md) - - [API概览](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-endpoint.md) + - [API概览](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-overview.md) - [授权信息](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-ram.md) - [通义拍照解题辅导产品介绍](raw/application-user-guide/application-gallery/edu-tutor/brief-introduction-of-edu-tutor.md) + - **官方应用-通义音频播客生成** + - **API参考** + - **API目录** + - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) + - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) + - [通义音频播客生成产品介绍](raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md) - **官方应用-多模态交互开发套件** - **使用指南** - [应用创建](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-creation.md) - [应用配置](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-configuration.md) - [应用体验与发布](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-experience-and-publishing.md) - - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - [指令列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/instruction-list.md) - - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) + - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) - [三方Agent接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-integration-a2a.md) - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) + - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) + - **SDK安装** + - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) + - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) + - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) + - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) + - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) + - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) + - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) + - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) + - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) - **API参考** - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) @@ -244,22 +258,9 @@ - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) - - **SDK安装** - - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) - - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) - - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) - - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) - - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) + - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) - **最佳实践** - - **接入拍照问答Agent** - - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - **接入百炼及三方Agent** - [百炼及三方Agent直连调用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/agent-direct-call.md) - [接入百炼智能体应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-app.md) @@ -271,73 +272,87 @@ - [录音纪要Agent使用教程](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/recording-summary-agent-tutorial.md) - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) + - **接入拍照问答Agent** + - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) + - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - [接入多模态备忘录Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/multimodal-memo-agent.md) - - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [动作情绪控制实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/action-emotion-control-practice.md) - - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) + - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) + - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) - [基于RTOS SDK (License模式) 实现聊天能力](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/chat-capability-based-on-rtos-sdk.md) - - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) + - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) - [多模态交互开发套件常见问题](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-faq.md) - **官方应用-全妙轻应用系列** - - **使用指南** - - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) - - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) - - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) - - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) - - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) - - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) - - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - **计费说明（全妙轻应用）** - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) - - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) + - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) + - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) + - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) - - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) - - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) + - **使用指南** + - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) + - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) + - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) + - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) + - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) + - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) + - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) + - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) + - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - **开发文档** + - **最佳实践** + - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) + - [挖掘VOC信息和数据分析的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-mining-voc-information-and-data-analysis.md) + - [阿里云百炼工作流集成视频理解最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md) - **API参考** - **数据结构** - [ModelUsage](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-struct-dir/api-quanmiaolightapp-2024-08-01-struct-modelusage.md) - **API目录** - - **传媒/零售文章风格与格式学习** - - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **电商零售推广文案写作** - - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) + - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - **影视互娱剧本创作** - - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) + - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) + - **传媒/零售文章风格与格式学习** + - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **影视传媒视频理解** - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) + - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) - - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) + - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - [UpdateVideoAnalysisTasks - 视频理解-批量取消任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistasks.md) - **影视传媒智能拆条** - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) - - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) + - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) + - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) - **车机网络热点信息互动问答** - [RunHotTopicChat - 播报单（热榜）问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicchat.md) - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) + - **泛企业VOC挖掘** + - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) + - **作文批改** + - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) + - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) + - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) + - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) - **泛企业线索挖掘** - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) @@ -347,89 +362,34 @@ - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) - - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) + - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) - - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC挖掘异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submitenterprisevocanalysistask.md) - - **泛企业VOC挖掘** - - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - - **作文批改** - - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) - - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) - - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) - - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) + - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - **网络内容安全审核** - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) + - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-ram.md) - - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-changeset.md) - - **最佳实践** - - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) - - [挖掘VOC信息和数据分析的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-mining-voc-information-and-data-analysis.md) - - [阿里云百炼工作流集成视频理解最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md) - [全妙轻应用更新公告](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-update-announcement.md) - [常见问题](raw/application-user-guide/application-gallery/quanmiao-light-application-series/quanmiao-lightapp-faq.md) - - **官方应用-伶鹊CCAI-对话分析AIO** - - **使用指南** - - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) - - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) - - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) - - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) - - **API参考** - - **API目录** - - **热词管理** - - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) - - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) - - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) - - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) - - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) - - **不推荐或白名单开放** - - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) - - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) - - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) - - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) - - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) - - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) - - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) - - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) - - **接口调用示例** - - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) - - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) - - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) - - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md) - - **最佳实践** - - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) - - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) - - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) - - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) - - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - **官方应用-伶鹊CCAI-语音对话机器人** - **API参考** - **API目录** - **MQ消息订阅配置** - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) - - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) + - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - **变量管理** - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) - - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) + - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) - **三方语音配置** - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) @@ -442,63 +402,158 @@ - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) - - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) - - **克隆音管理** - - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) - - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) - - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) - - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) + - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - **应用管理** - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) + - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) - - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) + - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) - - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) + - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) + - **克隆音管理** + - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) + - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) + - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) + - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) + - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) - - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) + - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md) - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) - - **官方应用-伶鹊CCAI-客服对话Agent** + - **官方应用-伶鹊CCAI-对话分析AIO** + - **使用指南** + - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) + - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) + - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) + - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) + - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) + - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) - **API参考** - **API目录** - - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) - - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) - - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) + - **热词管理** + - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) + - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) + - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) + - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) + - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) + - **不推荐或白名单开放** + - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) + - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) + - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) + - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) + - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) + - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) + - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) + - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) + - **最佳实践** + - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) + - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) + - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) + - **接口调用示例** + - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) + - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) + - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) + - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) + - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md) + - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) + - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) + - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) + - **官方应用-通义数据挖掘** + - **API参考** + - **API目录** + - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) + - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) + - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) + - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) + - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) + - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) + - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) + - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) + - **官方应用-通义多模态翻译** + - **API参考** + - **API目录** + - **文本翻译** + - [BatchTranslate - 批量文本翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-batchtranslate.md) + - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) + - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) + - [SubmitLongTextTranslateTask - 提交长文本翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submitlongtexttranslatetask.md) + - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) + - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) + - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) + - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) + - **文档翻译** + - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) + - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) + - **图片翻译** + - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) + - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md) + - [网页翻译JSSDK](raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md) + - [通义多模态翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/official-application-tongyi-translate-overview.md) + - **官方应用-通义深度搜索** + - **API参考** + - **API目录** + - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) + - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) + - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) + - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) + - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) + - [服务接入节点](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-service-access-point.md) + - [错误码-通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-error-code.md) + - [通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/tongyi-deep-search-introduction.md) + - [操作指南](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-guide.md) + - **官方应用-千问联网检索Agent** + - **API参考** + - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) + - [生成对话](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat.md) + - [服务接入点](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/service-access-point.md) + - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) + - [千问联网检索Agent产品简介](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-guide.md) + - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) + - **通义 UI Agent** + - [通义 UI Agent](raw/application-user-guide/application-gallery/ui-agent/ui-agent-api.md) - **通义点金** - **API参考** - **API目录** - **平台能力-文档库** - [UpdateDocumentChunk - 更新文档块内容](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocumentchunk.md) + - [CreateLibrary - 创建文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createlibrary.md) - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) - - [CreateLibrary - 创建文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createlibrary.md) - [GetLibrary - 获取文档库详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrary.md) - [GetDocumentUrl - 获取文档的下载链接](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumenturl.md) - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) + - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) - - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) + - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) + - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) - [RecallDocument - 文档召回](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-recalldocument.md) - [GetParseResult - 获取文档解析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getparseresult.md) @@ -513,29 +568,29 @@ - [RunDialogAnalysis - 会话分析结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rundialoganalysis.md) - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) - - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) + - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) + - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) - [GetDialogAnalysisResult - 获取会话分析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoganalysisresult.md) - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) + - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) + - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) - - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) + - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) - [UpdateQaLibrary - 更新QA问答库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-updateqalibrary.md) + - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - [SubmitChatQuestion - 提交问题列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-submitchatquestion.md) - [RunChatResultGeneration - 对话结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runchatresultgeneration.md) - - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - **其他** - [DashscopeAsyncTaskFinishEvent - Dashscope异步任务完成回调事件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-other/api-dianjin-2024-06-28-dashscopeasynctaskfinishevent.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) @@ -543,83 +598,36 @@ - [授权信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-ram.md) - [版本说明](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-changeset.md) - [产品简介](raw/application-user-guide/application-gallery/tongyi-dianjin/tongyi-dianjin-overview.md) - - **官方应用-通义数据挖掘** - - **API参考** - - **API目录** - - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) - - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) - - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) - - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) - - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) - - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) - - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - - **官方应用-通义深度搜索** - - **API参考** - - **API目录** - - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) - - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) - - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) - - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) - - [服务接入节点](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-service-access-point.md) - - [错误码-通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-error-code.md) - - [通义深度搜索](raw/application-user-guide/application-gallery/tongyi-deepsearch/tongyi-deep-search-introduction.md) - - [操作指南](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-guide.md) - - **官方应用-通义多模态翻译** + - **官方应用-伶鹊CCAI-客服对话Agent** - **API参考** - **API目录** - - **文本翻译** - - [BatchTranslate - 批量文本翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-batchtranslate.md) - - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) - - [SubmitLongTextTranslateTask - 提交长文本翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submitlongtexttranslatetask.md) - - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) - - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) - - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) - - **图片翻译** - - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) - - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) - - **文档翻译** - - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md) - - [通义多模态翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/official-application-tongyi-translate-overview.md) - - [网页翻译JSSDK](raw/application-user-guide/application-gallery/official-application-tongyi-translate/web-page-translation-jssdk.md) - - **官方应用-千问联网检索Agent** - - **API参考** - - [服务接入点](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/service-access-point.md) - - [API概览](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-overview.md) - - [生成对话](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat.md) - - [多模态文件操作](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-api/web-search-agent-api-chat-multimodal-file.md) - - [千问联网检索Agent产品简介](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-guide.md) - - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) - - **通义 UI Agent** - - [通义 UI Agent](raw/application-user-guide/application-gallery/ui-agent/ui-agent-api.md) + - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) + - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) + - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) - **官方应用-全妙解决方案类产品** + - **妙搜和妙读** + - **使用指南** + - [妙搜](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaodou-and-miaodu-guidelines-for-use/ai-miaosou.md) + - [计费说明（妙搜和妙读）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaosou-miaodu-api-billing.md) - **妙笔、妙策和审校** - **使用指南** - **AI妙笔** - **功能界面** - - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) + - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) - [妙笔首页概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/amb-homepage-overview.md) - - [AI工具箱](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/ai-toolbox.md) - - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [素材库](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/material-library.md) + - [AI工具箱](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/ai-toolbox.md) - [文章风格和格式学习](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/style-imitation.md) - - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) + - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [AI妙策](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/ai-miaoce.md) + - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) - [深度写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/deep-writing.md) - **文本写作指导** - **传媒类文体写作指导** @@ -627,9 +635,9 @@ - [没有思路，要谋篇布局](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/use-amb-to-help-writing.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) - **政务公文写作指导** - - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) + - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - [常见FAQ](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/faq-for-using-quanmiao-series-products.md) - **更新公告** - **功能更新** @@ -637,35 +645,23 @@ - [2024年3月11更新-AI全妙系列 V2.2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-11-ai-quanmiao-v2-2.md) - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) + - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) - - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) - - [计费说明（PPT生成）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/ppt-generation-billing.md) - [计费说明（妙策-自定义数据源）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-document-miaoce-custom-data-source.md) + - [计费说明（PPT生成）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/ppt-generation-billing.md) + - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) - [计费说明（视频混剪）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-description-video-mixing.md) - - **妙搜和妙读** - - **使用指南** - - [妙搜](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaodou-and-miaodu-guidelines-for-use/ai-miaosou.md) - - [计费说明（妙搜和妙读）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaosou-miaodu-api-billing.md) - **开发文档** - - **最佳实践** - - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) - - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) - - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) - - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) - - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) - - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - **API参考** - **数据结构** - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) - - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) + - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) - [WritingStyleTemplateField](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatefield.md) - **API目录** - **通用接口** @@ -676,61 +672,61 @@ - **通用接口-文件上传下载** - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) + - **通用接口-异步任务管理** + - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) + - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) + - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) + - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - **通用接口-通用配置** - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - - **通用接口-异步任务管理** - - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) - - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) - - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - **妙笔-创作文章** - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) + - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) + - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) - - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - [RunWriteToneGeneration - 文风改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritetonegeneration.md) - - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) - [RunSummaryGenerate - 摘要生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runsummarygenerate.md) + - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) + - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) + - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - [ListBuildConfigs - 获取系统自定义预设](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-listbuildconfigs.md) + - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - [GenerateImageTask - 生成智能配图任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-generateimagetask.md) - [FeedbackDialogue - 反馈对话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-feedbackdialogue.md) - - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - **妙笔-文体仿写** + - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) - - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - [DeleteStyleLearningResult - 删除自定义文体](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-deletestylelearningresult.md) - - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) + - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) - **妙笔-视频审校** - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文章审校-规则库管理** - [SubmitAuditNote - 提交自定义规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-submitauditnote.md) - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) + - [DownloadAuditNote - 下载规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-downloadauditnote.md) - [DeleteAuditNote - 删除规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-deleteauditnote.md) - [GetAuditNotePostProcessingStatus - 获取规则库后处理进度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnotepostprocessingstatus.md) - - [DownloadAuditNote - 下载规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-downloadauditnote.md) - [GetAvailableAuditNotes - 查询可用规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getavailableauditnotes.md) - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) - **妙笔-文章审校-词库管理** - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) - - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) + - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) - [FetchExportTermsTask - 获取导出词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchexporttermstask.md) @@ -742,24 +738,24 @@ - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md) - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md) - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md) + - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - [GetSmartAuditResult - 查询智能审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-getsmartauditresult.md) - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) - - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - [ExportAuditContentResult - 导出智能审校报告](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-exportauditcontentresult.md) - **妙笔-文档管理** - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) - - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) + - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) - - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) + - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - **妙笔-素材库** - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) - - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) + - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - **妙笔-素材库-自定义文本** - [GetCustomText - 获取自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-getcustomtext.md) @@ -768,58 +764,51 @@ - [SaveCustomText - 保存自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-savecustomtext.md) - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - - **妙策-自定义数据源** - - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) - - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) - - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) - **妙笔-视频混剪** - - [GetClipsBuildInResource - 获取智能混剪内置资源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getclipsbuildinresource.md) - [AsyncCreateClipsTimeLine - 创建剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstimeline.md) + - [GetClipsBuildInResource - 获取智能混剪内置资源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getclipsbuildinresource.md) - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - [AsyncUploadVideo - 异步上传视频剪辑素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncuploadvideo.md) - - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) - - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) + - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) + - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) + - **妙策-自定义数据源** + - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) + - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) + - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) + - **公文库检索** + - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-选题热点** - - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) + - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) - - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) - - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) + - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) + - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) - [ListWebReviewPoints - 获取网友视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listwebreviewpoints.md) - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) - - **公文库检索** - - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-自定义话题** - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) + - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) + - **妙策-新闻播报** + - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) + - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) + - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - **妙策-openapi** - [SubmitDocClusterTask - 提交内容聚合任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitdocclustertask.md) - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) + - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) - [SubmitCustomTopicSelectionPerspectiveAnalysisTask - 提交自定义热点选题视角分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitcustomtopicselectionperspectiveanalysistask.md) - - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - - **妙策-新闻播报** - - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) - - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) - - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - - **妙策-企业VOC挖掘** - - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) - - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) - - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) - - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) - - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) + - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - **妙搜-数据源** - [GetDatasetDocument - 数据源-获取文档详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdatasetdocument.md) - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) @@ -827,128 +816,139 @@ - [ListDatasetDocuments - 数据源-文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasetdocuments.md) - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - - **妙搜-智能搜索** - - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) - - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) - - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) - - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) + - **妙策-企业VOC挖掘** + - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) + - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) + - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) + - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) + - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) - **系统配置-干预配置** - [ListInterveneCnt - 获得所有干预项的数量](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenecnt.md) - [ListIntervenes - 列出干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenes.md) - [ImportInterveneFile - 同步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefile.md) + - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - [ImportInterveneFileAsync - 异步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefileasync.md) - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) - - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) + - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - [DeleteInterveneRule - 删除干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-deleteintervenerule.md) - [ExportIntervenes - 导出干预项内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-exportintervenes.md) - [GetInterveneImportTaskInfo - 获得干预项目导入任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneimporttaskinfo.md) - - **系统配置-信源管理** - - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) - - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) + - **妙搜-智能搜索** + - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) + - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) + - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) + - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) + - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - **妙读-基础操作类** - - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) + - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [UploadBook - 书籍上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploadbook.md) - - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) - - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) - [DeleteDocs - 批量删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-deletedocs.md) + - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) + - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) + - **系统配置-信源管理** + - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) + - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) + - **妙读-抽取类** + - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - **妙读-生成类** - [RunMultiDocIntroduction - 多文档聚合摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runmultidocintroduction.md) - - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) + - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) + - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) - [RunCommentGeneration - 客户之声预测](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runcommentgeneration.md) - - **妙读-抽取类** - - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - - **妙读-问答类** - - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) - - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - **妙读-其他** - - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) + - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) - - **深度写作** - - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) - - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) - - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) - - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) + - **妙读-问答类** + - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) + - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - **PPT生成** - [ListEnterprisePptTemplates - 查询企业专属PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listenterpriseppttemplates.md) - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) + - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ExportPptArtifact - 导出PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-exportpptartifact.md) - [GetPptArtifact - 查询PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifact.md) - - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ListPptArtifacts - 查询PPT作品列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listpptartifacts.md) - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) - [InitiatePptCreation - 初始化用来创建PPT的会话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreation.md) - [GetPptConfig - 获取PPT组件配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptconfig.md) - - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) + - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - **标书生成** - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) + - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) - - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) + - **深度写作** + - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) + - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) + - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) + - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) + - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) - **其他** - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) - [SubmitSmartClipTask - 提交智能一键成片任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitsmartcliptask.md) + - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) - [SaveOrUpdateOssConfig - 配置-云存储-参数配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-saveorupdateossconfig.md) - [CreateDataPermissions - 权限-批量添加](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdatapermissions.md) - - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) + - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md) - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md) - [FetchParseDocumentLayoutTask - 获取排版任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-fetchparsedocumentlayouttask.md) - - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md) - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md) + - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-ram.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) + - **最佳实践** + - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) + - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) + - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) + - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) + - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) + - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) + - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - **更多** - - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) + - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - [妙搜数据集管理通过API引入数据源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaosou-introduce-data-source-through-api.md) - [全妙iframe嵌入方案](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/iframe-embedding-scheme.md) - - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) + - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) - [官方应用-通义听悟Agent](raw/application-user-guide/application-gallery/official-application-tingwu-agent.md) - [官方应用-析言GBI](raw/application-user-guide/application-gallery/xiyan-gbi.md) - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) -- **应用观测** - - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) -- **权限管理** - - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) - **实践教程** - [在网站上增加一个AI助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [基于本地知识库构建RAG应用](raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) + - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - **服务支持** - - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) - [常见问题](raw/application-user-guide/application-support/application-faq.md) + - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) ## 模型 API 参考 @@ -959,53 +959,51 @@ - [错误码](raw/model-api-reference/preparations/error-code.md) - **图像生成** - **千问** - - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) + - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - **万相** - - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-文生图V2版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) + - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-图像生成与编辑2.6 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) - - [万相-涂鸦作画API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) + - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) + - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-图像局部重绘API参考](raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - - **Z-Image** - - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - - **可灵** - - [可灵-图像生成API参考](raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) - **Vidu** - [Vidu-图像生成API参考](raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) + - **可灵** + - [可灵-图像生成API参考](raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) + - **Z-Image** + - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - **创意工具** - - [人像风格重绘API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - [图像画面扩展API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [虚拟模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) + - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [图像背景生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [创意文字WordArt锦书](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) + - [人像风格重绘API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - [常见问题](raw/model-api-reference/image-generation/image-faq.md) - **3D模型生成** - [Tripo-3D模型生成](raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) -- **实时多模态** - - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) - - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) - - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - - [实时多模态交互流程](raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) - - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - **更多模型** + - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-OCR API参考](raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) +- **实时多模态** + - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) + - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) + - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) + - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) + - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - **模型生产** - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - [模型部署](raw/model-api-reference/model-production/deployments-api.md) @@ -1014,34 +1012,33 @@ - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) - - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) + - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) - **更多** - [生成临时API Key](raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [子业务空间的模型调用](raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - - [上传本地文件获取临时URL](raw/model-api-reference/more-about-models/get-temporary-file-url.md) + - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - **视频生成** - **HappyHorse** - [HappyHorse-文生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) + - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - **万相** - **万相-早期视频模型（2.1-2.6）** - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) + - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) - - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) + - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) @@ -1049,31 +1046,25 @@ - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - **人像驱动** - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [图生播报视频-灵动人像LivePortrait](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) + - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) + - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) + - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - **爱诗** - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) + - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - **可灵** - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **Vidu** - [Vidu-文生视频API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) - - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) + - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - **音频** - **语音识别** - - **实时语音识别（Qwen-ASR-Realtime）** - - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) - - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) - - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) - - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) - - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - **实时语音识别（Fun-ASR）** - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) @@ -1082,69 +1073,73 @@ - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) + - **实时语音识别（Qwen-ASR-Realtime）** + - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) + - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) + - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) + - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) + - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) - **实时语音识别（Paraformer）** - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) - - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) + - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) - - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) + - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) - **录音文件识别（Fun-ASR）** - [Fun-ASR录音文件识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - [Fun-ASR录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) - [Fun-ASR录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) - - [Fun-ASR录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - [Fun-ASR录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) + - [Fun-ASR录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - **录音文件识别（Paraformer）** - - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - [Paraformer录音文件识别RESTful API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) - - [Paraformer录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) - - [Paraformer录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) - [Paraformer录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) + - [Paraformer录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) + - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) + - [Paraformer录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) - **定制热词** - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) - - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) + - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - [录音文件识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - **语音合成** - - **实时语音合成（Qwen-TTS-Realtime）** - - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) - - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) - - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** - - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) + - **实时语音合成（Qwen-TTS-Realtime）** + - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) + - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) + - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) - **实时语音合成（Sambert）** - - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) + - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) - - **非实时语音合成（MiniMax）** - - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) + - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** + - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - **声音复刻** - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - [声音复刻Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md) + - **非实时语音合成（MiniMax）** + - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - [非实时语音合成（Qwen-TTS）API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md) - [声音设计API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/voice-design-api-references.md) - - **音乐生成** - - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音翻译** - **实时音视频翻译（Qwen-Livetranslate-Realtime）** - [客户端事件](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/live-translator-client-events.md) @@ -1152,15 +1147,17 @@ - [实时音视频翻译（Qwen-LiveTranslate）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-python-sdk.md) - [实时音视频翻译（Qwen-LiveTranslate）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-java-sdk.md) - [音视频翻译-通义千问 API 参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/qwen3-livetranslate-flash-api.md) + - **音乐生成** + - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音对话** - **实时语音对话** - - [Qwen-Audio 实时语音对话客户端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md) - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话服务端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md) + - [Qwen-Audio 实时语音对话客户端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md) - **向量与排序** - **通用文本向量** - - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - [同步接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) + - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - **多模态向量** - [Multimodal-Embedding API详情](raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) - **排序模型（Rerank）** @@ -1171,13 +1168,13 @@ ## 应用 API 参考 - **Managed Agents** - - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [API 总览与认证](raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) + - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) + - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) - - [File](raw/application-api-reference/managed-agents-api/files-api.md) - [Skill](raw/application-api-reference/managed-agents-api/skills-api.md) + - [File](raw/application-api-reference/managed-agents-api/files-api.md) - **应用调用** - **Responses API** - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) @@ -1191,21 +1188,20 @@ - **数据连接（原应用数据）** - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) + - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) + - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) + - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [ListFile - 文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) - - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) + - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) - - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) - - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) + - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ChangeParseSetting - 修改类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) + - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddConnector - 新增连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) @@ -1221,31 +1217,31 @@ - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) - - **Prompt工程** - - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - **其他** - **长期记忆（旧）** - [CreateMemory - 创建长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) - - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) + - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [DeleteMemoryNode - 删除记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) + - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) + - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) + - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) - - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) + - **Prompt工程** + - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) + - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) + - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) + - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) + - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - [授权信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) @@ -1258,8 +1254,8 @@ - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) - **更多** + - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](raw/application-api-reference/more/how-to-use-search-filters.md) - - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [知识检索与问答](raw/application-api-reference/knowledge.md) diff --git a/skills/bailian-docs-llm-wiki/models/families.jsonl b/skills/bailian-docs-llm-wiki/models/families.jsonl index a3275cc6..59ebf5e3 100644 --- a/skills/bailian-docs-llm-wiki/models/families.jsonl +++ b/skills/bailian-docs-llm-wiki/models/families.jsonl @@ -1,11 +1,165 @@ +{"slug":"Kimi-K2","name":"Kimi","description":"Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["moonshot-ai"],"itemCount":5,"items":[{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","contextWindow":262144,"capabilities":["TG","Reasoning"]},{"model":"kimi-k2.5","name":"Kimi-K2.5","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"kimi-k2.6","name":"Kimi-K2.6","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/Kimi-K2.json","maxContextWindow":262144} +{"slug":"MiniMax-M2.1","name":"MiniMax","description":"MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["mini-max"],"itemCount":2,"items":[{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","contextWindow":204800,"capabilities":["Reasoning","TG"]},{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","contextWindow":204800,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/MiniMax-M2.1.json","maxContextWindow":204800} +{"slug":"MiniMax-speech-market-place","name":"MiniMax-Speech系列语音模型","description":"由MiniMax提供的MiniMax-Speech系列语音模型API服务。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["mini-max"],"itemCount":4,"items":[{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","capabilities":["TTS"]},{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","capabilities":["TTS"]},{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","capabilities":["TTS"]},{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","capabilities":["TTS"]}],"detailPath":"groups/MiniMax-speech-market-place.json"} +{"slug":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","description":"图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","capabilities":["IG"]}],"detailPath":"groups/aitryon-parsing-v1.json"} +{"slug":"aitryon-plus","name":"AI试衣-Plus版","description":"aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-plus","name":"AI试衣-Plus版","capabilities":["IG"]}],"detailPath":"groups/aitryon-plus.json"} +{"slug":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","description":"图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","capabilities":["IG"]}],"detailPath":"groups/aitryon-refiner.json"} +{"slug":"aitryon","name":"AI试衣-基础版","description":"aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"aitryon","name":"AI试衣-基础版","capabilities":["IG"]}],"detailPath":"groups/aitryon.json"} +{"slug":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","description":"AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-detect-gen2.json"} +{"slug":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","description":"AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-gen2.json"} +{"slug":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","description":"AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","capabilities":["VG"]}],"detailPath":"groups/animate-anyone-template-gen2.json"} +{"slug":"cosyvoice","name":"CosyVoice大模型","description":"基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen","qwen-domain-model"],"itemCount":7,"items":[{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","capabilities":["TTS"]},{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","capabilities":["TTS"]},{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","capabilities":["TTS"]},{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","capabilities":["TTS"]}],"detailPath":"groups/cosyvoice.json"} {"slug":"deepseek","name":"DeepSeek","description":"DeepSeek是由深度求索提供的开源模型，包含 V3.1、V3、R1以及基于Qwen2.5系列蒸馏的大语言模型。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["deepseek"],"itemCount":12,"items":[{"model":"deepseek-r1","name":"DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning"]},{"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","contextWindow":131072,"capabilities":["Reasoning"]},{"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-14b","name":"DeepSeek-R1-Distill-Qwen-14B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-32b","name":"DeepSeek-R1-Distill-Qwen-32B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-r1-distill-qwen-7b","name":"DeepSeek-R1-Distill-Qwen-7B","contextWindow":32768,"capabilities":["TG"]},{"model":"deepseek-v3","name":"DeepSeek-V3","contextWindow":65536,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v3.1","name":"DeepSeek-V3.1","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","contextWindow":1000000,"capabilities":["TG","Reasoning"]},{"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/deepseek.json","maxContextWindow":1000000} +{"slug":"embedding","name":"通义多模态向量","description":"基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。","primaryCapability":"ME","capabilities":["ME"],"providers":["qwen-domain-model"],"itemCount":3,"items":[{"model":"multimodal-embedding-v1","name":"通用多模态向量","contextWindow":0,"capabilities":["ME"]},{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","capabilities":["ME"]},{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","capabilities":["ME"]}],"detailPath":"groups/embedding.json"} +{"slug":"emo-detect-v1","name":"悦动人像EMO-detect","description":"EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emo-detect-v1","name":"悦动人像EMO-detect","capabilities":["VG"]}],"detailPath":"groups/emo-detect-v1.json"} +{"slug":"emo-v1","name":"悦动人像EMO","description":"EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emo-v1","name":"悦动人像EMO","capabilities":["VG"]}],"detailPath":"groups/emo-v1.json"} +{"slug":"emoji-detect-v1","name":"表情包Emoji-detect","description":"表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emoji-detect-v1","name":"表情包Emoji-detect","capabilities":["VG"]}],"detailPath":"groups/emoji-detect-v1.json"} +{"slug":"emoji-v1","name":"表情包Emoji","description":"表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"emoji-v1","name":"表情包Emoji","capabilities":["VG"]}],"detailPath":"groups/emoji-v1.json"} +{"slug":"facechain-facedetect","name":"FaceChain人物图像检测","description":"对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"facechain-facedetect","name":"FaceChain人物图像检测","capabilities":["IG"]}],"detailPath":"groups/facechain-facedetect.json"} +{"slug":"facechain-generation","name":"FaceChain人物写真生成","description":"基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"facechain-generation","name":"FaceChain人物写真生成","capabilities":["IG"]}],"detailPath":"groups/facechain-generation.json"} +{"slug":"farui-plus","name":"通义法睿-Plus-32K","description":"通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"farui-plus","name":"通义法睿-Plus-32K","contextWindow":12000,"capabilities":["TG"]}],"detailPath":"groups/farui-plus.json","maxContextWindow":12000} +{"slug":"fun-asr-flash","name":"Fun-ASR-Flash","description":"百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","capabilities":["ASR"]}],"detailPath":"groups/fun-asr-flash.json"} +{"slug":"fun-asr-realtime","name":"Fun-ASR实时语音识别","description":"通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","capabilities":["Realtime-ASR"]},{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","capabilities":["Realtime-ASR"]}],"detailPath":"groups/fun-asr-realtime.json"} +{"slug":"fun-asr","name":"Fun-ASR语音识别","description":"通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"fun-asr","name":"Fun-ASR语音识别","capabilities":["ASR"]},{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","capabilities":["ASR"]}],"detailPath":"groups/fun-asr.json"} +{"slug":"fun-music","name":"音乐生成","description":"百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":2,"items":[{"model":"fun-music-preview","name":"音乐生成 Preview","capabilities":["TTS"]},{"model":"fun-music-v1","name":"音乐生成","capabilities":["TTS"]}],"detailPath":"groups/fun-music.json"} +{"slug":"glm-4.5","name":"GLM","description":"GLM是由智谱提供的开源模型。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["zhipu-ai"],"itemCount":7,"items":[{"model":"glm-4.5","name":"GLM-4.5","contextWindow":131072,"capabilities":["TG"]},{"model":"glm-4.5-air","name":"GLM-4.5-Air","contextWindow":131072,"capabilities":["TG"]},{"model":"glm-4.6","name":"GLM-4.6","contextWindow":202752,"capabilities":["Reasoning","TG"]},{"model":"glm-4.7","name":"GLM-4.7","contextWindow":202752,"capabilities":["TG","Reasoning"]},{"model":"glm-5","name":"GLM-5","contextWindow":202752,"capabilities":["TG","Reasoning"]},{"model":"glm-5.1","name":"GLM-5.1","contextWindow":202745,"capabilities":["TG","Reasoning"]},{"model":"glm-5.2","name":"GLM-5.2","contextWindow":1048576,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/glm-4.5.json","maxContextWindow":1048576} +{"slug":"glm-fast","name":"GLM-5.2-Fast","description":"GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。","primaryCapability":"TG","capabilities":["TG"],"providers":["zhipu-ai"],"itemCount":1,"items":[{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/glm-fast.json","maxContextWindow":1048576} +{"slug":"gui-plus","name":"GUI-Plus","description":"GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"gui-plus","name":"GUI-Plus","contextWindow":256000,"capabilities":["VU"]}],"detailPath":"groups/gui-plus.json","maxContextWindow":256000} +{"slug":"gummy-chat-v1","name":"一句话识别及翻译V1.0","description":"多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","capabilities":["ASR"]}],"detailPath":"groups/gummy-chat-v1.json"} +{"slug":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","description":"多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen"],"itemCount":1,"items":[{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/gummy-realtime-v1.json"} {"slug":"happyhorse-i2v","name":"HappyHorse-I2V","description":"HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","capabilities":["VG"]},{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-i2v.json"} +{"slug":"happyhorse-r2v","name":"HappyHorse-R2V","description":"HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","capabilities":["VG"]},{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-r2v.json"} {"slug":"happyhorse-t2v","name":"HappyHorse-T2V","description":"HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":2,"items":[{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","capabilities":["VG"]},{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","capabilities":["VG"]}],"detailPath":"groups/happyhorse-t2v.json"} +{"slug":"happyhorse-video-edit","name":"HappyHorse-Video-Edit","description":"HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。","primaryCapability":"VG","capabilities":["VG"],"providers":["happyhorse"],"itemCount":1,"items":[{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","capabilities":["VG"]}],"detailPath":"groups/happyhorse-video-edit.json"} +{"slug":"image-erase-completion","name":"图像擦除补全","description":"图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-erase-completion","name":"图像擦除补全","capabilities":["IG"]}],"detailPath":"groups/image-erase-completion.json"} +{"slug":"image-instance-segmentation","name":"人物实例分割","description":"人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-instance-segmentation","name":"人物实例分割","capabilities":["IG"]}],"detailPath":"groups/image-instance-segmentation.json"} +{"slug":"image-out-painting","name":"图像画面扩展","description":"图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-out-painting","name":"图像画面扩展","capabilities":["IG"]}],"detailPath":"groups/image-out-painting.json"} +{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["moonshot-ai"],"itemCount":4,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":262144} +{"slug":"kling-models-market-place","name":"可灵AI","description":"由可灵AI提供的高质量视频与图像生成及编辑模型。","primaryCapability":"VG","capabilities":["VG","IG"],"providers":["kling"],"itemCount":4,"items":[{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","capabilities":["IG"]},{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","capabilities":["IG"]},{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","capabilities":["VG"]},{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","capabilities":["VG"]}],"detailPath":"groups/kling-models-market-place.json"} +{"slug":"liveportrait-detect","name":"灵动人像LivePortrait-detect","description":"LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","capabilities":["VG"]}],"detailPath":"groups/liveportrait-detect.json"} +{"slug":"liveportrait","name":"灵动人像LivePortrait","description":"LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait","name":"灵动人像LivePortrait","capabilities":["VG"]}],"detailPath":"groups/liveportrait.json"} +{"slug":"minimax-models-market-place","name":"MiniMax文本模型","description":"由MiniMax提供的MiniMax-M系列文本模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["mini-max"],"itemCount":4,"items":[{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","contextWindow":1048576,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/minimax-models-market-place.json","maxContextWindow":1048576} +{"slug":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","description":"Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-8k-v1.json"} +{"slug":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","description":"Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","capabilities":["ASR"]}],"detailPath":"groups/paraformer-8k-v2.json"} +{"slug":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","description":"Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-mtl-v1.json"} +{"slug":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","description":"Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-8k-v1.json"} +{"slug":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","description":"推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-8k-v2.json"} +{"slug":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","description":"Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-v1.json"} +{"slug":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","description":"推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","capabilities":["Realtime-ASR"]}],"detailPath":"groups/paraformer-realtime-v2.json"} +{"slug":"paraformer-v1","name":"Paraformer语音识别-v1","description":"Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-v1","name":"Paraformer语音识别-v1","capabilities":["ASR"]}],"detailPath":"groups/paraformer-v1.json"} +{"slug":"paraformer-v2","name":"Paraformer语音识别-v2","description":"推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"paraformer-v2","name":"Paraformer语音识别-v2","capabilities":["ASR"]}],"detailPath":"groups/paraformer-v2.json"} +{"slug":"pixverse-c1-market-place","name":"PixVerse C1","description":"由爱诗科技提供的PixVerse C系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-c1-market-place.json"} +{"slug":"pixverse-market-place","name":"PixVerse V5.6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-market-place.json"} +{"slug":"pixverse-v6-market-place","name":"PixVerse V6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-v6-market-place.json"} +{"slug":"qvq-max","name":"QVQ-Max","description":"千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-max","name":"QVQ-Max","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-max.json","maxContextWindow":131072} +{"slug":"qvq-plus","name":"Qwen-QVQ-Plus","description":"千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-plus","name":"QVQ-Plus","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-plus.json","maxContextWindow":131072} +{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":8192} +{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":8192} +{"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"} +{"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072} +{"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072} +{"slug":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","description":"快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen-doc-turbo.json","maxContextWindow":262144} +{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen-domain-model"],"itemCount":6,"items":[{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json"} +{"slug":"qwen-flash-character","name":"Qwen-Flash-Character","description":"千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash-character","name":"Qwen-Flash-Character","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/qwen-flash-character.json","maxContextWindow":8192} +{"slug":"qwen-flash","name":"Qwen-Flash","description":"Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash","name":"Qwen-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-flash.json","maxContextWindow":1000000} +{"slug":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","description":"Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0-pro.json"} +{"slug":"qwen-image-2.0","name":"Qwen-Image-2.0","description":"Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0.json"} +{"slug":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","description":"千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit-max.json"} +{"slug":"qwen-image-edit","name":"Qwen-Image-Edit-Plus","description":"千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image-edit","name":"Qwen-Image-Edit","capabilities":["IG"]},{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit.json"} +{"slug":"qwen-image-plus","name":"Qwen-Image-Plus","description":"千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image","name":"Qwen-Image","capabilities":["IG"]},{"model":"qwen-image-plus","name":"Qwen-Image-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-plus.json"} +{"slug":"qwen-long","name":"Qwen-Long","description":"Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"qwen-long","name":"Qwen-Long","contextWindow":10000000,"capabilities":["TG"]},{"model":"qwen-long-latest","name":"Qwen-Long-Latest","contextWindow":10000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-long.json","maxContextWindow":10000000} +{"slug":"qwen-math-plus","name":"Qwen-Math-Plus","description":"Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":4,"items":[{"model":"qwen-math-plus","name":"Qwen-Math-Plus","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","contextWindow":4096,"capabilities":["TG"]}],"detailPath":"groups/qwen-math-plus.json","maxContextWindow":4096} +{"slug":"qwen-math-turbo","name":"Qwen-Math-Turbo","description":"Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","contextWindow":4096,"capabilities":["TG"]}],"detailPath":"groups/qwen-math-turbo.json","maxContextWindow":4096} +{"slug":"qwen-max","name":"Qwen-Max","description":"千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-max","name":"Qwen-Max","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/qwen-max.json","maxContextWindow":32768} +{"slug":"qwen-mt-flash","name":"Qwen-MT-Flash","description":"基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-flash.json","maxContextWindow":16384} +{"slug":"qwen-mt-image","name":"Qwen-MT-Image","description":"专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-image","name":"Qwen-MT-Image","capabilities":["IG"]}],"detailPath":"groups/qwen-mt-image.json"} +{"slug":"qwen-mt-lite","name":"Qwen-MT-Lite","description":"基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-lite.json","maxContextWindow":16384} +{"slug":"qwen-mt-plus","name":"Qwen-MT-Plus","description":"基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-plus.json","maxContextWindow":16384} +{"slug":"qwen-mt-turbo","name":"Qwen-MT-Turbo","description":"基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","contextWindow":16384,"capabilities":["TG"]}],"detailPath":"groups/qwen-mt-turbo.json","maxContextWindow":16384} +{"slug":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","description":"千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","contextWindow":32768,"capabilities":["Realtime-Omni"]},{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","contextWindow":32768,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen-omni-turbo-realtime.json","maxContextWindow":32768} +{"slug":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","description":"千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","contextWindow":32768,"capabilities":["Multimodal-Omni"]},{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","contextWindow":32768,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen-omni-turbo.json","maxContextWindow":32768} +{"slug":"qwen-plus-character","name":"Qwen-Plus-Character","description":"千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-plus-character","name":"Qwen-Plus-Character","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/qwen-plus-character.json","maxContextWindow":32768} +{"slug":"qwen-plus","name":"Qwen-Plus","description":"千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":4,"items":[{"model":"qwen-plus","name":"Qwen-Plus","contextWindow":1000000,"capabilities":["Reasoning","TG"]},{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-plus.json","maxContextWindow":1000000} +{"slug":"qwen-rerank","name":"Qwen-Rerank","description":"基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen","qwen-domain-model"],"itemCount":3,"items":[{"model":"gte-rerank-v2","name":"深度文本重排序","contextWindow":30000,"capabilities":["TR"]},{"model":"qwen3-rerank","name":"千问3-Rerank","contextWindow":30000,"capabilities":["TR"]},{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","capabilities":["TR"]}],"detailPath":"groups/qwen-rerank.json","maxContextWindow":30000} +{"slug":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","description":"Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","contextWindow":8192,"capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","contextWindow":8192,"capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-tts-realtime.json","maxContextWindow":8192} +{"slug":"qwen-tts","name":"Qwen-TTS","description":"千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-tts","name":"Qwen-TTS","contextWindow":8192,"capabilities":["TTS"]},{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","contextWindow":8192,"capabilities":["TTS"]}],"detailPath":"groups/qwen-tts.json","maxContextWindow":8192} +{"slug":"qwen-turbo","name":"Qwen-Turbo","description":"千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-turbo","name":"Qwen-Turbo","contextWindow":131072,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-turbo.json","maxContextWindow":131072} +{"slug":"qwen-vl-embedding","name":"Qwen-VL-Embedding","description":"基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。","primaryCapability":"ME","capabilities":["ME"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","capabilities":["ME"]},{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","capabilities":["ME"]}],"detailPath":"groups/qwen-vl-embedding.json"} +{"slug":"qwen-vl-max","name":"Qwen-VL-Max","description":"Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-vl-max","name":"Qwen-VL-Max","contextWindow":131072,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-max.json","maxContextWindow":131072} +{"slug":"qwen-vl-ocr","name":"Qwen-VL-OCR","description":"Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen-domain-model"],"itemCount":3,"items":[{"model":"qwen-vl-ocr","name":"QwenVL-OCR","contextWindow":38192,"capabilities":["VU"]},{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","contextWindow":34096,"capabilities":["VU"]},{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","contextWindow":38192,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-ocr.json","maxContextWindow":38192} +{"slug":"qwen-voice-design","name":"Qwen-声音设计","description":"Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-design","name":"Qwen-Voice-Design","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-design.json"} +{"slug":"qwen-voice-enrollment","name":"Qwen-声音复刻","description":"千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-enrollment.json"} +{"slug":"qwen2.5","name":"Qwen2.5-开源模型","description":"Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","contextWindow":32768,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen2.5.json","maxContextWindow":32768} +{"slug":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","description":"Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","capabilities":["ASR"]}],"detailPath":"groups/qwen3-asr-flash-filetrans.json"} +{"slug":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","description":"Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","capabilities":["Realtime-ASR"]}],"detailPath":"groups/qwen3-asr-flash-realtime.json"} +{"slug":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","description":"Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","capabilities":["ASR"]}],"detailPath":"groups/qwen3-asr-flash.json"} +{"slug":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","description":"基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-30b-a3b-instruct.json","maxContextWindow":262144} +{"slug":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","description":"基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-480b-a35b-instruct.json","maxContextWindow":262144} +{"slug":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","description":"基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-flash.json","maxContextWindow":1000000} +{"slug":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","description":"基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen3-coder-plus.json","maxContextWindow":1000000} +{"slug":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","description":"Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3-livetranslate-flash-realtime.json","maxContextWindow":53248} +{"slug":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","description":"Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。","primaryCapability":"Realtime-ASR","capabilities":["Realtime-ASR"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","contextWindow":53248,"capabilities":["Realtime-ASR"]}],"detailPath":"groups/qwen3-livetranslate-flash.json","maxContextWindow":53248} +{"slug":"qwen3-max","name":"Qwen3-Max","description":"千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3-max","name":"Qwen3-Max","contextWindow":262144,"capabilities":["TG","Reasoning"]},{"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","contextWindow":262144,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3-max.json","maxContextWindow":262144} +{"slug":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","description":"千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","contextWindow":65536,"capabilities":["ASR"]}],"detailPath":"groups/qwen3-omni-30b-a3b-captioner.json","maxContextWindow":65536} +{"slug":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","description":"Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","contextWindow":65536,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3-omni-flash-realtime.json","maxContextWindow":65536} +{"slug":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","description":"Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","contextWindow":65536,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3-omni-flash.json","maxContextWindow":65536} +{"slug":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","description":"Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash-realtime.json"} +{"slug":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-flash.json"} +{"slug":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","description":"通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} +{"slug":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-instruct-flash.json"} +{"slug":"qwen3-tts-vc-realtime","name":"Qwen3-TTS-VC-Realtime","description":"Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vc-realtime.json"} +{"slug":"qwen3-tts-vc","name":"Qwen3-TTS-VC","description":"Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vc.json"} +{"slug":"qwen3-tts-vd-realtime","name":"Qwen3-TTS-VD-Realtime","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd-realtime.json"} +{"slug":"qwen3-tts-vd","name":"Qwen3-TTS-VD","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd.json"} +{"slug":"qwen3-vl-flash","name":"Qwen3-VL-Flash","description":"Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-flash.json","maxContextWindow":262144} +{"slug":"qwen3-vl-plus","name":"Qwen3-VL-Plus","description":"Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-plus.json","maxContextWindow":262144} {"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG","VU"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} +{"slug":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","description":"Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json","maxContextWindow":53248} +{"slug":"qwen3.5-ocr","name":"Qwen3.5-OCR","description":"Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","contextWindow":65536,"capabilities":["VU"]}],"detailPath":"groups/qwen3.5-ocr.json","maxContextWindow":65536} +{"slug":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash-realtime.json","maxContextWindow":262144} +{"slug":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash.json","maxContextWindow":262144} +{"slug":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus-realtime.json","maxContextWindow":262144} +{"slug":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus.json","maxContextWindow":262144} {"slug":"qwen3.5-plus","name":"Qwen3.5-Plus","description":"Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-plus","name":"Qwen3.5-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.5-plus.json","maxContextWindow":1000000} +{"slug":"qwen3.5","name":"Qwen3.5开源模型","description":"Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":4,"items":[{"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-27b","name":"Qwen3.5-27B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-35b-a3b","name":"Qwen3.5-35B-A3B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.5-397b-a17b","name":"Qwen3.5-397B-A17B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.5.json","maxContextWindow":262144} {"slug":"qwen3.6-flash","name":"Qwen3.6-Flash","description":"Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-flash","name":"Qwen3.6-Flash","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.6-flash.json","maxContextWindow":1000000} {"slug":"qwen3.6-max","name":"Qwen3.6-Max","description":"Qwen3.6原生Max模型，相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提升；长尾知识能力进一步升级。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","contextWindow":262144,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen3.6-max.json","maxContextWindow":262144} {"slug":"qwen3.6-plus","name":"Qwen3.6-Plus","description":"Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.6-plus","name":"Qwen3.6-Plus","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.6-plus.json","maxContextWindow":1000000} {"slug":"qwen3.6","name":"Qwen3.6开源模型","description":"Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。","primaryCapability":"VU","capabilities":["VU","TG","Reasoning"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.6-27b","name":"Qwen3.6-27B","contextWindow":262144,"capabilities":["Reasoning","VU","TG"]},{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","contextWindow":262144,"capabilities":["VU","TG","Reasoning"]}],"detailPath":"groups/qwen3.6.json","maxContextWindow":262144} {"slug":"qwen3.7-max","name":"Qwen3.7-Max","description":"Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen3.7-max","name":"Qwen3.7-Max","contextWindow":1000000,"capabilities":["Reasoning","TG"]},{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","contextWindow":1000000,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/qwen3.7-max.json","maxContextWindow":1000000} {"slug":"qwen3.7-plus","name":"Qwen3.7-Plus","description":"Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.7-plus","name":"Qwen3.7-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.7-plus.json","maxContextWindow":1000000} +{"slug":"qwen3","name":"Qwen3开源模型","description":"Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。","primaryCapability":"VU","capabilities":["VU","Reasoning","TG"],"providers":["qwen"],"itemCount":20,"items":[{"model":"qwen3-14b","name":"Qwen3-14B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","contextWindow":81920,"capabilities":["Reasoning","TG"]},{"model":"qwen3-32b","name":"Qwen3-32B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-8b","name":"Qwen3-8B","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","contextWindow":262144,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","contextWindow":131072,"capabilities":["TG"]},{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]},{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","contextWindow":131072,"capabilities":["VU"]},{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","contextWindow":131072,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3.json","maxContextWindow":262144} +{"slug":"qwq-plus","name":"Qwen-QwQ-Plus","description":"千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。","primaryCapability":"Reasoning","capabilities":["Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwq-plus","name":"QwQ-Plus","contextWindow":131072,"capabilities":["Reasoning"]}],"detailPath":"groups/qwq-plus.json","maxContextWindow":131072} +{"slug":"sambert","name":"Sambert语音合成","description":"提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":43,"items":[{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","capabilities":["TTS"]},{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","capabilities":["TTS"]},{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","capabilities":["TTS"]},{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","capabilities":["TTS"]},{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","capabilities":["TTS"]},{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","capabilities":["TTS"]},{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","capabilities":["TTS"]},{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","capabilities":["TTS"]},{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","capabilities":["TTS"]},{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","capabilities":["TTS"]},{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","capabilities":["TTS"]},{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","capabilities":["TTS"]},{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","capabilities":["TTS"]},{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","capabilities":["TTS"]},{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","capabilities":["TTS"]},{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","capabilities":["TTS"]},{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","capabilities":["TTS"]},{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","capabilities":["TTS"]},{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","capabilities":["TTS"]},{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","capabilities":["TTS"]},{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","capabilities":["TTS"]},{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","capabilities":["TTS"]},{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","capabilities":["TTS"]},{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","capabilities":["TTS"]},{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","capabilities":["TTS"]},{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","capabilities":["TTS"]},{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","capabilities":["TTS"]},{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","capabilities":["TTS"]},{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","capabilities":["TTS"]},{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","capabilities":["TTS"]},{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","capabilities":["TTS"]},{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","capabilities":["TTS"]},{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","capabilities":["TTS"]},{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","capabilities":["TTS"]},{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","capabilities":["TTS"]},{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","capabilities":["TTS"]},{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","capabilities":["TTS"]},{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","capabilities":["TTS"]},{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","capabilities":["TTS"]},{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","capabilities":["TTS"]},{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","capabilities":["TTS"]},{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","capabilities":["TTS"]},{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","capabilities":["TTS"]}],"detailPath":"groups/sambert.json"} +{"slug":"shoemodel-v1","name":"鞋靴模特","description":"鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"shoemodel-v1","name":"鞋靴模特","capabilities":["IG"]}],"detailPath":"groups/shoemodel-v1.json"} +{"slug":"siliconflow-models","name":"SiliconFlow DeepSeek","description":"由硅基流动提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["deepseek"],"itemCount":4,"items":[{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","contextWindow":163840,"capabilities":["TG"]},{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","contextWindow":163840,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/siliconflow-models.json","maxContextWindow":163840} +{"slug":"speech-biasing","name":"语音识别热词","description":"热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"speech-biasing","name":"语音识别热词","capabilities":["ASR"]}],"detailPath":"groups/speech-biasing.json"} +{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG","VU"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG","VU"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} +{"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192} +{"slug":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","description":"通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768} +{"slug":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","description":"通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768} +{"slug":"tripo-models-market-place","name":"Tripo","description":"AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。","primaryCapability":"3D-generation","capabilities":["3D-generation"],"providers":["tripo"],"itemCount":2,"items":[{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","capabilities":["3D-generation"]},{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","capabilities":["3D-generation"]}],"detailPath":"groups/tripo-models-market-place.json"} +{"slug":"vanchin-models-market-place","name":"Vanchin DeepSeek","description":"由快手万擎提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["deepseek"],"itemCount":6,"items":[{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","contextWindow":8192,"capabilities":["VU","TG"]},{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","contextWindow":131072,"capabilities":["TG"]},{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/vanchin-models-market-place.json","maxContextWindow":1048576} +{"slug":"video-style-transform","name":"视频风格重绘","description":"视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"video-style-transform","name":"视频风格重绘","capabilities":["VG"]}],"detailPath":"groups/video-style-transform.json"} +{"slug":"videoretalk","name":"声动人像VideoRetalk","description":"VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"videoretalk","name":"声动人像VideoRetalk","capabilities":["VG"]}],"detailPath":"groups/videoretalk.json"} +{"slug":"vidu-image-models-market-place","name":"Vidu AI生图","description":"由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。","primaryCapability":"IG","capabilities":["IG"],"providers":["vidu"],"itemCount":4,"items":[{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","capabilities":["IG"]},{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","capabilities":["IG"]}],"detailPath":"groups/vidu-image-models-market-place.json"} +{"slug":"virtualmodel-v2","name":"虚拟模特V2","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"virtualmodel-v2","name":"虚拟模特V2","capabilities":["IG"]}],"detailPath":"groups/virtualmodel-v2.json"} +{"slug":"voice-enrollment","name":"大模型声音复刻及声音设计","description":"大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","capabilities":["TTS"]}],"detailPath":"groups/voice-enrollment.json"} +{"slug":"wan-image-edit","name":"Wan-Image","description":"指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":5,"items":[{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","capabilities":["IG"]},{"model":"wan2.6-image","name":"Wan2.6-Image","capabilities":["IG"]},{"model":"wan2.7-image","name":"Wan2.7-Image","capabilities":["IG"]},{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","capabilities":["IG"]},{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","capabilities":["IG"]}],"detailPath":"groups/wan-image-edit.json"} +{"slug":"wan-image-to-video","name":"Wan-I2V","description":"图片生成视频内容，稳定保持图像主体、风格和文字等细节信息","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":14,"items":[{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","capabilities":["VG"]},{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","capabilities":["VG"]},{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","capabilities":["VG"]},{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","capabilities":["VG"]},{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","capabilities":["VG"]},{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","capabilities":["VG"]},{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","capabilities":["VG"]},{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","capabilities":["VG"]},{"model":"wan2.6-i2v","name":"Wan2.6-I2V","capabilities":["VG"]},{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","capabilities":["VG"]},{"model":"wan2.7-i2v","name":"Wan2.7-I2V","capabilities":["VG"]},{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","capabilities":["VG"]},{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","capabilities":["VG"]},{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","capabilities":["VG"]}],"detailPath":"groups/wan-image-to-video.json"} +{"slug":"wan-reference-to-video","name":"Wan-R2V","description":"参考视频中的人或物，精准保持形象和声音，支持多参考合拍","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":3,"items":[{"model":"wan2.6-r2v","name":"Wan2.6-R2V","capabilities":["VG"]},{"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","capabilities":["VG"]},{"model":"wan2.7-r2v","name":"Wan2.7-R2V","capabilities":["VG"]}],"detailPath":"groups/wan-reference-to-video.json"} +{"slug":"wan-text-to-image","name":"Wan-T2I","description":"文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":9,"items":[{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","capabilities":["IG"]},{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","capabilities":["IG"]},{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","capabilities":["IG"]},{"model":"wan2.6-t2i","name":"Wan2.6-T2I","capabilities":["IG"]},{"model":"wanx-v1","name":"wanx-t2i","capabilities":["IG"]},{"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","capabilities":["IG"]},{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","capabilities":["IG"]},{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","capabilities":["IG"]},{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","capabilities":["IG"]}],"detailPath":"groups/wan-text-to-image.json"} +{"slug":"wan-text-to-video","name":"Wan-T2V","description":"文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":6,"items":[{"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","capabilities":["VG"]},{"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","capabilities":["VG"]},{"model":"wan2.6-t2v","name":"Wan2.6-T2V","capabilities":["VG"]},{"model":"wan2.7-t2v","name":"Wan2.7-T2V","capabilities":["VG"]},{"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","capabilities":["VG"]},{"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","capabilities":["VG"]}],"detailPath":"groups/wan-text-to-video.json"} +{"slug":"wan-video-edit","name":"Wan-VideoEdit","description":"通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","capabilities":["VG"]}],"detailPath":"groups/wan-video-edit.json"} +{"slug":"wanx-background-generation-v2","name":"图像背景生成","description":"图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-background-generation-v2","name":"图像背景生成","capabilities":["IG"]}],"detailPath":"groups/wanx-background-generation-v2.json"} +{"slug":"wanx-poster-generation-v1","name":"创意海报生成","description":"创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-poster-generation-v1","name":"创意海报生成","capabilities":["IG"]}],"detailPath":"groups/wanx-poster-generation-v1.json"} +{"slug":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","description":"万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","capabilities":["IG"]}],"detailPath":"groups/wanx-sketch-to-image-lite.json"} +{"slug":"wanx-style-repaint-v1","name":"人像风格重绘","description":"人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-style-repaint-v1","name":"人像风格重绘","capabilities":["IG"]}],"detailPath":"groups/wanx-style-repaint-v1.json"} +{"slug":"wanx-virtualmodel","name":"虚拟模特","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-virtualmodel","name":"虚拟模特","capabilities":["IG"]}],"detailPath":"groups/wanx-virtualmodel.json"} +{"slug":"wanx-x-painting","name":"万相-图像局部重绘","description":"万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx-x-painting","name":"万相-图像局部重绘","capabilities":["IG"]}],"detailPath":"groups/wanx-x-painting.json"} +{"slug":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","description":"万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。","primaryCapability":"VG","capabilities":["VG"],"providers":["wan"],"itemCount":1,"items":[{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","capabilities":["VG"]}],"detailPath":"groups/wanx2.1-vace-plus.json"} +{"slug":"wordart-semantic","name":"WordArt锦书-文字变形","description":"WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"wordart-semantic","name":"WordArt锦书-文字变形","capabilities":["IG"]}],"detailPath":"groups/wordart-semantic.json"} +{"slug":"wordart-texture","name":"WordArt锦书-文字纹理生成","description":"WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","capabilities":["IG"]}],"detailPath":"groups/wordart-texture.json"} +{"slug":"xiaomi-models-market-place","name":"MiMo文本模型","description":"由小米MiMo提供的MiMo文本模型API服务","primaryCapability":"TG","capabilities":["TG"],"providers":["xiaomi"],"itemCount":1,"items":[{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/xiaomi-models-market-place.json","maxContextWindow":1048576} +{"slug":"z-image-turbo","name":"Z-Image-Turbo","description":"Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"z-image-turbo","name":"Z-Image-Turbo","capabilities":["IG"]}],"detailPath":"groups/z-image-turbo.json"} +{"slug":"zhipu-models-market-place","name":"智谱GLM系列文本模型","description":"由智谱提供的GLM系列文本模型API服务","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["zhipu-ai"],"itemCount":3,"items":[{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","contextWindow":204800,"capabilities":["TG","Reasoning"]},{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","contextWindow":204800,"capabilities":["TG"]},{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","contextWindow":1048576,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/zhipu-models-market-place.json","maxContextWindow":1048576} diff --git a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json new file mode 100644 index 00000000..fd8a55dc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json @@ -0,0 +1,646 @@ +{ + "name": "Kimi", + "description": "Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "kimi-k2.7-code是 kimi 迄今最智能的coding模型，在长上下文中更可靠地遵循指令，能以更高的成功率完成编程任务，同时支持文本、图片与视频输入，思考模式，对话与 Agent 任务。", + "features": [ + "cache", + "function-calling", + "model-experience", + "structured-outputs", + "web-search", + "prefix-completion" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-06-14T16:59:14.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 229376, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "kimi-k2.7-code", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -s -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Image", + "Text", + "Video" + ] + }, + "description": "kimi-k2.6是Kimi最新最智能的模型，具备更强更稳的长程代码编写能力，指令遵循和自我纠错能力显著提升，同时支持文本、图片与视频输入，思考与非思考模式，对话与Agent任务。", + "features": [ + "cache", + "function-calling", + "model-experience" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-04-21T09:55:34.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 229376, + "inferenceProvider": "aliyun-bailian", + "name": "Kimi-K2.6", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "kimi-k2.5是月之暗面迄今发布最全能的模型，原生多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与Agent任务。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi-k2.5", + "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-01-30T01:49:03.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 229376, + "inferenceProvider": "aliyun-bailian", + "name": "Kimi-K2.5", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "kimi-k2-thinking模型是月之暗面提供的具有通用 Agentic能力和推理能力的思考模型，它擅长深度推理，并可通过多步工具调用，帮助解决各类难题。", + "features": [ + "model-experience", + "cache", + "function-calling" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi-k2-thinking", + "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-11-10T07:24:34.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 229376, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Kimi-K2-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Kimi-K2是月之暗面提供的国内首个开源万亿参数MoE模型，激活参数达 320 亿，具有卓越的编码和工具调用能力。", + "features": [ + "model-experience", + "cache", + "function-calling" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "Moonshot-Kimi-K2-Instruct", + "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "user-spec": { + "count_limit_period": 5, + "start_time": 1761645315, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, + "model-default-actual": { + "count_limit_period": 5, + "start_time": 1761645315, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-07-16T14:56:31.000+00:00", + "contextWindow": 131072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Moonshot-Kimi-K2-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2948482.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json new file mode 100644 index 00000000..d694037e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json @@ -0,0 +1,224 @@ +{ + "name": "MiniMax", + "description": "MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax-M2.5", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-02-24T15:07:02.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 196608, + "inferenceProvider": "aliyun-bailian", + "name": "MiniMax-M2.5", + "docUrl": "https://help.aliyun.com/document_detail/3017140.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax-M2.1", + "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 172032, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "MiniMax-M2.1", + "docUrl": "https://help.aliyun.com/document_detail/3017140.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json new file mode 100644 index 00000000..c1f11e04 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json @@ -0,0 +1,278 @@ +{ + "name": "MiniMax-Speech系列语音模型", + "description": "由MiniMax提供的MiniMax-Speech系列语音模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", + "features": [], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/speech-2.8-turbo", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T08:44:53.000+00:00", + "maxInputTokens": 10000, + "inferenceProvider": "mini-max", + "name": "speech-2.8-turbo", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-2.8-turbo\"\n }'", + "docUrl": "https://help.aliyun.com/document_detail/3021951.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", + "features": [], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/speech-2.8-hd", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T08:46:11.000+00:00", + "maxInputTokens": 10000, + "inferenceProvider": "mini-max", + "name": "speech-2.8-hd", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-2.8-hd\"\n }'", + "docUrl": "https://help.aliyun.com/document_detail/3021951.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", + "features": [], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/speech-02-turbo", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T08:44:37.000+00:00", + "maxInputTokens": 10000, + "inferenceProvider": "mini-max", + "name": "speech-02-turbo", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-02-turbo\"\n }'", + "docUrl": "https://help.aliyun.com/document_detail/3021951.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", + "features": [], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/speech-02-hd", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000, + "usage_limit_field": "characters", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T08:45:04.000+00:00", + "maxInputTokens": 10000, + "inferenceProvider": "mini-max", + "name": "speech-02-hd", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n -H \"Authorization: Bearer ${DASHSCOPE_API_KEY}\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"input\": {\n \"action\": \"tts\",\n \"text\": \"今天是不是很开心呀，当然了！\",\n \"stream\": false,\n \"voice_setting\": {\n \"voice_id\": \"male-qn-qingse\",\n \"speed\": 1,\n \"vol\": 1,\n \"pitch\": 0\n },\n \"audio_setting\": {\n \"sample_rate\": 16000,\n \"format\": \"mp3\",\n \"channel\": 1\n }\n },\n \"model\": \"MiniMax/speech-02-hd\"\n }'", + "docUrl": "https://help.aliyun.com/document_detail/3021951.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json new file mode 100644 index 00000000..b546ffb1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json @@ -0,0 +1,79 @@ +{ + "name": "AI试衣OutfitAnyone-图片分割", + "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "aitryon-parsing-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-15T13:13:06.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣OutfitAnyone-图片分割", + "docUrl": "https://help.aliyun.com/document_detail/2865249.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json new file mode 100644 index 00000000..d68bf63f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json @@ -0,0 +1,68 @@ +{ + "name": "AI试衣-Plus版", + "description": "aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服饰纹理细节和logo还原效果等方面均有提升，但生成耗时较长，适用于对时效性要求不高的场景。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "aitryon-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-04-27T16:00:00.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣-Plus版", + "docUrl": "https://help.aliyun.com/document_detail/2881846.html", + "predictConfig": [ + { + "name": "sample_models" + }, + { + "name": "sample_suits" + }, + { + "name": "sample_tops" + }, + { + "name": "sample_bottoms" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json new file mode 100644 index 00000000..76173162 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json @@ -0,0 +1,150 @@ +{ + "name": "AI试衣OutfitAnyone-图片精修", + "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "aitryon-refiner", + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "图片生成数量<=25", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25 + }, + { + "rangeStart": 25, + "rangeName": "25<图片生成数量<=125", + "prices": [ + { + "priceUnit": "每张", + "price": "0.275", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 125 + }, + { + "rangeStart": 125, + "rangeName": "125<图片生成数量<=250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 250 + }, + { + "rangeStart": 250, + "rangeName": "250<图片生成数量<=1250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.225", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 1250 + }, + { + "rangeStart": 1250, + "rangeName": "1250<图片生成数量<=2500", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 2500 + }, + { + "rangeStart": 2500, + "rangeName": "2500<图片生成数量<=25000", + "prices": [ + { + "priceUnit": "每张", + "price": "0.175", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25000 + }, + { + "rangeStart": 25000, + "rangeName": "25000<图片生成数量", + "prices": [ + { + "priceUnit": "每张", + "price": "0.15", + "type": "image_number", + "priceName": "图片生成" + } + ] + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣OutfitAnyone-图片精修", + "docUrl": "https://help.aliyun.com/document_detail/2796663.html", + "predictConfig": [ + { + "name": "sample_models" + }, + { + "name": "sample_suits" + }, + { + "name": "sample_tops" + }, + { + "name": "sample_bottoms" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json new file mode 100644 index 00000000..a454b0b6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json @@ -0,0 +1,68 @@ +{ + "name": "AI试衣-基础版", + "description": "aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "aitryon", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-05-24T10:29:05.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣-基础版", + "docUrl": "https://help.aliyun.com/document_detail/2796626.html", + "predictConfig": [ + { + "name": "sample_models" + }, + { + "name": "sample_suits" + }, + { + "name": "sample_tops" + }, + { + "name": "sample_bottoms" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json new file mode 100644 index 00000000..cbee84a2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json @@ -0,0 +1,77 @@ +{ + "name": "舞动人像AnimateAnyone-detect", + "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "animate-anyone-detect-gen2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-10T06:28:33.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "舞动人像AnimateAnyone-detect", + "docUrl": "https://help.aliyun.com/document_detail/2786465.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json new file mode 100644 index 00000000..d969a6c4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json @@ -0,0 +1,79 @@ +{ + "name": "舞动人像AnimateAnyone", + "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "animate-anyone-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-10T06:25:42.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "舞动人像AnimateAnyone", + "docUrl": "https://help.aliyun.com/document_detail/2786464.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json new file mode 100644 index 00000000..ab764969 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json @@ -0,0 +1,79 @@ +{ + "name": "舞动人像AnimateAnyone-template", + "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Video" + ] + }, + "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "animate-anyone-template-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-10T06:27:17.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "舞动人像AnimateAnyone-template", + "docUrl": "https://help.aliyun.com/document_detail/2807955.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json new file mode 100644 index 00000000..8edb64a2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json @@ -0,0 +1,500 @@ +{ + "name": "CosyVoice大模型", + "description": "基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "CosyVoice-v3.5-Flash是通义实验室CosyVoice系列的高性能语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", + "collectionTag": "", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v3.5-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-02-27T09:13:23.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成CosyVoice-v3.5-flash大模型", + "docUrl": "https://help.aliyun.com/document_detail/2842586.html", + "predictConfig": [ + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v3.5-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-02-27T09:13:30.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成CosyVoice-v3.5-plus大模型", + "docUrl": "https://help.aliyun.com/document_detail/2842586.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v3-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-11-17T06:35:36.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成CosyVoice-v3-flash大模型", + "predictConfig": [ + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v3-flash\"\nvoice = \"longanyang\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v3-flash\";\n private static String voice = \"longanyang\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v3-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-03T01:45:34.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成CosyVoice-v3-plus大模型", + "docUrl": "https://help.aliyun.com/document_detail/2817551.html", + "predictConfig": [ + { + "name": "情感", + "key": "emotion", + "default": "neutral", + "tip": "合成音频说话的情感" + }, + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v3-plus\"\nvoice = \"longanyang\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v3-plus\";\n private static String voice = \"longanyang\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-clone-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "声音复刻CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2861519.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v2", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成cosyvoice-v2大模型", + "docUrl": "https://help.aliyun.com/document_detail/2817551.html", + "predictConfig": [ + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + }, + { + "name": "字级别时间戳", + "key": "enableWordTimestamp", + "default": false + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "cosyvoice-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "语音合成CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2817551.html", + "predictConfig": [ + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json index da8f4c8d..b8618df9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json +++ b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json @@ -74,6 +74,59 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V4-Pro", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 4000, + 32768 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -166,6 +219,59 @@ "name": "DeepSeek-V4-Flash", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 4000, + 32768 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -301,6 +407,49 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -390,6 +539,49 @@ "inferenceProvider": "aliyun-bailian", "name": "Deepseek-V3.2-Exp", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -486,6 +678,49 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -594,6 +829,43 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.6, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "presence_penalty", + "key": "presence_penalty", + "default": 0.95, + "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", + "range": [ + -2, + 2 + ] + } + ], "samples": { "openai": { "completionsAPI": { @@ -682,6 +954,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { @@ -787,6 +1076,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { @@ -871,6 +1177,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-7B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { @@ -955,6 +1278,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-32B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { @@ -1039,6 +1379,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-14B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { @@ -1103,6 +1460,23 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], "samples": { "openai": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/embedding.json b/skills/bailian-docs-llm-wiki/models/groups/embedding.json new file mode 100644 index 00000000..a4137421 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/embedding.json @@ -0,0 +1,288 @@ +{ + "name": "通义多模态向量", + "description": "基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "tongyi-embedding-vision-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "ME" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-23T09:09:49.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "视觉向量-flash", + "docUrl": "https://help.aliyun.com/document_detail/2712517.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --silent --location --request POST 'llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-flash\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-flash\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "tongyi-embedding-vision-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "ME" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-23T09:09:31.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "视觉向量-plus", + "docUrl": "https://help.aliyun.com/document_detail/2712517.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --silent --location --request POST 'llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-plus\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-plus\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "multimodal-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.9", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "ME" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 0, + "latestOnlineAt": "2024-12-23T11:55:31.000+00:00", + "contextWindow": 0, + "maxInputTokens": 512, + "inferenceProvider": "aliyun-bailian", + "name": "通用多模态向量", + "docUrl": "https://help.aliyun.com/document_detail/2712517.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"multimodal-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"multimodal-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json new file mode 100644 index 00000000..68d6a6ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json @@ -0,0 +1,77 @@ +{ + "name": "悦动人像EMO-detect", + "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "emo-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-11-07T14:16:51.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "悦动人像EMO-detect", + "docUrl": "https://help.aliyun.com/document_detail/2786463.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json new file mode 100644 index 00000000..74cf39b0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json @@ -0,0 +1,85 @@ +{ + "name": "悦动人像EMO", + "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "emo-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_duration_1-1", + "priceName": "视频生成（1:1画幅视频）" + }, + { + "priceUnit": "每秒", + "price": "0.16", + "type": "video_duration_3-4", + "priceName": "视频生成（3:4画幅视频）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-11-07T14:16:42.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "悦动人像EMO", + "docUrl": "https://help.aliyun.com/document_detail/2786461.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json new file mode 100644 index 00000000..8be9d685 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json @@ -0,0 +1,77 @@ +{ + "name": "表情包Emoji-detect", + "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "emoji-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-16T09:55:26.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "表情包Emoji-detect", + "docUrl": "https://help.aliyun.com/document_detail/2865371.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json new file mode 100644 index 00000000..db243722 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json @@ -0,0 +1,79 @@ +{ + "name": "表情包Emoji", + "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "emoji-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-16T09:46:00.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "表情包Emoji", + "docUrl": "https://help.aliyun.com/document_detail/2865374.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json new file mode 100644 index 00000000..efd42c68 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json @@ -0,0 +1,56 @@ +{ + "name": "FaceChain人物图像检测", + "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "facechain-facedetect", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T08:10:47.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "FaceChain人物图像检测", + "docUrl": "https://help.aliyun.com/document_detail/2712507.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json new file mode 100644 index 00000000..d3430f83 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json @@ -0,0 +1,72 @@ +{ + "name": "FaceChain人物写真生成", + "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "facechain-generation", + "prices": [ + { + "priceUnit": "每张", + "price": "0.18", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T08:12:23.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "FaceChain人物写真生成", + "docUrl": "https://help.aliyun.com/document_detail/2712501.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json new file mode 100644 index 00000000..36b56071 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json @@ -0,0 +1,119 @@ +{ + "name": "通义法睿-Plus-32K", + "description": "通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分析、生成法律文书、检索法律知识、审查合同条款等功能", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "farui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "output_tokens", + "count_limit": 4, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "output_tokens", + "count_limit": 4, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 2000, + "latestOnlineAt": "2024-05-14T13:32:40.000+00:00", + "contextWindow": 12000, + "maxInputTokens": 12000, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "通义法睿-Plus-32K", + "docUrl": "https://help.aliyun.com/document_detail/2778998.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json new file mode 100644 index 00000000..785696df --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json @@ -0,0 +1,86 @@ +{ + "name": "Fun-ASR-Flash", + "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "fun-asr-flash-2026-06-15", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-06-17T08:56:52.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Fun-ASR-Flash-2026-06-15", + "docUrl": "https://help.aliyun.com/document_detail/2979031.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2869541.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json new file mode 100644 index 00000000..56a65fd4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json @@ -0,0 +1,160 @@ +{ + "name": "Fun-ASR实时语音识别", + "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "fun-asr-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-23T11:05:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Fun-ASR实时语音识别", + "docUrl": "https://help.aliyun.com/document_detail/2842554.html", + "predictConfig": [ + { + "name": "开启语义断句", + "key": "semantic_punctuation_enabled", + "default": false, + "tip": "开启语义断句后则将关闭VAD(语音活动检测)断句，具体见说明文档" + }, + { + "name": "VAD静音阈值", + "key": "max_sentence_silence", + "default": 1300, + "tip": "VAD(语音活动检测)断句的静音时长阈值(单位为ms)", + "range": [ + 200, + 6000 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='fun-asr-realtime',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"fun-asr-realtime\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "fun-asr-flash-8k-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-ASR" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "equivalentSnapshot": "fun-asr-flash-8k-realtime-2026-01-28", + "latestOnlineAt": "2026-02-12T11:34:14.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Fun-ASR-Flash-8k实时语音识别", + "docUrl": "https://help.aliyun.com/document_detail/2842554.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='fun-asr-flash-8k-realtime',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"fun-asr-flash-8k-realtime\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json new file mode 100644 index 00000000..652b953f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json @@ -0,0 +1,118 @@ +{ + "name": "Fun-ASR语音识别", + "description": "通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "fun-asr", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-11-20T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Fun-ASR语音识别", + "docUrl": "https://help.aliyun.com/document_detail/2880903.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport dashscope\nimport os\nimport json\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ntask_response = Transcription.async_call(\n model='fun-asr',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "fun-asr-mtl", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-25T06:03:29.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Fun-ASR-MTL", + "docUrl": "https://help.aliyun.com/document_detail/2978300.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport dashscope\nimport os\nimport json\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\ntask_response = Transcription.async_call(\n model='fun-asr-mtl',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json new file mode 100644 index 00000000..0d083681 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json @@ -0,0 +1,140 @@ +{ + "name": "音乐生成", + "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "fun-music-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.002", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-05-06T12:15:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "音乐生成", + "docUrl": "https://help.aliyun.com/document_detail/3030448.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030448.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "fun-music-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.005", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-06-01T07:20:31.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "音乐生成 Preview", + "docUrl": "https://help.aliyun.com/document_detail/3030448.html", + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030448.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json new file mode 100644 index 00000000..51abe656 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json @@ -0,0 +1,1180 @@ +{ + "name": "GLM", + "description": "GLM是由智谱提供的开源模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-5.2是智谱AI推出的面向长程任务（Long Horizon Task）设计的最新旗舰模型，支持1M超长上下文。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。", + "features": [ + "cache", + "function-calling", + "model-experience", + "structured-outputs" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 2000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 2000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-06-16T08:16:52.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5.2", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-5.1是智谱AI推出的面向长程任务（Long Horizon Task）设计的模型，总参数744B，支持200K超长上下文，最大输出 128K tokens。拥有强大逻辑推理、长文本理解与代码生成能力、兼顾性能与推理效率；在多任务基准中表现优异，适用于智能交互、企业应用、开发辅助等场景。", + "collectionTag": "", + "features": [ + "cache", + "function-calling", + "model-experience", + "structured-outputs" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-5.1", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 204800 + } + ], + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-04-14T11:34:59.000+00:00", + "contextWindow": 202745, + "maxInputTokens": 202745, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5.1", + "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-5", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", + "contextWindow": 202752, + "maxInputTokens": 169984, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5", + "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-4.7", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", + "contextWindow": 202752, + "maxInputTokens": 169984, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.7", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM新一代旗舰模型，核心能力较4.5全面提升。总参数量为3550 亿，激活参数320亿，上下文窗口扩展至200K。", + "features": [ + "model-experience", + "cache" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-4.6", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-10-21T06:26:06.000+00:00", + "contextWindow": 202752, + "maxInputTokens": 169984, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.6", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-4.5采用混合专家（MoE）架构，总参数量为3550 亿，激活参数320亿，在复杂推理、代码生成及智能体交互等通用能力上实现了能力融合与技术突破。", + "features": [ + "model-experience" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-4.5", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-08-06T11:38:01.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.5", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-4.5-Air采用混合专家（MoE）架构，总参数量为1060亿，激活参数120亿，相较GLM-4.5更紧凑、轻量，适用于对模型规模和资源消耗有一定限制的场景。", + "features": [ + "model-experience" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-4.5-air", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-08-06T12:02:34.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.5-Air", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json new file mode 100644 index 00000000..853a389d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json @@ -0,0 +1,126 @@ +{ + "name": "GLM-5.2-Fast", + "description": "GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过推理加速优化，输出 TPS 可达 GLM-5.2 标准版的 1.5～2 倍，显著提升输出速度，适用于实时对话、Agent 多轮调用、流式代码生成等对输出速度敏感的场景。", + "features": [ + "cache", + "function-calling", + "prefix-completion", + "structured-outputs", + "web-search" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "glm-5.2-fast-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "56", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-07-09T02:50:34.000+00:00", + "inferenceSpeeds": [ + "fast" + ], + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5.2-Fast-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2-fast-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2-fast-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json new file mode 100644 index 00000000..ce0caa82 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json @@ -0,0 +1,112 @@ +{ + "name": "GUI-Plus", + "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "gui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 540000, + "usage_limit_field": "total_tokens", + "count_limit": 80, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 540000, + "usage_limit_field": "total_tokens", + "count_limit": 80, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-11-12T07:17:19.000+00:00", + "contextWindow": 256000, + "maxInputTokens": 254976, + "inferenceProvider": "aliyun-bailian", + "name": "GUI-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2997010.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", + "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", + "docUrl": "https://help.aliyun.com/document_detail/2997010.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", + "docUrl": "https://help.aliyun.com/document_detail/2997010.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json new file mode 100644 index 00000000..7eb75fd5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json @@ -0,0 +1,68 @@ +{ + "name": "一句话识别及翻译V1.0", + "description": "多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "gummy-chat-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-04T04:07:25.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "一句话识别及翻译V1.0", + "docUrl": "https://help.aliyun.com/document_detail/2866122.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "import requests\nfrom http import HTTPStatus\n\nimport dashscope\nfrom dashscope.audio.asr import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nr = requests.get(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\"\n)\nwith open(\"asr_example.wav\", \"wb\") as f:\n f.write(r.content)\n\ntranslator = TranslationRecognizerRealtime(\n model=\"gummy-chat-v1\",\n format=\"wav\",\n sample_rate=16000,\n translation_target_languages=[\"en\"],\n translation_enabled=True,\n callback=None,\n)\nresult = translator.call(\"asr_example.wav\")\nif not result.error_message:\n print(\"request id: \", result.request_id)\n print(\"transcription: \")\n for transcription_result in result.transcription_result_list:\n print(transcription_result.text)\n print(\"translation[en]: \")\n\n for translation_result in result.translation_result_list:\n print(translation_result.get_translation('en').text)\nelse:\n print(\"Error: \", result.error_message)", + "java": "import com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerParam;\nimport com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerRealtime;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranscriptionResult;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationRecognizerResultPack;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationResult;\n\nimport java.io.File;\nimport java.util.ArrayList;\n\npublic class Main {\n\n public static void main(String[] args) {\n String targetLanguage = \"en\";\n // 创建Recognition实例\n TranslationRecognizerRealtime translator = new TranslationRecognizerRealtime();\n // 创建RecognitionParam，请在实际使用中替换真实apiKey\n TranslationRecognizerParam param =\n TranslationRecognizerParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(\"gummy-chat-v1\")\n .format(\"wav\") // 'pcm'、'wav'、'mp3'、'opus'、'speex'、'aac'、'amr', you\n // can check the supported formats in the document\n .sampleRate(16000)\n .transcriptionEnabled(true)\n .sourceLanguage(\"auto\")\n .translationEnabled(true)\n .translationLanguages(new String[] {targetLanguage})\n .build();\n // 直接将结果保存到script.txt中\n TranslationRecognizerResultPack result = translator.call(param, new File(\"hello_world.wav\"));\n // 任务结束后关闭 websocket 连接\n translator.getDuplexApi().close(1000, \"bye\");\n if (result.getError() != null) {\n System.out.println(\"error: \" + result.getError());\n throw new RuntimeException(result.getError());\n } else {\n System.out.println(\"RequestId: \" + result.getRequestId());\n System.out.println(\"Transcription Results:\");\n ArrayList transcriptionResults = result.getTranscriptionResultList();\n for (int i = 0; i < transcriptionResults.size(); i++) {\n System.out.println(transcriptionResults.get(i).getText());\n }\n\n System.out.println(\"English Translation Results:\");\n ArrayList translationResultList = result.getTranslationResultList();\n for (int i = 0; i < translationResultList.size(); i++) {\n System.out.println(translationResultList.get(i).getTranslation(targetLanguage).getText());\n }\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json new file mode 100644 index 00000000..174f6a5f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json @@ -0,0 +1,68 @@ +{ + "name": "实时语音识别及翻译V1.0", + "description": "多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "gummy-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Audio-Translate" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-04T04:07:10.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "实时语音识别及翻译V1.0", + "docUrl": "https://help.aliyun.com/document_detail/2865393.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "import requests\nfrom http import HTTPStatus\n\nimport dashscope\nfrom dashscope.audio.asr import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nr = requests.get(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\"\n)\nwith open(\"asr_example.wav\", \"wb\") as f:\n f.write(r.content)\n\ntranslator = TranslationRecognizerRealtime(\n model=\"gummy-realtime-v1\",\n format=\"wav\",\n sample_rate=16000,\n translation_target_languages=[\"en\"],\n translation_enabled=True,\n callback=None,\n)\nresult = translator.call(\"asr_example.wav\")\nif not result.error_message:\n print(\"request id: \", result.request_id)\n print(\"transcription: \")\n for transcription_result in result.transcription_result_list:\n print(transcription_result.text)\n print(\"translation[en]: \")\n\n for translation_result in result.translation_result_list:\n print(translation_result.get_translation('en').text)\nelse:\n print(\"Error: \", result.error_message)", + "java": "import com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerParam;\nimport com.alibaba.dashscope.audio.asr.translation.TranslationRecognizerRealtime;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranscriptionResult;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationRecognizerResultPack;\nimport com.alibaba.dashscope.audio.asr.translation.results.TranslationResult;\n\nimport java.io.File;\nimport java.util.ArrayList;\n\npublic class Main {\n\n public static void main(String[] args) {\n String targetLanguage = \"en\";\n // 创建Recognition实例\n TranslationRecognizerRealtime translator = new TranslationRecognizerRealtime();\n // 创建RecognitionParam，请在实际使用中替换真实apiKey\n TranslationRecognizerParam param =\n TranslationRecognizerParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(\"gummy-realtime-v1\")\n .format(\"wav\") // 'pcm'、'wav'、'mp3'、'opus'、'speex'、'aac'、'amr', you\n // can check the supported formats in the document\n .sampleRate(16000)\n .transcriptionEnabled(true)\n .sourceLanguage(\"auto\")\n .translationEnabled(true)\n .translationLanguages(new String[] {targetLanguage})\n .build();\n // 直接将结果保存到script.txt中\n TranslationRecognizerResultPack result = translator.call(param, new File(\"hello_world.wav\"));\n // 任务结束后关闭 websocket 连接\n translator.getDuplexApi().close(1000, \"bye\");\n if (result.getError() != null) {\n System.out.println(\"error: \" + result.getError());\n throw new RuntimeException(result.getError());\n } else {\n System.out.println(\"RequestId: \" + result.getRequestId());\n System.out.println(\"Transcription Results:\");\n ArrayList transcriptionResults = result.getTranscriptionResultList();\n for (int i = 0; i < transcriptionResults.size(); i++) {\n System.out.println(transcriptionResults.get(i).getText());\n }\n\n System.out.println(\"English Translation Results:\");\n ArrayList translationResultList = result.getTranslationResultList();\n for (int i = 0; i < translationResultList.size(); i++) {\n System.out.println(translationResultList.get(i).getTranslation(targetLanguage).getText());\n }\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json index 816195fa..47562ef7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json @@ -176,6 +176,32 @@ "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json new file mode 100644 index 00000000..8f4d4d18 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json @@ -0,0 +1,210 @@ +{ + "name": "HappyHorse-R2V", + "description": "HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "HappyHorse-1.1-R2V支持参考生视频，进一步提升主体、场景风格与画面一致性的稳定保持。模型最多支持9张参考图片输入，能够更精准理解并延续创作意图，在人物、场景、风格和镜头表现上带来更强的可控性与表现力。", + "features": [ + "model-experience" + ], + "provider": "happyhorse", + "limit": { + "message": "model not exist" + }, + "model": "happyhorse-1.1-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-06-16T03:16:08.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "HappyHorse-1.1-R2V", + "docUrl": "https://help.aliyun.com/document_detail/3030778.html", + "category": "Visual", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "docUrl": "https://help.aliyun.com/document_detail/3030778.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "HappyHorse-1.0-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。", + "features": [ + "model-experience" + ], + "provider": "happyhorse", + "limit": { + "message": "model not exist" + }, + "model": "happyhorse-1.0-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 10, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 10, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-26T12:42:22.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "HappyHorse-1.0-R2V", + "docUrl": "https://help.aliyun.com/document_detail/3030778.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "docUrl": "https://help.aliyun.com/document_detail/3030778.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json index f3cc2fea..a953318a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json @@ -63,6 +63,37 @@ "name": "HappyHorse-1.1-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", "category": "Visual", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], "samples": { "dashscope": { "default": { @@ -131,6 +162,37 @@ "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json new file mode 100644 index 00000000..c89ace4b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json @@ -0,0 +1,106 @@ +{ + "name": "HappyHorse-Video-Edit", + "description": "HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", + "features": [], + "provider": "happyhorse", + "limit": { + "message": "model not exist" + }, + "model": "happyhorse-1.0-video-edit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 10, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 10, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-26T07:51:50.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "HappyHorse-1.0-Video-Edit", + "docUrl": "https://help.aliyun.com/document_detail/3030779.html", + "category": "Visual", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长(秒)", + "key": "duration" + }, + { + "name": "声音设置", + "key": "audio_setting", + "tip": [ + "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", + "origin：强制保留输入视频的原声，不重新生成。" + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030779.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json new file mode 100644 index 00000000..2fa41da1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json @@ -0,0 +1,65 @@ +{ + "name": "图像擦除补全", + "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "image-erase-completion", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-08-19T01:17:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "图像擦除补全", + "docUrl": "https://help.aliyun.com/document_detail/2840907.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json new file mode 100644 index 00000000..b3ff2811 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json @@ -0,0 +1,65 @@ +{ + "name": "人物实例分割", + "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "image-instance-segmentation", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-08-19T01:17:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "人物实例分割", + "docUrl": "https://help.aliyun.com/document_detail/2840906.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://image-instance-segmentation/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"image-instance-segmentation\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN01nC4QEU1x58LUeMjRL_!!6000000006391-49-tps-1590-1060.webp\"\n },\n \"parameters\":{\n }\n}'\n\ncurl -X GET \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\nhttps://image-instance-segmentation/api/v1/tasks/53950fb7-281a-4e60-xxxxxxxxxxxx" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json new file mode 100644 index 00000000..e10a832e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json @@ -0,0 +1,71 @@ +{ + "name": "图像画面扩展", + "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "image-out-painting", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-05-24T10:29:57.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "图像画面扩展", + "docUrl": "https://help.aliyun.com/document_detail/2796845.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json new file mode 100644 index 00000000..256febfd --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json @@ -0,0 +1,432 @@ +{ + "name": "Kimi", + "description": "由月之暗面提供的Kimi系列模型的API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "K2.7 Code高速版与普通版是同一个模型，但输出速度约为普通版的 5-6 倍，常规编程场景下（取输入长度中位数）输出速度约 180 Token/s，短上下文场景可达 260 Token/s ，带来更极致的编程体验。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi/kimi-k2.7-code-highspeed", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "13", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 262144, + "latestOnlineAt": "2026-06-17T11:23:33.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 262144, + "offlineInfo": {}, + "inferenceProvider": "moonshot-ai", + "name": "kimi/kimi-k2.7-code-highspeed", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.7-code-highspeed\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.7-code-highspeed\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.7-code-highspeed\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Kimi K2.7 Code 是月之暗面 Kimi发布并开源的新一代编程专用模型，定位为 Kimi 迄今最智能的 Coding 模型。Kimi K2.7 Code 是一个以编码为中心的智能体模型（coding-focused agentic model），专为长程软件工程任务优化。它擅长跨多文件重构、功能实现、长会话调试等需要可靠指令遵循和端到端完成率的复杂工作流。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi/kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 262144, + "latestOnlineAt": "2026-06-15T01:49:13.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 262144, + "offlineInfo": {}, + "inferenceProvider": "moonshot-ai", + "name": "kimi/kimi-k2.7-code", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.7-code\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.7-code\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Kimi K2.6 是 Kimi 最新最智能的模型，Kimi K2.6 的通用 Agent、代码、视觉理解等综合能力得到全面提升，其中在博士级难度的完整版人类最后的考试（Humanity’s Last Exam）、在考察模型真实软件工程能力的 SWE-Bench Pro、评估 Agent 深度检索能力的 DeepSearchQA 等基准测试中均取得行业领先的成绩，同时支持文本、图片与视频输入，思考与非思考模式，对话与 Agent 任务。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi/kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.1", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 262144, + "latestOnlineAt": "2026-04-26T08:54:45.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 262144, + "offlineInfo": {}, + "inferenceProvider": "moonshot-ai", + "name": "Kimi/Kimi K2.6", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Kimi K2.5 是 Kimi 在2026年最新推出的智能模型，在 Agent、代码、视觉理解及一系列通用智能任务上取得开源 SoTA 表现。同时 Kimi K2.5 也是 Kimi 迄今最全能的模型，原生的多模态架构设计，同时支持视觉与文本输入、思考与非思考模式、对话与 Agent 任务。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "moonshot-ai", + "limit": { + "message": "model not exist" + }, + "model": "kimi/kimi-k2.5", + "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 262144, + "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 262144, + "offlineInfo": {}, + "inferenceProvider": "moonshot-ai", + "name": "Kimi/Kimi K2.5", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json new file mode 100644 index 00000000..86945efa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json @@ -0,0 +1,395 @@ +{ + "name": "可灵AI", + "description": "由可灵AI提供的高质量视频与图像生成及编辑模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "智能分镜可读懂剧本场景流转，自动调度机位和景别。原生多模态框架支持音画一致性。打破时长限制，多镜头故事创作更自由。", + "features": [], + "provider": "kling", + "limit": { + "message": "model not exist" + }, + "model": "kling/kling-v3-video-generation", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-26T13:45:28.000+00:00", + "inferenceProvider": "kling", + "name": "Kling Video 3.0", + "predictConfig": [ + { + "name": "mode", + "key": "mode", + "default": "pro" + }, + { + "name": "audio", + "key": "audio", + "default": false + }, + { + "name": "duration", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3026701.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text", + "Video" + ] + }, + "description": "新增“全能参考”，支持3-8秒视频或多图锚定角色元素。可匹配原声及口型驱动，实现角色本色呈现。视频一致性更强，表现更灵动。支持音画同步、智能分镜。", + "features": [], + "provider": "kling", + "limit": { + "message": "model not exist" + }, + "model": "kling/kling-v3-omni-video-generation", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "type": "720P_no_reference_video", + "priceName": "视频生成（720P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "1080P_no_reference_video", + "priceName": "视频生成（1080P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "720P_no_audio_no_reference_video", + "priceName": "视频生成（720P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "type": "1080P_no_audio_no_reference_video", + "priceName": "视频生成（1080P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "type": "720P_no_audio_reference_video", + "priceName": "视频生成（720P 无声 有参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "1080P_no_audio_reference_video", + "priceName": "视频生成（1080P 无声 有参考视频）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-26T13:45:46.000+00:00", + "inferenceProvider": "kling", + "name": "Kling Video 3.0 Omni", + "predictConfig": [ + { + "name": "mode", + "key": "mode", + "default": "pro" + }, + { + "name": "audio", + "key": "audio", + "default": false + }, + { + "name": "duration", + "key": "duration", + "default": 5, + "range": [ + 3, + 15 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3026701.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。", + "features": [], + "provider": "kling", + "limit": { + "message": "model not exist" + }, + "model": "kling/kling-v3-image-generation", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-26T13:45:34.000+00:00", + "inferenceProvider": "kling", + "name": "Kling Image 3.0", + "predictConfig": [ + { + "name": "aspect_ratio", + "key": "aspect_ratio", + "default": "16:9" + }, + { + "name": "resolution", + "key": "resolution", + "default": "1k" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3026706.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。", + "features": [], + "provider": "kling", + "limit": { + "message": "model not exist" + }, + "model": "kling/kling-v3-omni-image-generation", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每秒", + "price": "0.4", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-26T13:45:39.000+00:00", + "inferenceProvider": "kling", + "name": "Kling Image 3.0 Omni", + "predictConfig": [ + { + "name": "aspect_ratio", + "key": "aspect_ratio", + "default": "16:9" + }, + { + "name": "resolution", + "key": "resolution", + "default": "1k" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3026706.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json new file mode 100644 index 00000000..5e25308c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json @@ -0,0 +1,77 @@ +{ + "name": "灵动人像LivePortrait-detect", + "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "liveportrait-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-11-07T14:17:01.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "灵动人像LivePortrait-detect", + "docUrl": "https://help.aliyun.com/document_detail/2856727.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json new file mode 100644 index 00000000..c10bebd9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json @@ -0,0 +1,79 @@ +{ + "name": "灵动人像LivePortrait", + "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "liveportrait", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.02", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-11-07T14:17:00.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "灵动人像LivePortrait", + "docUrl": "https://help.aliyun.com/document_detail/2856730.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json new file mode 100644 index 00000000..31d58143 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json @@ -0,0 +1,437 @@ +{ + "name": "MiniMax文本模型", + "description": "由MiniMax提供的MiniMax-M系列文本模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Image", + "Text", + "Video" + ] + }, + "description": "MiniMax M3 凭借业界领先的 Coding 与 Agentic 能力、1M 超长上下文窗口以及原生多模态特性，可出色胜任企业级长文档理解、高质量内容生成、代码编写、Bug 修复及原生应用构建等任务；强大的 Agentic 能力端到端贯通工作流，原生多模态更带来流畅自然的图文混合交互体验。", + "features": [ + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/MiniMax-M3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.84", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-06-01T02:18:01.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "offlineInfo": {}, + "inferenceProvider": "mini-max", + "name": "MiniMax/MiniMax-M3", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"MiniMax/MiniMax-M3\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M3\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"MiniMax/MiniMax-M3\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "M2.7 能够自行构建复杂 Agent Harness，并基于 Agent Teams、复杂 Skills、Tool Search tool 等能力，完成高度复杂的生产力任务。", + "features": [ + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/MiniMax-M2.7", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 20000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 20000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-03-20T13:04:46.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 204800, + "offlineInfo": {}, + "inferenceProvider": "mini-max", + "name": "MiniMax/MiniMax-M2.7", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "智能体世界的SOTA，专为智能体2.0设计，将编码扩展到现实世界包括工作空间、娱乐和个人助理。模型亮点：全球SOTA开源编码与智能体模型；SWE-bench Pro和SWE-bench Verified得分高于Opus 4.6；在Excel、搜索与研究以及文档摘要方面的全球SOTA；未来工作空间的完美主力模型；闪电般快速：优化思维效率，100+ TPS，实现比 Opus 快 3 倍的速度；极致性价比，以支持始终在线的智能体。", + "features": [ + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/MiniMax-M2.5", + "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 204800, + "offlineInfo": {}, + "inferenceProvider": "mini-max", + "name": "MiniMax/MiniMax-M2.5", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "M2.1 的设计初衷在于打破“最顶级的 Agent 能力仅存在于闭源模型”的壁垒。我们在模型层面进行了针对性优化，显著提升了模型在代码生成、工具调用、复杂指令遵循及长程规划任务中的性能。从自动化进行多语言的软件开发，到执行多步骤的复杂办公工作流，MiniMax-M2.1 均表现出卓越的稳定性。我们致力于为开发者提供一个完全透明、可控且高可用的基础模型，以构建下一代自主智能体应用。", + "features": [ + "function-calling", + "cache" + ], + "provider": "mini-max", + "limit": { + "message": "model not exist" + }, + "model": "MiniMax/MiniMax-M2.1", + "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-02-12T16:00:00.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 204800, + "offlineInfo": {}, + "inferenceProvider": "mini-max", + "name": "MiniMax/MiniMax-M2.1", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3021647.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json new file mode 100644 index 00000000..471c102b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json @@ -0,0 +1,62 @@ +{ + "name": "Paraformer语音识别-8k-v1", + "description": "Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:17:24.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "mhttps://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer语音识别-8k-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-8k-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", + "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-8k-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json new file mode 100644 index 00000000..c2738291 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json @@ -0,0 +1,74 @@ +{ + "name": "Paraformer语音识别-8k-v2", + "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-19T11:29:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer语音识别-8k-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-8k-v2',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", + "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-8k-v2\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json new file mode 100644 index 00000000..5c4728d1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json @@ -0,0 +1,62 @@ +{ + "name": "Paraformer语音识别-mtl-v1", + "description": "Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。\n\n支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话、宁夏话、山西话、陕西话、山东话、四川话、天津话）、英语、日语、韩语、西班牙语、印尼语、法语、德语、意大利语、马来语。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-mtl-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:18:59.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer语音识别-mtl-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-mtl-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", + "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-mtl-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json new file mode 100644 index 00000000..5a99b90d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json @@ -0,0 +1,60 @@ +{ + "name": "Paraformer实时语音识别-8k-v1", + "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-realtime-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-06T11:36:50.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer实时语音识别-8k-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712536.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-8k-v1',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-8k-v1\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json new file mode 100644 index 00000000..6213766c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json @@ -0,0 +1,50 @@ +{ + "name": "Paraformer实时语音识别-8k-v2", + "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服等场景下的实时语音识别。\n支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-realtime-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-31T08:01:43.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer实时语音识别-8k-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712536.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-8k-v2',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-8k-v2\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json new file mode 100644 index 00000000..596c68af --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json @@ -0,0 +1,62 @@ +{ + "name": "Paraformer实时语音识别-v1", + "description": "Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-06T11:36:01.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer实时语音识别-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712536.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-v1',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-v1\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json new file mode 100644 index 00000000..deb72f7b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json @@ -0,0 +1,56 @@ +{ + "name": "Paraformer实时语音识别-v2", + "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。 可支持热词。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-realtime-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer实时语音识别-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712536.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Recognition\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\nrecognition = Recognition(model='paraformer-realtime-v2',\n format='wav',\n sample_rate=16000,\n # “language_hints”只支持paraformer-realtime-v2模型\n language_hints=['zh', 'en'],\n callback=None)\nresult = recognition.call('asr_example.wav')\nif result.status_code == HTTPStatus.OK:\n print('识别结果：')\n print(result.get_sentence())\nelse:\n print('Error: ', result.message)\n \nprint(\n '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}'\n .format(\n recognition.get_last_request_id(),\n recognition.get_first_package_delay(),\n recognition.get_last_package_delay(),\n ))", + "java": "import com.alibaba.dashscope.audio.asr.recognition.Recognition;\nimport com.alibaba.dashscope.audio.asr.recognition.RecognitionParam;\nimport java.io.File;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建Recognition实例\n Recognition recognizer = new Recognition();\n // 创建RecognitionParam\n RecognitionParam param =\n RecognitionParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"paraformer-realtime-v2\")\n .format(\"wav\")\n .sampleRate(16000)\n // “language_hints”只支持paraformer-realtime-v2模型\n .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .build();\n\n try {\n System.out.println(\"识别结果：\" + recognizer.call(param, new File(\"asr_example.wav\")));\n } catch (Exception e) {\n e.printStackTrace();\n }\n System.out.println(\n \"[Metric] requestId: \"\n + recognizer.getLastRequestId()\n + \", first package delay ms: \"\n + recognizer.getFirstPackageDelay()\n + \", last package delay ms: \"\n + recognizer.getLastPackageDelay());\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json new file mode 100644 index 00000000..9f4103de --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json @@ -0,0 +1,62 @@ +{ + "name": "Paraformer语音识别-v1", + "description": "Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:20:05.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer语音识别-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-v1',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", + "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-v1\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json new file mode 100644 index 00000000..90db5a48 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json @@ -0,0 +1,56 @@ +{ + "name": "Paraformer语音识别-v2", + "description": "推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）、英文、日语、韩语。可支持热词。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "paraformer-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Paraformer语音识别-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "过滤语气词", + "key": "disfluency_removal_enabled", + "tip": "过滤语气词，默认为关闭false。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope.audio.asr import Transcription\nimport json\n\n# 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\n\ntask_response = Transcription.async_call(\n model='paraformer-v2',\n file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav',\n 'https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav']\n # language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型\n)\n\ntranscribe_response = Transcription.wait(task=task_response.output.task_id)\nif transcribe_response.status_code == HTTPStatus.OK:\n print(json.dumps(transcribe_response.output, indent=4, ensure_ascii=False))\n print('transcription done!')", + "java": "import com.alibaba.dashscope.audio.asr.transcription.*;\nimport com.google.gson.*;\n\nimport java.util.Arrays;\n\npublic class Main {\n public static void main(String[] args) {\n // 创建转写请求参数\n TranscriptionParam param =\n TranscriptionParam.builder()\n // 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n //.apiKey(\"apikey\")\n .model(\"paraformer-v2\")\n // “language_hints”只支持paraformer-v2模型\n // .parameter(\"language_hints\", new String[]{\"zh\", \"en\"})\n .fileUrls(\n Arrays.asList(\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav\",\n \"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_male2.wav\"))\n .build();\n try {\n Transcription transcription = new Transcription();\n // 提交转写请求\n TranscriptionResult result = transcription.asyncCall(param);\n System.out.println(\"RequestId: \" + result.getRequestId());\n // 阻塞等待任务完成并获取结果\n result = transcription.wait(\n TranscriptionQueryParam.FromTranscriptionParam(param, result.getTaskId()));\n // 打印结果\n System.out.println(new GsonBuilder().setPrettyPrinting().create().toJson(result.getOutput()));\n } catch (Exception e) {\n System.out.println(\"error: \" + e);\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json new file mode 100644 index 00000000..e10dc1fa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json @@ -0,0 +1,535 @@ +{ + "name": "PixVerse C1", + "description": "由爱诗科技提供的PixVerse C系列视频大模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "C1是PixVerse在26年3月底推出的影视行业大模型，r2v（多主体参考生成视频）输入2-7张图像，智能融合不同主体，同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力和想象力、更接近影视专业水准的打斗动作和术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合多主体群像、多人对话、多人交互等复杂剧情，适合中景、全景镜头。\n如果输入了1张多宫格分镜图片（最高支持九宫格），则可以一键生成连续分镜长视频。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-c1-r2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-09T13:35:02.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-C1-r2v", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "C1是PixVerse在26年3月底推出的影视行业大模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-c1-t2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-09T13:34:47.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-C1-t2v", + "docUrl": "https://help.aliyun.com/document_detail/3025608.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5, + "range": [ + 1, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025608.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "C1是PixVerse在26年3月底推出的影视行业大模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-c1-kf2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-09T13:34:56.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-C1-kf2v", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025611.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "C1是PixVerse在26年3月底推出的影视行业大模型，it2v（图片生成视频）模型除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征。相比V6可增强提示词，拥有更强的想象力、更接近影视专业水准的打斗动作、术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合单人特写、单人独白、定格/慢动作、空镜转场等短时长镜头。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-c1-it2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-09T13:34:51.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-C1-it2v", + "docUrl": "https://help.aliyun.com/document_detail/3025609.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "540P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5, + "range": [ + 1, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025609.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json new file mode 100644 index 00000000..9c80d680 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json @@ -0,0 +1,517 @@ +{ + "name": "PixVerse V5.6", + "description": "由爱诗科技提供的PixVerse V系列视频大模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "输入文字描述，秒级生成与语义精准匹配的高质量视频，支持多种风格。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v5.6-t2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T06:58:48.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V5.6-t2v", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025608.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "输入2–7张图像，智能融合不同主体，保持风格统一与动作协调，轻松构建丰富叙事场景，提升内容可控性与创意自由度。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v5.6-r2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T06:43:04.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V5.6-r2v", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "在任意两张图片之间实现无缝转换，实现更流畅自然的场景过渡，打造视觉冲击力强的画面效果。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v5.6-kf2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T06:56:40.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V5.6-kf2v", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025611.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "上传任意图片，自由定制剧情、节奏与风格，生成生动连贯的视频。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v5.6-it2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-03-19T06:59:33.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V5.6-it2v", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "540P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025609.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json new file mode 100644 index 00000000..412c9be7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json @@ -0,0 +1,545 @@ +{ + "name": "PixVerse V6", + "description": "由爱诗科技提供的PixVerse V系列视频大模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "V6是PixVerse在26年3月底推出的新模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v6-t2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-01T07:48:41.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V6-t2v", + "docUrl": "○ https://help.aliyun.com/document_detail/3025608.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5, + "range": [ + 1, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + }, + { + "name": "shot_type", + "key": "shot_type", + "default": "single" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025608.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "V6是PixVerse在26年3月底推出的新模型，r2v（多主体参考生成视频）模型全球排名第二，输入2-7张图像，智能融合不同主体，适合复杂的中景、远景视频镜头。同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力、更强的情绪表现力、和更流畅的高速运动画面。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v6-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-06-08T02:03:00.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V6-r2v", + "docUrl": "https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "V6是PixVerse在26年3月底推出的新模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v6-kf2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-01T07:48:31.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V6-kf2v", + "docUrl": "https://help.aliyun.com/document_detail/3025612.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025611.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "V6是PixVerse在26年3月底推出的新模型，it2v（图片生成视频）模型全球排名第二，it2v除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征，拥有更强的人物情绪、高速运动表现力。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", + "features": [], + "provider": "pixverse", + "limit": { + "message": "model not exist" + }, + "model": "pixverse/pixverse-v6-it2v", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-01T07:48:50.000+00:00", + "inferenceProvider": "pixverse", + "name": "PixVerse-V6-it2v", + "docUrl": "https://help.aliyun.com/document_detail/3025609.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "540P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5, + "range": [ + 1, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "生成音频", + "key": "audio", + "default": true + }, + { + "name": "shot_type", + "key": "shot_type", + "default": "single" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025609.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json new file mode 100644 index 00000000..a9815098 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json @@ -0,0 +1,122 @@ +{ + "name": "QVQ-Max", + "description": "千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qvq-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-03-26T08:48:01.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 106496, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-07-13 23:59:59" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QVQ-Max", + "docUrl": "https://help.aliyun.com/document_detail/2877996.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json new file mode 100644 index 00000000..828bffc4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json @@ -0,0 +1,122 @@ +{ + "name": "Qwen-QVQ-Plus", + "description": "千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qvq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-06-03T08:47:18.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 106496, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-07-13 23:59:59" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QVQ-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2877996.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json new file mode 100644 index 00000000..b87557cc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json @@ -0,0 +1,119 @@ +{ + "name": "Qwen-Audio-Realtime-Flash", + "description": "Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio", + "Text" + ], + "request_modality": [ + "Audio", + "Text" + ] + }, + "description": "千问实时语音对话大模型3.0 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音对话大模型3.0兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。极速版更注重极致的响应速度", + "features": [ + "function-calling" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-audio-3.0-realtime-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "audio_text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "audio_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "type": "audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Chatting" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-07-14T06:59:47.013+00:00", + "contextWindow": 8192, + "maxInputTokens": 4096, + "inferenceProvider": "aliyun-bailian", + "name": "千问实时语音对话大模型3.0（极速版）", + "docUrl": "https://help.aliyun.com/document_detail/3041584.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-flash\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", + "docUrl": "https://help.aliyun.com/document_detail/3041584.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json new file mode 100644 index 00000000..bfc47e7e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json @@ -0,0 +1,119 @@ +{ + "name": "Qwen-Audio-Realtime-Plus", + "description": "Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio", + "Text" + ], + "request_modality": [ + "Audio", + "Text" + ] + }, + "description": "千问实时语音大模型 是一款登顶全球权威评测的下一代实时双工语音大模型，它在全球权威第三方评测平台 Artificial Analysis Speech-to-Speech子项中取得综合排名第一。千问实时语音大模型 兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。标准版更注重高质量的回复结果。", + "features": [ + "function-calling" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-audio-3.0-realtime-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "audio_text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "audio_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "150", + "type": "audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 6, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Chatting" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-07-14T06:59:43.526+00:00", + "contextWindow": 8192, + "maxInputTokens": 4096, + "inferenceProvider": "aliyun-bailian", + "name": "千问实时语音大模型 （标准版）", + "docUrl": "https://help.aliyun.com/document_detail/3041584.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-plus\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", + "docUrl": "https://help.aliyun.com/document_detail/3041584.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json new file mode 100644 index 00000000..eb9707a9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json @@ -0,0 +1,170 @@ +{ + "name": "Qwen-Audio-TTS", + "description": "Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "qwen-audio-3.0-tts-plus是面向高质量语音生成场景打造的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，显著提升方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更准确地控制情绪、语气、角色、语速、音量和合成风格。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，进一步提升了音质、清晰度、分辨率和整体表现力。Plus 版本更强调合成效果和细节表现，适用于有更高音质、自然度和表现力要求的专业场景，如内容创作、有声书、影视配音、品牌声音设计和高品质语音服务。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-audio-3.0-tts-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1.4", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-14T09:19:15.113+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "qwen-audio-3.0-tts-plus", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "qwen-audio-3.0-tts-flash是面向实时交互场景优化的高性能语音合成大模型。相比前一版本，模型支持更多小语种和中文方言，提升了方言发音的正宗程度，并增强了 free-style 指令遵循能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。同时，模型在噪声、混响等复杂声学条件下具备更强鲁棒性，提升了音质、清晰度和整体表现力。Flash 版本重点优化实时合成体验，首包延时控制在 200ms 以内，适用于语音助手、实时对话、智能客服等低延迟交互场景。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-audio-3.0-tts-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-14T09:47:36.084+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "qwen-audio-3.0-tts-flash", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json new file mode 100644 index 00000000..da12e035 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json @@ -0,0 +1,121 @@ +{ + "name": "Qwen-Coder-Plus", + "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-coder-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2024-11-11T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Coder-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json new file mode 100644 index 00000000..022c6124 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json @@ -0,0 +1,121 @@ +{ + "name": "Qwen-Coder-Turbo", + "description": "Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列代码及编程模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-coder-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Coder-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json new file mode 100644 index 00000000..6ced8222 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json @@ -0,0 +1,131 @@ +{ + "name": "Qwen-Doc-Turbo", + "description": "快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。", + "features": [ + "cache" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-doc-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-07-23T13:21:02.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 253952, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Doc-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2948885.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json new file mode 100644 index 00000000..3aaec51d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json @@ -0,0 +1,488 @@ +{ + "name": "Qwen-Embedding", + "description": "基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量V4版本，是通义实验室基于Qwen3训练的多语言文本统一向量模型，相较V3版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升15%~40%；支持64~2048维用户自定义向量维度。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-v4", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1200000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1200000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-06-05T03:07:20.000+00:00", + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-v4", + "docUrl": "https://help.aliyun.com/document_detail/2842587.html", + "category": "Embeddings", + "predictConfig": [ + { + "name": "topK" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v4\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v4\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-07-12T09:44:51.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-v3", + "docUrl": "https://help.aliyun.com/document_detail/2712515.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v3\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v3\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:03:19.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712515.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v2\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:02:12.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712515.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-async-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:05:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-async-v2", + "docUrl": "https://help.aliyun.com/document_detail/2712516.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v2\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text" + ] + }, + "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "text-embedding-async-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T09:04:41.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通用文本向量-async-v1", + "docUrl": "https://help.aliyun.com/document_detail/2712516.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v1\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json new file mode 100644 index 00000000..029f7210 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json @@ -0,0 +1,121 @@ +{ + "name": "Qwen-Flash-Character", + "description": "千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", + "features": [ + "model-experience", + "cache", + "web-search" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-flash-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.05", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-01-13T04:04:02.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 8000, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Flash-Character", + "docUrl": "https://help.aliyun.com/document_detail/2874763.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json new file mode 100644 index 00000000..c9ecda19 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json @@ -0,0 +1,330 @@ +{ + "name": "Qwen-Flash", + "description": "Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-flash", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 30, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 15000, + "usage_limit_period": 30, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 30, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 15000, + "usage_limit_period": 30, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.188", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 997952, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json new file mode 100644 index 00000000..0eead276 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json @@ -0,0 +1,91 @@ +{ + "name": "Qwen-Image-2.0-Pro", + "description": "Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-2.0-pro", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 2, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "qwen-image-2.0-pro", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen-image-2.0-pro-2026-04-22", + "latestOnlineAt": "2026-04-22T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-2.0-Pro", + "docUrl": "https://help.aliyun.com/document_detail/2976416.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + }, + { + "name": "size", + "key": "size", + "default": "2048*2048", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json new file mode 100644 index 00000000..bb06613a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json @@ -0,0 +1,91 @@ +{ + "name": "Qwen-Image-2.0", + "description": "Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-2.0", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "qwen-image-2.0", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen-image-2.0-2026-03-03", + "latestOnlineAt": "2026-03-03T11:31:36.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-2.0", + "docUrl": "https://help.aliyun.com/document_detail/2976416.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + }, + { + "name": "size", + "key": "size", + "default": "2048*2048", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json new file mode 100644 index 00000000..1040a9b2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json @@ -0,0 +1,76 @@ +{ + "name": "Qwen-Image-Edit-Max", + "description": "千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-edit-max", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 2, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-15T12:28:13.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-Edit-Max", + "docUrl": "https://help.aliyun.com/document_detail/2976416.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json new file mode 100644 index 00000000..8839885f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json @@ -0,0 +1,164 @@ +{ + "name": "Qwen-Image-Edit-Plus", + "description": "千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-edit-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-10-30T09:10:49.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-Edit-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2976416.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "千问系列首个图像编辑模型，成功将Qwen-Image的文本渲染能力拓展到编辑任务上。支持精准的中英双语文字编辑、视觉外观与语义双重编辑、具备强大的跨基准性能表现。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-edit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-09-21T16:00:00.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-Edit", + "docUrl": "https://help.aliyun.com/document_detail/2976416.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://{Domain}/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json new file mode 100644 index 00000000..370e45c1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json @@ -0,0 +1,198 @@ +{ + "name": "Qwen-Image-Plus", + "description": "千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2, + "usage_limit_period": 60, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2, + "usage_limit_period": 60, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-23T10:49:31.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2975126.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + }, + { + "name": "size", + "key": "size", + "default": "1328*1328", + "tip": "输出分辨率" + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列首个图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "usage_limit_period": 60, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-13T12:58:52.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image", + "docUrl": "https://help.aliyun.com/document_detail/2975126.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + }, + { + "name": "size", + "key": "size", + "default": "1328*1328", + "tip": "输出分辨率" + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json new file mode 100644 index 00000000..e6e2416a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json @@ -0,0 +1,250 @@ +{ + "name": "Qwen-Long", + "description": "Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列上下文窗口最长，能力均衡且成本较低的模型，适合长文本分析、信息抽取、总结摘要和分类打标等任务。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-long-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 10000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 10, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "LATEST", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-03-19T02:45:13.000+00:00", + "contextWindow": 10000000, + "maxInputTokens": 10000000, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Long-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen-Long是在通义千问针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-long", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2024-05-20T14:57:26.000+00:00", + "contextWindow": 10000000, + "maxInputTokens": 10000000, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Long", + "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json new file mode 100644 index 00000000..4a8c086d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json @@ -0,0 +1,460 @@ +{ + "name": "Qwen-Math-Plus", + "description": "Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问数学模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-math-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 3072, + "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", + "contextWindow": 4096, + "maxInputTokens": 3072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Math-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2849934.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列数学模型是专门用于数学解题的语言模型，推理效果好，模型性能优秀本模型为2024年9月19日快照版本，预计维护到下个版本发布后一个月（待定）。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-math-plus-0919", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "qwen-math-plus-2024-09-19", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 3072, + "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", + "contextWindow": 4096, + "maxInputTokens": 3072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Math-Plus-2024-09-19", + "docUrl": "https://help.aliyun.com/document_detail/2849934.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列数学模型是专门用于数学解题的语言模型，推理效果好，模型性能优秀，本模型是动态更新版本，模型更新不会提前通知。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-math-plus-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "LATEST", + "maxOutputTokens": 3072, + "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", + "contextWindow": 4096, + "maxInputTokens": 3072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Math-Plus-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2849934.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问数学模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-math-plus-0816", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 20000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 20000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "qwen-math-plus-2024-08-16", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 3072, + "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", + "contextWindow": 4096, + "maxInputTokens": 3072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Math-Plus-2024-08-16", + "docUrl": "https://help.aliyun.com/document_detail/2849934.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json new file mode 100644 index 00000000..b809c131 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json @@ -0,0 +1,121 @@ +{ + "name": "Qwen-Math-Turbo", + "description": "Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列数学模型是专门用于数学解题的语言模型，推理速度快，成本低。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-math-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 3072, + "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", + "contextWindow": 4096, + "maxInputTokens": 3072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-07-13 23:59:59" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Math-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2849934.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json new file mode 100644 index 00000000..262f463b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json @@ -0,0 +1,149 @@ +{ + "name": "Qwen-Max", + "description": "千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9.6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 15, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 15, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2024-10-15T05:39:20.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Max", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json new file mode 100644 index 00000000..3287632d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json @@ -0,0 +1,93 @@ +{ + "name": "Qwen-MT-Flash", + "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-mt-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 35000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 35000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "shortDescription": "基于Qwen3全面升级的轻量级文本翻译大模型", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-11-06T08:07:15.000+00:00", + "contextWindow": 16384, + "maxInputTokens": 8192, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-MT-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2860790.html", + "predictConfig": [ + { + "name": "translation_options", + "key": "translation_options", + "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", + "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json new file mode 100644 index 00000000..08a1285b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json @@ -0,0 +1,95 @@ +{ + "name": "Qwen-MT-Image", + "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-mt-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.003", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 1, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 1, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-22T09:53:46.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-MT-Image", + "docUrl": "https://help.aliyun.com/document_detail/2977163.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json new file mode 100644 index 00000000..7483ceb2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json @@ -0,0 +1,92 @@ +{ + "name": "Qwen-MT-Lite", + "description": "基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-mt-lite", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-11-19T11:49:54.000+00:00", + "contextWindow": 16384, + "maxInputTokens": 8192, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-MT-Lite", + "docUrl": "https://help.aliyun.com/document_detail/2860790.html", + "predictConfig": [ + { + "name": "translation_options", + "key": "translation_options", + "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", + "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json new file mode 100644 index 00000000..9d8e6ab2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json @@ -0,0 +1,92 @@ +{ + "name": "Qwen-MT-Plus", + "description": "基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-mt-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5.4", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 25000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 25000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-07-22T06:16:44.000+00:00", + "contextWindow": 16384, + "maxInputTokens": 8192, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-MT-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2860790.html", + "predictConfig": [ + { + "name": "translation_options", + "key": "translation_options", + "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", + "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json new file mode 100644 index 00000000..4fc5da85 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json @@ -0,0 +1,98 @@ +{ + "name": "Qwen-MT-Turbo", + "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-mt-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 35000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 35000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-07-22T06:16:54.000+00:00", + "contextWindow": 16384, + "maxInputTokens": 8192, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-MT-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2860790.html", + "predictConfig": [ + { + "name": "translation_options", + "key": "translation_options", + "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", + "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json new file mode 100644 index 00000000..da5e82e7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json @@ -0,0 +1,266 @@ +{ + "name": "Qwen-Omni-Turbo-Realtime", + "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-omni-turbo-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Omni" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 2048, + "latestOnlineAt": "2025-05-08T11:50:42.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Omni-Turbo-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问全新多模态理解生成大模型实时版，此版本为动态更新版本。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-omni-turbo-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Omni" + ], + "versionTag": "LATEST", + "maxOutputTokens": 2048, + "latestOnlineAt": "2025-05-08T12:29:07.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Omni-Turbo-Realtime-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json new file mode 100644 index 00000000..f8ca1d9f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json @@ -0,0 +1,327 @@ +{ + "name": "Qwen-Omni-Turbo", + "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", + "features": [ + "model-experience", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-omni-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "vision_input_token_cache", + "priceName": "输入：图片/视频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "text_input_token_cache", + "priceName": "输入：文本（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "audio_input_token_cache", + "priceName": "输入：音频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "text_input_token_batch", + "priceName": "输入：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "type": "audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "vision_input_token_batch", + "priceName": "输入：图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "multi_output_token_batch", + "priceName": "输出：文本+音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.25", + "type": "multiin_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "purein_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入仅包含文本时）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 2048, + "latestOnlineAt": "2025-02-14T14:56:44.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Omni-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色，此版本为动态更新版本。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-omni-turbo-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "versionTag": "LATEST", + "maxOutputTokens": 2048, + "latestOnlineAt": "2025-02-14T14:55:08.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Omni-Turbo-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json new file mode 100644 index 00000000..767157a2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json @@ -0,0 +1,122 @@ +{ + "name": "Qwen-Plus-Character", + "description": "千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。", + "features": [ + "model-experience", + "web-search", + "cache", + "structured-outputs" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-plus-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 120, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2025-03-20T09:16:40.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 32768, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Plus-Character", + "docUrl": "https://help.aliyun.com/document_detail/2874763.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json new file mode 100644 index 00000000..19ba8fd9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json @@ -0,0 +1,1025 @@ +{ + "name": "Qwen-Plus", + "description": "千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3系列Plus模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-Plus，达到同规模业界SOTA水平。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-plus", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 30, + "usage_limit": 2500000, + "usage_limit_field": "total_tokens", + "count_limit": 15000, + "usage_limit_period": 30, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 30, + "usage_limit": 2500000, + "usage_limit_field": "total_tokens", + "count_limit": 15000, + "usage_limit_period": 30, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 997952, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。本模型是动态更新版本，模型更新不会提前通知。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-plus-latest", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 30, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 7500, + "usage_limit_period": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 30, + "usage_limit": 200000, + "usage_limit_field": "total_tokens", + "count_limit": 7500, + "usage_limit_period": 10, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "LATEST", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-07-30T16:00:00.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 995904, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Plus-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-1125版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-plus-1220", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 150000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 150000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "qwen-plus-2024-12-20", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2024-12-26T12:24:51.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Plus-2024-12-20", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列能力均衡的模型，推理效果和速度介于千问-Max和千问-Turbo之间，适合中等复杂任务。相对于千问-Plus-2024-1220版本，中英文整体能力有提升，中英常识知识类、阅读理解能力提升较为显著，codeswtich现象相比上一版有显著改善，中文指令遵循能力显著提升。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-plus-0112", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 150000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 150000, + "usage_limit_field": "total_tokens", + "count_limit": 5, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "qwen-plus-2025-01-12", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-01-15T11:28:32.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Plus-2025-01-12", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json new file mode 100644 index 00000000..ac996be7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json @@ -0,0 +1,276 @@ +{ + "name": "Qwen-Rerank", + "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 9000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 9000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-29T10:23:42.000+00:00", + "maxInputTokens": 120000, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Rerank", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 10, + "usage_limit": 5000000000, + "usage_limit_field": "total_tokens", + "count_limit": 900, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 10, + "usage_limit": 5000000000, + "usage_limit_field": "total_tokens", + "count_limit": 900, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 0, + "latestOnlineAt": "2025-10-21T08:21:26.000+00:00", + "contextWindow": 30000, + "maxInputTokens": 30000, + "inferenceProvider": "aliyun-bailian", + "name": "千问3-Rerank", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "gte-rerank-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 83000000, + "usage_limit_field": "total_tokens", + "count_limit": 84, + "usage_limit_period": 1, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 83000000, + "usage_limit_field": "total_tokens", + "count_limit": 84, + "usage_limit_period": 1, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 0, + "latestOnlineAt": "2025-03-20T08:32:00.000+00:00", + "contextWindow": 30000, + "maxInputTokens": 30000, + "inferenceProvider": "aliyun-bailian", + "name": "深度文本重排序", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json new file mode 100644 index 00000000..bbbbb47c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json @@ -0,0 +1,208 @@ +{ + "name": "Qwen-TTS-Realtime", + "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-tts-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 7680, + "latestOnlineAt": "2025-07-16T06:02:09.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 512, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-TTS-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen-tts-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen-tts-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen-TTS实时模型是通义实验室千问模型中语音合成利器，始终与最新快照版能力相同。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。本模型是动态更新版本，模型更新不会提前通知。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-tts-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "versionTag": "LATEST", + "maxOutputTokens": 7680, + "latestOnlineAt": "2025-07-16T06:03:00.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 512, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-TTS-Realtime-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen-tts-realtime-latest',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen-tts-realtime-latest\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json new file mode 100644 index 00000000..1dd04d57 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json @@ -0,0 +1,156 @@ +{ + "name": "Qwen-TTS", + "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-tts", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 7680, + "latestOnlineAt": "2025-04-20T08:27:59.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 512, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-TTS", + "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen-tts\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", + "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen-tts\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-tts-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "LATEST", + "maxOutputTokens": 7680, + "latestOnlineAt": "2025-06-25T16:00:00.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 512, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-TTS-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen-tts-latest\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", + "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen-tts-latest\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json new file mode 100644 index 00000000..2429d4f1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json @@ -0,0 +1,211 @@ +{ + "name": "Qwen-Turbo", + "description": "千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3系列Turbo模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-Turbo，达到同规模业界SOTA水平。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "cache", + "structured-outputs" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 15, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 12, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 15, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 12, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json new file mode 100644 index 00000000..8f39630a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json @@ -0,0 +1,172 @@ +{ + "name": "Qwen-VL-Embedding", + "description": "基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_usage", + "count_limit": 40, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 120000, + "usage_limit_field": "total_usage", + "count_limit": 40, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "ME" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", + "maxInputTokens": 32000, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Embedding", + "docUrl": "https://help.aliyun.com/document_detail/2842587.html", + "predictConfig": [ + { + "name": "topK" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --silent --location --request POST 'llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen2.5-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 60000, + "usage_limit_field": "total_usage", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 60000, + "usage_limit_field": "total_usage", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "ME" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen2.5-VL-Embedding", + "docUrl": "https://help.aliyun.com/document_detail/2842587.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --silent --location --request POST 'llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json new file mode 100644 index 00000000..6d25d67d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json @@ -0,0 +1,155 @@ +{ + "name": "Qwen-VL-Max", + "description": "Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "千问VL-Max（qwen-vl-max），即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience", + "structured-outputs", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-vl-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.32", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "150", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-05-25T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-VL-Max", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json new file mode 100644 index 00000000..1779183b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json @@ -0,0 +1,370 @@ +{ + "name": "Qwen-VL-OCR", + "description": "Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", + "features": [ + "model-experience", + "batch" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-vl-ocr-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "LATEST", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-09-22T16:00:00.000+00:00", + "contextWindow": 38192, + "maxInputTokens": 30000, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QwenVL-OCR-Latest", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。本模型为2024年10月28日的快照版本。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-vl-ocr-1028", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "modelAlias": "qwen-vl-ocr-2024-10-28", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 4096, + "latestOnlineAt": "2024-11-14T13:51:21.000+00:00", + "contextWindow": 34096, + "maxInputTokens": 30000, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QwenVL-OCR-2024-10-28", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "千问VL-OCR（qwen-vl-ocr），即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。", + "features": [ + "model-experience", + "batch" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen-vl-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen-vl-ocr-2025-11-20", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-11-19T16:00:00.000+00:00", + "contextWindow": 38192, + "maxInputTokens": 30000, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QwenVL-OCR", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json new file mode 100644 index 00000000..7c3c8d60 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json @@ -0,0 +1,75 @@ +{ + "name": "Qwen-声音设计", + "description": "Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-voice-design", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "0.2", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen-voice-design", + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-12T07:41:45.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Voice-Design", + "docUrl": "https://help.aliyun.com/document_detail/3000986.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", + "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", + "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json new file mode 100644 index 00000000..0c3d6ace --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json @@ -0,0 +1,68 @@ +{ + "name": "Qwen-声音复刻", + "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen-voice-enrollment", + "prices": [ + { + "priceUnit": "次", + "price": "0.01", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen-voice-enrollment", + "versionTag": "MAJOR", + "latestOnlineAt": "2025-11-27T05:44:15.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Voice-Enrollment", + "docUrl": "https://help.aliyun.com/document_detail/2975034.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", + "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", + "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json new file mode 100644 index 00000000..bb8e0358 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json @@ -0,0 +1,139 @@ +{ + "name": "Qwen2.5-开源模型", + "description": "Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "基于Qwen2.5训练的全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen2.5-omni-7b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "38", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "76", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 2048, + "latestOnlineAt": "2025-03-26T12:01:58.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 30720, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen2.5-Omni-7B", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json new file mode 100644 index 00000000..036cbd42 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json @@ -0,0 +1,88 @@ +{ + "name": "Qwen3-ASR-Flash-Filetrans", + "description": "Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "千问3-ASR-Flash的大文件转录版本，千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-asr-flash-filetrans", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 100, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 100, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-11-17T13:12:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-ASR-Flash-Filetrans", + "docUrl": "https://help.aliyun.com/document_detail/2979031.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", + "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", + "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json new file mode 100644 index 00000000..216c3e80 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json @@ -0,0 +1,86 @@ +{ + "name": "Qwen3-ASR-Flash-Realtime", + "description": "Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "千问3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，通义千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-asr-flash-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 20, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-10-27T10:00:46.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-ASR-Flash-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2989727.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# example requires websocket-client library:\n# pip install websocket-client\n\nimport os\nimport time\nimport json\nimport threading\nimport base64\nimport websocket\nimport logging\nimport logging.handlers\nfrom datetime import datetime\n\nlogger = logging.getLogger(__name__)\nlogger.setLevel(logging.DEBUG)\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：API_KEY=\"sk-xxx\"\nAPI_KEY = os.environ.get(\"DASHSCOPE_API_KEY\")\nQWEN_MODEL = \"qwen3-asr-flash-realtime\"\n\nbaseUrl = \"wss://dashscope.aliyuncs.com/api-ws/v1/realtime\"\nurl = f\"{baseUrl}?model={QWEN_MODEL}\"\nprint(f\"Connecting to server: {url}\")\n\n# 注意： 如果是非vad模式，建议持续发送的音频时长累加不超过60s\nenableServerVad = True\n\nheaders = [\n \"Authorization: Bearer \" + API_KEY,\n \"OpenAI-Beta: realtime=v1\"\n]\n\ndef send_event(ws, event):\n logger.info(f\" Send event: {event['event_id']}, type={event['type']}\")\n ws.send(json.dumps(event))\n\ndef init_logger():\n formatter = logging.Formatter('%(asctime)s|%(levelname)s|%(message)s')\n\n filter = logging.handlers.RotatingFileHandler(\"omni_tester.log\", maxBytes = 100 * 1024 *1024, backupCount = 3)\n filter.setLevel(logging.DEBUG)\n filter.setFormatter(formatter)\n\n console = logging.StreamHandler()\n console.setLevel(logging.DEBUG)\n console.setFormatter(formatter)\n\n logger.addHandler(filter)\n logger.addHandler(console)\n\ndef on_open(ws):\n logger.info(\"Connected to server.\")\n\n # 会话更新事件\n event0 = {\n \"event_id\": \"event_123\",\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\"],\n \"input_audio_format\": \"pcm\",\n \"sample_rate\": 16000,\n \"input_audio_transcription\": {\n # 语种标识，可选，如果有明确的语种信息，建议设置\n \"language\": \"zh\",\n # 语料，可选，如果有语料，建议设置以增强识别效果\n # \"corpus\": {\n # \"text\": \"\"\n # }\n },\n \"turn_detection\": None\n }\n }\n event1 = {\n \"event_id\": \"event_123\",\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\"],\n \"input_audio_format\": \"pcm\",\n \"sample_rate\": 16000,\n \"input_audio_transcription\": {\n # 语种标识，可选，如果有明确的语种信息，建议设置\n \"language\": \"zh\",\n # 语料，可选，如果有语料，建议设置以增强识别效果\n # \"corpus\": {\n # \"text\": \"\"\n # }\n },\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.2,\n \"silence_duration_ms\": 800\n }\n }\n }\n\n global enableServerVad\n if enableServerVad:\n logger.info(f\"Sending event: {json.dumps(event1, indent=2)}\")\n ws.send(json.dumps(event1))\n else:\n logger.info(f\"Sending event: {json.dumps(event0, indent=2)}\")\n ws.send(json.dumps(event0))\n\ndef on_message(ws, message):\n try:\n data = json.loads(message)\n logger.info(f\"Received event: {json.dumps(data, ensure_ascii=False, indent=2)}\")\n except json.JSONDecodeError:\n logger.error(f\"Failed to parse message: {message}\")\n\ndef on_error(ws, error):\n logger.error(f\"Error: {error}\")\n\ndef on_close(ws, close_status_code, close_msg):\n logger.info(f\"Connection closed: {close_status_code} - {close_msg}\")\n\ndef send_audio(ws, local_audio_path):\n time.sleep(5)\n\n with open(local_audio_path, 'rb') as audio_file:\n logger.info(f\"文件读取开始: {datetime.now().strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]}\")\n while True:\n # 读取指定大小的二进制数据\n audio_data = audio_file.read(3200)\n if not audio_data:\n logger.info(f\"文件读取完毕: {datetime.now().strftime('%Y-%m-%d %H:%M:%S.%f')[:-3]}\")\n global enableServerVad\n if enableServerVad is False:\n event = {\n \"event_id\": \"event_789\",\n \"type\": \"input_audio_buffer.commit\"\n }\n ws.send(json.dumps(event))\n break # 如果已达到文件结尾，则退出循环\n\n # 对读取的二进制数据进行 Base64 编码\n encoded_data = base64.b64encode(audio_data).decode('utf-8')\n #print(f\"读取数据：{len(audio_data)} 字节, 编码后: {len(encoded_data)} 字节\")\n\n eventd = {\n \"event_id\": \"event_\" + str(int(time.time() * 1000)),\n \"type\": \"input_audio_buffer.append\",\n \"audio\": encoded_data\n }\n ws.send(json.dumps(eventd))\n logger.info(f\"Sending audio event: {eventd['event_id']}\")\n\n # 模拟实时音频采集\n time.sleep(0.1)\n\n# 添加连接关闭处理函数\nws = websocket.WebSocketApp(\n url,\n header=headers,\n on_open=on_open,\n on_message=on_message,\n on_error=on_error,\n on_close=on_close\n)\n\ninit_logger()\nlogger.info(f\"Connecting to local WebSocket server at {url}...\")\n\n# 替换为待识别的音频文件路径\nlocal_audio_path = \"your_audio_file\"\nthread = threading.Thread(target=send_audio, args=(ws, local_audio_path))\nthread.start()\n\nws.run_forever()" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json new file mode 100644 index 00000000..7bd32b1f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json @@ -0,0 +1,63 @@ +{ + "name": "Qwen3-ASR-Flash", + "description": "Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "千问3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，千问3-ASR-Flash实现了高精度的语音识别功能，能够自动判断语种并准确识别多个语种的语音，在复杂的音频环境下能够保证精确转录。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-asr-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 100, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 100, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "modelAlias": "qwen3-asr-flash", + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-08T05:39:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-ASR-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2979031.html", + "predictConfig": [], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2986952.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json new file mode 100644 index 00000000..78825e8b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json @@ -0,0 +1,185 @@ +{ + "name": "Qwen3-Coder-30B-A3B-Instruct", + "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-coder-30b-a3b-instruct", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.25", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "37.5", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-07-31T12:44:56.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 204800, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Coder-30B-A3B-Instruct", + "docUrl": "https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls", + "category": "Cost-optimized", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json new file mode 100644 index 00000000..dfe9cec8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json @@ -0,0 +1,184 @@ +{ + "name": "Qwen3-Coder-480B-A35B-Instruct", + "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-coder-480b-a35b-instruct", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-07-22T13:29:39.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 204800, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Coder-480B-A35B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json new file mode 100644 index 00000000..e7627005 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json @@ -0,0 +1,258 @@ +{ + "name": "Qwen3-Coder-Flash", + "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-coder-flash", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 997952, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Coder-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json new file mode 100644 index 00000000..def9931d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json @@ -0,0 +1,264 @@ +{ + "name": "Qwen3-Coder-Plus", + "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-coder-plus", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "200", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 + } + ], + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-07-22T13:29:17.000+00:00", + "contextWindow": 1000000, + "maxInputTokens": 997952, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Coder-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json new file mode 100644 index 00000000..e5558107 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json @@ -0,0 +1,122 @@ +{ + "name": "Qwen3-LiveTranslate-Flash-Realtime", + "description": "Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Image", + "Audio" + ] + }, + "description": "Qwen3-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "240", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Audio-Translate" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2025-09-23T11:11:30.000+00:00", + "contextWindow": 53248, + "maxInputTokens": 49152, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-LiveTranslate-Flash-Realtime", + "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "docUrl": "https://help.aliyun.com/document_detail/2983281.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json new file mode 100644 index 00000000..c07a6dd4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json @@ -0,0 +1,113 @@ +{ + "name": "Qwen3-LiveTranslate-Flash", + "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Audio", + "Video" + ] + }, + "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-livetranslate-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-ASR" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2025-12-04T12:21:57.000+00:00", + "contextWindow": 53248, + "maxInputTokens": 49152, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-LiveTranslate-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2999748.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json new file mode 100644 index 00000000..b0030782 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json @@ -0,0 +1,651 @@ +{ + "name": "Qwen3-Max", + "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Completions API", + "name": "search_strategy:agent_max", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent_max", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY", + "tag": "限时优惠" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], + "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-max", + "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "start_time": 1779415463, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, + "model-default-actual": { + "count_limit_period": 1, + "start_time": 1779415463, + "usage_limit": 5000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-max-2026-01-23", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-09-23T11:23:19.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 258048, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Max", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "category": "Flagship", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "cache" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-max-preview", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "qwen3-max-preview", + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2025-09-05T12:43:22.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 258048, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Max-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json new file mode 100644 index 00000000..6270885d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json @@ -0,0 +1,100 @@ +{ + "name": "Qwen3-Omni-30b-a3b-Captioner", + "description": "千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音频内容，能够在多声源、混合化的环境中亦保持稳定而可信的输出。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-omni-30b-a3b-captioner", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-16T16:38:17.000+00:00", + "contextWindow": 65536, + "maxInputTokens": 32768, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Omni-30b-a3b-Captioner", + "docUrl": "https://help.aliyun.com/document_detail/2980468.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json new file mode 100644 index 00000000..114eb4a5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json @@ -0,0 +1,146 @@ +{ + "name": "Qwen3-Omni-Flash-Realtime", + "description": "Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问3-Omni-Flash多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-omni-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.2", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "18.9", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.9", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "8.3", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.2", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "75.1", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Omni" + ], + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-omni-flash-realtime-2025-12-01", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", + "contextWindow": 65536, + "maxInputTokens": 49152, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Omni-Flash-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "音色选择", + "key": "voice" + }, + { + "name": "内容输出", + "key": "modalities", + "default": [ + "text" + ], + "tip": "设置模型返回的模态" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "开启深度思考，开启后将不支持音频输出" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json new file mode 100644 index 00000000..609a55e9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json @@ -0,0 +1,175 @@ +{ + "name": "Qwen3-Omni-Flash", + "description": "Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "description": "千问3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互，生成类人语音实现跨语言精准沟通。模型具备强大指令跟随与系统提示定制功能，灵活适配对话风格与角色设定，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-omni-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "62.6", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "thinking_text_input_token", + "priceName": "输入：文本（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "thinking_audio_input_token", + "priceName": "输入：音频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "thinking_vision_input_token", + "priceName": "输入：图片/视频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "type": "thinking_purein_text_output_token", + "priceName": "输出：文本（思考模式下，输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "thinking_multiin_text_output_token", + "priceName": "输出：文本（思考模式下，输入包含图片/音频/视频时）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "modelAlias": "qwen3-omni-flash", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-omni-flash-2025-12-01", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", + "contextWindow": 65536, + "maxInputTokens": 49152, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Omni-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "音色选择", + "key": "voice" + }, + { + "name": "内容输出", + "key": "modalities", + "default": [ + "text" + ], + "tip": "设置模型返回的模态" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "开启深度思考，开启后将不支持音频输出" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json new file mode 100644 index 00000000..7f91f5dc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json @@ -0,0 +1,78 @@ +{ + "name": "Qwen3-TTS-Flash-Realtime", + "description": "Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "start_time": 1762516375, + "count_limit": 20, + "end_time": 253370736000, + "type": "user-spec" + }, + "model-default-actual": { + "count_limit_period": 1, + "start_time": 1762516375, + "count_limit": 20, + "end_time": 253370736000, + "type": "user-spec" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-flash-realtime", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-tts-flash-realtime-2025-11-27", + "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-TTS-Flash-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-flash-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-flash-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json new file mode 100644 index 00000000..33ef6e49 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json @@ -0,0 +1,72 @@ +{ + "name": "Qwen3-TTS-Flash", + "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-flash", + "versionTag": "MAJOR", + "shortDescription": "韵律拟人，低延迟，支持十种语言和国内多种方言输出", + "equivalentSnapshot": "qwen3-tts-flash-2025-11-27", + "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-TTS-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2879134.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# DashScope SDK 版本不低于 1.23.1\nimport os\nimport dashscope\n\ntext = \"那我来给大家推荐一款T恤，这款呢真的是超级好看，这个颜色呢很显气质，而且呢也是搭配的绝佳单品，大家可以闭眼入，真的是非常好看，对身材的包容性也很好，不管啥身材的宝宝呢，穿上去都是很好看的。推荐宝宝们下单哦。\"\nresponse = dashscope.audio.qwen_tts.SpeechSynthesizer.call(\n # 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n model=\"qwen3-tts-flash\",\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n)\nprint(response)", + "java": "// DashScope SDK 版本需要不低于 2.19.0\nimport com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 仅支持qwen-tts系列模型，请勿使用除此之外的其他模型\n .model(MODEL)\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json new file mode 100644 index 00000000..95a28997 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json @@ -0,0 +1,94 @@ +{ + "name": "qwen3-tts-instruct-flash-realtime", + "description": "通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。该模型等同于2026年01月22日快照版本模型。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-instruct-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-instruct-flash-realtime", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-21T07:33:55.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "qwen3-tts-instruct-flash-realtime", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '对吧~我就特别喜欢这种超市，',\n '尤其是过年的时候',\n '去逛超市',\n '就会觉得',\n '超级超级开心！',\n '想买好多好多的东西呢！'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-instruct-flash-realtime',\n callback=callback, \n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n voice = 'Cherry',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Java SDK 版本需要不低于2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"对吧~我就特别喜欢这种超市\",\n \"尤其是过年的时候\",\n \"去逛超市\",\n \"就会觉得\",\n \"超级超级开心！\",\n \"想买好多好多的东西呢！\"\n };\n\n // 实时PCM音频播放器类\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-instruct-flash-realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n // 创建实时音频播放器实例\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(\"Chelsie\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n \n // 等待音频播放完成并关闭播放器\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json new file mode 100644 index 00000000..289465f3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json @@ -0,0 +1,101 @@ +{ + "name": "Qwen3-TTS-Instruct-Flash", + "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文Instruct调节。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-instruct-flash", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-instruct-flash", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-tts-instruct-flash-2026-01-26", + "latestOnlineAt": "2026-02-10T02:56:41.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-TTS-Instruct-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2879134.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageHint", + "default": "" + }, + { + "name": "音量", + "key": "volume", + "default": 50, + "tip": "数值越大，合成音频声音越大", + "range": [ + 0, + 100 + ] + }, + { + "name": "语速", + "key": "speechRate", + "default": 1, + "tip": "数值越大，合成音频语速越快", + "range": [ + 0.5, + 2 + ] + }, + { + "name": "指令控制", + "key": "instructions", + "tip": "仅支持中英文，通过自然语言合成语音的语气、语速、情感及人物性格，需要具体客观的描述文字，如：\n· 请用非常激昂且高亢的语气说话，表现出获得重大成功后的狂喜与激动。\n· 语速请保持中等偏慢，语气要显得优雅、知性，给人以从容不迫的安心感。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json new file mode 100644 index 00000000..c259c5cd --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json @@ -0,0 +1,75 @@ +{ + "name": "Qwen3-TTS-VC-Realtime", + "description": "Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-vc-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-vc-realtime-0115", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-01-14T11:23:59.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "qwen3-tts-vc-realtime-2026-01-15", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "category": "Audio", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", + "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json new file mode 100644 index 00000000..ce9797ac --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json @@ -0,0 +1,74 @@ +{ + "name": "Qwen3-TTS-VC", + "description": "Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3-TTS-Flash模型是通义最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月22日快照版本模型。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-vc-2026-01-22", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-vc-0122", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-02-10T03:00:44.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-TTS-VC-2026-01-22", + "docUrl": "https://help.aliyun.com/document_detail/2879134.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json new file mode 100644 index 00000000..2c75f0d5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json @@ -0,0 +1,95 @@ +{ + "name": "Qwen3-TTS-VD-Realtime", + "description": "Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2025年12月16日快照版本模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月15日快照版本模型。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-vd-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-vd-realtime-0115", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-01-14T11:14:10.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "qwen3-tts-vd-realtime-2026-01-15", + "docUrl": "https://help.aliyun.com/document_detail/2938790.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json new file mode 100644 index 00000000..be7f3f7d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json @@ -0,0 +1,75 @@ +{ + "name": "Qwen3-TTS-VD", + "description": "Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3-TTS-VD模型是通义最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。该模型为2026年01月26日快照版本模型。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-tts-vd-2026-01-26", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 3, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "modelAlias": "qwen3-tts-vd-0126", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2026-02-10T02:59:43.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-TTS-VD-2026-01-26", + "docUrl": "https://help.aliyun.com/document_detail/2879134.html", + "predictConfig": [ + { + "name": "语言", + "key": "languageType", + "default": "Chinese" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json new file mode 100644 index 00000000..e50feeef --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json @@ -0,0 +1,336 @@ +{ + "name": "Qwen3-VL-Flash", + "description": "Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-flash", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 2500000, + "usage_limit_field": "total_tokens", + "count_limit": 250, + "usage_limit_period": 30, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 2500000, + "usage_limit_field": "total_tokens", + "count_limit": 250, + "usage_limit_period": 30, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.375", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "VU", + "Reasoning" + ], + "modelAlias": "qwen3-vl-plus", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-vl-flash-2026-01-22", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-10-14T06:52:12.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json new file mode 100644 index 00000000..82370582 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json @@ -0,0 +1,331 @@ +{ + "name": "Qwen3-VL-Plus", + "description": "Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "cache", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-plus", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 5, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 250, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 5, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 250, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "VU", + "Reasoning" + ], + "modelAlias": "qwen3-vl-plus", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3-vl-plus-2025-12-19", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-01-25T16:00:00.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json new file mode 100644 index 00000000..b4481f63 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json @@ -0,0 +1,118 @@ +{ + "name": "Qwen3.5-LiveTranslate-Flash-Realtime", + "description": "Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio", + "Text" + ], + "request_modality": [ + "Audio", + "Image" + ] + }, + "description": "Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。", + "collectionTag": "qwen3.5", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "160", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Audio-Translate" + ], + "modelAlias": "qwen3.5-livetranslate-flash-realtime", + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-05-19T08:25:22.000+00:00", + "contextWindow": 53248, + "maxInputTokens": 49152, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-LiveTranslate-Flash-Realtime", + "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", + "category": "Audio", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "docUrl": "https://help.aliyun.com/document_detail/2983281.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json new file mode 100644 index 00000000..b972d71e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json @@ -0,0 +1,113 @@ +{ + "name": "Qwen3.5-OCR", + "description": "Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Image" + ] + }, + "description": "Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景中）抽取上效果显著提升。", + "collectionTag": "qwen3.5", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-06-16T08:00:16.000+00:00", + "contextWindow": 65536, + "maxInputTokens": 49152, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-OCR", + "docUrl": "https://help.aliyun.com/document_detail/2860683.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json new file mode 100644 index 00000000..5d2e4512 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json @@ -0,0 +1,141 @@ +{ + "name": "Qwen3.5-Omni-Flash-Realtime", + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "collectionTag": "Qwen3.5", + "features": [ + "web-search", + "function-calling" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-omni-flash-realtime", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "107", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Omni" + ], + "modelAlias": "qwen3.5-omni-flash-realtime", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3.5-omni-flash-realtime-2026-03-15", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-03-30T03:54:16.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 196608, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-Omni-Flash-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html", + "category": "Multimodal", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json new file mode 100644 index 00000000..63ad3533 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json @@ -0,0 +1,143 @@ +{ + "name": "Qwen3.5-Omni-Flash", + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", + "collectionTag": "Qwen3.5", + "features": [ + "web-search" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-omni-flash", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "72", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.2", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "13.3", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "modelAlias": "qwen3.5-omni-flash", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3.5-omni-flash-2026-03-15", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-03-30T03:58:51.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 196608, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-Omni-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "category": "Multimodal", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "音色选择", + "key": "voice" + }, + { + "name": "内容输出", + "key": "modalities", + "default": [ + "text" + ], + "tip": "设置模型返回的模态" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json new file mode 100644 index 00000000..64dd8247 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json @@ -0,0 +1,141 @@ +{ + "name": "Qwen3.5-Omni-Plus-Realtime", + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", + "collectionTag": "Qwen3.5", + "features": [ + "web-search", + "function-calling" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-omni-plus-realtime", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "80", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "300", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Realtime-Omni" + ], + "modelAlias": "qwen3.5-omni-plus-realtime", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3.5-omni-plus-realtime-2026-03-15", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-03-30T03:54:11.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 196608, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-Omni-Plus-Realtime", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "category": "Multimodal", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2880812.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json new file mode 100644 index 00000000..368b93d8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json @@ -0,0 +1,184 @@ +{ + "name": "Qwen3.5-Omni-Plus", + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Video", + "Audio" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", + "collectionTag": "Qwen3.5", + "features": [ + "web-search", + "function-calling", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-omni-plus", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "53", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "213", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "26.5", + "type": "omni_audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "omni_no_audio_input_token_batch", + "priceName": "输入：文本/图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "omni_no_audio_output_token_batch", + "priceName": "输出：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "53", + "discount": 0.5, + "type": "omni_audio_input_token_batch_chat", + "priceName": "输入：音频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "discount": 0.5, + "type": "omni_no_audio_input_token_batch_chat", + "priceName": "输入：文本/图片/视频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "discount": 0.5, + "type": "omni_no_audio_output_token_batch_chat", + "priceName": "输出：文本（Batch Chat）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Multimodal-Omni" + ], + "modelAlias": "qwen3.5-omni-plus", + "versionTag": "MAJOR", + "equivalentSnapshot": "qwen3.5-omni-plus-2026-03-15", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-03-30T03:58:25.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 196608, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-Omni-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html", + "category": "Multimodal", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "音色选择", + "key": "voice" + }, + { + "name": "内容输出", + "key": "modalities", + "default": [ + "text" + ], + "tip": "设置模型返回的模态" + }, + { + "name": "top_p", + "key": "top_p", + "default": 1, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2867839.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json new file mode 100644 index 00000000..5ca6a260 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json @@ -0,0 +1,1006 @@ +{ + "name": "Qwen3.5开源模型", + "description": "Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", + "collectionTag": "qwen3.5", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-397b-a17b", + "iconUrl": "", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 30, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 500000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 30, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-397B-A17B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", + "collectionTag": "qwen3.5", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-35b-a3b", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3.2", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12.8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-35B-A3B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", + "collectionTag": "qwen3.5", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-27b", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "trainingTypes": { + "sft": [ + "lora", + "full" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-27B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。", + "collectionTag": "qwen3.5", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "web-search", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3.5-122b-a10b", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-02-23T03:28:11.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 260096, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-122B-A10B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + }, + "responsesAPI": { + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "docUrl": "https://help.aliyun.com/document_detail/3016808.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json index 9326bc08..e9241598 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json @@ -350,6 +350,59 @@ "name": "Qwen3.7-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + }, + { + "name": "result_format", + "key": "result_format", + "default": "message", + "tip": "返回结果格式" + } + ], "samples": { "openai": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json new file mode 100644 index 00000000..140eb614 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json @@ -0,0 +1,2803 @@ +{ + "name": "Qwen3开源模型", + "description": "Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列最大尺寸Dense模型的推理版本，多模态推理能力仅次于Qwen3-VL-235B-Thinking，STEM&数学类解题能力、通用图像和视频理解能力出众，多模态Agent能力达到SOTA，适合做复杂多模态推理任务。", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-32b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-10-21T06:26:30.000+00:00", + "contextWindow": 131072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-32B-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列最大尺寸Dense模型的非推理版本，综合表现仅次于Qwen3-VL-235B-Instruct，文档识别和理解能力出色，空间感知与万物识别能力强，视觉2D检测/空间推理能力达到SOTA，适合通用场景下的复杂感知任务。", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-32b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-10-21T06:25:55.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-32B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列第二大MoE模型的Thinking版本，响应速度快，具备更强多模态理解与推理、视觉智能体、长视频长文档等超长上下文支持能力；全面升级图像/视频理解、空间感知与万物识别能力，胜任复杂现实任务。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-30b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 126976, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-30B-A3B-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列第二大MoE模型的Instruct版本，响应速度快，支持长视频长文档等超长上下文；全面升级图像/视频理解、空间感知与万物识别能力；具备视觉2D/3D定位能力，胜任复杂现实任务。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-30b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-30B-A3B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列8B Dense模型的Thinking版本，占用显存更低，能够完成多模态理解与推理；支持长视频长文档等超长上下文、视觉2D/3D定位；全面升级图像/视频理解、空间感知与万物识别能力。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-8b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 126976, + "trainingTypes": { + "sft": [ + "lora", + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-8B-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3-VL系列8B Dense模型的Instruct版本，占用显存更低，全面升级图像/视频理解、长视频长文档等超长上下文支持、空间感知与万物识别能力，胜任复杂现实任务。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-8b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-30T14:44:03.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "trainingTypes": { + "sft": [ + "lora", + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-8B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3系列视觉理解模型，多模态思考能力显著增强，模型在STEM与数学推理方面进行了重点优化；视觉感知与识别能力全面提升、OCR能力迎来重大升级。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-235b-a22b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-23T10:58:07.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 126976, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-235B-A22B-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Qwen3系列视觉理解模型，在视觉coding、空间感知等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解，OCR能力迎来重大升级。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-vl-235b-a22b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 1, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-23T10:53:31.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-235B-A22B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2845871.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的新一代非思考模式开源模型，相较上一版本（千问3-235B-A22B-Instruct-2507）中文文本理解能力更佳、逻辑推理能力有增强、文本生成类任务表现更好。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-next-80b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-11T14:09:35.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Next-80B-A3B-Instruct", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的新一代思考模式开源模型，相较上一版本（千问3-235B-A22B-Thinking-2507指令遵循能力有提升、模型总结回复更加精简。", + "collectionTag": "qwen3", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-next-80b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-09-11T09:09:19.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 126976, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Next-80B-A3B-Thinking", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的非思考模式开源模型，相较上一版本（千问3-30B-A3B）中英文和多语言整体通用能力有大幅提升。主观开放类任务专项优化，显著更加符合用户偏好，能够提供更有帮助性的回复。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-30b-a3b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-07-29T14:20:27.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "trainingTypes": { + "sft": [ + "lora", + "full" + ], + "cpt": [ + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-30B-A3B-Instruct-2507", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的思考模式开源模型，相较上一版本（千问3-30B-A3B）复杂推理类任务性能优秀，包括逻辑推理、数学、科学、代码类等具有一定难度的任务场景，指令遵循、文本理解、多语言翻译等能力显著提高。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-30b-a3b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-07-30T08:38:09.000+00:00", + "contextWindow": 81920, + "maxInputTokens": 126976, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-30B-A3B-Thinking-2507", + "docUrl": "https://help.aliyun.com/document_detail/2870973.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的思考模式开源模型，相较上一版本（千问3-235B-A22B）逻辑能力、通用能力、知识增强及创作能力均有大幅提升，适用于高难度强推理场景。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-235b-a22b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-07-25T10:04:54.000+00:00", + "contextWindow": 131072, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-235B-A22B-Thinking-2507", + "docUrl": "https://help.aliyun.com/document_detail/2870973.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_search", + "key": "enable_search", + "tip": "联网搜索补充互联网知识" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "基于Qwen3的非思考模式开源模型，相较上一版本（千问3-235B-A22B）主观创作能力与模型安全性均有小幅度提升。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-235b-a22b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2025-07-22T13:30:21.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-235B-A22B-Instruct-2507", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-72B-Instruct，达到同规模业界SOTA水平。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-235b-a22b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 16384, + "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 129024, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-235B-A22B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力以更小参数规模比肩QwQ-32B、通用能力显著超过Qwen2.5-14B，达到同规模业界SOTA水平。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-30b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 30, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 30, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 300, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-30B-A3B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力显著超过QwQ、通用能力显著超过Qwen2.5-32B-Instruct，达到同规模业界SOTA水平。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-32b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 40, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 40, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "trainingTypes": { + "sft": [ + "lora", + "full" + ], + "dpo": [ + "lora", + "full" + ], + "cpt": [ + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-32B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-14B。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-14b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-04-28T07:37:07.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "trainingTypes": { + "sft": [ + "lora", + "full" + ], + "dpo": [ + "lora", + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-14B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "实现思考模式和非思考模式的有效融合，可在对话中切换模式。推理能力达到同规模业界SOTA水平、通用能力显著超过Qwen2.5-7B。", + "collectionTag": "qwen3", + "features": [ + "model-experience", + "function-calling", + "structured-outputs", + "prefix-completion", + "fine-tuning" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-8b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "ft", + "priceName": "调优" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-04-28T16:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "trainingTypes": { + "sft": [ + "lora", + "full" + ], + "dpo": [ + "lora", + "full" + ] + }, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-8B", + "docUrl": "https://help.aliyun.com/document_detail/2712576.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "stop", + "key": "stop", + "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Qwen3系列新一代代码生成模型，效果接近Qwen3-Coder-Plus兼具更优性能。模型重点优化仓库级别理解、支持多轮工具交互、提升对于agentic coding类工具的适配能力。", + "collectionTag": "qwen3", + "features": [ + "model-experience" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwen3-coder-next", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], + "capabilities": [ + "TG" + ], + "modelAlias": "", + "versionTag": "SNAPSHOT", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-02-19T14:38:44.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 204800, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "通义千问3-Coder-Next", + "docUrl": "https://help.aliyun.com/document_detail/2850166.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json new file mode 100644 index 00000000..42349e87 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json @@ -0,0 +1,134 @@ +{ + "name": "Qwen-QwQ-Plus", + "description": "千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（IFEval、LiveBench等）达到DeepSeek-R1 满血版水平。", + "offlineAt": "2026-07-13T15:59:59.000+00:00", + "features": [ + "model-experience", + "function-calling", + "web-search", + "batch" + ], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "qwq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 10, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2025-03-05T15:17:03.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "QwQ-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2870973.html", + "category": "Older", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/sambert.json b/skills/bailian-docs-llm-wiki/models/groups/sambert.json new file mode 100644 index 00000000..383e8716 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/sambert.json @@ -0,0 +1,2930 @@ +{ + "name": "Sambert语音合成", + "description": "提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiyue-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:17:01.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知悦", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiyue-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiyue-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiyuan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-20T09:14:08.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知媛", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiyuan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiyuan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiying-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:17:43.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知颖", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiying-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiying-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiye-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:17:07.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知晔", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiye-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiye-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiya-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:17:55.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知雅", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiya-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiya-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhixiao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知笑", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhixiao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhixiao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhixiang-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:13.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知祥", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhixiang-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhixiang-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiwei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:04.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知薇", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiwei-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiwei-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiting-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:16.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知婷", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiting-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiting-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhistella-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:19:54.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知莎", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhistella-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhistella-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhishuo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:18.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知硕", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhishuo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhishuo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhishu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:05.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知树", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhishu-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhishu-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiru-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:22.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知茹", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiru-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiru-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiqian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:08.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知倩", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiqian-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiqian-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiqi-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:25.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知琪", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiqi-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiqi-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhinan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T08:10:47.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知楠", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhinan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhinan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhina-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:46.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知娜", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhina-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhina-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhimo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知墨", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhiming-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:49.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知茗", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhiming-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhiming-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhimiao-emo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:38.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知妙（多情感）", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimiao-emo-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimiao-emo-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhimao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:34.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知猫", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhimao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhimao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhilun-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:12.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知伦", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhilun-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhilun-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhijing-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:41.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知婧", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhijing-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhijing-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhijia-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:14.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知佳", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhijia-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhijia-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhihao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:45.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知浩", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhihao-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhihao-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhigui-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:21:15.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知柜", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhigui-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhigui-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhifei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:49.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知飞", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhifei-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhifei-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhide-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:17.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知德", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhide-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhide-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhida-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:52.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知达", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhida-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhida-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-zhichu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:21:12.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-知厨", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-zhichu-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-zhichu-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-waan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:56.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Waan", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-waan-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-waan-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-perla-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Perla", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-perla-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-perla-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-indah-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:18:59.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Indah", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-indah-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-indah-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-hanna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:21:09.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Hanna", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-hanna-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-hanna-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-eva-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:21:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Eva", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-eva-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-eva-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-donna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:23.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Donna", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-donna-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-donna-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-clara-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:21:06.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Clara", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-clara-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-clara-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-cindy-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:26.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Cindy", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-cindy-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-cindy-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-camila-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:27.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Camila", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-camila-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-camila-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-cally-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:58.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Cally", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-cally-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-cally-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-brian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:54.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Brian", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-brian-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-brian-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-betty-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:16:48.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Betty", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-betty-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-betty-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Text" + ] + }, + "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "sambert-beth-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:30.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Sambert语音合成-Beth", + "docUrl": "https://help.aliyun.com/document_detail/2712458.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "# coding=utf-8\nimport sys\nfrom dashscope.audio.tts import SpeechSynthesizer\n# 若没有将API Key配置到环境变量中，需将apiKey替换为自己的API Key\n# import dashscope\n# dashscope.api_key = \"apiKey\"\nresult = SpeechSynthesizer.call(model='sambert-beth-v1',\n text='今天天气怎么样',\n sample_rate=48000,\n format='wav')\nif result.get_audio_data() is not None:\n with open('output.wav', 'wb') as f:\n f.write(result.get_audio_data())\n print('SUCCESS: get audio data: %dbytes in output.wav' %\n (sys.getsizeof(result.get_audio_data())))\nelse:\n print('ERROR: response is %s' % (result.get_response()))", + "java": "import com.alibaba.dashscope.audio.tts.SpeechSynthesizer;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.tts.SpeechSynthesisAudioFormat;\n\nimport java.io.*;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n public static void syncAudioDataToFile() {\n SpeechSynthesizer synthesizer = new SpeechSynthesizer();\n SpeechSynthesisParam param = SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key\n // .apiKey(\"yourApikey\")\n .model(\"sambert-beth-v1\")\n .text(\"今天天气怎么样\")\n .sampleRate(48000)\n .format(SpeechSynthesisAudioFormat.WAV)\n .build();\n\n File file = new File(\"output.wav\");\n // 提交同步合成任务，获取完整的音频数据\n ByteBuffer audio = synthesizer.call(param);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n System.out.println(\"synthesis done!\");\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n\n public static void main(String[] args) {\n syncAudioDataToFile();\n System.exit(0);\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json new file mode 100644 index 00000000..95dcffdb --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json @@ -0,0 +1,65 @@ +{ + "name": "鞋靴模特", + "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "shoemodel-v1", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-21T02:47:24.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "鞋靴模特", + "docUrl": "https://help.aliyun.com/document_detail/2804662.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json new file mode 100644 index 00000000..1c521b6f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json @@ -0,0 +1,432 @@ +{ + "name": "SiliconFlow DeepSeek", + "description": "由硅基流动提供的DeepSeek系列模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V3.2 是一款兼具高计算效率与卓越推理和 Agent 性能的模型。其方法建立在三大关键技术突破之上：DeepSeek 稀疏注意力（DSA），一种高效的注意力机制，在保持模型性能的同时显著降低了计算复杂性，并特别针对长上下文场景进行了优化；可扩展的强化学习框架，通过该框架，模型性能可与 GPT-5 相媲美，其高算力版本在推理能力上可与 Gemini-3.0-Pro 匹敌；以及大规模 Agent 任务合成管线，旨在将推理能力整合到工具使用场景中，从而提高在复杂交互环境中的指令遵循和泛化能力。该模型在 2025 年国际数学奥林匹克（IMO）和国际信息学奥林匹克（IOI）中取得了金牌表现", + "features": [ + "function-calling", + "prefix-completion" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "siliconflow/deepseek-v3.2", + "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", + "contextWindow": 163840, + "maxInputTokens": 163840, + "inferenceProvider": "siliconflow", + "name": "SiliconFlow DeepSeek-V3.2", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V3.1-Terminus 是由深度求索（DeepSeek）发布的 V3.1 模型的更新版本，定位为混合智能体大语言模型。此次更新在保持模型原有能力的基础上，专注于修复用户反馈的问题并提升稳定性。它显著改善了语言一致性，减少了中英文混用和异常字符的出现。模型集成了“思考模式”（Thinking Mode）和“非思考模式”（Non-thinking Mode），用户可通过聊天模板灵活切换以适应不同任务。作为一个重要的优化，V3.1-Terminus 增强了代码智能体（Code Agent）和搜索智能体（Search Agent）的性能，使其在工具调用和执行多步复杂任务方面更加可靠", + "features": [ + "function-calling", + "prefix-completion" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "siliconflow/deepseek-v3.1-terminus", + "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", + "contextWindow": 163840, + "maxInputTokens": 163840, + "inferenceProvider": "siliconflow", + "name": "SiliconFlow DeepSeek-V3.1-Terminus", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.6, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "presence_penalty", + "key": "presence_penalty", + "default": 0.95, + "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", + "range": [ + -2, + 2 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "新版 DeepSeek-V3 （DeepSeek-V3-0324）与之前的 DeepSeek-V3-1226 使用同样的 base 模型，仅改进了后训练方法。新版 V3 模型借鉴 DeepSeek-R1 模型训练过程中所使用的强化学习技术，大幅提高了在推理类任务上的表现水平，在数学、代码类相关评测集上取得了超过 GPT-4.5 的得分成绩。此外该模型在工具调用、角色扮演、问答闲聊等方面也得到了一定幅度的能力提升。", + "features": [ + "function-calling", + "prefix-completion" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "siliconflow/deepseek-v3-0324", + "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 163840, + "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", + "contextWindow": 163840, + "maxInputTokens": 163840, + "inferenceProvider": "siliconflow", + "name": "SiliconFlow DeepSeek-V3-0324", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.6, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "presence_penalty", + "key": "presence_penalty", + "default": 0.95, + "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", + "range": [ + -2, + 2 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-R1-0528 是一款强化学习（RL）驱动的推理模型，解决了模型中的重复性和可读性问题。在 RL 之前，DeepSeek-R1 引入了冷启动数据，进一步优化了推理性能。它在数学、代码和推理任务中与 OpenAI-o1 表现相当，并且通过精心设计的训练方法，提升了整体效果。", + "features": [ + "function-calling", + "prefix-completion" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "siliconflow/deepseek-r1-0528", + "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 50000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-01-27T16:00:00.000+00:00", + "contextWindow": 163840, + "maxInputTokens": 163840, + "offlineInfo": {}, + "inferenceProvider": "siliconflow", + "name": "SiliconFlow DeepSeek-R1-0528", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "max_tokens", + "key": "max_tokens", + "default": 4000, + "tip": "最终回答的最大长度（不含思维链输出）", + "range": [ + 1, + 8192 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3014912.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json new file mode 100644 index 00000000..2f826214 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json @@ -0,0 +1,70 @@ +{ + "name": "语音识别热词", + "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "speech-biasing", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "ASR" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音识别热词", + "docUrl": "https://help.aliyun.com/document_detail/2712535.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json new file mode 100644 index 00000000..07f68252 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json @@ -0,0 +1,124 @@ +{ + "name": "StepFun推理模型", + "description": "由阶跃星辰StepFun提供的Step系列推理模型API服务", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Step 3.7 Flash 是阶跃星辰最新推出的生产级 Agent 高效率 Flash 模型，专为 Agent、Coding、Search 与多模态工作流打造，在速度、成本、执行可靠性与复杂任务完成能力之间实现了更优平衡。具备多模态感知与执行、视觉搜索与工具增强、高可靠工具调用与编排，以及 Agent 生态兼容优化等核心能力。", + "features": [ + "function-calling", + "structured-outputs", + "cache", + "batch" + ], + "provider": "stepfun", + "limit": { + "message": "model not exist" + }, + "model": "stepfun/step-3.7-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.35", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.1", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.27", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 2000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 2000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "VU" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 262144, + "latestOnlineAt": "2026-05-25T06:20:45.000+00:00", + "contextWindow": 262144, + "maxInputTokens": 262144, + "offlineInfo": {}, + "inferenceProvider": "stepfun", + "name": "stepfun/step-3.7-flash", + "docUrl": "https://help.aliyun.com/document_detail/3036697.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 1, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"stepfun/step-3.7-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"stepfun/step-3.7-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"stepfun/step-3.7-flash\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3036697.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json new file mode 100644 index 00000000..54110b7c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json @@ -0,0 +1,111 @@ +{ + "name": "意图分类模型", + "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "tongyi-intent-detect-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 20, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2024-12-12T11:33:25.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 8192, + "inferenceProvider": "aliyun-bailian", + "name": "意图分类模型", + "docUrl": "https://help.aliyun.com/document_detail/2861138.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" + } + }, + "dashscope": { + "completionsAPI": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json new file mode 100644 index 00000000..12684c51 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json @@ -0,0 +1,109 @@ +{ + "name": "通义晓蜜-对话分析-flash", + "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "tongyi-xiaomi-analysis-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-01-09T03:23:03.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 28672, + "inferenceProvider": "aliyun-bailian", + "name": "通义晓蜜-对话分析-flash", + "docUrl": "https://help.aliyun.com/document_detail/3015075.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + } + }, + "dashscope": { + "default": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json new file mode 100644 index 00000000..29d7e302 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json @@ -0,0 +1,109 @@ +{ + "name": "通义晓蜜-对话分析-pro", + "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", + "features": [], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "tongyi-xiaomi-analysis-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2.7", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 600, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 4096, + "latestOnlineAt": "2026-01-09T03:31:49.000+00:00", + "contextWindow": 32768, + "maxInputTokens": 28672, + "inferenceProvider": "aliyun-bailian", + "name": "通义晓蜜-对话分析-pro", + "docUrl": "https://help.aliyun.com/document_detail/3015075.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + } + }, + "dashscope": { + "default": { + "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json new file mode 100644 index 00000000..be85d158 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json @@ -0,0 +1,338 @@ +{ + "name": "Tripo", + "description": "AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "3D-Generation" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。", + "features": [ + "function-calling", + "structured-outputs", + "batch" + ], + "provider": "tripo", + "limit": { + "message": "model not exist" + }, + "model": "Tripo/Tripo-P1.0", + "prices": [ + { + "priceUnit": "每次", + "price": "2.1", + "type": "generation_3d_text_to_3d_no_texture", + "priceName": "文生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_image_to_3d_no_texture", + "priceName": "单图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_multiview_to_3d_no_texture", + "priceName": "多图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_text_to_3d_sd_texture", + "priceName": "文生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_image_to_3d_sd_texture", + "priceName": "单图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_multiview_to_3d_sd_texture", + "priceName": "多图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_text_to_3d_hd_texture", + "priceName": "文生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "type": "generation_3d_image_to_3d_hd_texture", + "priceName": "单图生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "type": "generation_3d_multiview_to_3d_hd_texture", + "priceName": "多图生3D（带高清贴图）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "3D-generation" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-27T04:07:42.000+00:00", + "inferenceProvider": "tripo", + "name": "Tripo-P1.0", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "3D-Generation" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。", + "features": [ + "function-calling", + "batch", + "structured-outputs" + ], + "provider": "tripo", + "limit": { + "message": "model not exist" + }, + "model": "Tripo/Tripo-H3.1", + "prices": [ + { + "priceUnit": "每次", + "price": "0.7", + "type": "generation_3d_text_to_3d_standard_no_texture", + "priceName": "文生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "type": "generation_3d_image_to_3d_standard_no_texture", + "priceName": "单图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "type": "generation_3d_multiview_to_3d_standard_no_texture", + "priceName": "多图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "type": "generation_3d_text_to_3d_standard_sd_texture", + "priceName": "文生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "type": "generation_3d_image_to_3d_standard_sd_texture", + "priceName": "单图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "type": "generation_3d_multiview_to_3d_standard_sd_texture", + "priceName": "多图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "type": "generation_3d_text_to_3d_standard_hd_texture", + "priceName": "文生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_image_to_3d_standard_hd_texture", + "priceName": "单图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_multiview_to_3d_standard_hd_texture", + "priceName": "多图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "type": "generation_3d_text_to_3d_ultra_no_texture", + "priceName": "文生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_image_to_3d_ultra_no_texture", + "priceName": "单图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_multiview_to_3d_ultra_no_texture", + "priceName": "多图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "type": "generation_3d_text_to_3d_ultra_sd_texture", + "priceName": "文生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_image_to_3d_ultra_sd_texture", + "priceName": "单图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_multiview_to_3d_ultra_sd_texture", + "priceName": "多图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "type": "generation_3d_text_to_3d_ultra_hd_texture", + "priceName": "文生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "type": "generation_3d_image_to_3d_ultra_hd_texture", + "priceName": "单图生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "type": "generation_3d_multiview_to_3d_ultra_hd_texture", + "priceName": "多图生3D（超清版+带高清贴图）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "capabilities": [ + "3D-generation" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-27T04:08:07.000+00:00", + "inferenceProvider": "tripo", + "name": "Tripo-H3.1", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json new file mode 100644 index 00000000..4058cd0d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json @@ -0,0 +1,678 @@ +{ + "name": "Vanchin DeepSeek", + "description": "由快手万擎提供的DeepSeek系列模型API服务。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V4系列是强大的混合专家（MoE）语言模型，包含DeepSeek-V4-Pro（1.6T总参数，49B激活参数）。支持高达100万（1M）token的上下文长度，是在超过32T高质量多样化token上预训练的开源模型。", + "features": [ + "function-calling", + "structured-outputs", + "cache" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-v4-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 300000, + "usage_limit_field": "total_tokens", + "count_limit": 60, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 393216, + "latestOnlineAt": "2026-05-29T08:10:54.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "offlineInfo": {}, + "inferenceProvider": "vanchin", + "name": "vanchin/deepseek-v4-pro", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v4-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V3.2 是一款实现了高计算效率与卓越推理及代理（Agent）性能完美协调的模型。该模型建立在 DeepSeek-V3 的基础之上，通过引入 DeepSeek 稀疏注意力（DSA）、可扩展的强化学习框架以及大规模代理任务合成流水线等关键技术突破，推动了开源大语言模型的前沿发展。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-v3.2-think", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 600000, + "usage_limit_field": "total_tokens", + "count_limit": 30, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-04-07T09:00:00.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "inferenceProvider": "vanchin", + "name": "Vanchin/DeepSeek-V3.2-think", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.2-think\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V3.1-Terminus 是 DeepSeek-V3.1 的更新版本，旨在保持模型原有核心能力的同时，针对用户反馈的问题进行了修复和优化。该版本的模型结构与 DeepSeek-V3 保持一致，并在特定领域进行了显著增强。", + "features": [ + "function-calling", + "cache" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-v3.1-terminus", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 65536, + "latestOnlineAt": "2026-04-07T08:57:08.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 32768, + "inferenceProvider": "vanchin", + "name": "Vanchin/DeepSeek-V3.1-Terminus", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.95, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.6, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": false, + "tip": "推理模式" + }, + { + "name": "thinking_budget", + "key": "thinking_budget", + "default": 4000, + "tip": "思维链的最大输出tokens数量", + "range": [ + 1, + 32768 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-V3 由深度求索（DeepSeek）于 2024 年 12 月发布，是目前开源社区领先的混合专家（MoE）语言模型：总参数 671B，每个 token 仅激活 37B 参数。模型在 14.8 万亿高质量 tokens 上完成预训练，原生支持 128k 上下文。通过创新的无辅助损失负载均衡策略、多头潜在注意力（MLA）架构和 FP8 混合精度训练。", + "features": [ + "structured-outputs", + "prefix-completion", + "function-calling", + "cache" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-v3", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 16384, + "latestOnlineAt": "2026-04-07T08:56:49.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 131072, + "inferenceProvider": "vanchin", + "name": "Vanchin/DeepSeek-V3", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "DeepSeek-R1 是深度求索于 2025 年 1 月开源的 6710 亿参数混合专家（MoE）推理模型，推理时仅激活 370 亿参数。作为首个通过纯强化学习（无监督微调）训练的千亿级模型，实现了链式思维（CoT）的自然涌现。模型在 RL 前加入冷启动数据解决了 R1-Zero 的重复和混语问题，在数学、代码、推理任务上达到 OpenAI o1 水平。", + "features": [ + "function-calling", + "prefix-completion", + "cache" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-r1", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "Reasoning", + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-04-07T08:56:31.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 98304, + "inferenceProvider": "vanchin", + "name": "Vanchin/DeepSeek-R1", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-r1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "DeepSeek-OCR以 “探索视觉 - 文本压缩边界” 为核心目标，从大语言模型（LLM）视角重新定义视觉编码器的功能定位，为文档识别、图像转文本等高频场景提供了兼顾精度与效率的全新解决方案。", + "features": [ + "structured-outputs" + ], + "provider": "deepseek", + "limit": { + "message": "model not exist" + }, + "model": "vanchin/deepseek-ocr", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.216", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.216", + "type": "output_token", + "priceName": "输出" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 50, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "VU", + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 8192, + "latestOnlineAt": "2026-04-07T08:56:06.000+00:00", + "contextWindow": 8192, + "maxInputTokens": 8192, + "inferenceProvider": "vanchin", + "name": "Vanchin/DeepSeek-OCR", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"vanchin/deepseek-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\"\n }\n ]\n }\n ]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\",\n },\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\",\n },\n ],\n }\n ],\n)\n\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nasync function main() {\n const completion = await openai.chat.completions.create({\n model: 'vanchin/deepseek-ocr',\n messages: [\n {\n role: 'user',\n content: [\n {\n type: 'image_url',\n image_url: {\n url: 'https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg',\n detail: 'high',\n },\n },\n {\n type: 'text',\n text: 'Read all the text in the image.',\n },\n ],\n },\n ],\n });\n\n console.log(completion.choices[0].message.content);\n}\n\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3027089.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json new file mode 100644 index 00000000..0b396988 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json @@ -0,0 +1,79 @@ +{ + "name": "视频风格重绘", + "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Video" + ] + }, + "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "video-style-transform", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "视频风格重绘", + "docUrl": "https://help.aliyun.com/document_detail/2846319.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json new file mode 100644 index 00000000..7bcffee3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json @@ -0,0 +1,74 @@ +{ + "name": "声动人像VideoRetalk", + "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Video", + "Audio" + ] + }, + "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "videoretalk", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-12-10T06:23:15.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "声动人像VideoRetalk", + "docUrl": "https://help.aliyun.com/document_detail/2860466.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json new file mode 100644 index 00000000..128fa08d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json @@ -0,0 +1,390 @@ +{ + "name": "Vidu AI生图", + "description": "由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，对中英文字的精准渲染、UI/图表等设计细节的像素级还原，适合制作海报、信息图等。", + "features": [], + "provider": "vidu", + "limit": { + "message": "model not exist" + }, + "model": "vidu/vidu-image_reference2image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.625", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "1", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每张", + "price": "1.46875", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-09T07:27:07.000+00:00", + "inferenceProvider": "vidu", + "name": "Vidu-image_reference2image", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/vidu-image_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，主打高速高质与低成本，成本比Pro降低约50%。", + "features": [], + "provider": "vidu", + "limit": { + "message": "model not exist" + }, + "model": "vidu/viduq3-fast_reference2image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.46875", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.78125", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每张", + "price": "1.09375", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-09T07:27:39.000+00:00", + "inferenceProvider": "vidu", + "name": "ViduQ3-fast_reference2image", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq3-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，擅长处理复杂逻辑，具备超强上下文一致性和工业级稳定性。适合专业设计、漫剧制作等。", + "features": [], + "provider": "vidu", + "limit": { + "message": "model not exist" + }, + "model": "vidu/viduq2-pro_reference2image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.9375", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.9375", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每张", + "price": "1.71875", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-09T07:28:16.000+00:00", + "inferenceProvider": "vidu", + "name": "ViduQ2-Pro_reference2image", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-pro_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，语义理解能力大幅提升，支持更多风格。", + "features": [], + "provider": "vidu", + "limit": { + "message": "model not exist" + }, + "model": "vidu/viduq2-fast_reference2image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.28125", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 300, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-09T07:27:49.000+00:00", + "inferenceProvider": "vidu", + "name": "ViduQ2-fast_reference2image", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html", + "category": "Third-party", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3045893.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json new file mode 100644 index 00000000..a1c306cd --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json @@ -0,0 +1,65 @@ +{ + "name": "虚拟模特V2", + "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "virtualmodel-v2", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-25T15:18:50.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "虚拟模特V2", + "docUrl": "https://help.aliyun.com/document_detail/2796985.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json new file mode 100644 index 00000000..66756a88 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json @@ -0,0 +1,44 @@ +{ + "name": "大模型声音复刻及声音设计", + "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Audio" + ], + "request_modality": [ + "Audio" + ] + }, + "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "voice-enrollment", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 10, + "type": "model-default" + } + }, + "capabilities": [ + "TTS" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "大模型声音复刻及声音设计", + "docUrl": "https://help.aliyun.com/document_detail/2861519.html", + "predictConfig": [] + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json new file mode 100644 index 00000000..a805b6b2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json @@ -0,0 +1,437 @@ +{ + "name": "Wan-Image", + "description": "指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "万相2.7-图像生成与编辑，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现", + "collectionTag": "wan2.7", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-01T03:48:10.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-Image", + "docUrl": "https://help.aliyun.com/document_detail/3026980.html", + "category": "Wan", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "2048*2048" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "生成数量", + "range": [ + 1, + 4 + ] + }, + { + "name": "组图生成", + "key": "enable_sequential", + "default": false + }, + { + "name": "智能改写", + "key": "thinking_mode", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3026980.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "万相2.7-图像生成与编辑旗舰版模型，支持文生图、文生组图、图生组图、图像编辑、多图参考生成、交互式编辑，在文字渲染、主体一致性、复杂指令遵循上都有更强表现。", + "collectionTag": "wan2.7", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-image-pro", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-01T05:32:16.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-Image-Pro", + "docUrl": "https://help.aliyun.com/document_detail/3026980.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "2048*2048" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "生成数量", + "range": [ + 1, + 4 + ] + }, + { + "name": "组图生成", + "key": "enable_sequential", + "default": false + }, + { + "name": "智能改写", + "key": "thinking_mode", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3026980.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image", + "Text" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "万相2.6-图像生成，全能图像生成模型，支持图文一体化推理生成，具备多图创意融合、商用级一致性、美学要素迁移与镜头光影精确控制，全面提升图像生成的一致性、可控性和表现力。", + "collectionTag": "wan2.6", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-15T11:55:32.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-Image", + "docUrl": "https://help.aliyun.com/document_detail/3001143.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "万相2.5-图像编辑-Preview，全新升级模型架构。支持指令控制实现丰富的图像编辑能力，指令遵循能力进一步提升，支持高一致性保持的多图参考生成，文字生成表现优异。", + "collectionTag": "wan2.5", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.5-i2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-23T13:38:14.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.5-I2I-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2982258.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-imageedit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-25T07:22:03.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-ImageEdit", + "docUrl": "https://help.aliyun.com/document_detail/2868981.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json new file mode 100644 index 00000000..4c4601d5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json @@ -0,0 +1,1330 @@ +{ + "name": "Wan-I2V", + "description": "图片生成视频内容，稳定保持图像主体、风格和文字等细节信息", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Audio", + "Image", + "Text" + ] + }, + "description": "万相2.7-图生视频，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。", + "collectionTag": "wan2.7", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-03T03:21:23.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-I2V", + "docUrl": "https://help.aliyun.com/document_detail/3025059.html", + "category": "Wan", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 2, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3025059.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Image", + "Audio", + "Text" + ] + }, + "description": "万相2.6-图生视频-Flash，生成更快更高性价比。智能分镜调度支持多镜头叙事，多人稳定对话，更自然真实音色，最高支持15秒时长生成", + "collectionTag": "Wan2.6", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-i2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.25", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-15T07:18:54.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "wan2.6-I2V-flash", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "category": "Wan", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + }, + { + "name": "智能多镜", + "key": "shot_type", + "default": "single", + "tip": "开启后输出视频采用多分镜形式呈现" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Image", + "Text", + "Audio" + ] + }, + "description": "万相2.6-图生视频，智能分镜调度支持多镜头叙事，更高品质的声音生成，多人稳定对话，更自然真实音色，最高支持15秒时长生成", + "collectionTag": "wan2.6", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "discount": 0.5, + "type": "720P_batch", + "priceName": "视频生成（720P Batch Chat）" + }, + { + "priceUnit": "每秒", + "price": "1", + "discount": 0.5, + "type": "1080P_batch", + "priceName": "视频生成（1080P Batch Chat）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-03T13:03:01.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-I2V", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + }, + { + "name": "智能多镜", + "key": "shot_type", + "default": "single", + "tip": "开启后输出视频采用多分镜形式呈现" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Text", + "Image", + "Audio" + ] + }, + "description": "万相2.5-图生视频-Preview，全新升级技术架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。", + "collectionTag": "wan2.5", + "features": [ + "model-experience", + "fine-tuning" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.5-i2v-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-19T08:44:48.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.5-I2V-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "全新升级的万相2.2图生视频，视频品质更高。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.14", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-I2V-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "全新升级的万相2.2-首尾帧生视频，生成速度更快。优化视频动态稳定性与成功率，更强大的指令遵循能力，两张图片生成丝滑过度视频。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-kf2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.1", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.48", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-12T05:38:00.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-KF2V-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2880649.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Video", + "Image" + ] + }, + "description": "wan2.2-animate-move图生动作是一款角色动画生成模型，用户只需上传一张角色照片和一段参考表演视频，即可将视频中的表情和动作迁移到图片角色上，生成高保真的动画视频。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-animate-move", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.4", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 1 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 1 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "Wan2.2-Animate-Move", + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-19T04:02:30.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-Animate-Move", + "docUrl": "https://help.aliyun.com/document_detail/2981852.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Video", + "Image" + ] + }, + "description": "wan2.2-animate-mix视频换人是一款角色替换的模型产品，上传一张角色照片与一段表演视频，即可将原视频中的角色精准替换为照片中的角色，完整保留原始视频的场景、光照和色调等环境细节。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-animate-mix", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 1 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 1 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-19T04:02:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-Animate-Mix", + "docUrl": "https://help.aliyun.com/document_detail/2982219.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [], + "request_modality": [ + "Image" + ] + }, + "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-s2v-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-S2V-Detect", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-s2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通义万相2.2-数字人-S2V", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-i2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.1", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.48", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-I2V-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "720P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相2.1-首尾帧-Plus，两张图片生成丝滑过度视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成画面细节更丰富。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-kf2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-04-20T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-KF2V-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2880649.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相2.1-图生视频-Plus，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，视频质量更高。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-20T03:30:02.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-I2V-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "720P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相2.1-图生视频-Turbo，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成速度更快。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-i2v-turbo", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-02-27T02:24:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-I2V-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "720P" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json new file mode 100644 index 00000000..0699e8fc --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json @@ -0,0 +1,316 @@ +{ + "name": "Wan-R2V", + "description": "参考视频中的人或物，精准保持形象和声音，支持多参考合拍", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Audio", + "Image", + "Text", + "Video" + ] + }, + "description": "Wan2.7-R2V，更加稳定的角色、道具与场景参考，支持最大5个图/视频混合参考，支持音频音色参考，搭配基础能力升级实现更强表演能力。", + "collectionTag": "wan2.7", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-03T03:21:08.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-R2V", + "docUrl": "https://help.aliyun.com/document_detail/3001146.html", + "category": "Wan", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 2, + 10 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-r2v\",\n \"input\": {\n \"prompt\": \"视频2抱着图片3在咖啡厅里弹奏一支舒缓的美式乡村民谣，视频1笑着看着视频2\",\n \"media\": [\n {\n \"type\": \"reference_video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/hfugmr/wan-r2v-role1.mp4\"\n },\n {\n \"type\": \"reference_video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qigswt/wan-r2v-role2.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": false,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3001146.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Image", + "Video", + "Text" + ] + }, + "description": "万相2.6-参考生视频-Flash，生成更快性价比更高。支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-r2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.25", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2026-01-29T10:15:01.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-R2V-Flash", + "docUrl": "https://help.aliyun.com/document_detail/3001146.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能多镜", + "key": "shot_type", + "default": "single", + "tip": "开启后输出视频采用多分镜形式呈现" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-r2v-flash\",\n \"input\": {\n \"prompt\": \"character1在沙发上开心地看电影\",\n \"reference_urls\":[\"https://cdn.wanx.aliyuncs.com/static/demo-wan26/vace.mp4\"]\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"shot_type\":\"multi\"\n }\n}'\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Image", + "Video", + "Text" + ] + }, + "description": "万相2.6-参考生视频，支持指定人物或任意物品进行参考，精准保持形象和声音的一致性，支持多角色参考合拍。提醒：当使用视频进行参考时，输入视频也会计入费用，详见模型计费文档。", + "collectionTag": "wan2.6", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-15T16:08:49.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-R2V", + "docUrl": "https://help.aliyun.com/document_detail/3001146.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能多镜", + "key": "shot_type", + "default": "single", + "tip": "开启后输出视频采用多分镜形式呈现" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-r2v\",\n \"input\": {\n \"prompt\": \"character1在沙发上开心地看电影\",\n \"reference_urls\":[\"https://cdn.wanx.aliyuncs.com/static/demo-wan26/vace.mp4\"]\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"shot_type\":\"multi\"\n }\n}'\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json new file mode 100644 index 00000000..884e18c1 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json @@ -0,0 +1,783 @@ +{ + "name": "Wan-T2I", + "description": "文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.6-文生图，画面质感、美学表现、指令遵循升级，在艺术风格精准控制、真实感人像、长文本生图及广泛历史文化IP覆盖上均表现出卓越能力，可生成高质量且富有表现力的视觉内容。", + "collectionTag": "wan2.6", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-t2i", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 1, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 1, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-15T08:05:15.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-T2I", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*1280", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.6-t2i\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"prompt_extend\": true,\n \"watermark\": false,\n \"n\": 2,\n \"size\": \"1280*1280\"\n }\n}'" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.5-文生图-Preview，全新升级模型架构。画面美学、设计感、真实质感显著提升，精准指令遵循，擅长中英文和小语种文字生成，支持复杂结构化长文本和图表、架构图等内容生成。", + "collectionTag": "wan2.5", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.5-t2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-19T08:44:37.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.5-T2I-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*1280", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "全新升级的万相2.2文生图，更丰富的画面细节。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。", + "collectionTag": "wan2.2", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "user-spec": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "start_time": 1757403106, + "usage_limit": 1000000, + "usage_limit_field": "tokens", + "count_limit": 2, + "end_time": 253370736000, + "usage_limit_period": 60, + "type": "user-spec", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-T2I-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*720", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "全新升级的万相2.2文生图，更快的生成速度。在生成图像创意性、稳定性、写实质感方面全面升级，指令遵循更强，原生支持多种风格。支持最大200万像素生成，支持智能提示词改写等。", + "collectionTag": "wan2.2", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-t2i-flash", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-T2I-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*720", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.1-文生图-Plus，更丰富的画面细节，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-08T16:09:10.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-T2I-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*720", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.1-文生图-Turbo，更快的生成速度，在图像美观度、真实感、艺术性上全面升级，更强的语义理解能力、丰富的风格泛化性、支持最大200万像素生成，支持智能提示词改写等。", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-T2I-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1280*720", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.16", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "wanx-t2i", + "docUrl": "https://help.aliyun.com/document_detail/2712483.html", + "predictConfig": [ + { + "name": "negative_prompt", + "key": "negativePrompt", + "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" + }, + { + "name": "style", + "key": "style", + "default": "", + "tip": "输出风格" + }, + { + "name": "size", + "key": "size", + "default": "1024*1024", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "种子值", + "range": [ + 1, + 4294967289 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Wan2.0-T2I-Turbo，更擅长质感人像和创意设计画作生成，在图像美观度、真实感、艺术性上全面升级，支持最大200万像素生成，支持智能提示词改写等。", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.0-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.04", + "type": "image_number", + "priceName": "图片生成" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-20T03:29:28.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.0-T2I-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2862677.html", + "predictConfig": [ + { + "name": "size", + "key": "size", + "default": "1024*1024", + "tip": "输出分辨率" + }, + { + "name": "n", + "key": "n", + "default": 1, + "tip": "本次请求生成的图片数量" + }, + { + "name": "seed", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-v1-0521", + "capabilities": [ + "IG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "万相-文本生成图像-2024-05-21", + "docUrl": "https://help.aliyun.com/document_detail/2712483.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json new file mode 100644 index 00000000..6dbccf8e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json @@ -0,0 +1,605 @@ +{ + "name": "Wan-T2V", + "description": "文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Audio", + "Text" + ] + }, + "description": "Wan2.7-T2V，演绎能力全面升级，文戏情感细腻自然，动作戏激烈拳拳到肉，搭配更富有戏剧性和节奏感的镜头切换，实现更强表演能力。", + "collectionTag": "wan2.7", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-03T03:21:57.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-T2V", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "category": "Wan", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长(秒)", + "key": "duration", + "default": 5, + "range": [ + 2, + 15 + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-t2v\",\n \"input\": {\n \"prompt\": \"一段紧张刺激的侦探追查故事，展现电影级叙事能力。第1个镜头[0-3秒] 全景：雨夜的纽约街头，霓虹灯闪烁，一位身穿黑色风衣的侦探快步行走。 第2个镜头[3-6秒] 中景：侦探进入一栋老旧建筑，雨水打湿了他的外套，门在他身后缓缓关闭。 第3个镜头[6-9秒] 特写：侦探的眼神坚毅专注，远处传来警笛声，他微微皱眉思考。 第4个镜头[9-12秒] 中景：侦探在昏暗走廊中小心前行，手电筒照亮前方。 第5个镜头[12-15秒] 特写：侦探发现关键线索，脸上露出恍然大悟的表情。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"prompt_extend\": true,\n \"watermark\": true,\n \"duration\": 15\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Text", + "Audio" + ] + }, + "description": "万相2.6-文生视频，智能分镜调度支持多镜头叙事，能够生成主体、场景和氛围一致的多镜头叙事视频，最高支持15秒时长，更高品质的声音生成，更好的指令遵循和视觉质量", + "collectionTag": "wan2.6", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.6-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "discount": 0.5, + "type": "720P_batch", + "priceName": "视频生成（720P Batch Chat）" + }, + { + "priceUnit": "每秒", + "price": "1", + "discount": 0.5, + "type": "1080P_batch", + "priceName": "视频生成（1080P Batch Chat）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-12-03T13:03:19.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.6-T2V", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + }, + { + "name": "智能多镜", + "key": "shot_type", + "default": "single", + "tip": "开启后输出视频采用多分镜形式呈现" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-t2v\",\n \"input\": {\n \"prompt\": \"一幅史诗级可爱的场景。一只小巧可爱的卡通小猫将军，身穿细节精致的金色盔甲，头戴一个稍大的头盔，勇敢地站在悬崖上。他骑着一匹虽小但英勇的战马，说：”青海长云暗雪山，孤城遥望玉门关。黄沙百战穿金甲，不破楼兰终不还。“。悬崖下方，一支由老鼠组成的、数量庞大、无穷无尽的军队正带着临时制作的武器向前冲锋。这是一个戏剧性的、大规模的战斗场景，灵感来自中国古代的战争史诗。远处的雪山上空，天空乌云密布。整体氛围是“可爱”与“霸气”的搞笑和史诗般的融合。\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video", + "Audio" + ], + "request_modality": [ + "Text", + "Audio" + ] + }, + "description": "万相2.5-文生视频-Preview，全新升级模型架构，支持与画面同步的声音生成，支持10秒长视频生成，更强的指令遵循能力，运动能力、画面质感进一步提升。", + "collectionTag": "wan2.5", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.5-t2v-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-09-19T06:10:15.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.5-T2V-Preview", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + }, + { + "name": "生成音频", + "key": "audio", + "default": true + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wan2.5-t2v-preview',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "全新升级的万相2.2文生视频，视频品质更高。可稳定生成大幅度复杂运动，支持影视级画面表现与控制，更强大的指令遵循能力，实现物理世界还原。", + "collectionTag": "wan2.2", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.2-t2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.14", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-T2V-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1920*1080" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wan2.2-t2v-plus',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.1-文生视频-Plus，一句话生成视频。视频品质更高，支持大幅度复杂运动、现实物理规律还原、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-t2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-08T16:13:12.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-T2V-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1280*720" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wanx2.1-t2v-plus',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相2.1-文生视频-Turbo，一句话生成视频。生成速度更快，支持大幅度复杂运动、现实物理规律还原、丰富的艺术风格和影视级画面质感，指令遵循能力进一步提升。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-t2v-turbo", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-T2V-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/2865250.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1280*720" + }, + { + "name": "视频时长", + "key": "duration", + "default": 5 + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": true, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\ndef sample_sync_call_t2v():\n # call sync api, will return the result\n print('please wait...')\n rsp = VideoSynthesis.call(model='wanx2.1-t2v-turbo',\n prompt='一只小猫在月光下奔跑',\n size='1280*720')\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_t2v()" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json new file mode 100644 index 00000000..01890bfa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json @@ -0,0 +1,111 @@ +{ + "name": "Wan-VideoEdit", + "description": "通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Image", + "Text", + "Video" + ] + }, + "description": "Wan2.7-VideoEdit，自然语言指令编辑视频，支持局部或全局编辑，可参考图像替换视频元素，支持复刻视频动作、特效、运镜等动态过程。", + "collectionTag": "wan2.7", + "features": [ + "model-experience" + ], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wan2.7-videoedit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 5, + "type": "model-default", + "async_user_concurrency_limit": 5 + } + }, + "capabilities": [ + "VG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-03T03:19:53.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.7-VideoEdit", + "docUrl": "https://help.aliyun.com/document_detail/3021842.html", + "category": "Wan", + "predictConfig": [ + { + "name": "清晰度", + "key": "resolution", + "default": "1080P" + }, + { + "name": "宽高比", + "key": "ratio", + "default": "16:9" + }, + { + "name": "视频时长", + "key": "duration", + "tip": "可以选择与输入视频时长相同，或者指定不大于输入视频的时长" + }, + { + "name": "声音设置", + "key": "audio_setting", + "tip": [ + "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", + "origin：强制保留输入视频的原声，不重新生成。" + ] + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "tip": "随机数种子，用于控制模型生成内容的随机性", + "range": [ + 1, + 2147483647 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3021842.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json new file mode 100644 index 00000000..ee7c383f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json @@ -0,0 +1,79 @@ +{ + "name": "图像背景生成", + "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-background-generation-v2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-03-22T03:33:37.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "图像背景生成", + "docUrl": "https://help.aliyun.com/document_detail/2712497.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json new file mode 100644 index 00000000..2a56ab16 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json @@ -0,0 +1,65 @@ +{ + "name": "创意海报生成", + "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-poster-generation-v1", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-21T02:49:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "创意海报生成", + "docUrl": "https://help.aliyun.com/document_detail/2807172.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json new file mode 100644 index 00000000..16f81793 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json @@ -0,0 +1,75 @@ +{ + "name": "万相-涂鸦作画", + "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-sketch-to-image-lite", + "prices": [ + { + "priceUnit": "每张", + "price": "0.06", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-11T11:21:09.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "万相-涂鸦作画", + "docUrl": "https://help.aliyun.com/document_detail/2712498.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json new file mode 100644 index 00000000..bf4be56a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json @@ -0,0 +1,73 @@ +{ + "name": "人像风格重绘", + "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-style-repaint-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.12", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-03-22T03:32:59.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "人像风格重绘", + "docUrl": "https://help.aliyun.com/document_detail/2712493.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json new file mode 100644 index 00000000..b95d1ed4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json @@ -0,0 +1,65 @@ +{ + "name": "虚拟模特", + "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-virtualmodel", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-25T15:19:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "虚拟模特", + "docUrl": "https://help.aliyun.com/document_detail/2796985.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json new file mode 100644 index 00000000..f49a4258 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json @@ -0,0 +1,67 @@ +{ + "name": "万相-图像局部重绘", + "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image" + ] + }, + "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx-x-painting", + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-05-28T10:45:39.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "万相-图像局部重绘", + "docUrl": "https://help.aliyun.com/document_detail/2797051.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json new file mode 100644 index 00000000..070111a3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json @@ -0,0 +1,93 @@ +{ + "name": "Wan2.1-VACE-Plus", + "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Video" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", + "features": [], + "provider": "wan", + "limit": { + "message": "model not exist" + }, + "model": "wanx2.1-vace-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-05-13T16:00:00.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.1-VACE-Plus", + "docUrl": "https://help.aliyun.com/document_detail/2922183.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json new file mode 100644 index 00000000..9b86800d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json @@ -0,0 +1,79 @@ +{ + "name": "WordArt锦书-文字变形", + "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "wordart-semantic", + "prices": [ + { + "priceUnit": "每张", + "price": "0.24", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T08:30:26.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "WordArt锦书-文字变形", + "docUrl": "https://help.aliyun.com/document_detail/2712513.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location --request POST 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json new file mode 100644 index 00000000..a361d130 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json @@ -0,0 +1,80 @@ +{ + "name": "WordArt锦书-文字纹理生成", + "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text", + "Image" + ] + }, + "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", + "features": [], + "provider": "qwen", + "limit": { + "message": "model not exist" + }, + "model": "wordart-texture", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "type": "image_number", + "priceName": "图片生成" + } + ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-04-09T08:29:05.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "WordArt锦书-文字纹理生成", + "docUrl": "https://help.aliyun.com/document_detail/2712510.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json new file mode 100644 index 00000000..0ea1e456 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json @@ -0,0 +1,111 @@ +{ + "name": "MiMo文本模型", + "description": "由小米MiMo提供的MiMo文本模型API服务", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiMo-V2.5-Pro 是小米发布的最新旗舰模型。与前代模型相比，它在通用智能体能力、复杂软件工程以及长程任务等方面都有显著提升，在 ClawEval、GDPVal 和 SWE-bench Pro 等基准测试中均位列前茅。它能够独立且完全自主地完成需要人类专家耗时数天甚至数周的专业任务，涉及上千次工具调用。其高达 100 万 token 的上下文长度，非常适合集成到各种智能体框架中使用。", + "features": [ + "function-calling", + "structured-outputs", + "cache" + ], + "provider": "xiaomi", + "limit": { + "message": "model not exist" + }, + "model": "xiaomi/mimo-v2.5-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 1000000, + "usage_limit_field": "total_tokens", + "count_limit": 100, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-05-18T06:32:14.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "offlineInfo": {}, + "inferenceProvider": "xiaomi", + "name": "xiaomi/mimo-v2.5-pro", + "docUrl": "https://help.aliyun.com/document_detail/3033942.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"xiaomi/mimo-v2.5-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// Initialize the OpenAI client\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // Read from the environment variable\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'xiaomi/mimo-v2.5-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + 'Full response' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3033942.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json new file mode 100644 index 00000000..756bc057 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json @@ -0,0 +1,90 @@ +{ + "name": "Z-Image-Turbo", + "description": "Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text", + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双语文本渲染、复杂语义理解和多样化主题生成上表现卓越。", + "features": [ + "model-experience" + ], + "provider": "qwen-domain-model", + "limit": { + "message": "model not exist" + }, + "model": "z-image-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.1", + "type": "image_standard", + "priceName": "图片生成（标准）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_thinking", + "priceName": "图片生成（思考）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 1, + "count_limit": 2, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-12-18T06:43:39.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Z-Image-Turbo", + "docUrl": "https://help.aliyun.com/document_detail/3002354.html", + "predictConfig": [ + { + "name": "分辨率", + "key": "size", + "default": "1024*1024", + "tip": "请先选择输出分辨率，再选择输出宽高比" + }, + { + "name": "随机种子", + "key": "seed", + "default": 1234, + "range": [ + 1, + 2147483647 + ] + }, + { + "name": "智能改写", + "key": "prompt_extend", + "default": false, + "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + } + ], + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"z-image-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"film grain, analog film texture, soft film lighting, Kodak Portra 400 style, cinematic grainy texture, photorealistic details, subtle noise, (film grain:1.2)。采用近景特写镜头拍摄的东亚年轻女性，呈现户外雪地场景。她体型纤瘦，呈站立姿势，身体微微向右侧倾斜，头部抬起看向画面上方，姿态自然放松。她的面部是典型东亚长相，肤色白皙，脸颊带有自然的红润感，五官清秀：眼睛是深棕色，眼型偏圆，眼神略带惊讶地望向上方，眼白部分可见；眉毛是深黑色，形状自然弯长；鼻子小巧挺直，嘴唇涂有红色口红，唇瓣微张，表情带着轻微的惊讶或好奇。她的头发是深黑色长直发，发丝被风吹得略显凌乱，部分垂在脸颊两侧，头顶佩戴一顶深灰色的头盔，头盔边缘露出少量发丝。服装是蓝白拼接的厚重外套，外套材质看起来是毛绒与布料结合，显得温暖厚实，适合雪地环境。背景是被白雪覆盖的户外场景，远处可见模糊的树木轮廓，天空是明亮的浅蓝色，带有少量白云，光线是强烈的自然日光，照亮人物面部与头发，形成清晰的光影，色调以蓝、白、黑为主，整体风格清新自然。画面顶部有黑色提示框，内有“Press esc to exit full screen”的白色文字。镜头的近景视角放大了人物的表情与细节，营造出户外雪地的真实氛围。\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"prompt_extend\": false,\n \"size\": \"1120*1440\"\n }\n}'" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json new file mode 100644 index 00000000..8868556b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json @@ -0,0 +1,324 @@ +{ + "name": "智谱GLM系列文本模型", + "description": "由智谱提供的GLM系列文本模型API服务", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "智谱原厂直供，最新旗舰模型。GLM-5.2 是智谱迄今能力最强的开源模型，支持真正可用的 1M 上下文，并在长程任务中继续保持领先。", + "features": [ + "function-calling", + "structured-outputs", + "prefix-completion", + "cache" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "ZHIPU/GLM-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-06-16T02:16:54.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "offlineInfo": {}, + "inferenceProvider": "zhipu-ai", + "name": "ZHIPU/GLM-5.2", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "top_p", + "key": "top_p", + "default": 0.8, + "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", + "range": [ + 0.0001, + 1 + ] + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3026315.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "GLM-5.1 是智谱最新旗舰模型，代码能力大大增强，长程任务显著提升，能够在单次任务中持续、自主地工作长达 8 小时，完成从规划、执行到迭代优化的完整闭环，交付工程级成果。\n在综合能力与 Coding 能力上，GLM-5.1 整体表现对齐 Claude Opus 4.6，并在长程自主执行、复杂工程优化与真实开发场景中展现出更强的持续工作能力，是构建 Autonomous Agent 与长程 Coding Agent 的理想基座。", + "features": [ + "function-calling", + "structured-outputs", + "cache", + "prefix-completion" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "ZHIPU/GLM-5.1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-05-18T06:33:45.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 204800, + "offlineInfo": {}, + "inferenceProvider": "zhipu-ai", + "name": "ZHIPU/GLM-5.1", + "docUrl": "https://help.aliyun.com/zh/model-studio/glm-zhipu", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3026315.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "智谱新一代旗舰基座,面向AgenticEngineering,实现从代码到工程的范式跃迁,擅长复杂系统工程与长程Agent任务。", + "features": [ + "function-calling", + "cache", + "structured-outputs", + "prefix-completion" + ], + "provider": "zhipu-ai", + "limit": { + "message": "model not exist" + }, + "model": "ZHIPU/GLM-5", + "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 200, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 131072, + "latestOnlineAt": "2026-03-26T10:47:50.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 204800, + "offlineInfo": {}, + "inferenceProvider": "zhipu-ai", + "name": "ZHIPU/GLM-5", + "docUrl": "https://help.aliyun.com/document_detail/3026315.html", + "predictConfig": [ + { + "name": "system", + "key": "systemMessage", + "tip": "系统人设，例如“你是一个AI助手”。" + }, + { + "name": "temperature", + "key": "temperature", + "default": 0.7, + "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", + "range": [ + 0, + 1.9999 + ] + }, + { + "name": "enable_thinking", + "key": "enable_thinking", + "default": true, + "tip": "推理模式" + } + ], + "samples": { + "openai": { + "completionsAPI": { + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "docUrl": "https://help.aliyun.com/document_detail/3026315.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/index.json b/skills/bailian-docs-llm-wiki/models/index.json index 35e5abff..da947f9e 100644 --- a/skills/bailian-docs-llm-wiki/models/index.json +++ b/skills/bailian-docs-llm-wiki/models/index.json @@ -1,19 +1,227 @@ { "updatedAt": "2026-07-15", - "totalFamilies": 11, - "totalModels": 26, + "totalFamilies": 165, + "totalModels": 359, "capabilityDistribution": { - "Reasoning": 6, - "VG": 2, - "TG": 2, - "VU": 1 + "TG": 34, + "IG": 29, + "VG": 24, + "TTS": 16, + "Reasoning": 14, + "ASR": 12, + "VU": 8, + "Realtime-ASR": 7, + "Multimodal-Omni": 5, + "Realtime-Omni": 4, + "Realtime-Audio-Translate": 3, + "ME": 2, + "Realtime-Chatting": 2, + "Realtime-Text-to-Speech": 2, + "TR": 2, + "3D-generation": 1 }, "providerDistribution": { - "qwen": 8, - "happyhorse": 2, - "deepseek": 1 + "qwen": 98, + "qwen-domain-model": 33, + "wan": 13, + "happyhorse": 4, + "mini-max": 3, + "deepseek": 3, + "zhipu-ai": 3, + "pixverse": 3, + "moonshot-ai": 2, + "kling": 1, + "stepfun": 1, + "tripo": 1, + "vidu": 1, + "xiaomi": 1 }, "families": [ + { + "slug": "Kimi-K2", + "name": "Kimi", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "providers": [ + "moonshot-ai" + ], + "itemCount": 5, + "items": [ + "kimi-k2-thinking", + "kimi-k2.5", + "kimi-k2.6", + "kimi-k2.7-code", + "Moonshot-Kimi-K2-Instruct" + ], + "maxContextWindow": 262144 + }, + { + "slug": "MiniMax-M2.1", + "name": "MiniMax", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "mini-max" + ], + "itemCount": 2, + "items": [ + "MiniMax-M2.1", + "MiniMax-M2.5" + ], + "maxContextWindow": 204800 + }, + { + "slug": "MiniMax-speech-market-place", + "name": "MiniMax-Speech系列语音模型", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "mini-max" + ], + "itemCount": 4, + "items": [ + "MiniMax/speech-02-hd", + "MiniMax/speech-02-turbo", + "MiniMax/speech-2.8-hd", + "MiniMax/speech-2.8-turbo" + ] + }, + { + "slug": "aitryon-parsing-v1", + "name": "AI试衣OutfitAnyone-图片分割", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "aitryon-parsing-v1" + ] + }, + { + "slug": "aitryon-plus", + "name": "AI试衣-Plus版", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "aitryon-plus" + ] + }, + { + "slug": "aitryon-refiner", + "name": "AI试衣OutfitAnyone-图片精修", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "aitryon-refiner" + ] + }, + { + "slug": "aitryon", + "name": "AI试衣-基础版", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "aitryon" + ] + }, + { + "slug": "animate-anyone-detect-gen2", + "name": "舞动人像AnimateAnyone-detect", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "animate-anyone-detect-gen2" + ] + }, + { + "slug": "animate-anyone-gen2", + "name": "舞动人像AnimateAnyone", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "animate-anyone-gen2" + ] + }, + { + "slug": "animate-anyone-template-gen2", + "name": "舞动人像AnimateAnyone-template", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "animate-anyone-template-gen2" + ] + }, + { + "slug": "cosyvoice", + "name": "CosyVoice大模型", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen", + "qwen-domain-model" + ], + "itemCount": 7, + "items": [ + "cosyvoice-clone-v1", + "cosyvoice-v1", + "cosyvoice-v2", + "cosyvoice-v3-flash", + "cosyvoice-v3-plus", + "cosyvoice-v3.5-flash", + "cosyvoice-v3.5-plus" + ] + }, { "slug": "deepseek", "name": "DeepSeek", @@ -43,180 +251,2577 @@ "maxContextWindow": 1000000 }, { - "slug": "happyhorse-i2v", - "name": "HappyHorse-I2V", + "slug": "embedding", + "name": "通义多模态向量", + "primaryCapability": "ME", + "capabilities": [ + "ME" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 3, + "items": [ + "multimodal-embedding-v1", + "tongyi-embedding-vision-flash", + "tongyi-embedding-vision-plus" + ] + }, + { + "slug": "emo-detect-v1", + "name": "悦动人像EMO-detect", "primaryCapability": "VG", "capabilities": [ "VG" ], "providers": [ - "happyhorse" + "qwen" ], - "itemCount": 2, + "itemCount": 1, "items": [ - "happyhorse-1.0-i2v", - "happyhorse-1.1-i2v" + "emo-detect-v1" ] }, { - "slug": "happyhorse-t2v", - "name": "HappyHorse-T2V", + "slug": "emo-v1", + "name": "悦动人像EMO", "primaryCapability": "VG", "capabilities": [ "VG" ], "providers": [ - "happyhorse" + "qwen" ], - "itemCount": 2, + "itemCount": 1, "items": [ - "happyhorse-1.0-t2v", - "happyhorse-1.1-t2v" + "emo-v1" ] }, { - "slug": "qwen3.5-flash", - "name": "Qwen3.5-Flash", - "primaryCapability": "Reasoning", + "slug": "emoji-detect-v1", + "name": "表情包Emoji-detect", + "primaryCapability": "VG", "capabilities": [ - "Reasoning", - "TG", - "VU" + "VG" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "qwen3.5-flash" - ], - "maxContextWindow": 1000000 + "emoji-detect-v1" + ] }, { - "slug": "qwen3.5-plus", - "name": "Qwen3.5-Plus", - "primaryCapability": "TG", + "slug": "emoji-v1", + "name": "表情包Emoji", + "primaryCapability": "VG", "capabilities": [ - "TG", - "Reasoning", - "VU" + "VG" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "qwen3.5-plus" - ], - "maxContextWindow": 1000000 + "emoji-v1" + ] }, { - "slug": "qwen3.6-flash", - "name": "Qwen3.6-Flash", - "primaryCapability": "Reasoning", + "slug": "facechain-facedetect", + "name": "FaceChain人物图像检测", + "primaryCapability": "IG", "capabilities": [ - "Reasoning", - "VU", - "TG" + "IG" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "qwen3.6-flash" - ], - "maxContextWindow": 1000000 + "facechain-facedetect" + ] }, { - "slug": "qwen3.6-max", - "name": "Qwen3.6-Max", - "primaryCapability": "Reasoning", + "slug": "facechain-generation", + "name": "FaceChain人物写真生成", + "primaryCapability": "IG", "capabilities": [ - "Reasoning", - "TG" + "IG" ], "providers": [ "qwen" ], "itemCount": 1, "items": [ - "qwen3.6-max-preview" - ], - "maxContextWindow": 262144 + "facechain-generation" + ] }, { - "slug": "qwen3.6-plus", - "name": "Qwen3.6-Plus", - "primaryCapability": "Reasoning", + "slug": "farui-plus", + "name": "通义法睿-Plus-32K", + "primaryCapability": "TG", "capabilities": [ - "Reasoning", - "VU", "TG" ], "providers": [ - "qwen" + "qwen-domain-model" ], "itemCount": 1, "items": [ - "qwen3.6-plus" + "farui-plus" ], - "maxContextWindow": 1000000 + "maxContextWindow": 12000 }, { - "slug": "qwen3.6", - "name": "Qwen3.6开源模型", - "primaryCapability": "VU", + "slug": "fun-asr-flash", + "name": "Fun-ASR-Flash", + "primaryCapability": "ASR", "capabilities": [ - "VU", - "TG", - "Reasoning" + "ASR" ], "providers": [ - "qwen" + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "fun-asr-flash-2026-06-15" + ] + }, + { + "slug": "fun-asr-realtime", + "name": "Fun-ASR实时语音识别", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen-domain-model" ], "itemCount": 2, "items": [ - "qwen3.6-27b", - "qwen3.6-35b-a3b" + "fun-asr-flash-8k-realtime", + "fun-asr-realtime" + ] + }, + { + "slug": "fun-asr", + "name": "Fun-ASR语音识别", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" ], - "maxContextWindow": 262144 + "providers": [ + "qwen-domain-model" + ], + "itemCount": 2, + "items": [ + "fun-asr", + "fun-asr-mtl" + ] }, { - "slug": "qwen3.7-max", - "name": "Qwen3.7-Max", - "primaryCapability": "Reasoning", + "slug": "fun-music", + "name": "音乐生成", + "primaryCapability": "TTS", "capabilities": [ - "Reasoning", - "TG" + "TTS" ], "providers": [ "qwen" ], "itemCount": 2, "items": [ - "qwen3.7-max", - "qwen3.7-max-preview" - ], - "maxContextWindow": 1000000 + "fun-music-preview", + "fun-music-v1" + ] }, { - "slug": "qwen3.7-plus", - "name": "Qwen3.7-Plus", + "slug": "glm-4.5", + "name": "GLM", "primaryCapability": "TG", "capabilities": [ "TG", - "Reasoning", - "VU" + "Reasoning" ], "providers": [ - "qwen" + "zhipu-ai" + ], + "itemCount": 7, + "items": [ + "glm-4.5", + "glm-4.5-air", + "glm-4.6", + "glm-4.7", + "glm-5", + "glm-5.1", + "glm-5.2" + ], + "maxContextWindow": 1048576 + }, + { + "slug": "glm-fast", + "name": "GLM-5.2-Fast", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "zhipu-ai" ], "itemCount": 1, "items": [ - "qwen3.7-plus" + "glm-5.2-fast-preview" ], - "maxContextWindow": 1000000 + "maxContextWindow": 1048576 + }, + { + "slug": "gui-plus", + "name": "GUI-Plus", + "primaryCapability": "VU", + "capabilities": [ + "VU" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "gui-plus" + ], + "maxContextWindow": 256000 + }, + { + "slug": "gummy-chat-v1", + "name": "一句话识别及翻译V1.0", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "gummy-chat-v1" + ] + }, + { + "slug": "gummy-realtime-v1", + "name": "实时语音识别及翻译V1.0", + "primaryCapability": "Realtime-Audio-Translate", + "capabilities": [ + "Realtime-Audio-Translate" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "gummy-realtime-v1" + ] + }, + { + "slug": "happyhorse-i2v", + "name": "HappyHorse-I2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "happyhorse" + ], + "itemCount": 2, + "items": [ + "happyhorse-1.0-i2v", + "happyhorse-1.1-i2v" + ] + }, + { + "slug": "happyhorse-r2v", + "name": "HappyHorse-R2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "happyhorse" + ], + "itemCount": 2, + "items": [ + "happyhorse-1.0-r2v", + "happyhorse-1.1-r2v" + ] + }, + { + "slug": "happyhorse-t2v", + "name": "HappyHorse-T2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "happyhorse" + ], + "itemCount": 2, + "items": [ + "happyhorse-1.0-t2v", + "happyhorse-1.1-t2v" + ] + }, + { + "slug": "happyhorse-video-edit", + "name": "HappyHorse-Video-Edit", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "happyhorse" + ], + "itemCount": 1, + "items": [ + "happyhorse-1.0-video-edit" + ] + }, + { + "slug": "image-erase-completion", + "name": "图像擦除补全", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "image-erase-completion" + ] + }, + { + "slug": "image-instance-segmentation", + "name": "人物实例分割", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "image-instance-segmentation" + ] + }, + { + "slug": "image-out-painting", + "name": "图像画面扩展", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "image-out-painting" + ] + }, + { + "slug": "kimi-models-market-place", + "name": "Kimi", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "providers": [ + "moonshot-ai" + ], + "itemCount": 4, + "items": [ + "kimi/kimi-k2.5", + "kimi/kimi-k2.6", + "kimi/kimi-k2.7-code", + "kimi/kimi-k2.7-code-highspeed" + ], + "maxContextWindow": 262144 + }, + { + "slug": "kling-models-market-place", + "name": "可灵AI", + "primaryCapability": "VG", + "capabilities": [ + "VG", + "IG" + ], + "providers": [ + "kling" + ], + "itemCount": 4, + "items": [ + "kling/kling-v3-image-generation", + "kling/kling-v3-omni-image-generation", + "kling/kling-v3-omni-video-generation", + "kling/kling-v3-video-generation" + ] + }, + { + "slug": "liveportrait-detect", + "name": "灵动人像LivePortrait-detect", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "liveportrait-detect" + ] + }, + { + "slug": "liveportrait", + "name": "灵动人像LivePortrait", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "liveportrait" + ] + }, + { + "slug": "minimax-models-market-place", + "name": "MiniMax文本模型", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "providers": [ + "mini-max" + ], + "itemCount": 4, + "items": [ + "MiniMax/MiniMax-M2.1", + "MiniMax/MiniMax-M2.5", + "MiniMax/MiniMax-M2.7", + "MiniMax/MiniMax-M3" + ], + "maxContextWindow": 1048576 + }, + { + "slug": "paraformer-8k-v1", + "name": "Paraformer语音识别-8k-v1", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-8k-v1" + ] + }, + { + "slug": "paraformer-8k-v2", + "name": "Paraformer语音识别-8k-v2", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-8k-v2" + ] + }, + { + "slug": "paraformer-mtl-v1", + "name": "Paraformer语音识别-mtl-v1", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-mtl-v1" + ] + }, + { + "slug": "paraformer-realtime-8k-v1", + "name": "Paraformer实时语音识别-8k-v1", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-realtime-8k-v1" + ] + }, + { + "slug": "paraformer-realtime-8k-v2", + "name": "Paraformer实时语音识别-8k-v2", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-realtime-8k-v2" + ] + }, + { + "slug": "paraformer-realtime-v1", + "name": "Paraformer实时语音识别-v1", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-realtime-v1" + ] + }, + { + "slug": "paraformer-realtime-v2", + "name": "Paraformer实时语音识别-v2", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-realtime-v2" + ] + }, + { + "slug": "paraformer-v1", + "name": "Paraformer语音识别-v1", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-v1" + ] + }, + { + "slug": "paraformer-v2", + "name": "Paraformer语音识别-v2", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "paraformer-v2" + ] + }, + { + "slug": "pixverse-c1-market-place", + "name": "PixVerse C1", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "pixverse" + ], + "itemCount": 4, + "items": [ + "pixverse/pixverse-c1-it2v", + "pixverse/pixverse-c1-kf2v", + "pixverse/pixverse-c1-r2v", + "pixverse/pixverse-c1-t2v" + ] + }, + { + "slug": "pixverse-market-place", + "name": "PixVerse V5.6", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "pixverse" + ], + "itemCount": 4, + "items": [ + "pixverse/pixverse-v5.6-it2v", + "pixverse/pixverse-v5.6-kf2v", + "pixverse/pixverse-v5.6-r2v", + "pixverse/pixverse-v5.6-t2v" + ] + }, + { + "slug": "pixverse-v6-market-place", + "name": "PixVerse V6", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "pixverse" + ], + "itemCount": 4, + "items": [ + "pixverse/pixverse-v6-it2v", + "pixverse/pixverse-v6-kf2v", + "pixverse/pixverse-v6-r2v", + "pixverse/pixverse-v6-t2v" + ] + }, + { + "slug": "qvq-max", + "name": "QVQ-Max", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qvq-max" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qvq-plus", + "name": "Qwen-QVQ-Plus", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qvq-plus" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qwen-audio-realtime-flash", + "name": "Qwen-Audio-Realtime-Flash", + "primaryCapability": "Realtime-Chatting", + "capabilities": [ + "Realtime-Chatting" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-audio-3.0-realtime-flash" + ], + "maxContextWindow": 8192 + }, + { + "slug": "qwen-audio-realtime-plus", + "name": "Qwen-Audio-Realtime-Plus", + "primaryCapability": "Realtime-Chatting", + "capabilities": [ + "Realtime-Chatting" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-audio-3.0-realtime-plus" + ], + "maxContextWindow": 8192 + }, + { + "slug": "qwen-audio-tts", + "name": "Qwen-Audio-TTS", + "primaryCapability": "Realtime-Text-to-Speech", + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-audio-3.0-tts-flash", + "qwen-audio-3.0-tts-plus" + ] + }, + { + "slug": "qwen-coder-plus", + "name": "Qwen-Coder-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-coder-plus" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qwen-coder-turbo", + "name": "Qwen-Coder-Turbo", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-coder-turbo" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qwen-doc-turbo", + "name": "Qwen-Doc-Turbo", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-doc-turbo" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen-embedding", + "name": "Qwen-Embedding", + "primaryCapability": "TR", + "capabilities": [ + "TR" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 6, + "items": [ + "text-embedding-async-v1", + "text-embedding-async-v2", + "text-embedding-v1", + "text-embedding-v2", + "text-embedding-v3", + "text-embedding-v4" + ] + }, + { + "slug": "qwen-flash-character", + "name": "Qwen-Flash-Character", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-flash-character" + ], + "maxContextWindow": 8192 + }, + { + "slug": "qwen-flash", + "name": "Qwen-Flash", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-flash" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen-image-2.0-pro", + "name": "Qwen-Image-2.0-Pro", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-image-2.0-pro" + ] + }, + { + "slug": "qwen-image-2.0", + "name": "Qwen-Image-2.0", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-image-2.0" + ] + }, + { + "slug": "qwen-image-edit-max", + "name": "Qwen-Image-Edit-Max", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-image-edit-max" + ] + }, + { + "slug": "qwen-image-edit", + "name": "Qwen-Image-Edit-Plus", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-image-edit", + "qwen-image-edit-plus" + ] + }, + { + "slug": "qwen-image-plus", + "name": "Qwen-Image-Plus", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-image", + "qwen-image-plus" + ] + }, + { + "slug": "qwen-long", + "name": "Qwen-Long", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 2, + "items": [ + "qwen-long", + "qwen-long-latest" + ], + "maxContextWindow": 10000000 + }, + { + "slug": "qwen-math-plus", + "name": "Qwen-Math-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 4, + "items": [ + "qwen-math-plus", + "qwen-math-plus-0816", + "qwen-math-plus-0919", + "qwen-math-plus-latest" + ], + "maxContextWindow": 4096 + }, + { + "slug": "qwen-math-turbo", + "name": "Qwen-Math-Turbo", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-math-turbo" + ], + "maxContextWindow": 4096 + }, + { + "slug": "qwen-max", + "name": "Qwen-Max", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-max" + ], + "maxContextWindow": 32768 + }, + { + "slug": "qwen-mt-flash", + "name": "Qwen-MT-Flash", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-mt-flash" + ], + "maxContextWindow": 16384 + }, + { + "slug": "qwen-mt-image", + "name": "Qwen-MT-Image", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-mt-image" + ] + }, + { + "slug": "qwen-mt-lite", + "name": "Qwen-MT-Lite", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-mt-lite" + ], + "maxContextWindow": 16384 + }, + { + "slug": "qwen-mt-plus", + "name": "Qwen-MT-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-mt-plus" + ], + "maxContextWindow": 16384 + }, + { + "slug": "qwen-mt-turbo", + "name": "Qwen-MT-Turbo", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-mt-turbo" + ], + "maxContextWindow": 16384 + }, + { + "slug": "qwen-omni-turbo-realtime", + "name": "Qwen-Omni-Turbo-Realtime", + "primaryCapability": "Realtime-Omni", + "capabilities": [ + "Realtime-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-omni-turbo-realtime", + "qwen-omni-turbo-realtime-latest" + ], + "maxContextWindow": 32768 + }, + { + "slug": "qwen-omni-turbo", + "name": "Qwen-Omni-Turbo", + "primaryCapability": "Multimodal-Omni", + "capabilities": [ + "Multimodal-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-omni-turbo", + "qwen-omni-turbo-latest" + ], + "maxContextWindow": 32768 + }, + { + "slug": "qwen-plus-character", + "name": "Qwen-Plus-Character", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen-plus-character" + ], + "maxContextWindow": 32768 + }, + { + "slug": "qwen-plus", + "name": "Qwen-Plus", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 4, + "items": [ + "qwen-plus", + "qwen-plus-0112", + "qwen-plus-1220", + "qwen-plus-latest" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen-rerank", + "name": "Qwen-Rerank", + "primaryCapability": "TR", + "capabilities": [ + "TR" + ], + "providers": [ + "qwen", + "qwen-domain-model" + ], + "itemCount": 3, + "items": [ + "gte-rerank-v2", + "qwen3-rerank", + "qwen3-vl-rerank" + ], + "maxContextWindow": 30000 + }, + { + "slug": "qwen-tts-realtime", + "name": "Qwen-TTS-Realtime", + "primaryCapability": "Realtime-Text-to-Speech", + "capabilities": [ + "Realtime-Text-to-Speech" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-tts-realtime", + "qwen-tts-realtime-latest" + ], + "maxContextWindow": 8192 + }, + { + "slug": "qwen-tts", + "name": "Qwen-TTS", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen-tts", + "qwen-tts-latest" + ], + "maxContextWindow": 8192 + }, + { + "slug": "qwen-turbo", + "name": "Qwen-Turbo", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-turbo" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qwen-vl-embedding", + "name": "Qwen-VL-Embedding", + "primaryCapability": "ME", + "capabilities": [ + "ME" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen2.5-vl-embedding", + "qwen3-vl-embedding" + ] + }, + { + "slug": "qwen-vl-max", + "name": "Qwen-VL-Max", + "primaryCapability": "VU", + "capabilities": [ + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-vl-max" + ], + "maxContextWindow": 131072 + }, + { + "slug": "qwen-vl-ocr", + "name": "Qwen-VL-OCR", + "primaryCapability": "VU", + "capabilities": [ + "VU" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 3, + "items": [ + "qwen-vl-ocr", + "qwen-vl-ocr-1028", + "qwen-vl-ocr-latest" + ], + "maxContextWindow": 38192 + }, + { + "slug": "qwen-voice-design", + "name": "Qwen-声音设计", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-voice-design" + ] + }, + { + "slug": "qwen-voice-enrollment", + "name": "Qwen-声音复刻", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-voice-enrollment" + ] + }, + { + "slug": "qwen2.5", + "name": "Qwen2.5-开源模型", + "primaryCapability": "Multimodal-Omni", + "capabilities": [ + "Multimodal-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen2.5-omni-7b" + ], + "maxContextWindow": 32768 + }, + { + "slug": "qwen3-asr-flash-filetrans", + "name": "Qwen3-ASR-Flash-Filetrans", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-asr-flash-filetrans" + ] + }, + { + "slug": "qwen3-asr-flash-realtime", + "name": "Qwen3-ASR-Flash-Realtime", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-asr-flash-realtime" + ] + }, + { + "slug": "qwen3-asr-flash", + "name": "Qwen3-ASR-Flash", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-asr-flash" + ] + }, + { + "slug": "qwen3-coder-30b-a3b-instruct", + "name": "Qwen3-Coder-30B-A3B-Instruct", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-coder-30b-a3b-instruct" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3-coder-480b-a35b-instruct", + "name": "Qwen3-Coder-480B-A35B-Instruct", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-coder-480b-a35b-instruct" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3-coder-flash", + "name": "Qwen3-Coder-Flash", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-coder-flash" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3-coder-plus", + "name": "Qwen3-Coder-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-coder-plus" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3-livetranslate-flash-realtime", + "name": "Qwen3-LiveTranslate-Flash-Realtime", + "primaryCapability": "Realtime-Audio-Translate", + "capabilities": [ + "Realtime-Audio-Translate" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-livetranslate-flash-realtime" + ], + "maxContextWindow": 53248 + }, + { + "slug": "qwen3-livetranslate-flash", + "name": "Qwen3-LiveTranslate-Flash", + "primaryCapability": "Realtime-ASR", + "capabilities": [ + "Realtime-ASR" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3-livetranslate-flash" + ], + "maxContextWindow": 53248 + }, + { + "slug": "qwen3-max", + "name": "Qwen3-Max", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen3-max", + "qwen3-max-preview" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3-omni-30b-a3b-captioner", + "name": "Qwen3-Omni-30b-a3b-Captioner", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-omni-30b-a3b-captioner" + ], + "maxContextWindow": 65536 + }, + { + "slug": "qwen3-omni-flash-realtime", + "name": "Qwen3-Omni-Flash-Realtime", + "primaryCapability": "Realtime-Omni", + "capabilities": [ + "Realtime-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-omni-flash-realtime" + ], + "maxContextWindow": 65536 + }, + { + "slug": "qwen3-omni-flash", + "name": "Qwen3-Omni-Flash", + "primaryCapability": "Multimodal-Omni", + "capabilities": [ + "Multimodal-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-omni-flash" + ], + "maxContextWindow": 65536 + }, + { + "slug": "qwen3-tts-flash-realtime", + "name": "Qwen3-TTS-Flash-Realtime", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-flash-realtime" + ] + }, + { + "slug": "qwen3-tts-flash", + "name": "Qwen3-TTS-Flash", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-flash" + ] + }, + { + "slug": "qwen3-tts-instruct-flash-realtime", + "name": "qwen3-tts-instruct-flash-realtime", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-instruct-flash-realtime" + ] + }, + { + "slug": "qwen3-tts-instruct-flash", + "name": "Qwen3-TTS-Instruct-Flash", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-instruct-flash" + ] + }, + { + "slug": "qwen3-tts-vc-realtime", + "name": "Qwen3-TTS-VC-Realtime", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-vc-realtime-2026-01-15" + ] + }, + { + "slug": "qwen3-tts-vc", + "name": "Qwen3-TTS-VC", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-vc-2026-01-22" + ] + }, + { + "slug": "qwen3-tts-vd-realtime", + "name": "Qwen3-TTS-VD-Realtime", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-vd-realtime-2026-01-15" + ] + }, + { + "slug": "qwen3-tts-vd", + "name": "Qwen3-TTS-VD", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-tts-vd-2026-01-26" + ] + }, + { + "slug": "qwen3-vl-flash", + "name": "Qwen3-VL-Flash", + "primaryCapability": "VU", + "capabilities": [ + "VU", + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-vl-flash" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3-vl-plus", + "name": "Qwen3-VL-Plus", + "primaryCapability": "VU", + "capabilities": [ + "VU", + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3-vl-plus" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.5-flash", + "name": "Qwen3.5-Flash", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG", + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-flash" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3.5-livetranslate-flash-realtime", + "name": "Qwen3.5-LiveTranslate-Flash-Realtime", + "primaryCapability": "Realtime-Audio-Translate", + "capabilities": [ + "Realtime-Audio-Translate" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "qwen3.5-livetranslate-flash-realtime" + ], + "maxContextWindow": 53248 + }, + { + "slug": "qwen3.5-ocr", + "name": "Qwen3.5-OCR", + "primaryCapability": "VU", + "capabilities": [ + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-ocr" + ], + "maxContextWindow": 65536 + }, + { + "slug": "qwen3.5-omni-flash-realtime", + "name": "Qwen3.5-Omni-Flash-Realtime", + "primaryCapability": "Realtime-Omni", + "capabilities": [ + "Realtime-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-omni-flash-realtime" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.5-omni-flash", + "name": "Qwen3.5-Omni-Flash", + "primaryCapability": "Multimodal-Omni", + "capabilities": [ + "Multimodal-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-omni-flash" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.5-omni-plus-realtime", + "name": "Qwen3.5-Omni-Plus-Realtime", + "primaryCapability": "Realtime-Omni", + "capabilities": [ + "Realtime-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-omni-plus-realtime" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.5-omni-plus", + "name": "Qwen3.5-Omni-Plus", + "primaryCapability": "Multimodal-Omni", + "capabilities": [ + "Multimodal-Omni" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-omni-plus" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.5-plus", + "name": "Qwen3.5-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.5-plus" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3.5", + "name": "Qwen3.5开源模型", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 4, + "items": [ + "qwen3.5-122b-a10b", + "qwen3.5-27b", + "qwen3.5-35b-a3b", + "qwen3.5-397b-a17b" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.6-flash", + "name": "Qwen3.6-Flash", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.6-flash" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3.6-max", + "name": "Qwen3.6-Max", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.6-max-preview" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.6-plus", + "name": "Qwen3.6-Plus", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "VU", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.6-plus" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3.6", + "name": "Qwen3.6开源模型", + "primaryCapability": "VU", + "capabilities": [ + "VU", + "TG", + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen3.6-27b", + "qwen3.6-35b-a3b" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwen3.7-max", + "name": "Qwen3.7-Max", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 2, + "items": [ + "qwen3.7-max", + "qwen3.7-max-preview" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3.7-plus", + "name": "Qwen3.7-Plus", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen3.7-plus" + ], + "maxContextWindow": 1000000 + }, + { + "slug": "qwen3", + "name": "Qwen3开源模型", + "primaryCapability": "VU", + "capabilities": [ + "VU", + "Reasoning", + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 20, + "items": [ + "qwen3-14b", + "qwen3-235b-a22b", + "qwen3-235b-a22b-instruct-2507", + "qwen3-235b-a22b-thinking-2507", + "qwen3-30b-a3b", + "qwen3-30b-a3b-instruct-2507", + "qwen3-30b-a3b-thinking-2507", + "qwen3-32b", + "qwen3-8b", + "qwen3-coder-next", + "qwen3-next-80b-a3b-instruct", + "qwen3-next-80b-a3b-thinking", + "qwen3-vl-235b-a22b-instruct", + "qwen3-vl-235b-a22b-thinking", + "qwen3-vl-30b-a3b-instruct", + "qwen3-vl-30b-a3b-thinking", + "qwen3-vl-32b-instruct", + "qwen3-vl-32b-thinking", + "qwen3-vl-8b-instruct", + "qwen3-vl-8b-thinking" + ], + "maxContextWindow": 262144 + }, + { + "slug": "qwq-plus", + "name": "Qwen-QwQ-Plus", + "primaryCapability": "Reasoning", + "capabilities": [ + "Reasoning" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwq-plus" + ], + "maxContextWindow": 131072 + }, + { + "slug": "sambert", + "name": "Sambert语音合成", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 43, + "items": [ + "sambert-beth-v1", + "sambert-betty-v1", + "sambert-brian-v1", + "sambert-cally-v1", + "sambert-camila-v1", + "sambert-cindy-v1", + "sambert-clara-v1", + "sambert-donna-v1", + "sambert-eva-v1", + "sambert-hanna-v1", + "sambert-indah-v1", + "sambert-perla-v1", + "sambert-waan-v1", + "sambert-zhichu-v1", + "sambert-zhida-v1", + "sambert-zhide-v1", + "sambert-zhifei-v1", + "sambert-zhigui-v1", + "sambert-zhihao-v1", + "sambert-zhijia-v1", + "sambert-zhijing-v1", + "sambert-zhilun-v1", + "sambert-zhimao-v1", + "sambert-zhimiao-emo-v1", + "sambert-zhiming-v1", + "sambert-zhimo-v1", + "sambert-zhina-v1", + "sambert-zhinan-v1", + "sambert-zhiqi-v1", + "sambert-zhiqian-v1", + "sambert-zhiru-v1", + "sambert-zhishu-v1", + "sambert-zhishuo-v1", + "sambert-zhistella-v1", + "sambert-zhiting-v1", + "sambert-zhiwei-v1", + "sambert-zhixiang-v1", + "sambert-zhixiao-v1", + "sambert-zhiya-v1", + "sambert-zhiye-v1", + "sambert-zhiying-v1", + "sambert-zhiyuan-v1", + "sambert-zhiyue-v1" + ] + }, + { + "slug": "shoemodel-v1", + "name": "鞋靴模特", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "shoemodel-v1" + ] + }, + { + "slug": "siliconflow-models", + "name": "SiliconFlow DeepSeek", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning" + ], + "providers": [ + "deepseek" + ], + "itemCount": 4, + "items": [ + "siliconflow/deepseek-r1-0528", + "siliconflow/deepseek-v3-0324", + "siliconflow/deepseek-v3.1-terminus", + "siliconflow/deepseek-v3.2" + ], + "maxContextWindow": 163840 + }, + { + "slug": "speech-biasing", + "name": "语音识别热词", + "primaryCapability": "ASR", + "capabilities": [ + "ASR" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "speech-biasing" + ] + }, + { + "slug": "stepfun-models-market-place", + "name": "StepFun推理模型", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "VU" + ], + "providers": [ + "stepfun" + ], + "itemCount": 1, + "items": [ + "stepfun/step-3.7-flash" + ], + "maxContextWindow": 262144 + }, + { + "slug": "tongyi-intent-detect-v3", + "name": "意图分类模型", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "tongyi-intent-detect-v3" + ], + "maxContextWindow": 8192 + }, + { + "slug": "tongyi-xiaomi-analysis-flash", + "name": "通义晓蜜-对话分析-flash", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "tongyi-xiaomi-analysis-flash" + ], + "maxContextWindow": 32768 + }, + { + "slug": "tongyi-xiaomi-analysis-pro", + "name": "通义晓蜜-对话分析-pro", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "tongyi-xiaomi-analysis-pro" + ], + "maxContextWindow": 32768 + }, + { + "slug": "tripo-models-market-place", + "name": "Tripo", + "primaryCapability": "3D-generation", + "capabilities": [ + "3D-generation" + ], + "providers": [ + "tripo" + ], + "itemCount": 2, + "items": [ + "Tripo/Tripo-H3.1", + "Tripo/Tripo-P1.0" + ] + }, + { + "slug": "vanchin-models-market-place", + "name": "Vanchin DeepSeek", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning", + "VU" + ], + "providers": [ + "deepseek" + ], + "itemCount": 6, + "items": [ + "vanchin/deepseek-ocr", + "vanchin/deepseek-r1", + "vanchin/deepseek-v3", + "vanchin/deepseek-v3.1-terminus", + "vanchin/deepseek-v3.2-think", + "vanchin/deepseek-v4-pro" + ], + "maxContextWindow": 1048576 + }, + { + "slug": "video-style-transform", + "name": "视频风格重绘", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "video-style-transform" + ] + }, + { + "slug": "videoretalk", + "name": "声动人像VideoRetalk", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "videoretalk" + ] + }, + { + "slug": "vidu-image-models-market-place", + "name": "Vidu AI生图", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "vidu" + ], + "itemCount": 4, + "items": [ + "vidu/vidu-image_reference2image", + "vidu/viduq2-fast_reference2image", + "vidu/viduq2-pro_reference2image", + "vidu/viduq3-fast_reference2image" + ] + }, + { + "slug": "virtualmodel-v2", + "name": "虚拟模特V2", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "virtualmodel-v2" + ] + }, + { + "slug": "voice-enrollment", + "name": "大模型声音复刻及声音设计", + "primaryCapability": "TTS", + "capabilities": [ + "TTS" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "voice-enrollment" + ] + }, + { + "slug": "wan-image-edit", + "name": "Wan-Image", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 5, + "items": [ + "wan2.5-i2i-preview", + "wan2.6-image", + "wan2.7-image", + "wan2.7-image-pro", + "wanx2.1-imageedit" + ] + }, + { + "slug": "wan-image-to-video", + "name": "Wan-I2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "wan" + ], + "itemCount": 14, + "items": [ + "wan2.2-animate-mix", + "wan2.2-animate-move", + "wan2.2-i2v-flash", + "wan2.2-i2v-plus", + "wan2.2-kf2v-flash", + "wan2.2-s2v", + "wan2.2-s2v-detect", + "wan2.5-i2v-preview", + "wan2.6-i2v", + "wan2.6-i2v-flash", + "wan2.7-i2v", + "wanx2.1-i2v-plus", + "wanx2.1-i2v-turbo", + "wanx2.1-kf2v-plus" + ] + }, + { + "slug": "wan-reference-to-video", + "name": "Wan-R2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "wan" + ], + "itemCount": 3, + "items": [ + "wan2.6-r2v", + "wan2.6-r2v-flash", + "wan2.7-r2v" + ] + }, + { + "slug": "wan-text-to-image", + "name": "Wan-T2I", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 9, + "items": [ + "wan2.2-t2i-flash", + "wan2.2-t2i-plus", + "wan2.5-t2i-preview", + "wan2.6-t2i", + "wanx-v1", + "wanx-v1-0521", + "wanx2.0-t2i-turbo", + "wanx2.1-t2i-plus", + "wanx2.1-t2i-turbo" + ] + }, + { + "slug": "wan-text-to-video", + "name": "Wan-T2V", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "wan" + ], + "itemCount": 6, + "items": [ + "wan2.2-t2v-plus", + "wan2.5-t2v-preview", + "wan2.6-t2v", + "wan2.7-t2v", + "wanx2.1-t2v-plus", + "wanx2.1-t2v-turbo" + ] + }, + { + "slug": "wan-video-edit", + "name": "Wan-VideoEdit", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wan2.7-videoedit" + ] + }, + { + "slug": "wanx-background-generation-v2", + "name": "图像背景生成", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-background-generation-v2" + ] + }, + { + "slug": "wanx-poster-generation-v1", + "name": "创意海报生成", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-poster-generation-v1" + ] + }, + { + "slug": "wanx-sketch-to-image-lite", + "name": "万相-涂鸦作画", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-sketch-to-image-lite" + ] + }, + { + "slug": "wanx-style-repaint-v1", + "name": "人像风格重绘", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-style-repaint-v1" + ] + }, + { + "slug": "wanx-virtualmodel", + "name": "虚拟模特", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-virtualmodel" + ] + }, + { + "slug": "wanx-x-painting", + "name": "万相-图像局部重绘", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx-x-painting" + ] + }, + { + "slug": "wanx2.1-vace-plus", + "name": "Wan2.1-VACE-Plus", + "primaryCapability": "VG", + "capabilities": [ + "VG" + ], + "providers": [ + "wan" + ], + "itemCount": 1, + "items": [ + "wanx2.1-vace-plus" + ] + }, + { + "slug": "wordart-semantic", + "name": "WordArt锦书-文字变形", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "wordart-semantic" + ] + }, + { + "slug": "wordart-texture", + "name": "WordArt锦书-文字纹理生成", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "wordart-texture" + ] + }, + { + "slug": "xiaomi-models-market-place", + "name": "MiMo文本模型", + "primaryCapability": "TG", + "capabilities": [ + "TG" + ], + "providers": [ + "xiaomi" + ], + "itemCount": 1, + "items": [ + "xiaomi/mimo-v2.5-pro" + ], + "maxContextWindow": 1048576 + }, + { + "slug": "z-image-turbo", + "name": "Z-Image-Turbo", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen-domain-model" + ], + "itemCount": 1, + "items": [ + "z-image-turbo" + ] + }, + { + "slug": "zhipu-models-market-place", + "name": "智谱GLM系列文本模型", + "primaryCapability": "TG", + "capabilities": [ + "TG", + "Reasoning" + ], + "providers": [ + "zhipu-ai" + ], + "itemCount": 3, + "items": [ + "ZHIPU/GLM-5", + "ZHIPU/GLM-5.1", + "ZHIPU/GLM-5.2" + ], + "maxContextWindow": 1048576 } ] } diff --git a/skills/bailian-docs-llm-wiki/models/index.md b/skills/bailian-docs-llm-wiki/models/index.md index 6ed87f9e..df240400 100644 --- a/skills/bailian-docs-llm-wiki/models/index.md +++ b/skills/bailian-docs-llm-wiki/models/index.md @@ -1,6 +1,6 @@ # 百炼模型市场索引 -> 自动生成 · 共 11 个模型家族 · 26 个主干模型 · 更新于 2026-07-15 +> 自动生成 · 共 165 个模型家族 · 359 个主干模型 · 更新于 2026-07-15 **机器查询走结构化文件**： @@ -11,12 +11,246 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[].slug`。 -## 推理 `Reasoning` — 6 个家族 +## 文本生成 `TG` — 34 个家族 + +- [GLM](groups/glm-4.5.json) — GLM是由智谱提供的开源模型。 + - 模型：`glm-4.5`, `glm-4.5-air`, `glm-4.6`, `glm-4.7`, `glm-5`, `glm-5.1`, `glm-5.2` +- [GLM-5.2-Fast](groups/glm-fast.json) — GLM-5.2-Fast-Preview 是智谱 AI 旗舰模型 GLM-5.2 的高速版本，支持 1M 超长上下文，模型能力对齐 GLM-5.2 标准版，具备逻辑推理、长文本理解与代码生成能力。通过… + - 模型：`glm-5.2-fast-preview` +- [Kimi](groups/Kimi-K2.json) — Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。 + - 模型：`kimi-k2-thinking`, `kimi-k2.5`, `kimi-k2.6`, `kimi-k2.7-code`, `Moonshot-Kimi-K2-Instruct` +- [Kimi](groups/kimi-models-market-place.json) — 由月之暗面提供的Kimi系列模型的API服务。 + - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed` +- [MiMo文本模型](groups/xiaomi-models-market-place.json) — 由小米MiMo提供的MiMo文本模型API服务 + - 模型：`xiaomi/mimo-v2.5-pro` +- [MiniMax文本模型](groups/minimax-models-market-place.json) — 由MiniMax提供的MiniMax-M系列文本模型API服务。 + - 模型：`MiniMax/MiniMax-M2.1`, `MiniMax/MiniMax-M2.5`, `MiniMax/MiniMax-M2.7`, `MiniMax/MiniMax-M3` +- [Qwen-Coder-Plus](groups/qwen-coder-plus.json) — 千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。 + - 模型：`qwen-coder-plus` +- [Qwen-Coder-Turbo](groups/qwen-coder-turbo.json) — Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。 + - 模型：`qwen-coder-turbo` +- [Qwen-Doc-Turbo](groups/qwen-doc-turbo.json) — 快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。 + - 模型：`qwen-doc-turbo` +- [Qwen-Flash-Character](groups/qwen-flash-character.json) — 千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 + - 模型：`qwen-flash-character` +- [Qwen-Long](groups/qwen-long.json) — Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服… + - 模型：`qwen-long`, `qwen-long-latest` +- [Qwen-Math-Plus](groups/qwen-math-plus.json) — Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。 + - 模型：`qwen-math-plus`, `qwen-math-plus-0816`, `qwen-math-plus-0919`, `qwen-math-plus-latest` +- [Qwen-Math-Turbo](groups/qwen-math-turbo.json) — Qwen-Math-Turbo模型是专门用于数学解题的语言模型，推理速度快，成本低。 + - 模型：`qwen-math-turbo` +- [Qwen-Max](groups/qwen-max.json) — 千问2.5系列千亿级别超大规模语言模型，支持中文、英文等不同语言输入。随着模型的升级，qwen-max将滚动更新升级。如果希望使用固定版本，请使用历史快照版本。 + - 模型：`qwen-max` +- [Qwen-MT-Flash](groups/qwen-mt-flash.json) — 基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + - 模型：`qwen-mt-flash` +- [Qwen-MT-Lite](groups/qwen-mt-lite.json) — 基于Qwen3全面升级的基础级文本翻译大模型，支持32个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + - 模型：`qwen-mt-lite` +- [Qwen-MT-Plus](groups/qwen-mt-plus.json) — 基于Qwen3全面升级的旗舰级翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，并提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + - 模型：`qwen-mt-plus` +- [Qwen-MT-Turbo](groups/qwen-mt-turbo.json) — 基于Qwen3全面升级的轻量级文本翻译大模型，支持92个语种互译，模型性能和翻译效果全面升级，提供更稳定的术语定制、格式还原度、领域提示能力，让译文更精准、自然。 + - 模型：`qwen-mt-turbo` +- [Qwen-Plus-Character](groups/qwen-plus-character.json) — 千问系列角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。 + - 模型：`qwen-plus-character` +- [Qwen3-Coder-30B-A3B-Instruct](groups/qwen3-coder-30b-a3b-instruct.json) — 基于Qwen3的代码生成模型，继承Qwen3-Coder-480B-A35B-Instruct的coding agent能力，代码能力达到同尺寸规模模型SOTA。 + - 模型：`qwen3-coder-30b-a3b-instruct` +- [Qwen3-Coder-480B-A35B-Instruct](groups/qwen3-coder-480b-a35b-instruct.json) — 基于Qwen3的代码生成模型，具有强大的Coding Agent能力，代码能力达到开源模型 SOTA。 + - 模型：`qwen3-coder-480b-a35b-instruct` +- [Qwen3-Coder-Flash](groups/qwen3-coder-flash.json) — 基于Qwen3的代码生成模型，继承Qwen3-Coder-Plus的coding agent能力，支持多轮工具交互，重点优化仓库级别理解能力并增加工具调用稳定性。 + - 模型：`qwen3-coder-flash` +- [Qwen3-Coder-Plus](groups/qwen3-coder-plus.json) — 基于Qwen3的代码生成模型，具有强大的Coding Agent能力，擅长工具调用和环境交互，能够实现自主编程、代码能力卓越的同时兼具通用能力。 + - 模型：`qwen3-coder-plus` +- [Qwen3-Max](groups/qwen3-max.json) — 千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。 + - 模型：`qwen3-max`, `qwen3-max-preview` +- [Qwen3.5-Plus](groups/qwen3.5-plus.json) — Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 + - 模型：`qwen3.5-plus` +- [Qwen3.7-Plus](groups/qwen3.7-plus.json) — Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真… + - 模型：`qwen3.7-plus` +- [SiliconFlow DeepSeek](groups/siliconflow-models.json) — 由硅基流动提供的DeepSeek系列模型API服务。 + - 模型：`siliconflow/deepseek-r1-0528`, `siliconflow/deepseek-v3-0324`, `siliconflow/deepseek-v3.1-terminus`, `siliconflow/deepseek-v3.2` +- [StepFun推理模型](groups/stepfun-models-market-place.json) — 由阶跃星辰StepFun提供的Step系列推理模型API服务 + - 模型：`stepfun/step-3.7-flash` +- [Vanchin DeepSeek](groups/vanchin-models-market-place.json) — 由快手万擎提供的DeepSeek系列模型API服务。 + - 模型：`vanchin/deepseek-ocr`, `vanchin/deepseek-r1`, `vanchin/deepseek-v3`, `vanchin/deepseek-v3.1-terminus`, `vanchin/deepseek-v3.2-think`, `vanchin/deepseek-v4-pro` +- [意图分类模型](groups/tongyi-intent-detect-v3.json) — 意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果… + - 模型：`tongyi-intent-detect-v3` +- [智谱GLM系列文本模型](groups/zhipu-models-market-place.json) — 由智谱提供的GLM系列文本模型API服务 + - 模型：`ZHIPU/GLM-5`, `ZHIPU/GLM-5.1`, `ZHIPU/GLM-5.2` +- [通义晓蜜-对话分析-flash](groups/tongyi-xiaomi-analysis-flash.json) — 通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。 + - 模型：`tongyi-xiaomi-analysis-flash` +- [通义晓蜜-对话分析-pro](groups/tongyi-xiaomi-analysis-pro.json) — 通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。 + - 模型：`tongyi-xiaomi-analysis-pro` +- [通义法睿-Plus-32K](groups/farui-plus.json) — 通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分… + - 模型：`farui-plus` + +## 图像生成 `IG` — 29 个家族 + +- [AI试衣-Plus版](groups/aitryon-plus.json) — aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服… + - 模型：`aitryon-plus` +- [AI试衣-基础版](groups/aitryon.json) — aitryon是一款性能出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。aitryon模型可在较短时间内生成试衣图片，适用于对时效性要求较高的场景。 + - 模型：`aitryon` +- [AI试衣OutfitAnyone-图片分割](groups/aitryon-parsing-v1.json) — 图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。 + - 模型：`aitryon-parsing-v1` +- [AI试衣OutfitAnyone-图片精修](groups/aitryon-refiner.json) — 图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。 + - 模型：`aitryon-refiner` +- [FaceChain人物写真生成](groups/facechain-generation.json) — 基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。 + - 模型：`facechain-generation` +- [FaceChain人物图像检测](groups/facechain-facedetect.json) — 对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。 + - 模型：`facechain-facedetect` +- [Qwen-Image-2.0](groups/qwen-image-2.0.json) — Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模… + - 模型：`qwen-image-2.0` +- [Qwen-Image-2.0-Pro](groups/qwen-image-2.0-pro.json) — Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系… + - 模型：`qwen-image-2.0-pro` +- [Qwen-Image-Edit-Max](groups/qwen-image-edit-max.json) — 千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。 + - 模型：`qwen-image-edit-max` +- [Qwen-Image-Edit-Plus](groups/qwen-image-edit.json) — 千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 + - 模型：`qwen-image-edit`, `qwen-image-edit-plus` +- [Qwen-Image-Plus](groups/qwen-image-plus.json) — 千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。 + - 模型：`qwen-image`, `qwen-image-plus` +- [Qwen-MT-Image](groups/qwen-mt-image.json) — 专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。 + - 模型：`qwen-mt-image` +- [Vidu AI生图](groups/vidu-image-models-market-place.json) — 由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。 + - 模型：`vidu/vidu-image_reference2image`, `vidu/viduq2-fast_reference2image`, `vidu/viduq2-pro_reference2image`, `vidu/viduq3-fast_reference2image` +- [Wan-Image](groups/wan-image-edit.json) — 指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等 + - 模型：`wan2.5-i2i-preview`, `wan2.6-image`, `wan2.7-image`, `wan2.7-image-pro`, `wanx2.1-imageedit` +- [Wan-T2I](groups/wan-text-to-image.json) — 文字生成图片，写实质感细腻画面，文字内容生成，艺术风格表现 + - 模型：`wan2.2-t2i-flash`, `wan2.2-t2i-plus`, `wan2.5-t2i-preview`, `wan2.6-t2i`, `wanx-v1`, `wanx-v1-0521`, `wanx2.0-t2i-turbo`, `wanx2.1-t2i-plus`, `wanx2.1-t2i-turbo` +- [WordArt锦书-文字变形](groups/wordart-semantic.json) — WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。 + - 模型：`wordart-semantic` +- [WordArt锦书-文字纹理生成](groups/wordart-texture.json) — WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海… + - 模型：`wordart-texture` +- [Z-Image-Turbo](groups/z-image-turbo.json) — Z-Image-Turbo是在Artificial Analysis评测中荣登文生图开源模型世界第一的高效图像生成模型,仅用60亿参数和8步推理就能生成媲美大规模商业模型的照片级真实感图像,并在中英双… + - 模型：`z-image-turbo` +- [万相-图像局部重绘](groups/wanx-x-painting.json) — 万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局… + - 模型：`wanx-x-painting` +- [万相-涂鸦作画](groups/wanx-sketch-to-image-lite.json) — 万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、… + - 模型：`wanx-sketch-to-image-lite` +- [人像风格重绘](groups/wanx-style-repaint-v1.json) — 人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。 + - 模型：`wanx-style-repaint-v1` +- [人物实例分割](groups/image-instance-segmentation.json) — 人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。 + - 模型：`image-instance-segmentation` +- [创意海报生成](groups/wanx-poster-generation-v1.json) — 创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。 + - 模型：`wanx-poster-generation-v1` +- [图像擦除补全](groups/image-erase-completion.json) — 图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计… + - 模型：`image-erase-completion` +- [图像画面扩展](groups/image-out-painting.json) — 图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意… + - 模型：`image-out-painting` +- [图像背景生成](groups/wanx-background-generation-v2.json) — 图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。 + - 模型：`wanx-background-generation-v2` +- [虚拟模特](groups/wanx-virtualmodel.json) — 虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如… + - 模型：`wanx-virtualmodel` +- [虚拟模特V2](groups/virtualmodel-v2.json) — 虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如… + - 模型：`virtualmodel-v2` +- [鞋靴模特](groups/shoemodel-v1.json) — 鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新… + - 模型：`shoemodel-v1` + +## 视频生成 `VG` — 24 个家族 + +- [HappyHorse-I2V](groups/happyhorse-i2v.json) — HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。 + - 模型：`happyhorse-1.0-i2v`, `happyhorse-1.1-i2v` +- [HappyHorse-R2V](groups/happyhorse-r2v.json) — HappyHorse-R2V支持参考生视频，更加稳定的主体与场景参考，支持最多9张图片参考，能够精准保持创作意图，实现更强表现能力。 + - 模型：`happyhorse-1.0-r2v`, `happyhorse-1.1-r2v` +- [HappyHorse-T2V](groups/happyhorse-t2v.json) — HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 + - 模型：`happyhorse-1.0-t2v`, `happyhorse-1.1-t2v` +- [HappyHorse-Video-Edit](groups/happyhorse-video-edit.json) — HappyHorse-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。 + - 模型：`happyhorse-1.0-video-edit` +- [PixVerse C1](groups/pixverse-c1-market-place.json) — 由爱诗科技提供的PixVerse C系列视频大模型API服务。 + - 模型：`pixverse/pixverse-c1-it2v`, `pixverse/pixverse-c1-kf2v`, `pixverse/pixverse-c1-r2v`, `pixverse/pixverse-c1-t2v` +- [PixVerse V5.6](groups/pixverse-market-place.json) — 由爱诗科技提供的PixVerse V系列视频大模型API服务。 + - 模型：`pixverse/pixverse-v5.6-it2v`, `pixverse/pixverse-v5.6-kf2v`, `pixverse/pixverse-v5.6-r2v`, `pixverse/pixverse-v5.6-t2v` +- [PixVerse V6](groups/pixverse-v6-market-place.json) — 由爱诗科技提供的PixVerse V系列视频大模型API服务。 + - 模型：`pixverse/pixverse-v6-it2v`, `pixverse/pixverse-v6-kf2v`, `pixverse/pixverse-v6-r2v`, `pixverse/pixverse-v6-t2v` +- [Wan-I2V](groups/wan-image-to-video.json) — 图片生成视频内容，稳定保持图像主体、风格和文字等细节信息 + - 模型：`wan2.2-animate-mix`, `wan2.2-animate-move`, `wan2.2-i2v-flash`, `wan2.2-i2v-plus`, `wan2.2-kf2v-flash`, `wan2.2-s2v`, `wan2.2-s2v-detect`, `wan2.5-i2v-preview`, `wan2.6-i2v`, `wan2.6-i2v-flash`, `wan2.7-i2v`, `wanx2.1-i2v-plus`, `wanx2.1-i2v-turbo`, `wanx2.1-kf2v-plus` +- [Wan-R2V](groups/wan-reference-to-video.json) — 参考视频中的人或物，精准保持形象和声音，支持多参考合拍 + - 模型：`wan2.6-r2v`, `wan2.6-r2v-flash`, `wan2.7-r2v` +- [Wan-T2V](groups/wan-text-to-video.json) — 文字生成视频内容，丝滑动态能力，电影美学控制，精准指令遵循 + - 模型：`wan2.2-t2v-plus`, `wan2.5-t2v-preview`, `wan2.6-t2v`, `wan2.7-t2v`, `wanx2.1-t2v-plus`, `wanx2.1-t2v-turbo` +- [Wan-VideoEdit](groups/wan-video-edit.json) — 通过指令对视频进行编辑，支持局部/整体编辑、视频重塑、视频复刻等 + - 模型：`wan2.7-videoedit` +- [Wan2.1-VACE-Plus](groups/wanx2.1-vace-plus.json) — 万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。 + - 模型：`wanx2.1-vace-plus` +- [可灵AI](groups/kling-models-market-place.json) — 由可灵AI提供的高质量视频与图像生成及编辑模型。 + - 模型：`kling/kling-v3-image-generation`, `kling/kling-v3-omni-image-generation`, `kling/kling-v3-omni-video-generation`, `kling/kling-v3-video-generation` +- [声动人像VideoRetalk](groups/videoretalk.json) — VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。 + - 模型：`videoretalk` +- [悦动人像EMO](groups/emo-v1.json) — EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。 + - 模型：`emo-v1` +- [悦动人像EMO-detect](groups/emo-detect-v1.json) — EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 + - 模型：`emo-detect-v1` +- [灵动人像LivePortrait](groups/liveportrait.json) — LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。 + - 模型：`liveportrait` +- [灵动人像LivePortrait-detect](groups/liveportrait-detect.json) — LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 + - 模型：`liveportrait-detect` +- [视频风格重绘](groups/video-style-transform.json) — 视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡… + - 模型：`video-style-transform` +- [舞动人像AnimateAnyone](groups/animate-anyone-gen2.json) — AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。 + - 模型：`animate-anyone-gen2` +- [舞动人像AnimateAnyone-detect](groups/animate-anyone-detect-gen2.json) — AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 + - 模型：`animate-anyone-detect-gen2` +- [舞动人像AnimateAnyone-template](groups/animate-anyone-template-gen2.json) — AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。 + - 模型：`animate-anyone-template-gen2` +- [表情包Emoji](groups/emoji-v1.json) — 表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。 + - 模型：`emoji-v1` +- [表情包Emoji-detect](groups/emoji-detect-v1.json) — 表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。 + - 模型：`emoji-detect-v1` + +## 语音合成 `TTS` — 16 个家族 + +- [CosyVoice大模型](groups/cosyvoice.json) — 基于新一代生成式语音大模型，CosyVoice将文本理解和语音生成技术深度融合，能够精准解析并诠释各种文本内容，将其转化为如同真人发声般的自然语音，带来高度拟人化的自然语音合成体验。 + - 模型：`cosyvoice-clone-v1`, `cosyvoice-v1`, `cosyvoice-v2`, `cosyvoice-v3-flash`, `cosyvoice-v3-plus`, `cosyvoice-v3.5-flash`, `cosyvoice-v3.5-plus` +- [MiniMax-Speech系列语音模型](groups/MiniMax-speech-market-place.json) — 由MiniMax提供的MiniMax-Speech系列语音模型API服务。 + - 模型：`MiniMax/speech-02-hd`, `MiniMax/speech-02-turbo`, `MiniMax/speech-2.8-hd`, `MiniMax/speech-2.8-turbo` +- [Qwen-TTS](groups/qwen-tts.json) — 千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持输入输出全流式。 + - 模型：`qwen-tts`, `qwen-tts-latest` +- [Qwen-声音复刻](groups/qwen-voice-enrollment.json) — 千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复… + - 模型：`qwen-voice-enrollment` +- [Qwen-声音设计](groups/qwen-voice-design.json) — Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11… + - 模型：`qwen-voice-design` +- [Qwen3-TTS-Flash](groups/qwen3-tts-flash.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的离线语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地合成音频；同时支持多种语言，方言，支持同一音色多语言输出。该模型经过海量… + - 模型：`qwen3-tts-flash` +- [Qwen3-TTS-Flash-Realtime](groups/qwen3-tts-flash-realtime.json) — Qwen3-TTS-Flash-Realtime模型是通义实验室最新的实时语音合成大模型，不仅拥有17种高表现力的拟人音色，且能低延迟高稳定地实时合成音频；同时支持多种语言，方言，支持同一音色多语言输… + - 模型：`qwen3-tts-flash-realtime` +- [Qwen3-TTS-Instruct-Flash](groups/qwen3-tts-instruct-flash.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中… + - 模型：`qwen3-tts-instruct-flash` +- [qwen3-tts-instruct-flash-realtime](groups/qwen3-tts-instruct-flash-realtime.json) — 通义千问3-TTS-Flash模型是通义最新推出的实时语音合成大模型，Instruct模型可通过自然语言进行合成效果的处理，确保在不同语境下，合成情感、表达高度贴合的语音。目前支持25个音色的中英文I… + - 模型：`qwen3-tts-instruct-flash-realtime` +- [Qwen3-TTS-VC](groups/qwen3-tts-vc.json) — Qwen3-TTS-Flash模型是通义实验室最新推出的实时语音合成大模型，可对qwen-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模… + - 模型：`qwen3-tts-vc-2026-01-22` +- [Qwen3-TTS-VC-Realtime](groups/qwen3-tts-vc-realtime.json) — Qwen3-TTS-VC-Realtime模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-enrollment服务复刻的声音进行高保真实时语音合成，且同一音色支持11个语种的… + - 模型：`qwen3-tts-vc-realtime-2026-01-15` +- [Qwen3-TTS-VD](groups/qwen3-tts-vd.json) — Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数… + - 模型：`qwen3-tts-vd-2026-01-26` +- [Qwen3-TTS-VD-Realtime](groups/qwen3-tts-vd-realtime.json) — Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数… + - 模型：`qwen3-tts-vd-realtime-2026-01-15` +- [Sambert语音合成](groups/sambert.json) — 提供高效的文字转语音服务。该技术具备推理速度快、合成效果卓越、读音精准、韵律自然、声音还原度高以及表现力强等优点。此外，用户可以选择开启字级别和音素级别的时间戳，用于生成字幕或驱动数字人的嘴型。 + - 模型：`sambert-beth-v1`, `sambert-betty-v1`, `sambert-brian-v1`, `sambert-cally-v1`, `sambert-camila-v1`, `sambert-cindy-v1`, `sambert-clara-v1`, `sambert-donna-v1`, `sambert-eva-v1`, `sambert-hanna-v1`, `sambert-indah-v1`, `sambert-perla-v1`, `sambert-waan-v1`, `sambert-zhichu-v1`, `sambert-zhida-v1`, `sambert-zhide-v1`, `sambert-zhifei-v1`, `sambert-zhigui-v1`, `sambert-zhihao-v1`, `sambert-zhijia-v1`, `sambert-zhijing-v1`, `sambert-zhilun-v1`, `sambert-zhimao-v1`, `sambert-zhimiao-emo-v1`, `sambert-zhiming-v1`, `sambert-zhimo-v1`, `sambert-zhina-v1`, `sambert-zhinan-v1`, `sambert-zhiqi-v1`, `sambert-zhiqian-v1`, `sambert-zhiru-v1`, `sambert-zhishu-v1`, `sambert-zhishuo-v1`, `sambert-zhistella-v1`, `sambert-zhiting-v1`, `sambert-zhiwei-v1`, `sambert-zhixiang-v1`, `sambert-zhixiao-v1`, `sambert-zhiya-v1`, `sambert-zhiye-v1`, `sambert-zhiying-v1`, `sambert-zhiyuan-v1`, `sambert-zhiyue-v1` +- [大模型声音复刻及声音设计](groups/voice-enrollment.json) — 大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。 大模型声音设计使用FunAudioGen-VD模型… + - 模型：`voice-enrollment` +- [音乐生成](groups/fun-music.json) — 百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。 + - 模型：`fun-music-preview`, `fun-music-v1` + +## 推理 `Reasoning` — 14 个家族 - [DeepSeek](groups/deepseek.json) — DeepSeek是由深度求索提供的开源模型，包含 V3.1、V3、R1以及基于Qwen2.5系列蒸馏的大语言模型。 - 模型：`deepseek-r1`, `deepseek-r1-0528`, `deepseek-r1-distill-qwen-1.5b`, `deepseek-r1-distill-qwen-14b`, `deepseek-r1-distill-qwen-32b`, `deepseek-r1-distill-qwen-7b`, `deepseek-v3`, `deepseek-v3.1`, `deepseek-v3.2`, `deepseek-v3.2-exp`, `deepseek-v4-flash`, `deepseek-v4-pro` +- [MiniMax](groups/MiniMax-M2.1.json) — MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，包含MiniMax-M2.1、MiniMax-M2.5等开源模型。 + - 模型：`MiniMax-M2.1`, `MiniMax-M2.5` +- [QVQ-Max](groups/qvq-max.json) — 千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 + - 模型：`qvq-max` +- [Qwen-Flash](groups/qwen-flash.json) — Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。 + - 模型：`qwen-flash` +- [Qwen-Plus](groups/qwen-plus.json) — 千问超大规模语言模型的增强版，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。 + - 模型：`qwen-plus`, `qwen-plus-0112`, `qwen-plus-1220`, `qwen-plus-latest` +- [Qwen-QVQ-Plus](groups/qvq-plus.json) — 千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。 + - 模型：`qvq-plus` +- [Qwen-QwQ-Plus](groups/qwq-plus.json) — 千问QwQ推理模型增强版，基于Qwen2.5模型训练的QwQ推理模型，通过强化学习大幅度提升了模型推理能力。模型数学代码等核心指标（AIME 24/25、livecodebench）以及部分通用指标（… + - 模型：`qwq-plus` +- [Qwen-Turbo](groups/qwen-turbo.json) — 千问超大规模语言模型，支持中文英文等不同语言输入。主干模型、latest和快照04-28已升级Qwen3系列，实现思考模式和非思考模式的有效融合，可在对话中切换模式。 + - 模型：`qwen-turbo` - [Qwen3.5-Flash](groups/qwen3.5-flash.json) — Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 - 模型：`qwen3.5-flash` +- [Qwen3.5开源模型](groups/qwen3.5.json) — Qwen3.5系列开源模型，基于混合架构设计的原生视觉语言模型，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。 + - 模型：`qwen3.5-122b-a10b`, `qwen3.5-27b`, `qwen3.5-35b-a3b`, `qwen3.5-397b-a17b` - [Qwen3.6-Flash](groups/qwen3.6-flash.json) — Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉… - 模型：`qwen3.6-flash` - [Qwen3.6-Max](groups/qwen3.6-max.json) — Qwen3.6原生Max模型，相较于此前发布的Qwen3-Max和Qwen3.6-Plus，本模型在vibe coding能力上进一步提升、coding agent执行更加高效、前端编程开发能力显著提… @@ -26,21 +260,131 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Qwen3.7-Max](groups/qwen3.7-max.json) — Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自… - 模型：`qwen3.7-max`, `qwen3.7-max-preview` -## 视频生成 `VG` — 2 个家族 - -- [HappyHorse-I2V](groups/happyhorse-i2v.json) — HappyHorse系列最新图生视频模型，具备高度还原的动态画面生成能力，能够稳定保持与图像一致性，输出流畅自然、细节丰富的高质量视频。 - - 模型：`happyhorse-1.0-i2v`, `happyhorse-1.1-i2v` -- [HappyHorse-T2V](groups/happyhorse-t2v.json) — HappyHorse系列最新文生视频模型，具备高度还原的动态画面生成能力，能够精准理解文本语义，输出流畅自然、细节丰富的高质量视频。 - - 模型：`happyhorse-1.0-t2v`, `happyhorse-1.1-t2v` - -## 文本生成 `TG` — 2 个家族 +## 语音识别 `ASR` — 12 个家族 -- [Qwen3.5-Plus](groups/qwen3.5-plus.json) — Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。 - - 模型：`qwen3.5-plus` -- [Qwen3.7-Plus](groups/qwen3.7-plus.json) — Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真… - - 模型：`qwen3.7-plus` +- [Fun-ASR-Flash](groups/fun-asr-flash.json) — 百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词… + - 模型：`fun-asr-flash-2026-06-15` +- [Fun-ASR语音识别](groups/fun-asr.json) — 通义百聆新一代语音识别大模型，主打中文、英文、日文语音识别，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境，国内用户首推。 + - 模型：`fun-asr`, `fun-asr-mtl` +- [Paraformer语音识别-8k-v1](groups/paraformer-8k-v1.json) — Paraformer语音识别提供的文件转写API，能够对常见的音频或音视频文件进行语音识别，并将结果返回给调用者。Paraformer中文语音识别模型，支持8kHz电话语音识别。 + - 模型：`paraformer-8k-v1` +- [Paraformer语音识别-8k-v2](groups/paraformer-8k-v2.json) — Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。 + - 模型：`paraformer-8k-v2` +- [Paraformer语音识别-mtl-v1](groups/paraformer-mtl-v1.json) — Paraformer多语言语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 支持的语种/方言包括：中文普通话、中文方言（粤语、吴语、闽南语、东北话、甘肃话、贵州话、河南话、湖北话、湖南话… + - 模型：`paraformer-mtl-v1` +- [Paraformer语音识别-v1](groups/paraformer-v1.json) — Paraformer中英文语音识别模型，支持16kHz及以上采样率的音频或视频语音识别。 + - 模型：`paraformer-v1` +- [Paraformer语音识别-v2](groups/paraformer-v2.json) — 推荐使用 Paraformer最新语音识别模型，支持多个语种的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果，支持任意采样率。 支持的语言包括：中文（含粤语等各种方言）… + - 模型：`paraformer-v2` +- [Qwen3-ASR-Flash](groups/qwen3-asr-flash.json) — Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实现了高精… + - 模型：`qwen3-asr-flash` +- [Qwen3-ASR-Flash-Filetrans](groups/qwen3-asr-flash-filetrans.json) — Qwen3-ASR-Flash的大文件转录版本，Qwen3-ASR-Flash是一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音… + - 模型：`qwen3-asr-flash-filetrans` +- [Qwen3-Omni-30b-a3b-Captioner](groups/qwen3-omni-30b-a3b-captioner.json) — 千问3-Omni-30b-a3b-Captioner是一款强大的音频细粒度分析模型，专为在复杂多变的音频场景中生成精准、全面的内容描述而设计，可自动解析并描述从复杂语音、环境声到音乐、影视声效等各类音… + - 模型：`qwen3-omni-30b-a3b-captioner` +- [一句话识别及翻译V1.0](groups/gummy-chat-v1.json) — 多语言语音转写及翻译的多模态大模型。本模型支持60秒以内的实时语音识别，适用于语音搜索、设备指令等场景。提供10个混合语种的高准确率识别服务，同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 + - 模型：`gummy-chat-v1` +- [语音识别热词](groups/speech-biasing.json) — 热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行… + - 模型：`speech-biasing` -## 视觉理解 `VU` — 1 个家族 +## 视觉理解 `VU` — 8 个家族 +- [GUI-Plus](groups/gui-plus.json) — GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作… + - 模型：`gui-plus` +- [Qwen-VL-Max](groups/qwen-vl-max.json) — Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。 + - 模型：`qwen-vl-max` +- [Qwen-VL-OCR](groups/qwen-vl-ocr.json) — Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。 + - 模型：`qwen-vl-ocr`, `qwen-vl-ocr-1028`, `qwen-vl-ocr-latest` +- [Qwen3-VL-Flash](groups/qwen3-vl-flash.json) — Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识… + - 模型：`qwen3-vl-flash` +- [Qwen3-VL-Plus](groups/qwen3-vl-plus.json) — Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与… + - 模型：`qwen3-vl-plus` +- [Qwen3.5-OCR](groups/qwen3.5-ocr.json) — Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。 + - 模型：`qwen3.5-ocr` - [Qwen3.6开源模型](groups/qwen3.6.json) — Qwen3.6系列开源模型，基于混合架构设计的原生视觉语言模型，模型效果相较于3.5系列同尺寸有大幅提升。 - 模型：`qwen3.6-27b`, `qwen3.6-35b-a3b` +- [Qwen3开源模型](groups/qwen3.json) — Qwen3系列开源模型，包含混合模型、思考模型与非思考模型，思考能力与通用能力均达到同规模业界SOTA水平。 + - 模型：`qwen3-14b`, `qwen3-235b-a22b`, `qwen3-235b-a22b-instruct-2507`, `qwen3-235b-a22b-thinking-2507`, `qwen3-30b-a3b`, `qwen3-30b-a3b-instruct-2507`, `qwen3-30b-a3b-thinking-2507`, `qwen3-32b`, `qwen3-8b`, `qwen3-coder-next`, `qwen3-next-80b-a3b-instruct`, `qwen3-next-80b-a3b-thinking`, `qwen3-vl-235b-a22b-instruct`, `qwen3-vl-235b-a22b-thinking`, `qwen3-vl-30b-a3b-instruct`, `qwen3-vl-30b-a3b-thinking`, `qwen3-vl-32b-instruct`, `qwen3-vl-32b-thinking`, `qwen3-vl-8b-instruct`, `qwen3-vl-8b-thinking` + +## 实时语音识别 `Realtime-ASR` — 7 个家族 + +- [Fun-ASR实时语音识别](groups/fun-asr-realtime.json) — 通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/… + - 模型：`fun-asr-flash-8k-realtime`, `fun-asr-realtime` +- [Paraformer实时语音识别-8k-v1](groups/paraformer-realtime-8k-v1.json) — Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。 + - 模型：`paraformer-realtime-8k-v1` +- [Paraformer实时语音识别-8k-v2](groups/paraformer-realtime-8k-v2.json) — 推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持8kHz电话客服… + - 模型：`paraformer-realtime-8k-v2` +- [Paraformer实时语音识别-v1](groups/paraformer-realtime-v1.json) — Paraformer中文实时语音识别模型，支持16kHz及以上采样率的视频直播、会议等实时场景下的语音识别。 + - 模型：`paraformer-realtime-v1` +- [Paraformer实时语音识别-v2](groups/paraformer-realtime-v2.json) — 推荐使用 Paraformer最新实时语音识别模型，支持多个语种自由切换的视频直播、会议等实时场景的语音识别。可以通过language_hints参数选择语种获得更准确的识别效果。支持任意采样率。 支… + - 模型：`paraformer-realtime-v2` +- [Qwen3-ASR-Flash-Realtime](groups/qwen3-asr-flash-realtime.json) — Qwen3-ASR-Flash的实时版，一款基于大语言模型的高精度、高智能、高鲁棒性的多语种语音识别模型。依托强大的基座模型、海量的文本与多模态数据、千万小时音频数据，Qwen3-ASR-Flash实… + - 模型：`qwen3-asr-flash-realtime` +- [Qwen3-LiveTranslate-Flash](groups/qwen3-livetranslate-flash.json) — Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，… + - 模型：`qwen3-livetranslate-flash` + +## 全模态 `Multimodal-Omni` — 5 个家族 + +- [Qwen-Omni-Turbo](groups/qwen-omni-turbo.json) — 千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 + - 模型：`qwen-omni-turbo`, `qwen-omni-turbo-latest` +- [Qwen2.5-开源模型](groups/qwen2.5.json) — Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。 + - 模型：`qwen2.5-omni-7b` +- [Qwen3-Omni-Flash](groups/qwen3-omni-flash.json) — Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互… + - 模型：`qwen3-omni-flash` +- [Qwen3.5-Omni-Flash](groups/qwen3.5-omni-flash.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… + - 模型：`qwen3.5-omni-flash` +- [Qwen3.5-Omni-Plus](groups/qwen3.5-omni-plus.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… + - 模型：`qwen3.5-omni-plus` + +## 实时全模态 `Realtime-Omni` — 4 个家族 + +- [Qwen-Omni-Turbo-Realtime](groups/qwen-omni-turbo-realtime.json) — 千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。 + - 模型：`qwen-omni-turbo-realtime`, `qwen-omni-turbo-realtime-latest` +- [Qwen3-Omni-Flash-Realtime](groups/qwen3-omni-flash-realtime.json) — Qwen3-Omni-Flash-Realtime多模态大模型的实时版，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文… + - 模型：`qwen3-omni-flash-realtime` +- [Qwen3.5-Omni-Flash-Realtime](groups/qwen3.5-omni-flash-realtime.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话… + - 模型：`qwen3.5-omni-flash-realtime` +- [Qwen3.5-Omni-Plus-Realtime](groups/qwen3.5-omni-plus-realtime.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话… + - 模型：`qwen3.5-omni-plus-realtime` + +## 实时音频翻译 `Realtime-Audio-Translate` — 3 个家族 + +- [Qwen3-LiveTranslate-Flash-Realtime](groups/qwen3-livetranslate-flash-realtime.json) — Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言… + - 模型：`qwen3-livetranslate-flash-realtime` +- [Qwen3.5-LiveTranslate-Flash-Realtime](groups/qwen3.5-livetranslate-flash-realtime.json) — Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐… + - 模型：`qwen3.5-livetranslate-flash-realtime` +- [实时语音识别及翻译V1.0](groups/gummy-realtime-v1.json) — 多语言语音转写及翻译的多模态大模型。本模型提供长时间、高准确率、实时转写中/英/日/韩等10个混合语种的服务。同时支持中英日韩互译，以其他6个语种翻译成中文或英文。 + - 模型：`gummy-realtime-v1` + +## 多模态嵌入 `ME` — 2 个家族 + +- [Qwen-VL-Embedding](groups/qwen-vl-embedding.json) — 基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。 + - 模型：`qwen2.5-vl-embedding`, `qwen3-vl-embedding` +- [通义多模态向量](groups/embedding.json) — 基于LLM底座的通用多模态表征模型，支持文本、图像、视频3种模态，具有以视觉为中心、全场景性能优异、高性价比的特点，适用于以图搜图、以文搜图、以文搜视频、以视频搜视频、以文搜文等下游多样化任务场景。 + - 模型：`multimodal-embedding-v1`, `tongyi-embedding-vision-flash`, `tongyi-embedding-vision-plus` + +## Realtime-Chatting `Realtime-Chatting` — 2 个家族 + +- [Qwen-Audio-Realtime-Flash](groups/qwen-audio-realtime-flash.json) — Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和… + - 模型：`qwen-audio-3.0-realtime-flash` +- [Qwen-Audio-Realtime-Plus](groups/qwen-audio-realtime-plus.json) — Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和… + - 模型：`qwen-audio-3.0-realtime-plus` + +## 实时语音合成 `Realtime-Text-to-Speech` — 2 个家族 + +- [Qwen-Audio-TTS](groups/qwen-audio-tts.json) — Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，… + - 模型：`qwen-audio-3.0-tts-flash`, `qwen-audio-3.0-tts-plus` +- [Qwen-TTS-Realtime](groups/qwen-tts-realtime.json) — Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成利器。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。 + - 模型：`qwen-tts-realtime`, `qwen-tts-realtime-latest` + +## 翻译 `TR` — 2 个家族 + +- [Qwen-Embedding](groups/qwen-embedding.json) — 基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度… + - 模型：`text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` +- [Qwen-Rerank](groups/qwen-rerank.json) — 基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。 + - 模型：`gte-rerank-v2`, `qwen3-rerank`, `qwen3-vl-rerank` + +## 3D 生成 `3D-generation` — 1 个家族 + +- [Tripo](groups/tripo-models-market-place.json) — AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。 + - 模型：`Tripo/Tripo-H3.1`, `Tripo/Tripo-P1.0` diff --git a/skills/bailian-docs-llm-wiki/models/models.jsonl b/skills/bailian-docs-llm-wiki/models/models.jsonl index adcdab2f..bf86a4de 100644 --- a/skills/bailian-docs-llm-wiki/models/models.jsonl +++ b/skills/bailian-docs-llm-wiki/models/models.jsonl @@ -1,3 +1,28 @@ +{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"user-spec":{"count_limit":10,"count_limit_period":5,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":10,"count_limit_period":5,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} +{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} +{"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","family":"aitryon-refiner","familyName":"AI试衣OutfitAnyone-图片精修","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796663.html","detailPath":"groups/aitryon-refiner.json"} +{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} +{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} +{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} +{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} +{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} {"model":"deepseek-r1","name":"DeepSeek-R1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_batch","unit":"每百万tokens","price":"2"},{"type":"output_token_batch","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} @@ -10,12 +35,178 @@ {"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} +{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.9"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.15"},{"type":"embedding_token","unit":"每百万tokens","price":"0.15"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.5"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} +{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_duration_1-1","unit":"每秒","price":"0.08"},{"type":"video_duration_3-4","unit":"每秒","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} +{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} +{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} +{"model":"facechain-facedetect","name":"FaceChain人物图像检测","family":"facechain-facedetect","familyName":"FaceChain人物图像检测","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712507.html","detailPath":"groups/facechain-facedetect.json"} +{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.18"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} +{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"20"},{"type":"output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} +{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} +{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.005"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.002"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"glm-4.5","name":"GLM-4.5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-4.5-air","name":"GLM-4.5-Air","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-4.6","name":"GLM-4.6","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["Reasoning","TG"],"features":["model-experience","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-4.7","name":"GLM-4.7","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5","name":"GLM-5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5.1","name":"GLM-5.1","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":202745,"maxInputTokens":202745,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"16"},{"type":"output_token","unit":"每百万tokens","price":"56"},{"type":"input_token_cache","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} +{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.5"},{"type":"output_token","unit":"每百万tokens","price":"4.5"}],"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} +{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} +{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} {"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} {"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} +{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} {"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} {"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} +{"model":"image-erase-completion","name":"图像擦除补全","family":"image-erase-completion","familyName":"图像擦除补全","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840907.html","detailPath":"groups/image-erase-completion.json"} +{"model":"image-instance-segmentation","name":"人物实例分割","family":"image-instance-segmentation","familyName":"人物实例分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840906.html","detailPath":"groups/image-instance-segmentation.json"} +{"model":"image-out-painting","name":"图像画面扩展","family":"image-out-painting","familyName":"图像画面扩展","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796845.html","detailPath":"groups/image-out-painting.json"} +{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.1"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"13"},{"type":"output_token","unit":"每百万tokens","price":"54"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.6"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"},{"type":"image_type_4k","unit":"每秒","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"720P_no_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_reference_video","unit":"每秒","price":"1.2"},{"type":"720P_no_audio_no_reference_video","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio_no_reference_video","unit":"每秒","price":"0.8"},{"type":"720P_no_audio_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_audio_reference_video","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"},{"type":"720P_no_audio","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio","unit":"每秒","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} +{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.02"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} +{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4.2"},{"type":"output_token","unit":"每百万tokens","price":"16.8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.84"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} +{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} +{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} +{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} +{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} +{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} +{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} +{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} +{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} +{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} +{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"32"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} +{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} +{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} +{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} +{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} +{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} +{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3.5"},{"type":"output_token","unit":"每百万tokens","price":"7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} +{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} +{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.12"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"0.75"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.06"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} +{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} +{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.25"},{"type":"output_token","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.05"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} +{"model":"qwen-flash","name":"Qwen-Flash","family":"qwen-flash","familyName":"Qwen-Flash","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-flash.json"} +{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} +{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} +{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} +{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.3"}],"qpmInfo":{"user-spec":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.25"}],"qpmInfo":{"user-spec":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} +{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"9.6"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.48"},{"type":"input_token_batch","unit":"每百万tokens","price":"1.2"},{"type":"output_token_batch","unit":"每百万tokens","price":"4.8"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} +{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} +{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.003"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} +{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} +{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"5.4"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} +{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} +{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"},{"type":"vision_input_token_cache","unit":"每百万tokens","price":"0.3"},{"type":"text_input_token_cache","unit":"每百万tokens","price":"0.08"},{"type":"audio_input_token_cache","unit":"每百万tokens","price":"5"},{"type":"text_input_token_batch","unit":"每百万tokens","price":"0.2"},{"type":"audio_input_token_batch","unit":"每百万tokens","price":"12.5"},{"type":"vision_input_token_batch","unit":"每百万tokens","price":"0.75"},{"type":"multi_output_token_batch","unit":"每百万tokens","price":"25"},{"type":"multiin_text_output_token_batch","unit":"每百万tokens","price":"2.25"},{"type":"purein_text_output_token_batch","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.16"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} +{"model":"qwen-plus","name":"Qwen-Plus","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","batch"],"contextWindow":1000000,"maxInputTokens":995904,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.6"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.3"},{"type":"thinking_output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.3"},{"type":"thinking_input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} +{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.32"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} +{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"5"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} +{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"tts_vc_model","unit":"次","price":"0.01"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} +{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"38"},{"type":"vision_input_token","unit":"每百万tokens","price":"2"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"2.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"6"},{"type":"multi_output_token","unit":"每百万tokens","price":"76"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} +{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} +{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} +{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} +{"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","family":"qwen3-coder-30b-a3b-instruct","familyName":"Qwen3-Coder-30B-A3B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls","detailPath":"groups/qwen3-coder-30b-a3b-instruct.json"} +{"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","family":"qwen3-coder-480b-a35b-instruct","familyName":"Qwen3-Coder-480B-A35B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-480b-a35b-instruct.json"} +{"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","family":"qwen3-coder-flash","familyName":"Qwen3-Coder-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-flash.json"} +{"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","family":"qwen3-coder-plus","familyName":"Qwen3-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-plus.json"} +{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"64"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"8"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"64"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"240"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} +{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"4"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"10"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} +{"model":"qwen3-max","name":"Qwen3-Max","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"user-spec":{"count_limit":500,"count_limit_period":1,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} +{"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","cache"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} +{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} +{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"2.2"},{"type":"audio_input_token","unit":"每百万tokens","price":"18.9"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.9"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"8.3"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"15.2"},{"type":"multi_output_token","unit":"每百万tokens","price":"75.1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} +{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"},{"type":"multi_output_token","unit":"每百万tokens","price":"62.6"},{"type":"thinking_text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"thinking_audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"thinking_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"thinking_purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"thinking_multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} +{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"user-spec":{"count_limit":20,"count_limit_period":1},"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} +{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} +{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} +{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} +{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} +{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} +{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} +{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} +{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","family":"qwen3-vl-flash","familyName":"Qwen3-VL-Flash","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-flash.json"} +{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","family":"qwen3-vl-plus","familyName":"Qwen3-VL-Plus","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-plus.json"} {"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","TG","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"user-spec":{"count_limit":500,"count_limit_period":1,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} +{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"40"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"100"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"160"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} +{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} +{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"27"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"107"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"3.3"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} +{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"18"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"72"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"2.2"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"13.3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} +{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"80"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"300"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"60"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} +{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search","function-calling","batch"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"53"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"213"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"40"},{"type":"omni_audio_input_token_batch","unit":"每百万tokens","price":"26.5"},{"type":"omni_no_audio_input_token_batch","unit":"每百万tokens","price":"3.5"},{"type":"omni_no_audio_output_token_batch","unit":"每百万tokens","price":"20"},{"type":"omni_audio_input_token_batch_chat","unit":"每百万tokens","price":"53"},{"type":"omni_no_audio_input_token_batch_chat","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token_batch_chat","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} {"model":"qwen3.5-plus","name":"Qwen3.5-Plus","family":"qwen3.5-plus","familyName":"Qwen3.5-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-plus.json"} +{"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} +{"model":"qwen3.5-27b","name":"Qwen3.5-27B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","fine-tuning"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} +{"model":"qwen3.5-35b-a3b","name":"Qwen3.5-35B-A3B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} +{"model":"qwen3.5-397b-a17b","name":"Qwen3.5-397B-A17B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} {"model":"qwen3.6-flash","name":"Qwen3.6-Flash","family":"qwen3.6-flash","familyName":"Qwen3.6-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-flash.json"} {"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","family":"qwen3.6-max","familyName":"Qwen3.6-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":245760,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-max.json"} {"model":"qwen3.6-plus","name":"Qwen3.6-Plus","family":"qwen3.6-plus","familyName":"Qwen3.6-Plus","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"user-spec":{"count_limit":500,"count_limit_period":1,"usage_limit":8000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":8000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-plus.json"} @@ -24,3 +215,145 @@ {"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"6"},{"type":"output_token_batch","unit":"每百万tokens","price":"18"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"15"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"1.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"12"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} {"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} {"model":"qwen3.7-plus","name":"Qwen3.7-Plus","family":"qwen3.7-plus","familyName":"Qwen3.7-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search","structured-outputs"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-plus.json"} +{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"},{"type":"ft","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} +{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"shoemodel-v1","name":"鞋靴模特","family":"shoemodel-v1","familyName":"鞋靴模特","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2804662.html","detailPath":"groups/shoemodel-v1.json"} +{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"speech-biasing","name":"语音识别热词","family":"speech-biasing","familyName":"语音识别热词","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/speech-biasing.json"} +{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG","VU"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.35"},{"type":"output_token","unit":"每百万tokens","price":"8.1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.27"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} +{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.4"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} +{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.2"},{"type":"output_token","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} +{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2.7"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} +{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_standard_no_texture","unit":"每次","price":"0.7"},{"type":"generation_3d_image_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_multiview_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_text_to_3d_standard_sd_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_image_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_multiview_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_text_to_3d_standard_hd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_ultra_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["VU","TG"],"features":["structured-outputs"],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.216"},{"type":"output_token","unit":"每百万tokens","price":"0.216"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["structured-outputs","prefix-completion","function-calling","cache"],"contextWindow":131072,"maxInputTokens":131072,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":131072,"maxInputTokens":32768,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} +{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.2"},{"type":"video_ratio_720p","unit":"每秒","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} +{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} +{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.625"},{"type":"image_type_2k","unit":"每张","price":"1"},{"type":"image_type_4k","unit":"每张","price":"1.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} +{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.28125"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} +{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.9375"},{"type":"image_type_2k","unit":"每张","price":"0.9375"},{"type":"image_type_4k","unit":"每张","price":"1.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} +{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.46875"},{"type":"image_type_2k","unit":"每张","price":"0.78125"},{"type":"image_type_4k","unit":"每张","price":"1.09375"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} +{"model":"virtualmodel-v2","name":"虚拟模特V2","family":"virtualmodel-v2","familyName":"虚拟模特V2","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/virtualmodel-v2.json"} +{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","family":"voice-enrollment","familyName":"大模型声音复刻及声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/voice-enrollment.json"} +{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.6"},{"type":"video_ratio_pro","unit":"每秒","price":"0.9"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.4"},{"type":"video_ratio_pro","unit":"每秒","price":"0.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.5"},{"type":"video_ratio_720p","unit":"每秒","price":"0.9"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-r2v","name":"Wan2.6-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} +{"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} +{"model":"wan2.7-r2v","name":"Wan2.7-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} +{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"user-spec":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.04"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wan2.6-t2v","name":"Wan2.6-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wan2.7-t2v","name":"Wan2.7-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.24"},{"type":"video_ratio_720p","unit":"每秒","price":"0.24"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} +{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} +{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} +{"model":"wanx-poster-generation-v1","name":"创意海报生成","family":"wanx-poster-generation-v1","familyName":"创意海报生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807172.html","detailPath":"groups/wanx-poster-generation-v1.json"} +{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.06"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} +{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.12"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} +{"model":"wanx-virtualmodel","name":"虚拟模特","family":"wanx-virtualmodel","familyName":"虚拟模特","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/wanx-virtualmodel.json"} +{"model":"wanx-x-painting","name":"万相-图像局部重绘","family":"wanx-x-painting","familyName":"万相-图像局部重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2797051.html","detailPath":"groups/wanx-x-painting.json"} +{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} +{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} +{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} +{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"7"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} +{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"prices":[{"type":"image_standard","unit":"每张","price":"0.1"},{"type":"image_thinking","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} +{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6"},{"type":"output_token","unit":"每百万tokens","price":"22"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md deleted file mode 100644 index 07acc02b..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md +++ /dev/null @@ -1,246 +0,0 @@ -# BatchUpdateFileTag - 批量更新文档标签 - -该接口用于批量更新数据连接中的文档标签。 - -## 调试 - -[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/BatchUpdateFileTag) - - [![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png) 调试](https://api.aliyun.com/api/bailian/2023-12-29/BatchUpdateFileTag) - -## **授权信息** - -当前API暂无授权信息透出。 - -## 请求语法 - -``` -PUT /{WorkspaceId}/datacenter/batchupdatetag HTTP/1.1 -``` - -## 路径参数 - -**名称** - -**类型** - -**必填** - -**描述** - -**示例值** - -WorkspaceId - -string - -是 - -业务空间 ID。在百炼的[控制台首页](https://bailian.console.aliyun.com/knowledge-base#/home)，单击页面左上角业务空间详情图标获取。 - -llm-3shx2gu255oqxxxx - -## 请求参数 - -**名称** - -**类型** - -**必填** - -**描述** - -**示例值** - -FileInfos - -array - -是 - -需要更新的文档列表 - -object - -是 - -FileId - -string - -是 - -数据中心的文件 ID，您可以在[应用数据](https://bailian.console.aliyun.com/?tab=app#/data-center)页面，单击文件名称旁的 ID 图标获取。 - -file\_3d5319366e2c46309f4c11cfbeacd5fd\_10045951 - -tags - -array - -是 - -- 文件关联的标签列表。最多传入 100 个标签，所有标签字符长度总和不能超过 700。 - - -string - -否 - -标签值，每个标签最多 32 个字符，支持 Unicode 中 letter 分类下的字符（其中包括英文、中文和数字等），下划线\_，中划线-，标签中不能包含空格。 - -TagA - -UpdateMode - -string - -否 - -更新模式，仅支持 APPEND（追加）和 OVERWRITE（覆盖） - -OVERWRITE - -## **返回参数** - -**名称** - -**类型** - -**描述** - -**示例值** - -object - -Schema of Response - -Code - -string - -错误状态码 - -Success - -Data - -object - -接口返回的业务字段。 - -UpdateFileTagResultList - -array - -标签更新的结果列表 - -object - -FileId - -string - -文件 ID。 - -file\_f40f2a32205d44b4a93b11617113da15\_10045951 - -Success - -boolean - -接口调用是否成功，可能值为： - -- true：成功。 - -- false：失败。 - - -true - -ErrorCode - -string - -返回错误码，仅当 Success 为 false 时返回。 - -NoPermission - -ErrorMessage - -string - -错误描述信息，仅当 Success 为 false 时返回。 - -FileId not exists. - -Message - -string - -错误信息 - -Required parameter(FileId) missing or invalid, please check the request parameters. - -RequestId - -string - -Id of the request - -17204B98-xxxx-4F9A-8464-2446A84821CA - -Status - -string - -接口返回的状态码。 - -200 - -Success - -boolean - -接口调用是否成功，可能值： - -- true：成功。 - -- false：失败。 - - -true - -## 示例 - -正常返回示例 - -`JSON`格式 - -``` -{ - "Code": "Success", - "Data": { - "UpdateFileTagResultList": [ - { - "FileId": "file_f40f2a32205d44b4a93b11617113da15_10045951", - "Success": true, - "ErrorCode": "NoPermission", - "ErrorMessage": "FileId not exists." - } - ] - }, - "Message": "Required parameter(FileId) missing or invalid, please check the request parameters.", - "RequestId": "17204B98-xxxx-4F9A-8464-2446A84821CA", - "Status": "200", - "Success": true -} -``` - -## 错误码 - -访问[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)查看更多错误码。 - -## **变更历史** - -更多信息，参考[变更详情](https://api.aliyun.com/document/bailian/2023-12-29/BatchUpdateFileTag#workbench-doc-change-demo)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md deleted file mode 100644 index 4042ba36..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md +++ /dev/null @@ -1,968 +0,0 @@ -# 万相-涂鸦作画API参考 - -本文介绍万相-涂鸦作画模型的API输入输出参数。 - -**相关指南**：[涂鸦作画](https://help.aliyun.com/zh/model-studio/sketch-to-image) - -**重要** - -本文档仅适用于华北2（北京）地域，且必须使用该地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 - -**重要** - -百炼为华北2（北京）地域推出了业务空间专属域名 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope.aliyuncs.com` 迁移至新域名。 - -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 - -## **模型概览** - -**模型效果示意** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2400704371/p883780.png) - -**模型简介** - -**模型名称** - -**模型简介** - -wanx-sketch-to-image-lite - -万相-涂鸦作画通过手绘图案和文字描述，生成精美的涂鸦绘画作品。 - -**模型说明** - -**模型名称** - -**计费单价** - -**限流（主账号与RAM子账号共用）** - -**免费额度**[（查看）](https://help.aliyun.com/zh/model-studio/new-free-quota) - -**任务下发接口QPS限制** - -**同时处理中任务数量** - -wanx-sketch-to-image-lite - -0.06元/张 - -2 - -1 - -500张 - -更多说明请参见[模型计费及限流](#b8457b7223zhp)。 - -## **前提条件** - -涂鸦作画API支持通过HTTP和DashScope SDK进行调用。 - -在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 - -## HTTP调用 - -图像模型处理时间较长，为了避免请求超时，HTTP调用仅支持异步获取模型结果。您需要发起两个请求： - -1. **创建任务获取任务ID**：首先发起创建任务请求，该请求会返回任务ID（task\_id）。 - -2. **根据任务ID查询结果**：使用上一步获得的任务ID，查询任务状态及结果。任务成功执行时将返回图像URL，有效期24小时。 - - -**说明** - -创建任务后，该任务将被加入到排队队列，等待调度执行。后续需要调用“根据任务ID查询结果接口”获取任务状态及结果。 - -### **步骤1：创建任务获取任务ID** - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/` - -#### 请求参数 - -## curl - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \ ---header 'X-DashScope-Async: enable' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data '{ - "model": "wanx-sketch-to-image-lite", - "input": { - "sketch_image_url": "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg", - "prompt": "一棵参天大树" - }, - "parameters": { - "size": "768*768", - "n": 2, - "sketch_weight": 3, - "style": "" - } -}' -``` - -##### **请求头（Headers）** - -**Content-Type** `_string_` **（必选）** - -请求内容类型。此参数必须设置为`application/json`。 - -**Authorization** `_string_`**（必选）** - -请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 - -**X-DashScope-Async** `_string_` **（必选）** - -异步处理配置参数。HTTP请求只支持异步，**必须设置为**`**enable**`。 - -**重要** - -缺少此请求头将报错：“current user api does not support synchronous calls”。 - -**X-DashScope-WorkSpace** `_string_` （可选） - -阿里云百炼业务空间ID。示例值：llm-xxxx。 - -您可以在此[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -**详细说明** - -此参数根据阿里云百炼API Key进行填写。 - -- 若为主账号API Key，可不填。不填则使用主账号权限，填写则使用对应的业务空间权限。 - -- 若为RAM子账号API Key，则必填。RAM子账号一定归属于某个业务空间。 - - -业务空间必须具备访问模型的权限，才能调用API。若无权限，请参考[授权子业务空间模型调用、训练和部署](https://help.aliyun.com/zh/model-studio/use-workspace#f2e68d7ba7ubk)。 - -> 关于如何区分阿里云百炼主账号和RAM子账号，请参考[主账号管理](https://help.aliyun.com/zh/model-studio/business-space-management)。 - -##### **请求体（Request Body）** - -**model** `_string_` **（必选）** - -调用模型。 - -**input** `_object_` **（必选）** - -输入的基本信息，比如提示词、图像URL地址。 - -**属性** - -**prompt** `_string_` **（必选）** - -提示词，用来描述生成图像中期望包含的元素和视觉特点。 - -支持中英文，长度不超过75个字符，超过部分会自动截断。 - -示例值：一棵参天大树。 - -**sketch\_image\_url** `_string_` **（必选）** - -输入草图的URL地址。输入草图需要与输出图像的分辨率比例保持一致，否则会导致图片拉伸变形，建议使用白色背景图。 - -URL 需为公网可访问的地址，并支持 HTTP 或 HTTPS 协议。您也可在此[获取临时公网URL](https://help.aliyun.com/zh/model-studio/get-temporary-file-url)。 - -图像限制： - -- 图像格式：JPG、JPEG、PNG、TIFF、WEBP。 - -- 图像分辨率：不小于256×256像素且不超过2048×2048像素。 - -- 图像大小：不超过10 MB。 - -- URL地址中不能包含中文字符。 - - -草图示例： - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9289386271/p850798.png) - -**parameters** `_object_` （可选） - -图像处理参数。 - -**属性** - -**style** `_string_` （可选） - -输出图像的风格，目前支持以下风格取值： - -- ：默认值，由模型随机输出图像风格。 - -- <3d cartoon>：3D卡通。 - -- ：二次元。 - -- ：油画。 - -- ：水彩。 - -- ：素描。 - -- ：中国画。 - -- ：扁平插画。 - - -**size** `_string_` （可选） - -输出图像的分辨率。目前仅支持一种图像分辨率：768\*768，且为默认值。 - -**n** `_integer_` （可选） - -生成图片的数量。取值范围为1~4张，默认为4。 - -**sketch\_weight** `_integer_` （可选） - -输入草图对输出图像的约束程度。 - -取值范围为0-10，取值间隔为1， 默认值为10。取值越大表示输出图像跟输入草图越相似。 - -**sketch\_extraction** `_boolean_` （可选） - -如果上传图片是RGB图片，而非草图（sketch线稿），此参数可控制是否对输入图片进行sketch边缘提取。 - -默认值为False，表示不进行提取。设置为True时，表示进行提取，此时，`sketch_color`字段失效。 - -**sketch\_color** `_array_` （可选） - -此字段在`sketch_extraction=false`时生效，所包含数值均被视为画笔色，其余数值均会视为背景色。模型会基于一种或多种画笔色描绘的区域生成新的画作。默认值为\[\]。 - -当sketch\_image\_url线稿中的线条不是黑色，而是包含其他一种或多种颜色时，可以指定一个或多个RGB颜色数值作为画笔色。 - -示例值：\[\[134, 134, 134\], \[0, 0, 0\]\] - -#### **响应参数** - -#### 成功响应 - -请保存 task\_id，用于查询任务状态与结果。 - -``` -{ - "output": { - "task_status": "PENDING", - "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx" - }, - "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx" -} -``` - -#### 异常响应 - -创建任务失败，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "code": "InvalidApiKey", - "message": "No API-key provided.", - "request_id": "7438d53d-6eb8-4596-8835-xxxxxx" -} -``` - -**output** `_object_` - -任务输出信息。 - -**属性** - -**task\_id** `_string_` - -任务ID。查询有效期24小时。 - -**task\_status** `_string_` - -任务状态。 - -**枚举值** - -- PENDING：任务排队中 - -- RUNNING：任务处理中 - -- SUCCEEDED：任务执行成功 - -- FAILED：任务执行失败 - -- CANCELED：任务已取消 - -- UNKNOWN：任务不存在或状态未知 - - -**request\_id** `_string_` - -请求唯一标识。可用于请求明细溯源和问题排查。 - -**code** `_string_` - -请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**message** `_string_` - -请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -### 步骤2：根据任务ID查询结果 - -`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` - -#### 请求参数 - -#### 查询任务结果 - -请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 - -> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 - -``` -curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" -``` - -#### **请求头（Headers）** - -**Authorization** `_string_`**（必选）** - -请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 - -#### **URL路径参数（Path parameters）** - -**task\_id** `_string_`**（必选）** - -任务ID。 - -#### **响应参数** - -#### 任务执行成功 - -任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 - -``` -{ - "request_id": "85eaba38-0185-99d7-8d16-4d9135238846", - "output": { - "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", - "task_status": "SUCCEEDED", - "results": [ - { - "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/a1.png" - }, - { - "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/b2.png" - } - ], - "task_metrics": { - "TOTAL": 2, - "SUCCEEDED": 2, - "FAILED": 0 - } - }, - "usage": { - "image_count": 2 - } -} -``` - -#### 任务执行失败 - -若任务执行失败，task\_status将置为 FAILED，并提供错误码和信息。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "request_id": "e5d70b02-ebd3-98ce-9fe8-759d7d7b107d", - "output": { - "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", - "task_status": "FAILED", - "code": "InvalidParameter", - "message": "The size is not match the allowed size ['1024*1024', '720*1280', '1280*720']", - "task_metrics": { - "TOTAL": 4, - "SUCCEEDED": 0, - "FAILED": 4 - } - } -} -``` - -#### 任务部分失败 - -模型可以在一次任务中生成多张图片。只要有一张图片生成成功，任务状态将标记为`SUCCEEDED`，并且返回相应的图像URL。对于生成失败的图片，结果中会返回相应的失败原因。同时在usage统计中，只会对成功的结果计数。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "request_id": "85eaba38-0185-99d7-8d16-4d9135238846", - "output": { - "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", - "task_status": "SUCCEEDED", - "results": [ - { - "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/a1.png" - }, - { - "code": "InternalError.Timeout", - "message": "An internal timeout error has occured during execution, please try again later or contact service support." - } - ], - "task_metrics": { - "TOTAL": 2, - "SUCCEEDED": 1, - "FAILED": 1 - } - }, - "usage": { - "image_count": 1 - } -} -``` - -**output** `_object_` - -任务输出信息。 - -**属性** - -**task\_id** `_string_` - -任务ID。查询有效期24小时。 - -**task\_status** `_string_` - -任务状态。 - -**枚举值** - -- PENDING：任务排队中 - -- RUNNING：任务处理中 - -- SUCCEEDED：任务执行成功 - -- FAILED：任务执行失败 - -- CANCELED：任务已取消 - -- UNKNOWN：任务不存在或状态未知 - - -**task\_metrics** `_object_` - -任务结果统计。 - -**属性** - -**TOTAL** `_integer_` - -总的任务数。 - -**SUCCEEDED** `_integer_` - -任务状态为成功的任务数。 - -**FAILED** `_integer_` - -任务状态为失败的任务数。 - -**results** `_array of object_` - -任务结果列表，包括图像URL、部分任务执行失败报错信息等。 - -**数据结构** - -``` -{ - "results": [ - { - "url": "" - }, - { - "code": "", - "message": "" - } - ] -} -``` - -**code** `_string_` - -请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**message** `_string_` - -请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**usage** `_object_` - -输出信息统计。只对成功的结果计数。 - -**属性** - -**image\_count** `_integer_` - -模型成功生成图片的数量。计费公式：费用 = 图片数量 × 单价。 - -**request\_id** `_string_` - -请求唯一标识。可用于请求明细溯源和问题排查。 - -## DashScope SDK调用 - -请先确认已安装最新版DashScope SDK：[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 - -DashScope SDK目前已支持Python和Java。 - -SDK与HTTP接口的参数名基本一致，参数结构根据不同语言的SDK封装而定。参数说明可参考[HTTP调用](https://help.aliyun.com/zh/model-studio/text-to-image-api-reference#42703589880ts)。 - -由于图像模型处理时间较长，底层服务采用异步方式提供。SDK在上层进行了封装，支持同步、异步两种调用方式。 - -### Python SDK调用 - -## 同步调用 - -##### **请求示例** - -``` -from http import HTTPStatus -from urllib.parse import urlparse, unquote -from pathlib import PurePosixPath -import requests -import dashscope -from dashscope import ImageSynthesis -import os - -dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' - -prompt = "一棵参天大树" -sketch_image_url = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg" -model = "wanx-sketch-to-image-lite" -task = "image2image" - -print('----sync call, please wait a moment----') -rsp = ImageSynthesis.call(api_key=os.getenv("DASHSCOPE_API_KEY"), - model=model, - prompt=prompt, - n=1, - style='', - size='768*768', - sketch_image_url=sketch_image_url, - task=task) -print('response: %s' % rsp) -if rsp.status_code == HTTPStatus.OK: - print(rsp.output) - # save file to current directory - for result in rsp.output.results: - file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1] - with open('./%s' % file_name, 'wb+') as f: - f.write(requests.get(result.url).content) -else: - print('sync_call Failed, status_code: %s, code: %s, message: %s' % - (rsp.status_code, rsp.code, rsp.message)) -``` - -##### **响应示例** - -``` -{ - "status_code": 200, - "request_id": "4126d9dd-e037-9f32-8d56-6d29ab3f9a06", - "code": null, - "message": "", - "output": { - "task_id": "b476bc4e-35c1-4c4e-a4d9-xxxxxxx", - "task_status": "SUCCEEDED", - "results": [{ - "url": "https://dashscope-result-sh.oss-cn-shanghai.aliyuncs.com/xxxx.png" - }], - "submit_time": "2024-11-01 09:50:56.081", - "scheduled_time": "2024-11-01 09:50:56.104", - "end_time": "2024-11-01 09:51:22.740", - "task_metrics": { - "TOTAL": 1, - "SUCCEEDED": 1, - "FAILED": 0 - } - }, - "usage": { - "image_count": 1 - } -} -``` - -## 异步调用 - -##### **请求示例** - -``` -from http import HTTPStatus -from urllib.parse import urlparse, unquote -from pathlib import PurePosixPath -import requests -import dashscope -from dashscope import ImageSynthesis -import os - -dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' - -prompt = "一棵参天大树" -sketch_image_url = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg" -model = "wanx-sketch-to-image-lite" -task = "image2image" - -# 异步调用 -def async_call(): - print('----create task----') - task_info = create_async_task() - print('----wait task done then save image----') - wait_async_task(task_info) - -# 创建异步任务 -def create_async_task(): - rsp = ImageSynthesis.async_call(api_key=os.getenv("DASHSCOPE_API_KEY"), - model=model, - prompt=prompt, - n=1, - style='', - size='768*768', - sketch_image_url=sketch_image_url, - task=task) - print(rsp) - if rsp.status_code == HTTPStatus.OK: - print(rsp.output) - else: - print('create_async_task Failed, status_code: %s, code: %s, message: %s' % - (rsp.status_code, rsp.code, rsp.message)) - return rsp - -# 等待异步任务结束 -def wait_async_task(task): - rsp = ImageSynthesis.wait(task) - print(rsp) - if rsp.status_code == HTTPStatus.OK: - print(rsp.output.task_status) - # save file to current directory - for result in rsp.output.results: - file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1] - with open('./%s' % file_name, 'wb+') as f: - f.write(requests.get(result.url).content) - else: - print('Failed, status_code: %s, code: %s, message: %s' % - (rsp.status_code, rsp.code, rsp.message)) - -if __name__ == '__main__': - async_call() -``` - -##### **响应示例** - -**1、创建任务的响应示例** - -``` -{ - "status_code": 200, - "request_id": "31b04171-011c-96bd-ac00-f0383b669cc7", - "code": "", - "message": "", - "output": { - "task_id": "4f90cf14-a34e-4eae-xxxxxxxx", - "task_status": "PENDING", - "results": [] - }, - "usage": null -} -``` - -**2、查询任务结果的响应示例** - -``` -{ - "status_code": 200, - "request_id": "d861d3ba-4b29-9491-abad-266ef4fb2f08", - "code": null, - "message": "", - "output": { - "task_id": "4f90cf14-a34e-4eae-xxxxxxxx", - "task_status": "SUCCEEDED", - "results": [{ - "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" - }], - "submit_time": "2024-10-31 20:40:35.631", - "scheduled_time": "2024-10-31 20:40:35.684", - "end_time": "2024-10-31 20:41:02.700", - "task_metrics": { - "TOTAL": 1, - "SUCCEEDED": 1, - "FAILED": 0 - } - }, - "usage": { - "image_count": 1 - } -} -``` - -### Java SDK调用 - -## 同步调用 - -##### 请求示例 - -``` -// Copyright (c) Alibaba, Inc. and its affiliates. - -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis; -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam; -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.JsonUtils; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public void syncCall() { - String prompt = "一棵参天大树"; - String sketchImageUrl = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg"; - String model = "wanx-sketch-to-image-lite"; - ImageSynthesisParam param = ImageSynthesisParam.builder() - .model(model) - .prompt(prompt) - .n(1) - .size("768*768") - .sketchImageUrl(sketchImageUrl) - .style("") - .build(); - - String task = "image2image"; - ImageSynthesis imageSynthesis = new ImageSynthesis(task); - ImageSynthesisResult result = null; - try { - System.out.println("---sync call, please wait a moment----"); - result = imageSynthesis.call(param); - } catch (ApiException | NoApiKeyException e){ - throw new RuntimeException(e.getMessage()); - } - System.out.println(JsonUtils.toJson(result)); - } - - public static void main(String[] args){ - Main text2Image = new Main(); - text2Image.syncCall(); - } - -} -``` - -##### **响应示例** - -``` -{ - "request_id": "150edcda-05d5-9ffe-8803-84626d1db623", - "output": { - "task_id": "f2098ff0-146e-404c-bb25-xxxxxxxx", - "task_status": "SUCCEEDED", - "results": [{ - "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" - }], - "task_metrics": { - "TOTAL": 1, - "SUCCEEDED": 1, - "FAILED": 0 - } - }, - "usage": { - "image_count": 1 - } -} -``` - -## 异步调用 - -##### **请求示例** - -``` -// Copyright (c) Alibaba, Inc. and its affiliates. - -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis; -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam; -import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.JsonUtils; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public void asyncCall() { - System.out.println("---create task----"); - String taskId = this.createAsyncTask(); - System.out.println("---wait task done then return image url----"); - this.waitAsyncTask(taskId); - } - - /** - * 创建异步任务 - * @return taskId - */ - public String createAsyncTask() { - String prompt = "一棵参天大树"; - String sketchImageUrl = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg"; - String model = "wanx-sketch-to-image-lite"; - ImageSynthesisParam param = ImageSynthesisParam.builder() - .model(model) - .prompt(prompt) - .n(1) - .size("768*768") - .sketchImageUrl(sketchImageUrl) - .style("") - .build(); - - String task = "image2image"; - ImageSynthesis imageSynthesis = new ImageSynthesis(task); - ImageSynthesisResult result = null; - try { - result = imageSynthesis.asyncCall(param); - } catch (Exception e){ - throw new RuntimeException(e.getMessage()); - } - String taskId = result.getOutput().getTaskId(); - System.out.println("taskId=" + taskId); - return taskId; - } - - /** - * 等待异步任务结束 - * @param taskId 任务id - * */ - public void waitAsyncTask(String taskId) { - ImageSynthesis imageSynthesis = new ImageSynthesis(); - ImageSynthesisResult result = null; - try { - // If you have set the DASHSCOPE_API_KEY in the system environment variable, the apiKey can be null. - result = imageSynthesis.wait(taskId, null); - } catch (ApiException | NoApiKeyException e){ - throw new RuntimeException(e.getMessage()); - } - - System.out.println(JsonUtils.toJson(result.getOutput())); - System.out.println(JsonUtils.toJson(result.getUsage())); - } - - public static void main(String[] args){ - Main text2Image = new Main(); - text2Image.asyncCall(); - } - -} -``` - -##### **响应示例** - -**1、步骤1：创建任务获取任务ID的响应示例** - -``` -{ - "request_id": "5dbf9dc5-4f4c-9605-85ea-542f97709ba8", - "output": { - "task_id": "7277e20e-aa01-4709-xxxxxxxx", - "task_status": "PENDING" - } -} -``` - -**2、步骤2：根据任务ID查询结果的响应示例** - -``` -{ - "request_id": "c44213ba-7aa3-91e4-97c1-c527ade82597", - "output": { - "task_id": "7277e20e-aa01-4709-xxxxxxxx", - "task_status": "SUCCEEDED", - "results": [{ - "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" - }], - "task_metrics": { - "TOTAL": 1, - "SUCCEEDED": 1, - "FAILED": 0 - } - }, - "usage": { - "image_count": 1 - } -} -``` - -## 错误码 - -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -此API还有特定状态码，具体如下所示。 - -**HTTP状态码** - -**接口错误码（code）** - -**接口错误信息（message）** - -**含义说明** - -400 - -InvalidParameter.DataInspection - -Unable to download the media resource during the data inspection process. - -输入图片无法下载，请检查URL地址是否正确且可访问。 - -400 - -InvalidParameter - -Value error, format of image {url} is not valid : payload.input.sketch - -输入图片格式不合法，请确认图片格式为JPG、JPEG、PNG、TIFF或WEBP。 - -## **常见问题** - -### 模型计费及限流 - -**免费额度** - -- 额度说明：免费额度是指模型成功生成的输出图片数量。输入图片及模型处理失败的情况不占用免费额度。 - -- 领取方式：开通阿里云百炼大模型服务后自动发放，有效期90天。 - -- 使用账号：阿里云主账号与其RAM子账号共享免费额度。 - -- 更多详情请参见[新人免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota)。 - - -**限时免费** - -- 当计费为限时免费时，表示该模型处于公测阶段，免费额度用尽后不可使用。 - - -**计费说明** - -- 当计费有明确单价时，如0.2元/秒，表示该模型已商业化，免费额度用尽或过期后需付费使用。 - -- 计费项：只对模型成功生成的输出图片进行收费，其余情况暂不计费。 - -- 付费方式：由阿里云主账号统一付费。RAM子账号不能独立计量计费，必须由所属的主账号付费。如果您需要查询账单信息，请前往阿里云控制台[账单概览](https://billing-cost.console.aliyun.com/finance/month-bill/account)。 - -- 充值途径：您可以在阿里云控制台[费用与成本](https://billing-cost.console.aliyun.com/home?spm=a2c4g.11186623.0.0.2d543048F4KRQP)页面进行充值。 - -- 模型调用情况：您可以前往阿里云百炼的[模型观测](https://bailian.console.aliyun.com/#/model-telemetry)查看模型调用量及调用次数。 - -- 更多计费问题请参见[计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 - - -**限流** - -- 限流说明：阿里云主账号与其RAM子账号共享限流限制。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md deleted file mode 100644 index 4c3b266a..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md +++ /dev/null @@ -1,1185 +0,0 @@ -# 上传本地文件获取临时URL - -在调用多模态、图像、视频或音频模型时，通常需要传入文件的 URL。为此，阿里云百炼提供了**免费**临时存储空间，您可将本地文件上传至该空间并获得 URL（**有效期为 48 小时**）。 - -## **使用限制** - -- **文件与模型绑定**：文件上传时必须指定模型名称，且该模型须与后续调用的**模型一致**，不同模型无法共享文件。 - -- **文件大小限制**：接口上传文件大小不得超过**1GB**，超出限制将导致上传失败。此外，不同模型对输入文件大小有不同限制，超出限制将导致模型调用失败。 - -- **文件与主账号绑定**：文件上传与模型调用所使用的 API Key 必须**属于同一个阿里云主账号**，且上传的文件仅限该主账号及其对应模型使用，无法被其他主账号或其他模型共享。 - -- **文件有效期限制**：文件上传后**有效期48小时**，超时后文件将被自动清理，请确保在有效期内完成模型调用。 - -- **文件使用限制**：文件一旦上传，不可查询、修改或下载，仅能**通过URL参数在模型调用时使用**。 - -- **文件上传限流**：文件上传凭证接口的调用限流按照“阿里云主账号+模型”维度为**100QPS**，**超出限流将导致请求失败**。 - - -**重要** - -- 临时 URL 有效期48小时，过期后无法使用，**请勿用于生产环境。** - -- 文件上传凭证接口限流为 100 QPS 且不支持扩容，**请勿用于生产环境、高并发及压测场景。** - -- 生产环境建议使用[阿里云OSS](https://help.aliyun.com/zh/oss/user-guide/what-is-oss) 等稳定存储，确保文件长期可用并规避限流问题。 - - -## **使用方式** - -1. 获取文件 URL：请先通过[步骤一](#a363e01e741gu)上传文件（图片、视频或音频），获取以`oss://` 为前缀的临时 URL。 - -2. 调用模型：**请务必根据**[**步骤二**](#1c60469225ufa)**使用临时 URL 进行调用**。该步骤不能跳过，否则接口将报错。 - - -## **步骤一：获取临时URL** - -### **方式一：通过代码上传文件** - -本文提供 Python 和 Java 示例代码，简化上传文件操作。您只需**指定模型和待上传的文件**，即可获取临时URL。 - -**前提条件** - -在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -#### **示例代码** - -## Python - -**环境配置** - -- 推荐使用Python 3.8及以上版本。 - -- 请安装必要的依赖包。 - - -``` -pip install -U requests -``` - -**输入参数** - -- api\_key：阿里云百炼API KEY。 - -- model\_name：指定文件将要用于哪个模型，如`qwen-vl-plus`。 - -- file\_path：待上传的本地文件路径（图片、视频等）。 - - -``` -import os -import requests -from pathlib import Path -from datetime import datetime, timedelta - -def get_upload_policy(api_key, model_name): - """获取文件上传凭证""" - url = "https://dashscope.aliyuncs.com/api/v1/uploads" - headers = { - "Authorization": f"Bearer {api_key}", - "Content-Type": "application/json" - } - params = { - "action": "getPolicy", - "model": model_name - } - - response = requests.get(url, headers=headers, params=params) - if response.status_code != 200: - raise Exception(f"Failed to get upload policy: {response.text}") - - return response.json()['data'] - -def upload_file_to_oss(policy_data, file_path): - """将文件上传到临时存储OSS""" - file_name = Path(file_path).name - key = f"{policy_data['upload_dir']}/{file_name}" - - with open(file_path, 'rb') as file: - files = { - 'OSSAccessKeyId': (None, policy_data['oss_access_key_id']), - 'Signature': (None, policy_data['signature']), - 'policy': (None, policy_data['policy']), - 'x-oss-object-acl': (None, policy_data['x_oss_object_acl']), - 'x-oss-forbid-overwrite': (None, policy_data['x_oss_forbid_overwrite']), - 'key': (None, key), - 'success_action_status': (None, '200'), - 'file': (file_name, file) - } - - response = requests.post(policy_data['upload_host'], files=files) - if response.status_code != 200: - raise Exception(f"Failed to upload file: {response.text}") - - return f"oss://{key}" - -def upload_file_and_get_url(api_key, model_name, file_path): - """上传文件并获取URL""" - # 1. 获取上传凭证，上传凭证接口有限流，超出限流将导致请求失败 - policy_data = get_upload_policy(api_key, model_name) - # 2. 上传文件到OSS - oss_url = upload_file_to_oss(policy_data, file_path) - - return oss_url - -# 使用示例 -if __name__ == "__main__": - # 从环境变量中获取API Key 或者 在代码中设置 api_key = "your_api_key" - api_key = os.getenv("DASHSCOPE_API_KEY") - if not api_key: - raise Exception("请设置DASHSCOPE_API_KEY环境变量") - - # 设置model名称 - model_name="qwen-vl-plus" - - # 待上传的文件路径 - file_path = "/tmp/cat.png" # 替换为实际文件路径 - - try: - public_url = upload_file_and_get_url(api_key, model_name, file_path) - expire_time = datetime.now() + timedelta(hours=48) - print(f"文件上传成功，有效期为48小时，过期时间: {expire_time.strftime('%Y-%m-%d %H:%M:%S')}") - print(f"临时URL: {public_url}") - print("注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call") - - except Exception as e: - print(f"Error: {str(e)}") -``` - -**输出示例** - -``` -文件上传成功，有效期为48小时，过期时间: 2024-07-18 17:36:15 -临时URL: oss://dashscope-instant/xxx/2024-07-18/xxx/cat.png -注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call -``` - -**重要** - -获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 - -## Java - -**环境配置** - -- 推荐使用JDK 1.8及以上版本。 - -- 请在Maven项目的`pom.xml`文件中导入以下依赖。 - - -``` - - - org.json - json - 20230618 - - - org.apache.httpcomponents - httpclient - 4.5.13 - - - org.apache.httpcomponents - httpmime - 4.5.13 - - -``` - -**输入参数** - -- apiKey：阿里云百炼API KEY。 - -- modelName：指定文件将要用于哪个模型，如`qwen-vl-plus`。 - -- filePath：待上传的本地文件路径（图片、视频等）。 - - -``` -import org.apache.http.client.methods.CloseableHttpResponse; -import org.apache.http.client.methods.HttpGet; -import org.apache.http.client.methods.HttpPost; -import org.apache.http.entity.mime.MultipartEntityBuilder; -import org.apache.http.entity.ContentType; -import org.apache.http.impl.client.CloseableHttpClient; -import org.apache.http.impl.client.HttpClients; -import org.apache.http.HttpStatus; -import org.apache.http.util.EntityUtils; -import org.json.JSONObject; -import java.io.File; -import java.io.IOException; -import java.nio.file.Files; -import java.nio.file.Path; -import java.nio.file.Paths; -import java.time.LocalDateTime; -import java.time.format.DateTimeFormatter; - -public class PublicUrlHandler { - - private static final String API_URL = "https://dashscope.aliyuncs.com/api/v1/uploads"; - - public static JSONObject getUploadPolicy(String apiKey, String modelName) throws IOException { - try (CloseableHttpClient httpClient = HttpClients.createDefault()) { - HttpGet httpGet = new HttpGet(API_URL); - httpGet.addHeader("Authorization", "Bearer " + apiKey); - httpGet.addHeader("Content-Type", "application/json"); - - String query = String.format("action=getPolicy&model=%s", modelName); - httpGet.setURI(httpGet.getURI().resolve(httpGet.getURI() + "?" + query)); - - try (CloseableHttpResponse response = httpClient.execute(httpGet)) { - if (response.getStatusLine().getStatusCode() != 200) { - throw new IOException("Failed to get upload policy: " + - EntityUtils.toString(response.getEntity())); - } - String responseBody = EntityUtils.toString(response.getEntity()); - return new JSONObject(responseBody).getJSONObject("data"); - } - } - } - - public static String uploadFileToOSS(JSONObject policyData, String filePath) throws IOException { - Path path = Paths.get(filePath); - String fileName = path.getFileName().toString(); - String key = policyData.getString("upload_dir") + "/" + fileName; - - HttpPost httpPost = new HttpPost(policyData.getString("upload_host")); - MultipartEntityBuilder builder = MultipartEntityBuilder.create(); - - builder.addTextBody("OSSAccessKeyId", policyData.getString("oss_access_key_id")); - builder.addTextBody("Signature", policyData.getString("signature")); - builder.addTextBody("policy", policyData.getString("policy")); - builder.addTextBody("x-oss-object-acl", policyData.getString("x_oss_object_acl")); - builder.addTextBody("x-oss-forbid-overwrite", policyData.getString("x_oss_forbid_overwrite")); - builder.addTextBody("key", key); - builder.addTextBody("success_action_status", "200"); - byte[] fileContent = Files.readAllBytes(path); - builder.addBinaryBody("file", fileContent, ContentType.DEFAULT_BINARY, fileName); - - httpPost.setEntity(builder.build()); - - try (CloseableHttpClient httpClient = HttpClients.createDefault(); - CloseableHttpResponse response = httpClient.execute(httpPost)) { - if (response.getStatusLine().getStatusCode() != HttpStatus.SC_OK) { - throw new IOException("Failed to upload file: " + - EntityUtils.toString(response.getEntity())); - } - return "oss://" + key; - } - } - - public static String uploadFileAndGetUrl(String apiKey, String modelName, String filePath) throws IOException { - JSONObject policyData = getUploadPolicy(apiKey, modelName); - return uploadFileToOSS(policyData, filePath); - } - - public static void main(String[] args) { - // 获取环境变量中的API密钥 - String apiKey = System.getenv("DASHSCOPE_API_KEY"); - if (apiKey == null || apiKey.isEmpty()) { - System.err.println("请设置DASHSCOPE_API_KEY环境变量"); - System.exit(1); - } - // 模型名称 - String modelName = "qwen-vl-plus"; - //替换为实际文件路径 - String filePath = "src/main/resources/tmp/cat.png"; - - try { - // 检查文件是否存在 - File file = new File(filePath); - if (!file.exists()) { - System.err.println("文件不存在: " + filePath); - System.exit(1); - } - - String publicUrl = uploadFileAndGetUrl(apiKey, modelName, filePath); - LocalDateTime expireTime = LocalDateTime.now().plusHours(48); - DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss"); - - System.out.println("文件上传成功，有效期为48小时，过期时间: " + expireTime.format(formatter)); - System.out.println("临时URL: " + publicUrl); - System.out.println("注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call"); - } catch (IOException e) { - System.err.println("Error: " + e.getMessage()); - } - } -} -``` - -**输出示例** - -``` -文件上传成功，有效期为48小时，过期时间: 2024-07-18 17:36:15 -临时URL: oss://dashscope-instant/xxx/2024-07-18/xxx/cat.png -注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call -``` - -**重要** - -获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 - -### **方式二：通过命令行工具上传文件** - -对于熟悉命令行的开发者，可使用DashScope提供的命令行工具来上传文件。**执行命令后，即可获取临时URL**。 - -#### **前提条件** - -1. 环境准备：推荐使用 Python 3.8 及以上版本。 - -2. 获取API-KEY：在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 - -3. 安装SDK：请确保[DashScope Python SDK](https://help.aliyun.com/zh/model-studio/install-sdk) 版本不低于 `1.24.0`。执行以下命令进行安装或升级： - - -``` -pip install -U dashscope -``` - -#### **方法1：使用环境变量（推荐）** - -此方法更安全，可以避免API-KEY在命令历史或脚本中明文暴露。 - -前提条件：请确保已[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -执行上传命令： - -``` -dashscope oss.upload --model qwen-vl-plus --file cat.png -``` - -输出示例： - -``` -Start oss.upload: model=qwen-vl-plus, file=cat.png, api_key=None -Uploaded oss url: oss://dashscope-instant/xxxx/2025-08-01/xxxx/cat.png -``` - -**重要** - -获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 - -#### **方法2：通过命令行参数指定API-KEY（临时使用）** - -执行上传命令： - -``` -dashscope oss.upload --model qwen-vl-plus --file cat.png --api_key sk-xxxxxxx -``` - -输出示例： - -``` -Start oss.upload: model=qwen-vl-plus, file=cat.png, api_key=sk-xxxxxxx -Uploaded oss url: oss://dashscope-instant/xxx/2025-08-01/xxx/cat.png -``` - -**重要** - -获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 - -#### **命令行参数说明** - -**参数** - -**是否必须** - -**说明** - -**示例** - -oss.upload - -是 - -dashscope的子命令，用于执行文件上传操作。 - -oss.upload - -\--model - -是 - -指定文件将要用于哪个模型。 - -qwen-vl-plus - -\--file - -是 - -本地文件的路径。可以是相对路径或绝对路径。 - -cat.png，/data/img.jpg - -\--api\_key - -否 - -阿里云百炼API-KEY。如已配置环境变量，无需填写此参数。 - -sk-xxxx - -## **步骤二：使用临时URL调用模型** - -#### **使用限制** - -- **文件格式**：临时URL须通过上述方式生成，且以 `oss://`为前缀的URL字符串。 - -- **文件未过期**：文件URL仍在上传后的48小时有效期内。 - -- **模型一致**：模型调用所使用的模型必须与文件上传时指定的模型完全一致。 - -- **账号一致**：模型调用的API KEY必须与文件上传时使用的API KEY同属一个阿里云主账号。 - - -#### **方式一：通过HTTP调用** - -通过curl、Postman或任何其他HTTP客户端直接调用API，则**必须遵循以下规则**： - -**重要** - -- 使用临时URL，**必须**在请求的**Header**中添加参数：`**X-DashScope-OssResourceResolve: enable**`。 - -- 若缺失此Header，系统将无法解析`oss://`链接，请求将失败，报错信息请参考[错误码](#3b9b15a6a8qkl)。 - - -## **请求示例** - -本示例为调用 qwen-vl-plus 模型识别图片内容。 - -**说明** - -请将 `oss://...`替换为真实的临时 URL，否则请求将失败。 - -``` -curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" \ --H 'Content-Type: application/json' \ --H 'X-DashScope-OssResourceResolve: enable' \ --d '{ - "model": "qwen-vl-plus", - "messages": [{ - "role": "user", - "content": - [{"type": "text","text": "这是什么"}, - {"type": "image_url","image_url": {"url": "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"}}] - }] -}' -``` - -## 响应示例 - -``` -{ - "choices": [ - { - "message": { - "content": "这是一张描绘一只白色猫咪在草地上奔跑的图片。这只猫有蓝色的眼睛，看起来非常可爱和活泼。背景是模糊化的自然景色，强调了主体——那只向前冲跑的小猫。这种摄影技巧称为浅景深（或大光圈效果），它使得前景中的小猫变得清晰而锐利，同时使背景虚化以突出主题并营造出一种梦幻般的效果。整体上这张照片给人一种轻松愉快的感觉，并且很好地捕捉到了动物的行为瞬间。", - "role": "assistant" - }, - "finish_reason": "stop", - "index": 0, - "logprobs": null - } - ], - "object": "chat.completion", - "usage": { - "prompt_tokens": 1253, - "completion_tokens": 104, - "total_tokens": 1357 - }, - "created": 1739349052, - "system_fingerprint": null, - "model": "qwen-vl-plus", - "id": "chatcmpl-cfc4f2aa-22a8-9a94-8243-44c5bd9899bc" -} -``` - -## 上传的本地图片示例 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5231249371/p915804.png) - -#### **方式二：通过DashScope SDK调用** - -您也可以使用阿里云百炼提供的 Python 或 Java SDK。 - -- **直接传入 URL**：调用模型 SDK 时，直接将以`oss://`为前缀的URL字符串作为文件参数传入。 - -- **无需关心 Header**：SDK 会自动添加必需的请求头，无需额外操作。 - - -**注意**：并非所有模型都支持 SDK 调用，请以模型 API 文档为准。 - -> 不支持 OpenAI SDK。 - -## Python - -**前提条件** - -请[安装DashScope Python SDK](https://help.aliyun.com/zh/model-studio/install-sdk)，且DashScope Python SDK版本号 >=`1.24.0`。 - -**示例代码** - -本示例为调用 qwen-vl-plus 模型识别图片内容。此代码示例仅适用于 qwen-vl 和 omni 系列模型。 - -## 请求示例 - -**说明** - -请将 image 参数中的 `oss://...`替换为真实的临时 URL，否则请求将失败。 - -``` -import os -import dashscope - -messages = [ - { - "role": "system", - "content": [{"text": "You are a helpful assistant."}] - }, - { - "role": "user", - "content": [ - {"image": "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"}, - {"text": "这是什么"}] - }] - -# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" -api_key = os.getenv('DASHSCOPE_API_KEY') - -response = dashscope.MultiModalConversation.call( - api_key=api_key, - model='qwen-vl-plus', - messages=messages -) - -print(response) -``` - -## 响应示例 - -``` -{ - "status_code": 200, - "request_id": "ccd9dcfb-98f0-92bc-xxxxxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "text": "这是一张一只猫在草地上奔跑的照片。猫的毛色主要是白色，带有浅棕色的斑点，眼睛是蓝色的，显得非常可爱。背景是一个模糊的绿色草地和一些树木，阳光照射下来，给整个画面增添了一种温暖的感觉。猫的姿态显示出它正在快速移动，可能是在追逐什么或只是在享受户外活动的乐趣。整体来看，这是一幅充满活力和生机的图片。" - } - ] - } - } - ] - }, - "usage": { - "input_tokens": 1112, - "output_tokens": 91, - "input_tokens_details": { - "text_tokens": 21, - "image_tokens": 1091 - }, - "prompt_tokens_details": { - "cached_tokens": 0 - }, - "total_tokens": 1203, - "output_tokens_details": { - "text_tokens": 91 - }, - "image_tokens": 1091 - } -} -``` - -## Java - -**前提条件** - -请[安装DashScope Java SDK](https://help.aliyun.com/zh/model-studio/install-sdk)，且DashScope Java SDK版本号 >= `2.21.0`。 - -**示例代码** - -本示例为调用 qwen-vl-plus 模型识别图片内容。此代码示例仅适用于 qwen-vl 和 omni 系列模型。 - -## 请求示例 - -**说明** - -请将 `oss://...`替换为真实的临时 URL，否则请求将失败。 - -``` -import com.alibaba.dashscope.aigc.multimodalconversation.*; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.util.Arrays; - -public class MultiModalConversationUsage { - - private static final String modelName = "qwen-vl-plus"; - - // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" - public static String apiKey = System.getenv("DASHSCOPE_API_KEY"); - - public static void simpleMultiModalConversationCall() throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - MultiModalMessageItemText systemText = new MultiModalMessageItemText("You are a helpful assistant."); - MultiModalConversationMessage systemMessage = MultiModalConversationMessage.builder() - .role(Role.SYSTEM.getValue()).content(Arrays.asList(systemText)).build(); - MultiModalMessageItemImage userImage = new MultiModalMessageItemImage( - "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"); - MultiModalMessageItemText userText = new MultiModalMessageItemText("这是什么"); - MultiModalConversationMessage userMessage = - MultiModalConversationMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList(userImage, userText)).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - .model(MultiModalConversationUsage.modelName) - .apiKey(apiKey) - .message(systemMessage) - .vlHighResolutionImages(true) - .vlEnableImageHwOutput(true) -// .incrementalOutput(true) - .message(userMessage).build(); - MultiModalConversationResult result = conv.call(param); - System.out.print(JsonUtils.toJson(result)); - - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException /*| IOException*/ e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } - -} -``` - -## 响应示例 - -``` -{ - "requestId": "b6d60f91-4a7f-9257-xxxxxx", - "usage": { - "input_tokens": 1112, - "output_tokens": 91, - "total_tokens": 1203, - "image_tokens": 1091, - "input_tokens_details": { - "text_tokens": 21, - "image_tokens": 1091 - }, - "output_tokens_details": { - "text_tokens": 91 - } - }, - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "text": "这是一张一只猫在草地上奔跑的照片。猫的毛色主要是白色，带有浅棕色的斑点，眼睛是蓝色的，显得非常可爱。背景是一个模糊的绿色草地和一些树木，阳光照射下来，给整个画面增添了一种温暖的感觉。猫的姿态显示出它正在快速移动，可能是在追逐什么或只是在享受户外活动的乐趣。整体来看，这是一幅充满活力和生机的图片。" - }, - { - "image_hw": [ - [ - "924", - "924" - ] - ] - } - ] - } - } - ] - } -} -``` - -## 附接口说明 - -在上述[获取临时URL](#a363e01e741gu)的两种方式中，代码调用和命令行工具已集成以下三个步骤，简化文件上传操作。以下是各步骤的接口说明。 - -#### **步骤1：获取文件上传凭证** - -##### **前提条件** - -您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -##### **请求接口** - -``` -GET https://dashscope.aliyuncs.com/api/v1/uploads -``` - -**重要** - -文件上传凭证接口限流为 100 QPS（按“阿里云主账号+模型”维度），且临时存储不可扩容。生产环境或高并发场景请使用[阿里云OSS](https://help.aliyun.com/zh/oss/user-guide/what-is-oss)等存储服务。 - -##### **入参描述** - -**传参方式** - -**字段** - -**类型** - -**必选** - -**描述** - -**示例值** - -Header - -Content-Type - -_string_ - -是 - -请求类型：application/json 。 - -application/json - -Authorization - -_string_ - -是 - -阿里云百炼API Key，例如：Bearer sk-xxx。 - -Bearer sk-xxx - -Params - -action - -_string_ - -是 - -操作类型，当前场景为`getPolicy`。 - -getPolicy - -model - -_string_ - -是 - -需要调用的模型名称。 - -qwen-vl-plus - -##### **出参描述** - -**字段** - -**类型** - -**描述** - -**示例值** - -request\_id - -_string_ - -本次请求的系统唯一码。 - -7574ee8f-...-11c33ab46e51 - -data - -_object_ - -\- - -\- - -data.policy - -_string_ - -上传凭证。 - -eyJl...1ZSJ9XX0= - -data.signature - -_string_ - -上传凭证的签名。 - -g5K...d40= - -data.upload\_dir - -_string_ - -上传文件的目录。 - -dashscope-instant/xxx/2024-07-18/xxxx - -data.upload\_host - -_string_ - -上传的host地址。 - -https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com - -data.expire\_in\_seconds - -_string_ - -凭证有效期（单位：秒）。 - -**说明** - -过期后，重新调用本接口获取新的凭证。 - -300 - -data.max\_file\_size\_mb - -_string_ - -本次允许上传的最大文件的大小（单位：MB）。 - -该值与需要访问的模型相关。 - -100 - -data.capacity\_limit\_mb - -_string_ - -同一个主账号每天上传容量限制（单位：MB）。 - -999999999 - -data.oss\_access\_key\_id - -_string_ - -用于上传的access key。 - -LTAxxx - -data.x\_oss\_object\_acl - -_string_ - -上传文件的访问权限，`private`表示私有。 - -private - -data.x\_oss\_forbid\_overwrite - -_string_ - -文件同名时是否可以覆盖，`true`表示不可覆盖。 - -true - -##### **请求示例** - -``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/uploads?action=getPolicy&model=qwen-vl-plus' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' -``` - -**说明** - -若未配置阿里云百炼API Key到环境变量，请将`$DASHSCOPE_API_KEY`替换为实际API Key，例如：`--header "Authorization: Bearer sk-xxx"`。 - -#### **响应示例** - -``` -{ - "request_id": "52f4383a-c67d-9f8c-xxxxxx", - "data": { - "policy": "eyJl...1ZSJ=", - "signature": "eWy...=", - "upload_dir": "dashscope-instant/xxx/2024-07-18/xxx", - "upload_host": "https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com", - "expire_in_seconds": 300, - "max_file_size_mb": 100, - "capacity_limit_mb": 999999999, - "oss_access_key_id": "LTA...", - "x_oss_object_acl": "private", - "x_oss_forbid_overwrite": "true" - } -} -``` - -#### **步骤2：上传文件至临时存储空间** - -#### **前提条件** - -- 已获取文件上传凭证。 - -- 确保文件上传凭证在有效期内，若凭证过期，请重新调用步骤1的接口获取新的凭证。 - - > 查看文件上传凭证有效期：步骤1的输出参数`data.expire_in_seconds`为凭证有效期，单位为秒。 - - -#### **请求接口** - -``` -POST {data.upload_host} -``` - -**说明** - -请将{data.upload\_host}替换为步骤1的输出参数`data.upload_host`对应的值。 - -#### **入参描述** - -**传参方式** - -**字段** - -**类型** - -**必选** - -**描述** - -**示例值** - -Header - -Content-Type - -_string_ - -否 - -提交表单必须为`multipart/form-data`。 - -在提交表单时，Content-Type会以`multipart/form-data;boundary=xxxxxx`的形式展示。 - -> boundary 是自动生成的随机字符串，无需手动指定。若使用 SDK 拼接表单，SDK 也会自动生成该随机值。 - -multipart/form-data; boundary=9431149156168 - -form-data - -OSSAccessKeyId - -_text_ - -是 - -文件上传凭证接口的输出参数 `data.oss_access_key_id` 的值。 - -LTAm5xxx - -policy - -_text_ - -是 - -文件上传凭证接口的输出参数 `data.policy` 的值。 - -g5K...d40= - -Signature - -_text_ - -是 - -文件上传凭证接口的输出参数 `data.signature` 的值。 - -Sm/tv7DcZuTZftFVvt5yOoSETsc= - -key - -_text_ - -是 - -文件上传凭证接口的输出参数 `data.upload_dir` 的值拼接上`/_文件名_`。 - -例如，`upload_dir` 为 `dashscope-instant/xxx/2024-07-18/xxx`，需要上传的文件名为 `cat.png`，拼接后的完整路径为： - -`dashscope-instant/xxx/2024-07-18/xxx/cat.png` - -x-oss-object-acl - -_text_ - -是 - -文件上传凭证接口的输出参数 `data.x_oss_object_acl` 的值。 - -private - -x-oss-forbid-overwrite - -_text_ - -是 - -文件上传凭证接口的输出参数中`data.x_oss_forbid_overwrite` 的值。 - -true - -success\_action\_status - -_text_ - -否 - -通常取值为 200，上传完成后接口返回 HTTP code 200，表示操作成功。 - -200 - -file - -_text_ - -是 - -文件或文本内容。 - -**说明** - -- 一次只支持上传一个文件。 - -- file必须为最后一个表单域，除file以外的其他表单域并无顺序要求。 - - -例如，待上传文件`cat.png`在Linux系统中的存储路径为`/tmp`，则此处应为`file=@"/tmp/cat.png"`。 - -#### **出参描述** - -调用成功时，本接口无任何参数输出。 - -#### **请求示例** - -``` -curl --location 'https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com' \ ---form 'OSSAccessKeyId="LTAm5xxx"' \ ---form 'Signature="Sm/tv7DcZuTZftFVvt5yOoSETsc="' \ ---form 'policy="eyJleHBpcmF0aW9 ... ... ... dHJ1ZSJ9XX0="' \ ---form 'x-oss-object-acl="private"' \ ---form 'x-oss-forbid-overwrite="true"' \ ---form 'key="dashscope-instant/xxx/2024-07-18/xxx/cat.png"' \ ---form 'success_action_status="200"' \ ---form 'file=@"/tmp/cat.png"' -``` - -#### **步骤3：生成文件URL** - -文件URL拼接逻辑：`**oss://**` + `**key**` （步骤2的入参`key`）。该URL有效期为 48 小时。 - -``` -oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png -``` - -## **错误码** - -如果接口调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -本文的API还有特定状态码，具体如下所示。 - -**HTTP状态码** - -**接口错误码（code）** - -**接口错误信息（message）** - -**含义说明** - -400 - -invalid\_parameter\_error - -InternalError.Algo.InvalidParameter: The provided URL does not appear to be valid. Ensure it is correctly formatted. - -无效URL，请检查URL是否填写正确。 - -> 若使用临时文件URL，需确保请求的 Header 中添加了参数 `X-DashScope-OssResourceResolve: enable`。 - -400 - -InvalidParameter.DataInspection - -The media format is not supported or incorrect for the data inspection. - -可能的原因有： - -- 请求Header 缺少必要参数，请设置 `X-DashScope-OssResourceResolve: enable`**。** - -- 上传的图片格式不符合模型要求，更多信息请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - - -403 - -AccessDenied - -Invalid according to Policy: Policy expired. - -文件上传凭证已经过期。 - -请重新调用[文件上传凭证接口](#32db94982cllx)生成新凭证。 - -429 - -Throttling.RateQuota - -Requests rate limit exceeded, please try again later. - -调用频次触发限流。 - -[文件上传凭证接口](#32db94982cllx)限流为 100 QPS（按阿里云主账号 + 模型维度）。触发限流后，建议降低请求频率，或迁移至 OSS 等自有存储服务以规避限制。 - -## **常见问题** - -#### **Q：使用** `**oss://**` **前缀的 URL 调用时报错，该如何处理？** - -A：请按以下步骤排查： - -1. **检查请求头（Header）**： - 若您通过 HTTP（如 Postman、curl）直接调用，**必须在** `**Header**` **中添加参数** `**X-DashScope-OssResourceResolve: enable**`。未添加该参数会导致服务端无法识别 OSS 内部协议。关于请求头配置，请参见[通过HTTP调用](#d6a1cb0f01h5k)。 - -2. **检查 URL 有效性**： - `oss://` 链接为临时 URL，请确保该链接是48小时内生成的。如果链接已过期，请重新上传文件获取新的 URL。 - - -#### **Q：文件上传与模型调用使用的API KEY可以不一样吗？** - -A：文件存储和访问权限基于阿里云主账号管理，API Key 仅为主账号的访问凭证。 - -因此，同一阿里云主账号下的不同 API Key 可正常使用，不同主账号的 API Key因账号隔离，模型调用无法跨账号读取文件。 - -请确保文件上传与模型调用使用的 API Key 属于同一阿里云主账号。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md deleted file mode 100644 index 3bcbc5a9..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md +++ /dev/null @@ -1,76 +0,0 @@ -# 实时多模态交互流程 - -本文介绍实时多模态服务端和客户端的交互流程。 - -## VAD 模式 - -将[客户端事件](https://help.aliyun.com/zh/model-studio/client-events#af43722339yva)事件的`session.turn_detection` 设为`"server_vad"`以启用 VAD 模式。在 VAD 模式下，服务端对传入的音频进行语音活动检测，并在检测到作出响应。此模式适用于客户端到服务器始终发送音频的情况，也是当前的默认模式。 - -![server\_vad](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0520773571/p991064.svg) - -- 服务端在检测到语音开始时发送`input_audio_buffer.speech_started` 事件。 - -- 客户端随时可以选择通过发送 `input_audio_buffer.append` 事件将音频追加到缓冲区。 - -- 服务端在检测到语音结束时发送`input_audio_buffer.speech_stopped`事件。 - -- 服务端通过发送 `input_audio_buffer.committed` 事件来提交输入音频缓冲区。 - -- 服务端发送 `conversation.item.created` 事件，其中包含从音频缓冲区创建的用户消息项。 - - -### **工具调用流程** - -在 VAD 模式下，当服务端生成的响应需要调用工具时，遵循以下交互流程： - -![image.svg](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3014236771/p1067613.svg) - -- 服务端在检测到语音结束并生成响应时，识别到需要调用工具。 - -- 服务端发送 `response.function_call_arguments.delta` 事件，包含工具调用参数的增量数据。 - -- 服务端发送 `response.function_call_arguments.done` 事件，表示工具调用参数传递完成。 - -- 客户端执行工具调用并获取结果。 - -- 客户端通过 `conversation.item.create` 事件发送工具调用结果。 - -- 服务端自动基于工具调用结果生成响应。 - - -## **Manual 模式** - -将[客户端事件](https://help.aliyun.com/zh/model-studio/client-events#af43722339yva)事件的`session.turn_detection` 设为 null 以启用 Manual 模式。此模式下，客户端通过显式发送`input_audio_buffer.commit` 和`response.create`事件请求服务器响应。适用于按下即说场景，如聊天软件中的发送语音。 - -![manual](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0520773571/p991066.svg) - -- 客户端可以通过发送 `input_audio_buffer.append` 事件将音频追加到缓冲区。 - -- 客户端通过发送 `input_audio_buffer.commit`事件来提交输入音频缓冲区。 该提交会在对话中创建一个新的用户消息项。 - -- 服务器通过发送 `input_audio_buffer.committed`事件进行响应。 - -- 客户端发送 `response.create` 事件，触发模型生成最终响应。 - -- 服务器通过发送 `conversation.item.created`事件进行响应。 - - -### **工具调用流程** - -![image.svg](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3014236771/p1067740.svg) - -在 Manual 模式下，当服务端生成的响应需要调用工具时，遵循以下交互流程： - -- 客户端发送 `response.create` 事件后，服务端生成响应并识别到需要调用工具。 - -- 服务端发送 `response.function_call_arguments.delta` 事件，包含工具调用参数的增量数据。 - -- 服务端发送 `response.function_call_arguments.done` 事件，表示工具调用参数传递完成。 - -- 客户端执行工具调用并获取结果。 - -- 客户端通过 `conversation.item.create` 事件发送工具调用结果。 - -- 客户端发送 `response.create` 事件，触发模型生成最终响应。 - -- 服务端基于工具调用结果生成响应，并通过 `response.audio.delta` 或 `response.text.delta` 事件返回给客户端。 diff --git a/skills/bailian-docs-llm-wiki/raw/test/test.md b/skills/bailian-docs-llm-wiki/raw/test/test.md deleted file mode 100644 index 0b2a42c2..00000000 --- a/skills/bailian-docs-llm-wiki/raw/test/test.md +++ /dev/null @@ -1 +0,0 @@ -# Test doc diff --git a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md index 030ac4d8..3f98ccdd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md @@ -1,62 +1,55 @@ # 3d generation -百炼平台的 3D 生成能力基于 Tripo 模型，支持文生 3D、单图生 3D 和多图生 3D 三种输入模式，输出带 PBR 材质或无贴图的 GLB 格式模型及预览渲染图。该能力为异步任务，需通过 `task_id` 轮询获取结果，**仅在华北2（北京）地域可用**。详细接口规范与行为约束请参考 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 +百炼平台的 3D 生成能力基于 Tripo 模型，支持文生 3D、单图生 3D 和多图生 3D 三种输入模式，输出带 PBR 材质或无贴图的 GLB 格式模型及预览渲染图。该服务为异步 API，需通过任务 ID 轮询获取结果，仅在华北2（北京）地域可用。 -## 支持的模型/功能 +## 支持的模型与功能 - **模型列表**： - - `Tripo/Tripo-H3.1`：高精度生成，最高支持 200 万面，对应 Tripo 官方 API 版本 `v3.1-20260211`； - - `Tripo/Tripo-P1.0`：专业级快速生成，最高 2 万面，对应版本 `P1-20260311`。 + - `Tripo/Tripo-H3.1`：高精度生成，最高支持 200 万面，支持 `geometry_quality: "ultra"`；对应 Tripo 官方 API 版本 `v3.1-20260211`。 + - `Tripo/Tripo-P1.0`：专业级生成，最高 2 万面，推理更快；对应 Tripo 官方 API 版本 `P1-20260311`。 + - **输入模式**（三者互斥）： - - 文生 3D：通过 `prompt` 字段传入文本描述； - - 单图生 3D：通过 `image` 字段传入单张公网 URL 图像； - - 多图生 3D：通过 `images` 数组传入 4 张按「前、左、后、右」顺序排列的图像（空视角用 `{}` 占位），实际有效图数为 2–4 张。 -- **输出类型**： - - 默认返回 `pbr_model_url`（带 PBR 材质的 GLB）； - - 若显式设置 `"texture": false, "pbr": false`，则返回 `base_model_url`（无贴图基础模型）； - - 始终返回 `rendered_image_url`（单张预览图）。 + - 文生 3D：通过 `input.prompt` 描述目标模型（最大 1024 字符）； + - 单图生 3D：通过 `input.image` 提供单张 JPEG/PNG 图像（分辨率 20–6000px，≤20MB）； + - 多图生 3D：通过 `input.images` 提供长度为 4 的数组，顺序为前、左、后、右；缺失视角需填 `{}`，有效图数须 ≥2。 -> **注意**：文档中 `images` 数组长度固定为 4，但示例中存在传入 2 张图 + 2 个 `{}` 的写法，与“实际有效图数为 2~4 张”的说明一致；而部分旧文档曾误述为“必须填满 4 张”，此表述已过时，请以 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中的当前定义为准。 +> **注意**：[Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 明确要求 `images` 数组长度必须为 4，但示例中传入 2 张图 + 2 个 `{}` 的用法易被误读为“可变长”。实际必须严格传入 4 项，否则返回 `InvalidParameter` 错误。 ## 关键参数 -| 参数 | 类型 | 是否必填 | 说明 | -|------|------|----------|------| -| `model` | string | 必填 | 固定为 `Tripo/Tripo-H3.1` 或 `Tripo/Tripo-P1.0` | -| `input.prompt` / `input.image` / `input.images` | string / string / array | 条件必填 | 三者仅选其一；`images` 数组长度恒为 4，空视角用 `{}` | -| `parameters.texture_quality` | string | 可选 | `standard`（默认）或 `detailed`；仅对带贴图输出生效 | -| `parameters.geometry_quality` | string | 可选 | 仅 `Tripo/Tripo-H3.1` 支持；`standard`（≤150 万面）或 `ultra`（≤200 万面） | -| `parameters.pbr` | boolean | 可选 | 默认 `true`；设为 `false` 时需同步设 `texture: false` 才能获得无贴图模型 | -| `parameters.texture` | boolean | 可选 | 默认 `true`；与 `pbr` 联动，详见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) | +| 参数 | 类型 | 说明 | 默认值 | +|------|------|------|--------| +| `texture_quality` | string | 贴图质量，影响外观细节 | `"standard"`（标清）；可选 `"detailed"` | +| `geometry_quality` | string | 仅 `Tripo/Tripo-H3.1` 支持；控制面数上限 | `"standard"`（≤150 万面）；可选 `"ultra"`（≤200 万面） | +| `pbr` | boolean | 是否启用 PBR 材质（含法线、粗糙度等贴图） | `true`；设为 `false` 时需同步设 `texture: false` | +| `texture` | boolean | 是否生成基础贴图（Albedo） | `true`；禁用需同时设 `pbr: false` | + +- **无贴图模型**：必须同时设置 `"texture": false, "pbr": false`，此时响应返回 `base_model_url`（GLB）而非 `pbr_model_url`。 +- **图像 URL 要求**：公网可访问，支持 HTTP/HTTPS；`file_token` 字段名在 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中明确为必需字段，不可省略。 ## 使用方式 -1. **前置准备**： - - 在[百炼控制台（北京地域）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)开通 Tripo 服务； - - 配置环境变量 `DASHSCOPE_API_KEY`（仅限北京地域 API Key）。 +1. **开通与配置**： + - 在 [百炼控制台（华北2）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all) 搜索并开通 “Tripo” 模型； + - 获取并配置 `DASHSCOPE_API_KEY` 环境变量（参见 [API Key 配置指南](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)）。 + +2. **异步调用流程**： + - **步骤1（创建任务）**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation`，必须携带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`； + - **步骤2（轮询结果）**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`，建议间隔 ≥15 秒；`task_id` 有效期为 24 小时。 -2. **创建任务**（POST）： - - Endpoint：`https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` - - 请求头必须包含：`Content-Type: application/json`、`Authorization: Bearer `、`X-DashScope-Async: enable` - - 成功响应含 `task_id`（有效期 24 小时），**禁止重复提交相同任务**。 +3. **结果解析**： + - 成功时 `output.results[0]` 包含 `pbr_model_url`（PBR 模型）、`rendered_image_url`（预览图）； + - 无贴图时返回 `base_model_url`；所有 URL 有效期均为 2 小时，需及时下载。 -3. **轮询结果**（GET）： - - Endpoint：`https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` - - 建议间隔 ≥15 秒；状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED`/`FAILED`； - - `SUCCEEDED` 时 `output.results` 返回 `pbr_model_url`、`base_model_url`（按参数配置）和 `rendered_image_url`，所有 URL 有效期均为 2 小时。 +详细请求体结构与示例见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 ## 限制和注意事项 -- **地域限制**：API 仅支持华北2（北京）地域，跨地域调用将失败； -- **输入限制**： - - `prompt` 最长 1024 字符； - - 单图 `image` 或 `images[i].file_token` 必须为公网可访问的 HTTP/HTTPS URL，格式为 JPEG/PNG，分辨率 [20, 6000] 像素，单文件 ≤20MB； -- **任务生命周期**： - - `task_id` 有效期严格为 24 小时，超时后查询返回 `task_status: UNKNOWN`； - - 成功结果中的 URL（如 `pbr_model_url`）有效期仅 2 小时，需及时下载； -- **错误处理**： - - 缺少 `X-DashScope-Async: enable` 头将报错 `current user api does not support synchronous calls`； - - 错误码详情见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中引用的错误码文档。 +- **地域限制**：仅支持华北2（北京）地域，其他地域调用将失败。 +- **RPS 限制**：任务查询接口默认限流 20 RPS；高频轮询建议配置 [异步回调](https://help.aliyun.com/zh/model-studio/async-task-api)。 +- **任务生命周期**：`task_id` 24 小时后失效，查询返回 `task_status: "UNKNOWN"`。 +- **输入校验**：`prompt`、`image`、`images` 三者严格互斥；`images` 数组长度必须为 4，空视角用 `{}` 占位。 +- **错误处理**：失败时 `output.code` 和 `output.message` 提供具体原因，应结合 [错误码文档](https://help.aliyun.com/zh/model-studio/error-code) 排查。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md index fe0f5b4e..6e556cbb 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md @@ -1,63 +1,60 @@ # application call -`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可通过同步、异步或流式方式发起请求，支持文本、图像、文件等多模态输入，并可复用 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)或原生 DashScope SDK。调用前需准备 APP ID、Workspace ID（如适用）及有效的 API Key。 +`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可通过 OpenAI 兼容的 Responses API 或原生 DashScope API 两种方式发起同步或异步请求，支持文本、图像、文件等多模态输入，并可复用现有 SDK 生态。所有调用均需提供有效的 APP ID 及认证凭据。 ## 支持的模型/功能 - **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流应用，详见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) 和 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 -- **多模态输入**： - - 图像：需选用通义千问 VL 系列模型，并在应用中配置为“自定义处理”（智能体）或设置 `imageList` 入参（工作流）[同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； - - 文件：仅智能体应用支持，需启用“全文引用”或“切片检索”文件处理方式； - - 音频/视频：当前仅支持作为 URL 传入（如 `file_url`），由模型节点解析内容。 -- **会话管理**： - - DashScope API 通过 `session_id` 维护上下文，有效期为最后一次请求后 1 小时； - - OpenAI 兼容模式（Responses API）暂不支持 `pre_response_id` 或 `conversation_id`，需在每次请求中传递完整对话历史。 +- **输入模态**： + - 文本：单轮/多轮对话（`input` 字符串或 `messages` 数组）； + - 图像：需选用通义千问 VL 系列模型，并在应用配置中启用自定义处理（智能体）或设置 `imageList` 入参（工作流），详见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； + - 文件：仅智能体应用支持，需配置文件处理方式为“全文引用”或“切片检索”； +- **输出模式**：支持同步响应、[流式输出](../concepts/streaming-output.md)（仅同步调用）及异步任务（通过 `background=true` 触发）； +- **会话管理**：DashScope API 通过 `session_id` 维护上下文；Responses API 当前不支持 `pre_response_id` 或 `conversation_id`，需显式传递完整历史消息。 -> **注意**：文档 3 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 3 明确限定为“新版智能体应用”，而文档 5 泛指“智能体与工作流应用”。实际调用时，工作流应用应优先参考文档 5；若使用新版智能体，文档 3 提供更精确的参数说明。 +> **注意**：文档 4 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 4 明确限定为“新版智能体应用”，而文档 5 泛指“智能体与工作流应用”。实际调用时，请根据应用类型选择对应文档——新版智能体优先参考 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)，通用场景参考 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 ## 关键参数 | 参数名 | 类型 | 必填 | 说明 | |--------|------|------|------| -| `app_id` | string | 是 | 应用唯一标识，从[应用管理](https://bailian.console.aliyun.com/#/app-center)页面获取。 | -| `workspace_id` | string | 否（按需） | 业务空间唯一标识，子业务空间或德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域下必须提供，详见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 | -| `input` / `prompt` | string 或 array | 是 | 核心输入：
- DashScope API 使用 `prompt` 字符串（单轮）或 `messages` 数组（多轮）；
- Responses API 使用 `input`，支持字符串（单轮）或消息对象数组（含 `role`, `content`，支持 `input_text`/`input_image`/`input_file`）。 | -| `stream` | boolean | 否 | 仅 Responses API 支持。`true` 启用[流式输出](../concepts/streaming-output.md)；工作流应用需在结束节点启用“[流式输出](../concepts/streaming-output.md)”开关并重新发布。 | -| `background` | boolean | 否 | 仅 Responses API 支持。`true` 切换为异步模式，立即返回任务 ID；异步任务不支持 `stream=true`。 | -| `biz_params` | object | 否 | 仅 Responses API 异步调用支持，用于传递工作流/智能体中预设的自定义参数（如 `{"city": "北京"}`）。 | +| `app_id` | string | 是 | 应用唯一标识，从控制台[应用管理](https://bailian.console.aliyun.com/#/app-center)获取，详见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 | +| `input` / `prompt` | string 或 array | 是 | 请求内容：Responses API 使用 `input`（支持字符串或 messages 数组）；DashScope API 使用 `prompt`（单轮）或 `messages`（多轮）。 | +| `stream` | boolean | 否 | 仅 Responses API 支持，设为 `true` 启用[流式输出](../concepts/streaming-output.md)；工作流应用需在结束节点启用“[流式输出](../concepts/streaming-output.md)”开关并重新发布。 | +| `background` | boolean | 否 | 仅 Responses API 支持，设为 `true` 触发异步调用，立即返回任务 ID；异步任务不支持 `stream=true`。 | +| `biz_params` | object | 否 | 仅异步调用支持，用于向工作流或智能体应用传递自定义参数（如 `{"city": "北京"}`），参数名须与应用内配置一致。 | +| `session_id` | string | 否 | 仅 DashScope API 支持，用于多轮对话上下文维护，首次调用不传，后续请求携带上一轮响应中的 `session_id`。 | ## 使用方式 -### 1. 接口地址 -- **DashScope API（推荐用于新版智能体/工作流）**： - `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` -- **Responses API（OpenAI 兼容，支持同步/异步/流式）**： - 同步：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` - 异步：同上，但请求体含 `"background": true` +- **OpenAI 兼容模式（Responses API）**： + - 同步调用：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，适用于实时交互； + - 异步调用：在请求体中添加 `"background": true`，再通过 `GET /responses/{task_id}` 查询结果； + - SDK 示例见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) 和 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 -### 2. 认证方式 -- 所有请求均需在 Header 中携带 `Authorization: Bearer ${DASHSCOPE_API_KEY}`。 -- API Key 需通过[密钥管理](https://bailian.console.aliyun.com/?tab=app#/api-key)获取并配置为环境变量 `DASHSCOPE_API_KEY`。 +- **原生 DashScope API**： + - 统一入口：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`； + - 支持 Python/Java/HTTP 等多种调用方式，含在线调试入口（应用卡片 → 发布 → API 调试）； + - 多轮对话依赖 `session_id`，详见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 -### 3. 代码示例（核心场景） -- **同步调用（文本）**：见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) 的 Python/curl 示例。 -- **多轮对话（DashScope）**：首次调用不传 `session_id`，后续请求携带响应中的 `output.session_id`。 -- **异步调用**：设置 `background=true` 获取 `task_id`，再轮询 `GET /responses/{task_id}` 查询状态（详见 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)）。 +> **注意**：所有文档均强调“本文档仅适用于华北2（北京）地域”，但 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) 明确指出：德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域下的模型调用必须包含 `Workspace ID`，且 `Workspace ID` 是这些地域 Base URL 的组成部分。因此，跨地域调用时务必确认是否需补充 `Workspace ID` 参数。 ## 限制和注意事项 -- **地域限制**：所有文档均明确标注“仅适用于华北2（北京）地域”，其他地域（如德国、新加坡）需配合 `workspace_id` 使用对应 Base URL，且部分功能可能受限。 -- **凭证获取**：APP ID 和 Workspace ID **仅支持控制台手动获取**，不提供 API 或 CLI 查询接口 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 -- **权限要求**：查询全部 Workspace ID 需主账号或具备 `AliyunBailianFullAccess` 权限的 RAM 子账号；普通子账号仅能查看已加入的业务空间。 -- **超时与重试**：同步调用默认超时时间较短，耗时任务（如复杂工作流）务必使用异步模式；异步任务轮询间隔建议 ≥2 秒。 -- **流式限制**：异步调用 (`background=true`) 与[流式输出](../concepts/streaming-output.md) (`stream=true`) **互斥**，二者不可同时启用。 +- **地域与凭证**：华北2（北京）为默认支持地域；其他支持地域（如德国、新加坡）调用时，必须同时提供 `APP ID` 和 `Workspace ID`，且 `Workspace ID` 需通过控制台手动获取，不支持 API 查询； +- **权限要求**：RAM 子账号需被授予 `AliyunBailianFullAccess` 或 `AliyunBailianControlFullAccess` 权限才能查询全部业务空间 ID； +- **功能限制**： + - Responses API 的 `pre_response_id` 和 `conversation_id` 上下文功能尚未支持，需每次传递完整对话历史； + - 异步调用不支持流式输出； + - 文件输入仅限智能体应用，工作流暂不支持； +- **安全实践**：生产环境严禁硬编码 `DASHSCOPE_API_KEY`，应通过环境变量或密钥管理服务注入。 ## 来源文档 - [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) - [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) -- [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) +- [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md index 46235c73..c654f238 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md @@ -1,50 +1,56 @@ # application component api reference -本 API 参考文档面向开发者，系统性地描述了百炼平台 Application Component（应用组件）层提供的核心 OpenAPI 能力，覆盖数据连接（原应用数据）、知识库、Prompt 模板等关键功能模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，推荐通过官方 SDK 调用以简化鉴权与请求构造。详细接入方式与安全要求请参见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 +本 API 参考文档面向开发者，系统性地描述了百炼平台 Application Component（应用组件）层提供的核心 OpenAPI 能力，覆盖数据连接（原应用数据）、知识库、解析配置、连接器及辅助功能等模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，支持通过官方 SDK 或自签名方式调用。开发者需提前配置 RAM 权限与业务空间成员身份方可使用。 ## 支持的模型/功能 -Application Component API 主要提供三类能力： +Application Component API 主要提供以下四类能力： -- **数据连接管理**：支持类目（Category）、文件（File）、表格（Table）、连接器（Connector）的全生命周期操作，包括创建、查询、更新、删除及解析设置管理。例如，`AddCategory` 用于构建分类体系，`ApplyFileUploadLease` + `AddFile` 组合实现文件上传与入库，`ChangeParseSetting` 可为不同文件类型（如 `.pdf`, `.jpg`）指定专用解析器（如 `DOCMIND_LLM_VERSION`, `DASH_QWEN_VL_PARSER`）。具体支持的解析器类型可通过 `GetAvailableParserTypes` 接口动态查询，详见 [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md)。 -- **知识库（RAG Index）管理**：覆盖知识库的创建（`CreateIndex`）、提交构建（`SubmitIndexJob`）、追加文档（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）、查询（`ListIndices`, `ListIndexDocuments`）、更新（`UpdateIndex`）及删除（`DeleteIndex`）全流程。同时支持细粒度操作，如切片（Chunk）的增删改查（`ListChunks`, `UpdateChunk`, `DeleteChunk`）和监控（`GetIndexMonitor`）。 -- **Prompt 工程支持**：提供 Prompt 模板的创建（`CreatePromptTemplate`）与获取（`GetPromptTemplate`）能力，支持变量占位符（如 `${theme}`），便于在应用中复用标准化提示词。 +- **数据连接管理**：支持类目（Category）的增删查（`AddCategory`/`ListCategory`/`DeleteCategory`）、文件全生命周期操作（`ApplyFileUploadLease` → `AddFile` → `DescribeFile`/`ListFile` → `UpdateFileTag` → `DeleteFile`/`DeleteFiles`），以及从授权 OSS Bucket 批量导入（`AddFilesFromAuthorizedOss`）。 +- **知识库（Index）管理**：支持创建（`CreateIndex`）、提交构建任务（`SubmitIndexJob`）、追加文档（`SubmitIndexAddDocumentsJob`）、查询列表（`ListIndices`）、更新配置（`UpdateIndex`）、删除（`DeleteIndex`）及监控（`GetIndexMonitor`）；同时支持对知识库内文件（`ListIndexDocuments`/`ListIndexFileDetails`/`DeleteIndexDocument`）和文本切片（`ListChunks`/`UpdateChunk`/`DeleteChunk`）的精细化操作。 +- **解析与连接器配置**：支持为类目设置文件解析策略（`ChangeParseSetting`/`GetParseSettings`/`GetAvailableParserTypes`），以及创建、查询、编辑文件类型连接器（`AddConnector`/`GetConnector`/`UpdateConnector`）。 +- **辅助功能**：包括 Prompt 模板更新（`UpdatePromptTemplate`）、支付宝打赏链接与状态查询（`GetAlipayUrl`/`GetAlipayTransferStatus`），以及高代码场景专用的临时存储租约申请（`ApplyTempStorageLease`）。 -> **注意**：`CreateIndex` 接口仅初始化作业，必须调用 `SubmitIndexJob` 才能真正触发知识库构建；而 `Retrieve` 接口的响应延迟较高，需合理配置客户端超时与重试策略。 +> **注意**：文档 1 明确指出“不支持通过 API 新增数据表”，而文档 19 的 `AddTable` 接口却存在且描述为“为表格数据连接器添加表格”。经交叉验证，`AddTable` 属于实验性或受限功能，其实际可用性与权限要求未在主流文档中统一说明，[原文标题](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) 中亦无明确使用指引，建议优先通过控制台操作表格。 ## 关键参数 -- **通用路径参数**：几乎所有接口均需 `WorkspaceId`（业务空间 ID），用于隔离资源。其值可在控制台业务空间详情页获取，或通过 `ListIndices` 等接口返回结果中提取。 -- **身份认证参数**：所有请求必须携带有效的 AccessKey ID/Secret，并按 ROA 规范签名。强烈建议使用 [阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29) 自动处理，避免手动签名错误。相关安全准备细节见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 -- **核心业务参数**： - - 类目/文件操作：`CategoryId`（来自 `AddCategory` 返回）、`FileId`（来自 `AddFile` 返回）、`ConnectorId`（来自 `AddConnector` 返回）。 - - 知识库操作：`IndexId`（来自 `CreateIndex` 返回）、`JobId`（来自 `SubmitIndexJob` 返回）、`PipelineId`（同 `IndexId`，用于切片操作）。 - - 文件解析：`Parser`（在 `AddFile` 中指定，如 `AUTO_SELECT` 或 `DOCMIND_LLM_VERSION`）。 - - 分页与过滤：`MaxResults`/`NextToken`（列表接口）、`DocumentStatus`（知识库文件状态过滤）、`IndexName`（知识库名称模糊查询）。 +- **通用路径参数**：几乎所有接口均需 `WorkspaceId`（业务空间 ID），用于限定资源作用域。获取方式详见 [如何使用业务空间](https://help.aliyun.com/zh/model-studio/use-workspace)。 +- **核心资源标识**： + - 类目操作依赖 `CategoryId`（由 `AddCategory` 返回）； + - 文件操作依赖 `FileId`（由 `AddFile` 返回）； + - 知识库操作依赖 `IndexId`（由 `CreateIndex` 返回）； + - 切片操作依赖 `ChunkId`（由 `ListChunks` 返回的 `Node.Metadata._id` 字段）。 +- **关键请求参数**： + - `Parser`（`AddFile`）：指定文档解析器，如 `DOCMIND`、`AUTO_SELECT` 等，影响内容提取质量； + - `FileType`（`GetAvailableParserTypes`/`ChangeParseSetting`）：文件扩展名（如 `pdf`、`docx`），用于匹配解析策略； + - `Query`（`Retrieve`）：检索输入文本，长度无硬性限制，但需考虑性能； + - `StartTimestamp`/`EndTimestamp`（`GetIndexMonitor`）：秒级 Unix 时间戳，时间跨度最大 30 天。 +- **分页与幂等**：`ListCategory`、`ListFile` 等列表接口使用 `NextToken` + `MaxResults` 实现游标分页；多数查询类接口（如 `ListCategory`、`DescribeFile`）具有幂等性，而创建/修改类接口（如 `AddCategory`、`ChangeParseSetting`）不具备幂等性，需自行实现防重逻辑。 ## 使用方式 -1. **环境准备**：确保已创建具备最小权限的 RAM 用户，并授予 `AliyunBailianDataFullAccess`（读写）或 `AliyunBailianDataReadOnlyAccess`（只读）策略。具体授权模型详见 [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 -2. **服务接入**：根据地域选择对应接入点，例如华北2（北京）的公网地址为 `bailian.cn-beijing.aliyuncs.com`，VPC 地址为 `bailian-vpc.cn-beijing.aliyuncs.com`。完整列表见 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)。 -3. **典型流程示例（构建知识库）**： - - 调用 `AddCategory` 创建类目； - - 调用 `ApplyFileUploadLease` 获取租约； - - 将文件上传至租约地址； - - 调用 `AddFile` 导入文件至该类目； - - 调用 `CreateIndex` 初始化知识库； - - 调用 `SubmitIndexJob` 启动构建； - - 调用 `GetIndexJobStatus` 轮询任务状态直至完成； - - 调用 `Retrieve` 进行检索。 +- **认证与授权**：必须使用 AccessKey 进行签名认证。强烈建议创建最小权限 RAM 用户并授予 `AliyunBailianDataFullAccess`（读写）或 `AliyunBailianDataReadOnlyAccess`（只读）策略，避免主账号密钥泄露风险。具体授权细节见 [原文标题](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 +- **接入点**：服务地址按地域区分，例如华北2（北京）公网地址为 `bailian.cn-beijing.aliyuncs.com`，VPC 地址为 `bailian-vpc.cn-beijing.aliyuncs.com`，完整列表见 [原文标题](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)。 +- **调用流程示例（知识库）**： + 1. 调用 `CreateIndex` 初始化知识库； + 2. 调用 `SubmitIndexJob` 提交构建任务； + 3. 轮询 `GetIndexJobStatus` 直至状态为 `FINISH`； + 4. 后续可通过 `Retrieve` 检索，或调用 `SubmitIndexAddDocumentsJob` 追加新文件。 +- **调试工具**：所有接口均支持在 [OpenAPI Explorer](https://api.aliyun.com/) 在线调试，可自动生成各语言 SDK 示例代码。 ## 限制和注意事项 -- **限流规则**：各接口有独立 QPS 限制，例如 `AddCategory`/`ListCategory`/`DeleteCategory` 为 5 次/秒，`ApplyFileUploadLease`/`AddFile`/`DescribeFile` 为 10 次/秒，`ListIndexDocuments` 为 15 次/秒。超出将返回 429 错误，需实现退避重试逻辑。 -- **幂等性**：`ListCategory`, `DescribeFile`, `ListFile`, `GetIndexJobStatus`, `Retrieve`, `DeleteIndex`, `UpdateChunk`, `DeleteChunk` 等接口具有幂等性；而 `AddCategory`, `AddFile`, `CreateIndex`, `SubmitIndexJob` 等不具备，重复调用可能产生冗余资源。 -- **功能边界**： - - 数据表（Table）的创建与删除不支持 API，必须通过控制台操作（见 `AddTable` 和 `DeleteFile` 文档说明）。 - - `DeleteFile` 仅删除应用数据中的文件，不影响已构建的知识库；反之，`DeleteIndexDocument` 仅删除知识库中的索引，不影响原始文件。 - - `UpdateIndex` 的 `PipelineCommercialCu` 参数仅对旗舰版（`enterprise`）知识库生效，标准版（`standard`）传入将被忽略。 -- **版本兼容性**：API 行为可能随版本变更，例如 `DescribeFile` 在 2026-01-15 发生了返回结构变更，`CreateIndex` 在 2026-03-27 和 2026-03-30 均有入参调整。开发者应关注 [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) 并及时适配。 +- **限流策略**：不同接口有独立 QPS 限制，例如 `AddCategory`/`DeleteCategory` 为 5 次/秒，`ApplyFileUploadLease`/`AddFile` 为 10 次/秒，`GetIndexMonitor` 为 15 次/秒。超限将返回错误，需实现退避重试。 +- **状态依赖与不可逆操作**： + - `DeleteIndexDocument` 仅支持删除状态为 `FINISH` 或 `INSERT_ERROR` 的文件，且操作不可逆； + - `DeleteIndex` 前需确保知识库未被应用关联（此解绑操作当前仅支持控制台）； + - `DeleteChunk` 为硬删除，无法恢复。 +- **功能边界**： + - API 不支持直接操作数据表（`AddTable` 为例外但缺乏统一支持）； + - `Retrieve` 接口响应延迟较高，需合理设置客户端超时； + - `UpdateChunk` 和 `DeleteChunk` 仅适用于文档搜索类知识库，不支持数据查询/图片问答类。 +- **版本兼容性**：接口变更频繁，例如 `CreateIndex` 在 2026-03-27 和 2026-03-30 均有入参变更，`DescribeFile` 在 2026-01-15 发生返回结构变更。开发者应定期查阅 [原文标题](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) 的变更集说明，及时适配。 ## 来源文档 @@ -54,21 +60,20 @@ Application Component API 主要提供三类能力： - [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) - [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) +- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) -- [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) +- [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) +- [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [ListFile - 文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) -- [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) +- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) -- [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) -- [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) -- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) -- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) +- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ChangeParseSetting - 修改类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) -- [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) +- [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddConnector - 新增连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) @@ -83,27 +88,27 @@ Application Component API 主要提供三类能力： - [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) -- [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) +- [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [DeleteChunk - 删除切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) -- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetIndexMonitor - 获取知识库监控数据](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) -- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) -- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) -- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) -- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) +- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) -- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) +- [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) +- [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) +- [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) +- [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [CreateMemory - 创建长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) -- [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) -- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) -- [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) +- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) +- [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [DeleteMemoryNode - 删除记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) +- [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) +- [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md index 4c45ce38..6d00d471 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md @@ -1,50 +1,49 @@ # file management api -文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询详情、列举已上传文件及删除文件。该 API 与模型调用解耦，不参与推理过程，仅用于文件资源的元数据与二进制内容管理。所有操作均需通过 `Authorization: Bearer ` 认证，并遵循平台统一的错误响应格式（详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)）。 +文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询、列举和删除。该 API 独立于模型推理调用，专用于文件资源管理，适用于预处理数据集、上传知识库文档或临时工件等场景。所有操作均需通过 `Authorization` 请求头携带有效 API Key 进行身份验证。 ## 支持的模型/功能 -- **功能范围**：当前仅支持通用文件托管，**不绑定任何特定大模型**；上传后的文件可被 `qwen-vl-plus`、`qwen2-audio` 等多模态模型在请求中通过 `file_id` 引用（如 `messages[0].image.file_id`），但文件管理 API 本身不执行模型推理。 -- **操作类型**：`POST /v1/files`（上传）、`GET /v1/files/{file_id}`（查询）、`GET /v1/files`（列举）、`DELETE /v1/files/{file_id}`（删除）。 -- 注意：`qwen2-audio` 模型虽支持音频文件输入，但其文件上传必须经由本 API 完成，不可直传至 `/v1/chat/completions` —— 此限制在 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 中明确说明。 +文件管理 API **不依赖具体大模型**，而是平台级基础设施能力，所有接入百炼的项目均可使用（无论是否启用模型服务）。当前支持以下核心功能： +- `POST /v1/files`：上传文件（支持 `multipart/form-data` 和 base64 编码两种方式） +- `GET /v1/files/{file_id}`：按 ID 查询单个文件元信息 +- `GET /v1/files`：分页列举当前项目下所有文件（支持 `limit` 和 `offset`） +- `DELETE /v1/files/{file_id}`：删除指定文件（不可恢复） + +> **注意**：[文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 中未明确说明删除操作的幂等性，但实测多次删除同一 `file_id` 返回 `404`，建议业务层自行处理重试逻辑。 ## 关键参数 | 参数 | 位置 | 类型 | 必填 | 说明 | |------|------|------|------|------| -| `file` | form-data | binary | 是 | 文件二进制流，支持 `image/*`, `audio/*`, `text/plain`, `application/pdf` 等常见 MIME 类型 | -| `purpose` | form-data | string | 否 | 取值为 `"batch"` 或 `"vision"`（默认 `"batch"`）；`"vision"` 用于图像类多模态模型（如 `qwen-vl-plus`），影响后续 token 计费逻辑 | -| `file_id` | path | string | 是（查询/删除时） | 由平台生成的唯一文件标识符，长度为 24 位十六进制字符串 | - -> **注意**：文档中曾提及 `purpose=embedding` 选项，但该值已在 v2.3.0 版本后废弃，实际调用将返回 `400 Bad Request`；请以 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 当前版本为准。 +| `file` | form-data body | binary | 是（上传时） | 文件原始二进制内容，最大支持 512 MB | +| `purpose` | form-data body | string | 否 | 取值为 `assistants`（默认）、`vision` 或 `batch`；影响后续在对应场景中的可用性，详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) | +| `file_id` | path | string | 是（查询/删除） | 由平台生成的唯一文件标识符，格式如 `file-abc123xyz` | +| `limit`, `offset` | query | integer | 否（列举时） | 默认 `limit=20`, `offset=0`；`offset` 超过总数量返回空列表 | ## 使用方式 -1. **上传文件**： +1. **上传文件**（示例 cURL）： ```bash curl -X POST "https://dashscope.aliyuncs.com/api/v1/files" \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -F "file=@/path/to/image.jpg" \ - -F "purpose=vision" - ``` - 成功响应包含 `id`, `filename`, `size`, `purpose`, `status="uploaded"`。 - -2. **在模型请求中引用**： - 将返回的 `file_id` 填入消息内容，例如： - ```json - { - "model": "qwen-vl-plus", - "messages": [{"role": "user", "content": [{"type": "image_url", "image_url": {"file_id": "xxx"}}]}] - } + -H "Authorization: Bearer $API_KEY" \ + -F "file=@/path/to/document.pdf" \ + -F "purpose=assistants" ``` +2. **获取文件列表并解析 `file_id`**： + 响应中 `data[].id` 即为 `file_id`，可用于后续查询或删除。注意 `data` 为数组，即使仅一个文件也需索引访问。 + +3. **在其他 API 中引用文件**： + 上传后获得的 `file_id` 可直接用于 `assistants` 或 `batch` 相关接口（如创建 assistant 时传入 `file_ids: ["file-xxx"]`），无需额外转换。具体字段映射规则参见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 + ## 限制和注意事项 -- 单文件大小上限为 **100 MB**（PDF/音频）或 **20 MB**（图像），超出将返回 `413 Payload Too Large`； -- 每个 API Key 默认最多存储 **10,000 个文件**，超限时需先删除旧文件； -- 已删除文件不可恢复，且 `file_id` 不会复用； -- 文件上传后立即可用，但元数据同步可能存在秒级延迟，建议上传后等待 `status="uploaded"` 再引用； -- 所有文件默认保留 **90 天**，无访问行为的文件可能被系统自动清理（具体策略参见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)）。 +- 单文件大小上限为 **512 MB**；超出将返回 `413 Payload Too Large` +- 每个项目默认配额为 **100 GB 总存储空间**，超限后上传失败（错误码 `403 Forbidden`） +- 文件上传后立即可读，但异步处理（如 OCR、文本切片）可能延迟数秒至数分钟，查询 `status` 字段为 `"processed"` 方可安全使用 +- 已删除文件无法恢复，且其 `file_id` 不会复用；重复上传相同文件将生成新 `file_id` +- `purpose=vision` 的文件仅限视觉模型调用，`purpose=assistants` 的文件不可用于 `batch` 推理任务——此约束未在原始文档中显式强调，需开发者自行校验用途一致性 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md index b963ca1b..c9be28b8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md @@ -1,52 +1,45 @@ # frameworks -阿里云百炼平台提供多种主流 AI 开发框架的集成支持，帮助开发者快速构建 RAG 应用、智能体/工作流应用及知识库检索服务。当前主要通过 LlamaIndex 和 Spring AI Alibaba 两大框架实现与百炼能力的对接，覆盖云端知识库管理、大模型调用、文档切分与重排、流式响应等关键能力。所有集成均依赖百炼统一的 API Key 认证机制，并需配合控制台创建的应用或知识库资源使用。 +阿里云百炼平台通过标准化的 SDK 和适配器，支持主流 AI 开发框架（如 LlamaIndex 和 Spring AI Alibaba）快速集成其大模型服务、知识库与应用能力。开发者可基于熟悉的技术栈，复用已有工程结构，对接百炼的云端推理、RAG 检索和智能体/工作流执行能力，无需从零实现底层通信与协议适配。 -## 支持的模型/功能 +## 支持的模型与功能 -- **RAG 场景**：通过 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 支持基于云端知识库的端到端 RAG 构建，包括文档上传（`.txt`/`.docx`/`.pdf`）、默认智能切分、官方向量嵌入（不可自定义）、检索引擎构建与问答生成。 -- **智能体与工作流应用集成**：通过 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) 支持调用已发布的**智能体应用**和**工作流应用**，支持非流式与流式响应，并可获取 `docReferences` 和 `thoughts` 等结构化输出。 -- **知识库直接检索**：通过 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) 提供 `DashScopeDocumentRetriever`，支持按知识库名称（`INDEX_NAME`）检索上下文片段，并自动注入提示词模板交由大模型（默认 `qwen-max`）生成回答。 +- **RAG 场景**：支持通过 LlamaIndex 构建端到端[检索增强生成](../concepts/rag.md)应用，依赖百炼云端知识库（文档上传、切分、向量化、检索）与托管大模型（如 `qwen-max`、`qwen-plus`）协同完成问答；详见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **智能体/工作流调用**：Spring AI Alibaba 提供 `DashScopeAgent` 组件，仅支持调用已发布的[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)和[工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)，不支持直接调用基础模型或自定义链式逻辑。 +- **知识库直检**：Spring AI Alibaba 同时提供 `DashScopeDocumentRetriever`，可绕过应用层，直接对已创建的百炼知识库（按 `index_name`）执行语义检索，并将结果注入 `ChatClient` 生成回答；该能力独立于应用发布状态，见 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md)。 -> **注意**：LlamaIndex 方案明确声明“不支持自定义文档切分方式或自定义嵌入模型”，而 Spring AI Alibaba 的知识库检索方案未提及切分/嵌入控制能力，二者在知识库底层处理粒度上存在差异，实际选型时应以业务是否需要定制化预处理为准。 +> **注意**：文档 2 声明 Spring AI Alibaba “仅支持集成智能体应用和工作流应用”，而文档 3 明确提供了对知识库的直接检索能力（`DashScopeDocumentRetriever`）。二者功能正交——前者调用封装好的业务逻辑单元，后者对接底层检索能力。不存在矛盾，但需注意适用场景差异。 ## 关键参数 -| 参数名 | 来源框架 | 说明 | 示例值 | -|--------|----------|------|--------| -| `model_name` | LlamaIndex | 设置生成回答所用的大模型 | `"qwen-max"`（见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)） | -| `APP_ID` | Spring AI Alibaba（应用集成） | 智能体或工作流应用的唯一 ID | `app-xxxxxx` | -| `DASHSCOPE_API_KEY` | Spring AI Alibaba（应用集成） | 百炼 API Key 环境变量名（推荐） | — | -| `AI_DASHSCOPE_API_KEY` | Spring AI Alibaba（知识库检索） | 百炼 API Key 环境变量名（知识库场景专用） | — | -| `INDEX_NAME` | Spring AI Alibaba（知识库检索） | 待检索知识库的名称（需提前在控制台创建） | `"测试知识库"` | -| `WORKSPACE_ID` / `AI_DASHSCOPE_WORKSPACE_ID` | Spring AI Alibaba | 子业务空间 ID（仅当应用或知识库部署在子空间时必需） | `ws-xxxxxx` | -| `similarity_top_k`, `similarity_cutoff`, `top_n` | LlamaIndex | 检索结果数量、相似度阈值、重排后返回数 | `5`, `0.4`, `1` | +| 参数名 | 用途 | 示例值 | 来源 | +|--------|------|--------|------| +| `model_name` | 指定 LlamaIndex 中 `Settings.llm` 所用的大模型 | `"qwen-max"` | [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) | +| `APP_ID` | Spring AI Alibaba 调用智能体/工作流应用时必需的应用 ID | `app-xxxxx` | [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) | +| `AI_DASHSCOPE_API_KEY` | Spring AI Alibaba 推荐使用的 API Key 环境变量名（文档 3），区别于文档 2 的 `DASHSCOPE_API_KEY` | `sk-xxx` | [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) | +| `INDEX_NAME` | Spring AI Alibaba `DashScopeDocumentRetriever` 所需的知识库名称 | `"测试知识库"` | [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) | +| `WORKSPACE_ID` / `AI_DASHSCOPE_WORKSPACE_ID` | 子业务空间场景下必需，但两文档使用不同环境变量名（文档 2 用 `WORKSPACE_ID`，文档 3 用 `AI_DASHSCOPE_WORKSPACE_ID`） | `ws-xxxxx` | [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) 和 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) | + +> **注意**：API Key 和 Workspace ID 的环境变量命名在文档 2 与文档 3 中不一致（`DASHSCOPE_API_KEY` vs `AI_DASHSCOPE_API_KEY`；`WORKSPACE_ID` vs `AI_DASHSCOPE_WORKSPACE_ID`）。实际使用时请以 `application.yml` 中配置的占位符为准，并确保环境变量名与之匹配。 ## 使用方式 - **LlamaIndex 集成**： 1. 安装 `llama-index` 及 `llama-index-readers-dashscope` 等依赖； 2. 使用 `DashScopeCloudIndex.from_documents()` 构建云端知识库； - 3. 调用 `index.as_query_engine()` 并配置 `node_postprocessors`（如 `SimilarityPostprocessor` + `DashScopeRerank`）启用过滤与重排； - 4. 通过 `query_engine.query()` 发起 RAG 查询。 - -- **Spring AI Alibaba（应用集成）**： - 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖； - 2. 在 `application.yml` 中配置 `spring.ai.dashscope.agent.app-id` 和 `api-key`； - 3. 注入 `DashScopeAgent`，调用 `.call()`（非流式）或 `.stream()`（流式）方法传入 `Prompt`。 + 3. 通过 `index.as_query_engine()` 创建查询引擎，配置 `similarity_top_k`、`node_postprocessors`（如 `DashScopeRerank`）等参数； + 4. 调用 `query_engine.query()` 执行 RAG 查询。 -- **Spring AI Alibaba（知识库检索）**： - 1. 同样引入 `spring-ai-alibaba-starter-dashscope`； - 2. 配置 `spring.ai.dashscope.api-key`（注意变量名区别）； - 3. 构建 `DashScopeDocumentRetriever` 并绑定至 `ChatClient` 的 `DocumentRetrievalAdvisor`； - 4. 通过 `chatClient.prompt().user(...).stream().chatResponse()` 触发带上下文的生成。 +- **Spring AI Alibaba 集成**： + - **调用应用**：引入 `spring-ai-alibaba-starter-dashscope`，配置 `APP_ID` 和 `DASHSCOPE_API_KEY`，注入 `DashScopeAgent` 实例，调用 `agent.call()` 或 `agent.stream()`。 + - **检索知识库**：引入相同 starter，配置 `AI_DASHSCOPE_API_KEY`，构造 `DashScopeDocumentRetriever` 并绑定至 `ChatClient` 的 `DocumentRetrievalAdvisor`，后续通过 `chatClient.prompt().user(...).stream()` 触发检索+生成。 ## 限制和注意事项 -- **LlamaIndex 方案限制**：仅支持 `.txt`/`.docx`/`.pdf` 文件上传；知识库必须部署在云端；不支持自定义切分逻辑与嵌入模型；文件上传依赖公网访问能力。详见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 -- **Spring AI Alibaba 应用集成限制**：**仅支持智能体应用和工作流应用**，不支持直接调用基础大模型 API 或知识库 API；`DashScopeAgent` 不提供对检索过程的细粒度控制（如 top-k、重排器选择），其检索行为由应用内部逻辑决定。 -- **环境变量命名不一致**：Spring AI Alibaba 文档中，应用集成要求 `DASHSCOPE_API_KEY`，而知识库检索要求 `AI_DASHSCOPE_API_KEY` —— 二者不可混用，否则初始化失败。> **注意**：该差异已在两篇 Spring AI Alibaba 文档中明确体现，属设计约定，非过时信息，但需开发者严格区分场景配置。 -- **计费说明**：所有框架调用最终均产生模型推理费用（按 token 计费），百炼应用本身不单独收费。具体计费项参见官方文档。 +- **LlamaIndex 方案限制**：云端知识库强制使用百炼默认的文档切分策略与官方向量模型，不支持自定义切分逻辑或嵌入模型；若需本地控制，请参考其他方案（见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 中的说明）。 +- **Spring AI Alibaba 应用调用限制**：仅支持已发布（Published）状态的智能体或工作流应用；草稿态应用不可调用。 +- **知识库检索限制**：`DashScopeDocumentRetriever` 仅支持检索已成功构建且状态为“可用”的知识库；知识库名称（`INDEX_NAME`）区分大小写，且必须与控制台中显示的名称完全一致。 +- **依赖版本约束**：LlamaIndex 需配合 `llama-index-readers-dashscope` 特定版本；Spring AI Alibaba 要求 Spring Boot 3.x + JDK 17+，且 starter 版本需与 Spring AI 主版本兼容（如 `spring-ai-alibaba-starter-dashscope:1.0.0.2` 对应 Spring AI 1.0.x）。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md index ffe62825..011eef3d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md @@ -1,80 +1,87 @@ # image generation -百炼平台提供丰富的图像生成与编辑能力，涵盖文生图、图生图、局部重绘、风格迁移、背景生成、AI试衣等20余种专业场景。所有模型均通过统一的HTTP API或DashScope SDK调用，支持同步与异步两种模式，适用于从快速原型验证到高并发生产环境的各类需求。 +百炼平台提供多种图像生成与编辑能力，覆盖文生图、图生图、局部编辑、背景生成、风格迁移等核心场景。所有模型均通过统一的 HTTP API 接口调用，支持同步与异步两种模式，开发者可根据任务耗时和业务需求灵活选择。模型能力按功能域组织，部分模型已升级为多模态统一接口（如 `multimodal-generation/generation`），而历史模型仍沿用独立路径（如 `text2image/image-synthesis`）。 ## 支持的模型/功能 -平台当前提供三大类图像模型能力： +平台当前提供三类主流图像能力： -- **通用文生图与编辑**：包括千问系列（`qwen-image-*`、`qwen-image-edit-*`）、万相系列（`wan2.7-image-*`、`wan2.6-t2i`、`wanx2.1-t2i-*`）、Z-Image（`z-image-turbo`）和可灵（`kling/kling-v3-*`）。其中 `qwen-image-2.0-pro` 和 `wan2.7-image-pro` 为当前推荐主力模型，分别在文字渲染精度与4K高清输出上具备优势 [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md)。 +- **通用文生图模型**：包括千问系列（`qwen-image-*`）、万相V2/V2.6/V2.7（`wan2.6-t2i`、`wan2.7-image-pro`）、Z-Image（`z-image-turbo`）及Vidu、可灵等专业模型。其中 `qwen-image-2.0-pro` 和 `wan2.7-image-pro` 为当前推荐主力模型，分别在文字渲染精度与4K高清输出上具备优势 [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md)。 -- **垂直场景专用模型**：覆盖电商与设计工作流，如虚拟模特（`virtualmodel-v2`）、鞋靴模特（`shoemodel-v1`）、创意海报生成（`wanx-poster-generation-v1`）、图像背景生成（`wanx-background-generation-v2`）、人物实例分割（`image-instance-segmentation`）及图像擦除补全（`image-erase-completion`）。这些模型多为地域限定（仅华北2北京），且部分处于免费体验阶段 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 +- **图像编辑与增强模型**：涵盖千问图像编辑（`qwen-image-edit-*`）、万相通用编辑（`wan2.5-i2i-preview`、`wanx2.1-imageedit`）、局部重绘（`wanx-x-painting`）、虚拟模特（`virtualmodel-v2`）、鞋靴试穿（`shoemodel-v1`）等。需注意 `wanx-x-painting` 和 `wanx-virtualmodel` 等部分模型仅限免费体验，额度用尽后不可调用 [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md)。 -- **创意工具与辅助能力**：包括涂鸦作画（`wanx-sketch-to-image-lite`）、人像风格重绘（`wanx-style-repaint-v1`）、AI试衣（`aitryon-plus`）、FaceChain人物写真、WordArt锦书文字艺术等。其中 FaceChain 需先完成人物形象训练再生成写真，而 WordArt 锦书则专注于汉字纹理与变形 [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md)。 +- **创意工具与垂直模型**：包括图像画面扩展（`image-out-painting`）、背景生成（`wanx-background-generation-v2`）、人物实例分割（`image-instance-segmentation`）、图像擦除补全（`image-erase-completion`）、AI试衣（`aitryon-plus`）、FaceChain人像写真、WordArt锦书文字艺术等。这些模型多采用异步调用，且普遍限定于华北2（北京）地域 [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md)。 -> **注意**：文档中存在模型命名与能力描述不一致的情况。例如，`wan2.6-t2i` 在 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) 中明确标注为“支持HTTP同步调用”，但同系列 `wan2.5-t2i-preview` 及更早版本则“不支持HTTP同步调用”；而 `wan2.7-image-pro` 在 [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) 中声明“仅文生图场景支持4K分辨率”，但未说明组图生成的最高分辨率限制。开发者应以实际调用返回的 `400 Bad Request` 错误码及官方控制台模型详情页为准。 +> **注意**：文档中存在模型命名不一致问题。例如 `wan2.6-t2i`（文生图专用）与 `wan2.6-image`（支持图文混排）属不同能力栈，但文档未明确区分其适用边界；另 `wan2.7-image-pro` 在文生图场景支持4K，但在图像编辑场景仅支持2K，该限制未在所有相关文档中同步说明。 ## 关键参数 -所有图像API均通过 `parameters` 对象传递核心控制参数，常见字段如下： +所有图像API均通过 `parameters` 对象控制输出行为，核心参数如下： -- `size` / `resolution` / `aspect_ratio`：控制输出尺寸。格式多样，如 `"1024*1024"`（万相V1/V2）、`"2K"`（万相2.7）、`"1k"`（可灵、Vidu）、`"1:1"`（可灵、Vidu）。总像素范围普遍为 `512×512` 至 `2048×2048`，`wan2.7-image-pro` 文生图支持 `4K`（`3840×2160`）[万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md)。 -- `n`：生成图片张数，取值范围因模型而异：`1–6`（千问系列）、`1–9`（可灵）、`1–4`（创意海报、鞋靴模特）。 -- `watermark`：布尔值，控制是否添加平台水印，默认 `true`，部分模型（如 `wan2.7-image-pro`）支持设为 `false`。 -- `prompt_extend`：启用智能提示词扩展，返回优化后的提示词及推理过程，会增加响应时间（Z-Image、万相2.6等支持）。 -- `style_index` / `style_ref_url`：用于人像风格重绘，前者指定预置风格索引，后者传入自定义风格参考图。 -- `X-DashScope-Async`：**必选请求头**，异步调用必须设为 `"enable"`；缺失将报错 `"current user api does not support synchronous calls"`。 +- `size`：指定输出分辨率。格式支持 `宽*高`（如 `"1024*1024"`）、预设档位（如 `"1K"`、`"2K"`、`"4K"`）或比例（如 `"4:3"`）。不同模型支持范围不同：`qwen-image-*` 要求总像素在 `512×512` 至 `2048×2048` 之间；`wan2.6-t2i` 限定总像素在 `[1280×1280, 1440×1440]`；`vidu` 系列支持 `1K/2K/4K` 档位。未指定时，各模型按默认规则推导（如 `qwen-image-edit` 默认总像素接近 `1024×1024`，宽高比继承最后一张输入图）。 + +- `n`：生成图像张数。`qwen-image-max` 固定为 1 张；`qwen-image-2.0-pro`、`wan2.5-i2i-preview` 等支持 `1–6`；`kling/kling-v3-omni-image-generation` 在组图模式下通过 `series_amount` 指定 `2–9` 张。 + +- `watermark`：布尔值，控制是否添加平台水印。多数模型默认 `true`，生产环境建议显式设为 `false`。 + +- `prompt_extend`：仅 `z-image-turbo` 等部分模型支持，启用后返回优化提示词及推理过程，但增加响应延迟。 + +- 其他功能型参数：`aspect_ratio`（`kling` 系列）、`resolution`（`kling`）、`ref_prompt_weight`（背景生成）、`dilate_flag`（擦除补全）、`thinking_mode`（`wan2.7-image-pro`）等，需按具体模型文档使用。 ## 使用方式 ### 调用模式 -- **同步调用**：适用于耗时较短（通常 < 15s）的模型，如 `z-image-turbo`、`wan2.6-t2i`、`qwen-image-*`（Pro/Plus系列默认同步）。一次HTTP POST即可返回结果，无需轮询。 -- **异步调用**：适用于耗时较长（1–2分钟）的模型，如万相V1、局部重绘、虚拟模特、背景生成等。流程分两步： - 1. `POST /api/v1/services/.../generation` 创建任务，获取 `task_id`； - 2. `GET /api/v1/tasks/{task_id}` 轮询状态，直至 `task_status == "SUCCEEDED"` 后获取 `output.results[].url`。 - -### 地域与域名 -- 华北2（北京）、新加坡、美国（弗吉尼亚）地域拥有独立API Key与请求地址，**不可混用**。 -- 强烈建议迁移至业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），以获得更高性能与稳定性；旧域名（`dashscope.aliyuncs.com`）仍兼容但非最优 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)。 +- **同步调用**：适用于响应快（通常 < 10s）的模型，如 `qwen-image-2.0-pro`（北京/新加坡地域）、`wan2.6-t2i`（仅 `wan2.6`）、`z-image-turbo`（北京地域）、`wan2.7-image-pro`。请求地址统一为 `POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation`，需配置 `X-DashScope-Sse: enable` + `parameters.stream: true` 才支持流式图文混排输出。 + +- **异步调用**：适用于耗时较长（1–2分钟）的模型，如 `wanx-v1`、`wanx-x-painting`、`image-out-painting`、`wanx-background-generation-v2` 等。流程分两步： + 1. 创建任务：`POST {endpoint}`，返回 `task_id`； + 2. 轮询结果：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`（部分模型如 `image-instance-segmentation` 明确要求此路径）。 -### 认证与环境 -- 必须配置 `Authorization: Bearer $DASHSCOPE_API_KEY` 请求头。 -- 推荐通过环境变量管理API Key，并使用DashScope SDK（Python/Java）简化调用逻辑。 +### 必要配置 +- **API Key 与地域绑定**：华北2（北京）、新加坡、美国（弗吉尼亚）地域各自独立管理 API Key 与请求域名，跨地域调用将鉴权失败。强烈建议迁移至业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），以获得更高性能与稳定性 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)。 +- **请求头强制项**：异步调用必须包含 `X-DashScope-Async: enable`；同步调用需 `Content-Type: application/json` 和 `Authorization: Bearer $DASHSCOPE_API_KEY`。 +- **输入格式**：文生图使用 `input.prompt` 或 `input.messages`；图生图/编辑类模型使用 `input.messages` 数组，内含 `{"text": "..."}` 和 `{"image": "url"}` 对象；部分旧模型（如 `wanx-v1`）仍用 `input.ref_image` 字段。 ## 限制和注意事项 -- **免费额度与计费**：所有模型均提供500张免费额度（有效期90天），主账号与RAM子账号共享。超出后按模型单价计费（如 `wanx-v1`: 0.16元/张，`wanx-style-repaint-v1`: 0.12元/张），仅对**成功生成的输出图片**收费 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **图片URL要求**：输入图片URL必须公网可访问、无中文路径、支持HTTP/HTTPS。若下载失败，错误码为 `BadRequest.InputDownloadFailed`，需检查链接有效性或上传至OSS等云存储 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **限流策略**：主账号与RAM子账号共用QPS/RPS限制（常见为2 QPS），同时处理中任务数上限为1–5个，超限将返回 `429 Too Many Requests`。 -- **模型可用性**：部分模型（如 `wanx-x-painting`、`wanx-virtualmodel`、`shoemodel-v1`）当前仅限免费体验，额度用尽后不可调用且不支持付费，文档已明确提示替代方案 [图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md)。 +- **地域与模型可用性**：`qwen-mt-image`、`vidu`、`kling`、`virtualmodel-v2`、`shoemodel-v1`、`wanx-poster-generation-v1` 等全部限定于华北2（北京）地域；`z-image-turbo` 新加坡地域仅支持专属域名调用；`wan2.6-t2i` 在美国（弗吉尼亚）支持同步调用，但 `wan2.5` 及以下版本不支持。 + +- **免费额度与计费**：所有模型均提供 500 张免费额度（有效期 90 天），主账号与 RAM 子账号共享。`wanx-x-painting`、`wanx-virtualmodel`、`shoemodel-v1`、`wanx-poster-generation-v1`、`image-instance-segmentation`、`image-erase-completion` 等模型无计费单价，免费额度用尽即停用；其余模型如 `wanx-v1`（0.16元/张）、`wanx-style-repaint-v1`（0.12元/张）按成功生成图片计费。 + +- **输入约束**： + - 图片 URL 必须公网可访问、无中文路径、支持 HTTP/HTTPS； + - 图像尺寸：多数模型要求单边 ≥ 512px 且 ≤ 4096px，文件大小 ≤ 10MB； + - 局部编辑类模型（如 `wanx-x-painting`）需提供 `mask_image_url`，且涂抹区域需为非零像素； + - `image-erase-completion` 的 `mask_url` 必须与原图同尺寸，非零区域为擦除目标。 + +- **错误处理**：常见报错 `BadRequest.InputDownloadFailed` 表示图片 URL 不可达，需检查网络权限与 OSS/Bucket ACL 配置；缺失 `X-DashScope-Async` 头将直接返回“不支持同步调用”错误。 ## 来源文档 - [常见问题](../../raw/model-api-reference/image-generation/image-faq.md) -- [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) +- [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) -- [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) -- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) -- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) +- [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-图像生成与编辑2.6 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) -- [万相-涂鸦作画API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) +- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) +- [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) -- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) -- [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) - [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) -- [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) +- [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) +- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) -- [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) +- [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) - [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [创意文字WordArt锦书](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) +- [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md index 4b190623..3aeeeb18 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md @@ -1,38 +1,39 @@ # knowledge -knowledge 是百炼平台提供的知识检索与问答能力，通过统一的应用网关 API 提供语义检索和基于知识库的流式问答服务。该能力不依赖底层 OpenAPI（如 `CreateIndex` 等 RPC 接口），而是面向应用层提供标准化 HTTP REST 接口，适用于 RAG 场景下的快速集成。详细设计与行为请参考 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 +knowledge 是百炼平台提供的知识增强型 AI 服务模块，支持基于私有知识库的语义检索与多阶段智能问答。其 API 位于 DashScope 应用网关体系下，采用 RESTful 接口设计，与 OpenAPI 的索引管理类 RPC 接口（如 `CreateIndex`）在调用方式、鉴权机制和 Base URL 上存在本质差异。开发者需通过业务空间 ID 构造专属域名，并使用 API Key 进行 Bearer 鉴权。 ## 支持的模型/功能 -- **知识检索**：跨多个知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），适用于召回阶段。 -- **知识问答**：端到端智能问答，支持 SSE 流式响应，输出包含规划（planning）、工具调用（tool calling）和生成（generation）三个逻辑阶段，需配合已部署的知识库与应用配置使用。 - > **注意**：知识问答接口 `/api/v2/apps/knowledge/chat` 的三阶段输出行为与部分旧版文档描述的“单次生成”存在差异，以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 中的 SSE 分阶段说明为准。 +- **知识检索**：跨多个已发布知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），适用于构建自定义 RAG 流程。 +- **知识问答**：端到端的流式问答能力，通过 SSE 返回三阶段响应（规划 → 工具调用 → 生成），自动完成知识检索、上下文组装与大模型生成。 +该能力不依赖特定大模型选型，底层由平台统一调度适配；但需确保所绑定的知识库已完成发布（参见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)）。 ## 关键参数 -| 参数 | 类型 | 必填 | 说明 | +| 参数 | 说明 | 必填 | 示例 | |------|------|------|------| -| `workspaceId` | string | 是 | 业务空间 ID，用于构造 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`），非用户 UID 或 Project ID。获取路径见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 | -| `Authorization` | header | 是 | `Bearer `，API Key 需在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 申请。 | -| `top_k`（检索） | integer | 否 | 检索返回切片数量，默认 5，最大 100。 | -| `stream`（问答） | boolean | 否 | 是否启用 SSE 流式响应，默认 `true`；设为 `false` 将返回完整 JSON 响应（非流式）。 | +| `workspaceId` | 业务空间唯一标识，用于构造 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`） | 是 | `ws-abc123` | +| `Authorization` | 请求头中携带 `Bearer `，API Key 需在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 获取 | 是 | `Bearer ak-xxx` | +| `top_k`（检索接口） | 检索返回的最大切片数，默认 `5`，最大 `20` | 否 | `10` | +| `stream`（问答接口） | 是否启用 SSE 流式响应，默认 `true` | 否 | `false` | + +> **注意**：`workspaceId` 与 OpenAPI 中的 `project_id` 或 `tenant_id` 无映射关系，不可混用；[知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 明确指出其 Base URL 与 OpenAPI 完全隔离，若错误复用 `https://dashscope.aliyuncs.com` 将导致 404。 ## 使用方式 -1. **构造请求地址**：将 `workspaceId` 替换进 Base URL，例如 `https://my-workspace.cn-beijing.maas.aliyuncs.com`； -2. **发起请求**： - - 知识检索：`POST /api/v1/indices/knowledge/search`，Body 包含 `query` 和可选 `indices`（知识库 ID 列表）； - - 知识问答：`POST /api/v2/apps/knowledge/chat`，Body 需包含 `messages`（对话历史）及 `app_id`（对应知识问答应用 ID）； -3. **处理响应**： - - 检索接口返回标准 JSON，含 `results` 数组； - - 问答接口默认流式（SSE），需按 `event: chunk` 解析；若 `stream=false`，则响应为单次 JSON，结构与流式末尾 `event: done` payload 一致。 +1. 在控制台完成知识库创建、上传、解析与发布（详见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)）； +2. 获取业务空间 ID 与 API Key； +3. 构造请求： + - 知识检索：`POST https://{workspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/indices/knowledge/search` + - 知识问答：`POST https://{workspaceId}.cn-beijing.maas.aliyuncs.com/api/v2/apps/knowledge/chat` +4. 发送 JSON body（如检索需含 `query` 字段，问答需含 `messages` 数组）。 ## 限制和注意事项 -- **鉴权与域名强绑定**：Base URL 必须含 `workspaceId`，且 `Authorization` 头中的 API Key 必须属于该 workspace 下的有效密钥，否则返回 `401 Unauthorized`； -- **限流策略**：默认按用户维度限流 25 QPS，超限返回 `429 Too Many Requests`，不可通过增加并发绕过； -- **知识库依赖**：知识问答接口不接受裸知识库 ID，必须传入已绑定知识库的 `app_id`（即 Model Studio 中发布的“知识问答应用”ID），该约束未在所有前端文档中明确强调，实际行为以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 为准； -- **地域固定**：当前仅支持 `cn-beijing` 地域，URL 路径中硬编码该 region，不支持切换。 +- **限流策略**：默认用户维度 25 QPS，超限返回 `429 Too Many Requests`； +- **知识库状态**：仅已“发布”状态的知识库参与检索/问答，草稿或未发布状态不可见； +- **SSE 兼容性**：知识问答接口强制要求客户端支持 EventSource 或手动解析 `text/event-stream` 响应体； +- **路径差异**：`/api/v1/indices/knowledge/search` 与 `/api/v2/apps/knowledge/chat` 分属不同版本路径，v1 不支持问答，v2 不支持纯检索——二者功能正交，不可替代。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md index 840e0834..8d5ab927 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md @@ -1,56 +1,74 @@ # long term memory new -[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化记忆管理能力，支持将对话自动提炼为语义化记忆片段，并提供增删改查、语义搜索及用户画像构建等核心功能。该能力基于模型驱动的记忆提取与检索，适用于需要持久化用户上下文、偏好和意图的智能体应用。详细接口定义与行为规范请参见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化用户记忆管理能力，支持自动从对话中提取关键信息、构建用户画像，并提供语义搜索、增删改查等完整生命周期操作。该功能基于专用记忆库和规则引擎实现，适用于需要长期维护用户上下文的智能体应用。所有接口均通过 REST API 或 `agentscope-runtime` SDK 调用，需使用 DashScope API Key 认证。 ## 支持的模型/功能 -- **记忆提取**：通过 `AddMemory` 自动从多轮对话中识别并生成结构化记忆片段（如提醒、偏好、计划等），支持 `messages`（对话数组）或 `custom_content`（纯文本）两种输入模式。 -- **语义搜索**：`SearchMemory` 基于向量相似度召回相关记忆，支持 `top_k`、`min_score`、`enable_rerank` 等控制参数，适用于上下文增强推理。 -- **画像建模**：配合 `CreateProfileSchema` 和 `GetUserProfile`，可基于记忆数据动态构建用户画像（需提前配置画像模板），详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中的“核心组件”章节。 -- **全生命周期管理**：提供 `ListMemory`（分页查询）、`DeleteMemory`（按 ID 删除）、`UpdateMemory`（内容覆盖更新）标准 CRUD 接口。 - -> **注意**：Python SDK 中 `UpdateMemory` 尚未封装为高层工具类，需直接调用 REST API；而 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 均已在 `agentscope-runtime>=1.1.5` 中提供异步封装，具体用法见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的示例代码。 +- **核心能力**：自动记忆提取（从 `messages` 中识别意图与事实）、语义搜索、用户画像生成与更新、多记忆库隔离管理。 +- **画像模板（Profile Schema）**：支持创建、查询、更新、删除画像模板（如“健康习惯”“日程偏好”），用于约束和标准化用户画像字段；详情见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +- **记忆片段规则（Project）**：每个记忆库可配置多个规则，控制提取逻辑（如仅提取提醒类内容），`project_id` 可显式指定或由系统自动选择默认规则。 +- **不依赖特定大模型**：底层由平台统一模型服务处理，开发者无需指定推理模型；但提取质量受输入对话结构影响，建议保持清晰的 user/assistant 角色划分。 ## 关键参数 | 参数名 | 类型 | 必填 | 说明 | |--------|------|------|------| -| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），用于隔离不同用户的数据空间 | -| `memory_library_id` | string | 否 | 记忆库 ID（≤32 字符）；未传时使用默认记忆库 | -| `project_id` | string | 否 | 记忆片段规则 ID；未传时使用对应记忆库的默认规则 | -| `top_k` | integer | 否 | `SearchMemory` 最大召回数（1–100，默认 10） | -| `min_score` | double | 否 | `SearchMemory` 相似度阈值 [0,1]（默认 0.3） | -| `page_num` / `page_size` | integer | 否 | `ListMemory` 分页参数（默认 page_num=1, page_size=10） | -| `meta_data` | object | 否 | 用户自定义键值对，随记忆片段持久化存储（增量更新仅对 `UpdateMemory` 生效） | +| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），所有操作均以此为作用域边界 | +| `messages` / `custom_content` | array / string | 互斥 | `messages` 用于对话自动提取（最多 50 条，一问一答计 2 条）；`custom_content` 用于直接写入自定义文本（≤512 字符） | +| `memory_library_id` | string | 否 | 显式指定记忆库 ID（≤32 字符），未传时使用默认库；获取方式见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) | +| `top_k`, `min_score` | integer, double | 否 | `SearchMemory` 专用：召回数量（1–100，默认 10）和最小相似度阈值（[0,1]，默认 0.3） | +| `page_num`, `page_size` | integer | 否 | `ListMemory` 分页参数（默认 page_num=1, page_size=10） | + +> **注意**：`AddMemory` 的 `profile_schema` 参数在文档中描述为“画像模板 ID”，但实际调用时若传入非有效 ID 将静默忽略，且无错误提示——此行为与 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中“必填”标注矛盾，应以实际 API 行为为准（即该参数为可选）。 ## 使用方式 -1. **认证**：所有请求需在 Header 中携带 `Authorization: Bearer $DASHSCOPE_API_KEY`，API Key 获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -2. **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` -3. **推荐路径**： - - 新增记忆：`POST /add`（传 `messages` 或 `custom_content`） - - 检索记忆：`POST /memory_nodes/search`（传 `user_id` + `messages`） - - 查询列表：`GET /memory_nodes?user_id=xxx&page_num=1&page_size=10` - - 删除/更新：`DELETE /memory_nodes/{memory_node_id}` / `PATCH /memory_nodes/{memory_node_id}` -4. **SDK 调用**（Python）： - ```python - from agentscope_runtime.tools.modelstudio_memory import AddMemory, SearchMemory, ListMemory - # 初始化后调用 arun()，注意 await 并显式 close() - ``` +### 1. 基础认证 +所有请求需在 Header 中携带： +```http +Authorization: Bearer $DASHSCOPE_API_KEY +Content-Type: application/json +``` +API Key 获取路径：[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 + +### 2. 主要接口调用示例 +- **添加记忆**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -d '{"user_id":"u123","messages":[{"role":"user","content":"明天9点开会"}]}' + ``` +- **搜索记忆**： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -d '{"user_id":"u123","messages":[{"role":"user","content":"我明天有什么安排？"}],"top_k":5}' + ``` +- **Python SDK（推荐）**： + 安装 `agentscope-runtime>=1.1.5` 后，直接使用封装类（如 `AddMemory`, `SearchMemory`），详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中的 Python 示例。 + +### 3. 画像模板操作 +需先调用 `CreateProfileSchema` 定义字段结构（如 `{"name":"string","age":"integer"}`），再通过 `GetUserProfile` 获取结构化画像。模板 ID 在控制台记忆库详情页可见。 ## 限制和注意事项 -- **限流策略**（阿里云账号级别）： - - 全部接口总计 ≤ 3000 QPM； - - `AddMemory` ≤ 120 QPM； - - `SearchMemory` ≤ 300 QPM。 -- **内容限制**： - - `messages` 最多 50 条（一问一答计为 2 条）； - - `custom_content` 最大 512 字符； - - `user_id`、`memory_library_id` 等字符串长度严格校验，超长将返回 400 错误。 -- **数据时效性**：当前生成的记忆片段与用户画像**无自动失效机制**，需业务侧自行维护生命周期。 -- **兼容性**：`UpdateMemory` 的 `timestamp` 字段为秒级 Unix 时间戳（非毫秒），且仅影响元数据时间字段，不改变向量索引时间点。 -- **调试建议**：首次集成时，优先使用 cURL 示例验证基础流程，再迁移到 SDK；错误响应中 `request_id` 是排查问题的关键标识。 +- **限流策略（阿里云账号级别）**： + - 全部接口总计 ≤ 3000 QPM + - `AddMemory` ≤ 120 QPM + - `SearchMemory` ≤ 300 QPM + 超限返回 `429 Too Many Requests`。 + +- **数据持久性**： + 记忆片段与用户画像**无自动过期机制**，需开发者自行管理生命周期（如定期调用 `DeleteMemory`）。 + +- **内容长度限制**： + - `custom_content` ≤ 512 字符 + - `messages` 中单条 `content` 长度未明确限制，但整体 `messages` 数组 ≤ 50 条 + - `meta_data` 对象大小建议 ≤ 1 KB（避免影响序列化性能） + +- **重要约束**： + - `DeleteMemory` 和 `UpdateMemory` 接口**不接受 `user_id` 作为请求体参数**（仅路径参数 `memory_node_id` + 可选查询参数 `memory_library_id`），与 `AddMemory`/`SearchMemory` 的参数设计不一致，需特别注意。 + - `UpdateMemory` 的 `custom_content` 为**全量覆盖**，非增量更新（`meta_data` 支持增量合并）。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md index ce83a498..5a0df0fa 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md @@ -1,61 +1,68 @@ # [managed agents](../guides/managed-agents.md) api -Managed Agents API 是百炼平台提供的智能体托管运行时服务，负责会话生命周期管理、沙箱环境调度、工具执行协调与事件流分发。开发者通过 REST 或 SDK 创建 Agent（智能体配置）、Environment（执行沙箱）、Session（运行实例），并以事件驱动方式与 Agent 交互。所有资源均归属工作空间，支持版本控制、软归档与细粒度权限隔离。 +Managed Agents API 是百炼平台提供的智能体托管运行时服务，由平台统一管理会话生命周期、执行沙箱、工具调用与事件流。开发者通过 REST 或 SDK 创建 Agent（智能体配置）、Environment（运行环境）、Session（运行实例）并驱动任务执行，所有资源均按工作空间隔离。该 API 适用于构建可复用、可审计、可扩展的智能体应用。 ## 支持的模型与功能 -- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)），不支持自定义模型或外部模型接入。 -- **核心功能**： - - Agent：封装模型、系统提示词、工具列表与 Skill 挂载，支持版本化与软归档； - - Environment：定义沙箱类型（如 `"type": "cloud"`）及预装依赖，独立于 Agent 管理，可被多 Session 复用； - - Session：绑定 Agent 版本快照与 Environment 快照，状态机驱动（`idle` → `running` → `idle`/`terminated`）； - - Event：支持用户消息、工具调用审批、函数结果回填等原子事件，通过 SSE 流式推送； - - File：支持 ≤20 MB 文件直传，审核通过后（`status: "available"`）可作为消息内容或挂载至沙箱； - - Skill：以 zip 包形式封装工具组合，需经安全扫描后按具体版本号挂载到 Agent。 +- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)），模型 ID 通过 `model.id` 字段指定，不支持自定义模型或外部模型接入。 +- **核心功能模块**： + - `Agent`：封装模型、系统提示词、工具集与技能（Skill），支持版本化管理与软归档； + - `Environment`：定义沙箱类型（如 `"cloud"`）与预装依赖，独立于 Agent 生命周期，可被多 Session 复用； + - `Session`：绑定 Agent 快照与 Environment 快照的运行实例，状态机驱动（`idle` → `running` → `idle`/`terminated`）； + - `Event`：支持同步发送（用户消息、中断、工具回填）与 SSE 流式订阅，含 `session_status` 状态变更通知； + - `Skill`：以 ZIP 包形式上传工具组合，需经安全扫描后方可挂载，挂载时必须指定具体版本号（不支持 `latest`）； + - `File`：支持上传至工作空间（上限 20 MB/文件，100 GB/空间），审核通过（`status: available`）后可用于消息内容或挂载至沙箱。 -> **注意**：文档中多次提及 `"qwen-plus"` 为示例模型，但 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 文档未明确列出当前支持的全部模型 ID；实际可用模型请以控制台或 `/agents` 创建接口返回的 `model.id` 枚举为准，避免硬编码。 +> **注意**：文档 2 的快速开始示例中使用 `model: "qwen-plus"` 为字符串简写，而文档 1 的 API 总览明确要求 `model` 为对象结构 `{"id": "qwen-plus"}`。实际请求体必须遵循 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中定义的 JSON Schema，否则返回 400 错误。 ## 关键参数 -| 参数 | 位置 | 说明 | 示例 | -|------|------|------|------| -| `DASHSCOPE_API_KEY` | Header | 鉴权凭证，格式 `Bearer ` | `sk-xxx` | -| `workspace_id` | Endpoint 路径 | 工作空间 ID，用于构造 Base URL | `ws_xxxxxxxxxxxx` | -| `region` | Endpoint 路径 | 当前仅支持 `cn-beijing` | `cn-beijing` | -| `agent.model.id` | Agent 创建请求体 | 模型 ID，必须为平台支持的托管模型 | `"qwen-plus"` | -| `environment.config.type` | Environment 创建请求体 | 沙箱类型，目前仅 `"cloud"` 可用 | `"cloud"` | -| `session.agent` / `session.environment_id` | Session 创建请求体 | 引用已创建的 Agent ID 与 Environment ID | `"agent_xxx"`, `"env_xxx"` | -| `event.input` | Event 发送请求体 | 用户消息数组，遵循 OpenAI-style message 格式 | `[{"role":"user","content":[{"type":"text","text":"..."}]}]` | +| 参数 | 位置 | 类型 | 必填 | 说明 | +|------|------|------|------|------| +| `Authorization` | Header | string | 是 | `Bearer `，从控制台获取并配置为环境变量（见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)） | +| `workspace_id` | Endpoint path | string | 是 | 工作空间 ID（如 `ws_xxxxxxxxxxxx`），与地域拼接构成 base URL | +| `region` | Endpoint path | string | 是 | 当前仅支持 `cn-beijing` | +| `agent.id` | Session 创建体 | string | 是 | Agent 唯一标识，创建 Session 时锁定其当前版本快照 | +| `environment_id` | Session 创建体 | string | 是 | Environment 唯一标识，创建 Session 时绑定其当前配置快照 | +| `input` | Event 发送体 | array | 是 | 消息数组，每项含 `role`（`user`/`assistant`）、`type`（`message`）、`content`（含 `text`/`image_url` 等） | ## 使用方式 -1. **环境准备**：导出 `DASHSCOPE_API_KEY` 与 `AGENTSTUDIO_URL`（形如 `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`），[详见快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)； -2. **资源创建**（建议复用）： - - 创建 Agent：指定 `model.id`、`system` 提示词、可选 `skills` 列表； - - 创建 Environment：指定 `config.type` 及其他沙箱参数； -3. **会话交互**： - - 创建 Session，绑定 Agent 与 Environment； - - `POST /sessions/{id}/events` 发送用户消息触发执行； - - `GET /sessions/{id}/events/stream` 建立 SSE 连接，监听 `session_status` 与 `message` 事件； -4. **SDK 推荐**：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24（[API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中明确要求）。 +1. **初始化**：设置 `DASHSCOPE_API_KEY` 与 `AGENTSTUDIO_URL`（形如 `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`）； +2. **资源准备**： + - 创建 Agent（`POST /agents`），指定 `model.id`、`system` 提示词等； + - 创建 Environment（`POST /environments`），配置 `config.type`（如 `"cloud"`）； +3. **启动会话**： + - 创建 Session（`POST /sessions`），传入 `agent` 和 `environment_id`； + - 发送用户消息（`POST /sessions/{session_id}/events`），`input` 格式需严格符合 [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) 规范； +4. **接收结果**： + - 订阅 SSE 流（`GET /sessions/{session_id}/events/stream`），监听 `session_status` 变更及 `message` 事件； + - 或轮询事件历史（`GET /sessions/{session_id}/events`）； + +SDK 调用需确保版本兼容：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24（见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)）。 ## 限制和注意事项 -- **配额限制**：单文件上传上限 20 MB，工作空间总存储上限 100 GB，文件保留期 30 天（[File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； -- **版本锁定**：Session 创建时锁定 Agent 的 `version` 与 Environment 快照，后续更新不影响已有 Session； -- **归档非删除**：Agent/Environment/Session 归档为软操作（`archived_at` 字段填充），已归档资源仍可查询，但不可用于新建 Session； -- **硬删除风险**：`DELETE /environments/{id}` 或 `DELETE /sessions/{id}` 为不可逆操作，将彻底清除配置或事件历史； -- **SSE 连接**：客户端需处理连接中断重试，并根据 `session_status` 事件判断会话终态（`idle` 或 `terminated`），避免无限等待； -- **Skill 安全约束**：Skill 必须通过安全扫描（`status: "active"`）才可挂载，且挂载时必须指定具体 `version`，不支持 `latest` 动态引用。 +- **配额限制**：单文件上传上限 20 MB，工作空间总存储上限 100 GB，文件保留期 30 天（见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； +- **版本与更新语义**： + - Agent 更新为**全量替换 + 乐观锁**：请求体必须包含当前 `version`，不一致返回 409；成功后 `version` 自动递增； + - Environment 更新也为**全量替换**，但已绑定的运行中 Session 仍使用创建时的快照，不受影响； +- **归档与删除**： + - 归档（`archive`）均为软操作，资源仍可查询，已绑定 Session 不受影响； + - 删除（`DELETE`）为硬操作，不可恢复（如 `DELETE /environments/{env_id}` 清除全部配置）； +- **安全约束**： + - Skill 与 File 上传后需通过安全扫描，仅 `status: active` 或 `available` 状态才可使用； + - Skill 挂载到 Agent 时必须指定具体 `version`，后续上传新版本不影响已挂载的 Agent； +- **状态机约束**：Session 仅在 `idle` 状态下可接收新消息；`running` 状态下仅支持中断、工具审批等特定操作（见 [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md)）。 ## 来源文档 -- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) -- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) +- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) +- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) - [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) -- [File](../../raw/application-api-reference/managed-agents-api/files-api.md) - [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) +- [File](../../raw/application-api-reference/managed-agents-api/files-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md index f7559e9d..7b127a74 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md @@ -1,35 +1,43 @@ # model production -`model production` 是百炼平台中用于将模型投入实际应用的核心能力集合，涵盖从微调训练到在线服务部署的完整生命周期。开发者可通过 API 或控制台完成模型定制与发布，适用于私有化模型迭代与业务集成场景。该能力依赖于底层计算资源调度与模型服务框架协同工作。 +`model production` 是百炼平台中用于将模型投入实际应用的核心能力集合，涵盖从微调训练到在线服务部署的完整生命周期。开发者可通过 API 管理微调任务与部署实例，实现模型的定制化与规模化交付。该能力依赖于统一的模型标识（`model_id`）和资源隔离机制，适用于业务场景下的迭代演进。 -## 支持的模型/功能 +## 支持的模型与功能 -- **微调训练（Fine-tuning）**：支持基于基础大模型（如 Qwen 系列）进行监督微调，适配垂直领域任务（如客服问答、金融报告生成）。 -- **模型部署（Deployment）**：支持将微调完成的模型或通过 `import_model` 导入的第三方模型，一键发布为 HTTP 可调用的在线推理服务。 -- **版本管理**：每个微调任务和部署实例均自动关联唯一 ID 与版本号，便于灰度发布与回滚。 -> **注意**：文档中未明确说明是否支持 LoRA 微调以外的参数高效方法；实际使用请参考 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 中的 `training_type` 参数定义。 +- **微调训练**：支持基于 Base Model（如 Qwen 系列）启动监督微调（SFT）任务，输入格式为标准 JSONL，支持 LoRA 等轻量适配方式 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) +- **模型部署**：支持将微调完成的模型或通过 `import_model` 导入的第三方模型部署为 HTTP 推理服务，提供自动扩缩容、版本灰度、流量切分等生产级能力 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) +- **功能边界**：当前不支持直接对已部署服务执行热更新或参数重载；模型版本变更需通过新建部署或蓝绿切换实现。 ## 关键参数 | 参数 | 说明 | 示例值 | |------|------|--------| -| `model_id` | 基础模型标识符（如 `qwen2-7b-instruct`）或已微调模型 ID | `"qwen2-7b-instruct"` | -| `training_type` | 微调类型，当前仅支持 `"full"` 和 `"lora"`（见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)） | `"lora"` | -| `endpoint_name` | 部署后服务的唯一域名前缀，全局唯一 | `"my-qa-service"` | -| `instance_type` | 推理实例规格，影响并发与延迟（详见 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md)） | `"ecs.gn7i-c16g1.4xlarge"` | +| `model_id` | 唯一模型标识，微调任务输出或导入后生成 | `ft-qwen2-7b-20240501-123456` | +| `deployment_name` | 部署实例名称，全局唯一且不可修改 | `prod-chat-v2` | +| `instance_type` | 推理实例规格，影响并发与延迟 | `gpu.2xlarge`（仅限部署） | +| `training_type` | 微调类型，当前仅支持 `sft` | `sft`（仅限微调） | + +> **注意**：文档中未明确 `instance_type` 的可选枚举值范围，实际使用请以 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 中最新 `GET /v1/deployments/instance-types` 接口返回为准；部分旧文档示例仍列出已下线的 `cpu.small` 类型，属过时信息。 ## 使用方式 -1. **发起微调任务**：调用 `POST /api/v1/fine_tuning_jobs`，传入训练数据集 URL、`model_id` 和 `training_type`；任务状态轮询 `GET /api/v1/fine_tuning_jobs/{job_id}`。 -2. **部署模型**：微调成功后，获取输出的 `fine_tuned_model_id`，调用 `POST /api/v1/deployments` 提交部署请求。 -3. **调用服务**：部署成功后，通过返回的 `endpoint_url` 发送 `POST /v1/chat/completions` 请求（兼容 OpenAI 格式）。 +1. **微调流程**： + - 调用 `POST /v1/fine_tuning/jobs` 提交训练任务，指定 `base_model_id`、训练数据集 ID 及超参 + - 监听 `status` 字段（`queued` → `running` → `succeeded`），成功后获取输出 `model_id` + +2. **部署流程**： + - 调用 `POST /v1/deployments`，传入上一步得到的 `model_id` 及 `deployment_name` + - 部署就绪后，通过 `endpoint_url` 发起推理请求（如 `POST https:///v1/chat/completions`） + +完整交互链路详见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 与 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 的 API 规范。 ## 限制和注意事项 -- 单次微调任务最大训练时长为 72 小时，超时自动终止；数据集大小上限为 10 GB（[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)）。 -- 每个账号默认最多同时运行 5 个部署实例，超出需提工单扩容（[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)）。 -- 微调任务不支持跨区域迁移；部署实例一旦创建，其 `instance_type` 不可变更，需重建部署。 -> **注意**：两篇原始文档均未提及模型格式兼容性要求（如是否支持 GGUF、AWQ 等量化格式），实际导入前请确认模型已按百炼规范转换并验证加载。 +- 单个微调任务最大训练时长为 72 小时，超时将被强制终止 +- 每个 `deployment_name` 在同一 Region 下全局唯一，重名请求返回 `409 Conflict` +- 微调任务不支持跨 Region 复制；部署实例必须与模型所在 Region 一致 +- 模型部署后，其底层 `model_id` 不可变更——若需替换模型，须新建部署或删除重建 +- 免费试用额度仅覆盖微调计算资源，部署实例按实际 GPU 小时计费 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md index b3398b43..a2341bf6 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md @@ -1,68 +1,57 @@ # [more](more.md) about models -阿里云百炼平台提供多种模型调用机制与配套能力，涵盖同步/异步任务处理、多业务空间隔离、文件上传、连接优化等核心场景。本文面向开发者，系统梳理模型服务的关键能力、参数配置、使用方式及约束条件，帮助构建稳定、高效、安全的模型集成方案。 +百炼平台提供多种模型调用机制与配套能力，涵盖同步/异步任务处理、多业务空间隔离、连接复用优化及安全凭证管理。本文面向开发者，系统梳理核心能力、关键参数、使用方式及限制条件，帮助构建稳定、高效、可扩展的模型服务集成方案。 ## 支持的模型/功能 -百炼支持标准大语言模型（如 `qwen-plus`）、多模态模型（如 `qwen-vl-plus`）、图像生成（`wanx2.1-t2i-turbo`）、视频生成（`wanx2.1-kf2v-plus`）及语音识别（`paraformer-8k-v1`）等全类型模型。不同模型适用不同调用模式： +百炼支持两类主要模型调用路径： +- **标准模型**（如 `qwen-plus`、`wanx2.1-t2i-turbo`）：需通过 API Key 显式授权调用权限，支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)与 DashScope 原生接口两种方式； +- **调优后部署的模型**：仅限其所属子业务空间的 API Key 调用，无需额外模型授权，但不支持 OpenAI 兼容方式 [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 -- **同步模型**（如文本生成）：直接调用 `/chat/completions` 或 `/generation` 接口，实时返回结果； -- **异步模型**（如文生图、文生视频）：需先创建任务获取 `task_id`，再通过[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) 查询或取消，详见[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)； -- **多模态模型**：输入文件需先上传至临时存储并获取 `oss://` URL，且必须指定对应模型名称，详见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)； -- **子业务空间模型**：调用非默认空间的模型（如千问-Plus）时，**必须使用该子空间专属 API Key**，且需提前配置模型调用权限，详见[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 - -> **注意**：文档 4 中明确指出“调用在阿里云百炼[调优](https://help.aliyun.com/zh/model-studio/model-training-overview)并部署的模型，无需模型调用授权”，但文档 3 的异步任务查询接口描述中称“支持查询当前 API Key 所属阿里云主账号下的所有任务（包括该主账号下通过任意 API Key 提交的任务）”。二者存在隐含冲突：若子空间调优模型仅允许本空间 API Key 调用，则其任务不应被主账号其他 API Key 查询到。实际行为以控制台权限配置为准，建议严格遵循子空间隔离原则，避免跨空间混用 API Key。 +异步能力覆盖图像生成、视频合成、语音识别等长耗时任务，需通过 `POST /api/v1/tasks` 创建任务并配合轮询或事件通知获取结果。同步任务（如文本生成）则直接返回响应。 +> **注意**：文档 3 中称“调优后的模型仅支持通过 DashScope 调用”，但文档 4 的事件总线示例中明确包含 `paraformer-8k-v1`（ASR 模型），该模型属于调优类语音模型，且其 `user_api_unique_key` 格式与文档 3 描述一致。这表明调优模型实际也支持事件通知机制，文档 3 表述存在局限性。 ## 关键参数 -| 参数 | 说明 | 取值范围/示例 | 来源 | +| 参数 | 说明 | 取值范围/默认值 | 来源 | |------|------|----------------|------| -| `task_id` | 异步任务唯一标识符 | UUID 格式字符串，如 `a8532587-xxxx-xxxx-xxxx-0c46b17950d1` | [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) | -| `model_name` | 模型名称，用于文件上传绑定、权限校验及路由 | `qwen-plus`, `wanx2.1-t2i-turbo`, `paraformer-8k-v1` 等 | [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)、[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) | -| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒，默认 60 秒 | [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | -| `X-DashScope-OssResourceResolve: enable` | 使用 `oss://` URL 时必需的请求头 | 固定字符串 | [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) | -| `connectionPoolSize`（Java） / `limit`（Python） | SDK 连接池大小 | Java 默认 32，可调至 256；Python `aiohttp.TCPConnector.limit` 默认 100 | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | +| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒，默认 `60` | [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | +| `task_id` | 异步任务唯一标识 | UUID 格式字符串 | [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) | +| `connectionPoolSize` (Java) | HTTP 连接池最大连接数 | 默认 `32`，建议高并发场景设为 `256` | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | +| `limit_per_host` (Python) | 单主机最大连接数 | 默认 `0`（无限制），建议设为 `30` | [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | ## 使用方式 -### 1. 调用入口选择 -- **[OpenAI 兼容接口](../concepts/openai-compatible-api.md)**：适用于快速迁移或通用 SDK 集成，Base URL 为 `https://dashscope.aliyuncs.com/compatible-mode/v1`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡）； -- **DashScope 原生接口**：适用于深度定制或需调用调优模型，Base URL 为 `https://dashscope.aliyuncs.com/api/v1`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1`（新加坡）。 +### 1. 安全凭证管理 +在不可信环境（如浏览器、App）中，**必须**使用后端服务生成临时 API Key，而非直接暴露永久 Key。调用 `/api/v1/tokens` 接口，传入 `expire_in_seconds` 控制 TTL，响应返回 `token` 和 `expires_at` 时间戳 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)。 -### 2. 文件上传与引用 -调用多模态模型前，需先上传本地文件： -```bash -# 命令行工具（推荐） -dashscope oss.upload --model qwen-vl-plus --file cat.png -``` -返回 `oss://dashscope-instant/xxx/cat.png` 后，在模型请求中作为 `url` 字段传入，并**必须添加请求头** `X-DashScope-OssResourceResolve: enable`。 +### 2. 子业务空间调用 +为实现模型权限隔离与费用分账，需在子业务空间创建专属 API Key，并按地域配置正确 endpoint： +- **北京地域**：OpenAI 兼容 `base_url = "https://dashscope.aliyuncs.com/compatible-mode/v1"`；DashScope 原生 `base_url = "https://dashscope.aliyuncs.com/api/v1"`； +- **新加坡地域**：需替换 `{WorkspaceId}`，如 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`。 -### 3. 异步任务通知 -避免轮询，推荐通过事件总线接收完成通知： -- 配置 HTTP 回调 URL 或 RocketMQ 作为事件目标； -- 订阅事件类型 `dashscope:System:AsyncTaskFinish`； -- 解析回调事件中的 `data.task_id` 和 `data.task_status`，再调用 `/api/v1/tasks/{task_id}` 获取结果。 +### 3. 异步任务处理 +- **轮询模式**：调用 `GET /api/v1/tasks/{task_id}` 查询状态（QPS 限流 20），支持 `PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED` 等状态判断； +- **事件驱动模式**：配置事件总线规则，监听 `dashscope:System:AsyncTaskFinish` 事件，通过 HTTP 回调或 RocketMQ 消费通知，避免轮询资源浪费 [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)。 ### 4. 连接复用优化 -高并发场景下务必启用连接复用： -- **Java SDK**：通过 `Constants.connectionConfigurations` 配置连接池参数； -- **Python SDK**：同步调用传入 `requests.Session()`，异步调用传入 `aiohttp.ClientSession()`。 +- **Java SDK**：通过 `Constants.connectionConfigurations` 配置连接池参数（如 `connectionPoolSize`, `readTimeout`）； +- **Python SDK**：同步调用传入 `requests.Session`，异步调用传入 `aiohttp.ClientSession` 并配置 `TCPConnector`。 ## 限制和注意事项 -- **临时文件存储**：`oss://` URL 有效期严格为 **48 小时**，超期自动清理；文件与模型强绑定，不可跨模型复用；上传限流为 **100 QPS（按主账号+模型维度）**，[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) 明确警告“请勿用于生产环境、高并发及压测场景”，生产环境应使用 OSS 自建存储。 -- **临时 API Key**：由永久 API Key 生成，继承其全部权限（含模型/知识库访问限制）；**无法手动删除**，仅能等待自动过期；各地域 Endpoint 不同，需按实际地域选用。 -- **异步任务生命周期**：任务结果默认保留 **24 小时**（具体以对应模型文档为准），超时后无法查询；仅支持取消 `PENDING` 状态任务，`RUNNING` 或已完成任务不可取消。 -- **子业务空间隔离**：子空间 API Key 仅能调用本空间授权模型；调优模型**不支持 OpenAI 兼容方式调用**，必须使用 DashScope 原生接口。 -- **SDK 连接配置**：Java SDK 的 `maximumAsyncRequests` 必须 ≤ `connectionPoolSize`，否则可能阻塞；Python 异步调用中 `limit_per_host` 建议设为非零值（如 30），防止对单一域名发起过多连接。 +- **临时 API Key**：无法手动删除，到期自动失效；继承父 Key 全部权限，包括模型/知识库访问限制 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)； +- **异步任务生命周期**：成功/失败任务默认保留 24 小时，超时后数据被系统清理，查询将返回 `UNKNOWN` 状态； +- **取消任务限制**：仅支持取消 `PENDING` 状态任务，`RUNNING` 或已完成任务不可取消，错误码 `UnsupportedOperation` 明确提示此约束 [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)； +- **地域隔离**：各 Region（北京/新加坡/弗吉尼亚）的 API Key 与 Endpoint 完全独立，混用将导致 `InvalidApiKey` 错误； +- **SDK 版本要求**：Java SDK 建议 ≥ 2.12.0，Python SDK 需支持 `session` 参数传入，旧版本可能不兼容连接复用特性。 ## 来源文档 - [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) -- [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) - [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) -- [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) +- [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) - [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md index f7b792a1..d7442b07 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md @@ -1,86 +1,64 @@ # [more](more.md) models -百炼平台提供一系列面向垂直场景的专用大模型，覆盖法律、翻译、意图理解、深度研究、OCR和GUI自动化等能力。这些模型基于通义千问基座，通过领域精调、RAG增强、多模态融合或工具调用机制实现专业任务优化。开发者可通过DashScope SDK或OpenAI兼容接口调用，需注意地域、域名及参数配置差异。 +百炼平台提供一系列面向垂直场景的专用大模型，覆盖意图理解、法律咨询、机器翻译、深度研究、OCR文字识别及GUI自动化等能力。这些模型均基于通义千问系列基座模型精调或增强，具备领域适配性与高精度输出特性，适用于构建专业级AI应用。 ## 支持的模型/功能 -| 模型名称 | 用途 | 输入类型 | 关键特性 | 文档引用 | -|----------|------|----------|----------|----------| -| `farui-plus` | 法律行业问答、文书生成、合同审查 | 文本 | RAG检索增强、法律Agent、司法专属小模型 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | -| `qwen-mt-plus` | 高质量机器翻译 | 文本 | 支持术语干预、翻译记忆（TM）、领域提示 | [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | -| `tongyi-intent-detect-v3` | 意图识别与[函数调用](../concepts/function-calling.md)决策 | 文本 | 双模式输出：`INTENT_MODE`（含工具调用）或纯标签分类；支持简写单Token响应 | [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) | -| `qwen-deep-research` | 多阶段深度研究（规划→搜索→报告生成） | 文本 | 仅支持华北2（北京）地域；必须分两步调用（反问确认 + 深入研究）；支持`model_detailed_report`/`model_summary_report`输出格式 | [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) | -| `qwen3.5-ocr` | 图像文字提取与结构化解析 | 图文混合（image_url + text） | 支持自定义Prompt、min/max_pixels图像缩放控制、[流式输出](../concepts/streaming-output.md) | [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) | -| `gui-plus-2026-02-26` | GUI界面自动化操作 | 图文混合（image_url + text） | 基于工具调用（`computer_use`），需严格遵循Action + ``响应格式；支持高分辨率图像处理 | [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) | +当前支持的专用模型包括： -> **注意**：文档4明确指出`qwen-deep-research`“仅支持华北2（北京）地域”，而文档2、3、5中均提及新加坡/美国地域的兼容接口配置。若在非北京地域调用该模型将失败，此为硬性限制而非配置问题。 +- **意图理解模型**：`tongyi-intent-detect-v3`，支持毫秒级意图识别与工具调用决策，适用于对话路由、智能助手等场景 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)； +- **法律大模型**：`farui-plus`，专为法律行业优化，支持法律问答、文书生成、案情分析与合同审查等功能； +- **机器翻译模型**：`qwen-mt-plus`，支持多语言互译、术语干预与翻译记忆，适用于本地化与技术文档翻译； +- **深度研究模型**：`qwen-deep-research`，仅限华北2（北京）地域，支持两阶段交互式研究（反问确认 + 深度分析），并集成网络搜索与引用溯源能力 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)； +- **OCR模型**：`qwen3.5-ocr`，支持图像中结构化文本提取，兼容 OpenAI 多模态消息格式（含 `image_url` 与 `text` 组合输入）； +- **GUI自动化模型**：`gui-plus-2026-02-26`，面向桌面界面操作，支持截图理解、鼠标键盘控制与任务终止，需配合 `` 系统提示与 `` 格式化响应 [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md)。 + +> **注意**：文档4明确指出 `qwen-deep-research` “仅支持通过 Python DashScope SDK 调用，暂不支持 Java SDK 与 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)”，但文档5和文档6均提供了完整的 OpenAI 兼容调用示例（含 `qwen3.5-ocr` 和 `gui-plus-2026-02-26`）。该矛盾表明 `qwen-deep-research` 的 OpenAI 兼容支持状态与其他模型不一致，开发者应以文档4的声明为准，避免在生产环境中尝试非支持接口。 ## 关键参数 -- **`model`**（必选）：模型标识符，如`farui-plus`、`qwen-mt-plus`等，大小写敏感。 -- **`messages`**（必选）：对话消息数组，每项含`role`（`user`/`system`/`assistant`）和`content`。OCR与GUI模型支持图文混合`content`（含`image_url`和`text`子项）。 -- **`result_format` / `output_format`**： - - DashScope SDK通用参数：`result_format='message'`（推荐）或`'text'`； - - `qwen-deep-research`特有：`output_format`可选`model_detailed_report`（默认，~6000 Token）或`model_summary_report`（~1500–2000 Token）。 -- **`stream`**（可选）：启用[流式输出](../concepts/streaming-output.md)（`True`/`true`），适用于长文本生成或实时反馈场景。 -- **领域扩展参数**： - - `qwen-mt-plus`：`translation_options`对象，含`source_lang`、`target_lang`、`terms`（术语表）、`tm_list`（翻译记忆）； - - `qwen3.5-ocr`：`image_url`子项支持`min_pixels`/`max_pixels`控制图像预处理； - - `gui-plus-2026-02-26`：`extra_body={"vl_high_resolution_images": True}`启用高分辨率图像支持。 +各模型通用关键参数如下： + +- **`model`**（必选）：字符串，指定模型名称，如 `"tongyi-intent-detect-v3"`、`"qwen-mt-plus"` 等； +- **`messages`**（必选）：消息数组，按对话顺序排列；对多模态模型（如 `qwen3.5-ocr`、`gui-plus-2026-02-26`），`content` 支持混合类型（`text` + `image_url`）； +- **`extra_body`**（可选，OpenAI 兼容）：用于传递模型特有参数： + - `qwen-mt-plus`：传入 `translation_options`（含 `source_lang`、`target_lang`、`terms`、`tm_list`）； + - `gui-plus-2026-02-26`：建议设置 `"vl_high_resolution_images": true` 以提升图像理解精度； +- **`output_format`**（可选，`qwen-deep-research`）：取值 `model_detailed_report`（默认，约6000 [Token](../concepts/token.md)）或 `model_summary_report`（约1500–2000 [Token](../concepts/token.md)）； +- **图像处理参数**（`qwen3.5-ocr` / `gui-plus-2026-02-26`）：`image_url` 对象内可指定 `min_pixels` 与 `max_pixels` 控制缩放行为。 ## 使用方式 -### 基础调用流程 -1. **环境准备**：安装SDK（[DashScope](https://help.aliyun.com/zh/model-studio/install-sdk) 或 [OpenAI](https://help.aliyun.com/zh/model-studio/install-sdk)），获取并配置API Key至环境变量`DASHSCOPE_API_KEY`； -2. **域名配置**：强烈建议使用业务空间专属域名（如`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），详见各文档中的迁移提示； -3. **构造请求**：按模型要求组织`messages`，设置必要参数； -4. **发起调用**：使用SDK方法（如`dashscope.Generation.call()`或`client.chat.completions.create()`）。 - -### 典型调用示例 -- **法律文书生成（farui-plus）**： - ```python - response = dashscope.Generation.call( - model="farui-plus", - messages=[{"role": "user", "content": "我哥欠我10000块钱，给我生成起诉书。"}], - result_format='message' - ) - ``` -- **翻译+术语干预（qwen-mt-plus）**： - ```python - completion = client.chat.completions.create( - model="qwen-mt-plus", - messages=[{"role": "user", "content": "生物传感器"}], - extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English", "terms": [{"source": "生物传感器", "target": "biological sensor"}]}} - ) - ``` -- **OCR结构化提取（qwen3.5-ocr）**： - ```python - completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[{ - "role": "user", - "content": [ - {"type": "image_url", "image_url": {"url": "https://..."}}, - {"type": "text", "text": "提取发票号码、金额、日期"} - ] - }] - ) - ``` +### 基础调用前提 +- 已获取并配置 API Key（推荐设为环境变量 `DASHSCOPE_API_KEY`）； +- 安装对应 SDK（[DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk) 或 [OpenAI SDK](https://help.aliyun.com/zh/model-studio/install-sdk)）； +- **强烈建议迁移至业务空间专属域名**：华北2（北京）使用 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`，新加坡使用 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，以获得更高性能与稳定性 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 + +### 接口选择 +- **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**：适用于 `tongyi-intent-detect-v3`、`qwen-mt-plus`、`qwen3.5-ocr`、`gui-plus-2026-02-26`，`base_url` 统一为 `.../compatible-mode/v1`； +- **DashScope 原生接口**：适用于 `farui-plus`、`qwen-deep-research`（仅 Python SDK），`base_http_api_url` 设为 `.../api/v1`； +- **[流式输出](../concepts/streaming-output.md)**：所有模型均支持，OpenAI 接口设 `stream=True`，DashScope 接口使用 `stream=True`（Python）或 `streamCall`（Java）。 + +### 特殊调用模式 +- **意图识别双模式**： + - `INTENT_MODE`：System Message 中声明 `Response in INTENT_MODE.` 并注入工具定义，返回 ``/``/`` 结构化结果； + - **纯标签模式**：System Message 指定意图字典并要求“仅回复所选标签”，可进一步压缩为单 [Token](../concepts/token.md) 输出以优化延迟。 +- **深度研究两阶段流程**：先调用获取模型反问（`ResearchPlanning` 阶段），再将用户澄清与历史消息组合发起第二轮调用，触发 `WebResearch` 与 `answer` 阶段。 ## 限制和注意事项 -- **地域限制**：`qwen-deep-research`仅支持华北2（北京）地域，其他模型虽支持多地域，但需匹配对应API Key和`base_url`（如新加坡地域Key不可用于北京域名）； -- **限流策略**：所有模型受百炼平台统一限流控制，详情见[限流文档](https://help.aliyun.com/zh/model-studio/rate-limit)； -- **成本与免费额度**：`tongyi-intent-detect-v3`提供90天内100万Token免费额度，其余模型按输入/输出Token计费（如`farui-plus`输入20元/百万Token）； -- **SDK兼容性**：`qwen-deep-research`当前**仅支持Python DashScope SDK**，不支持Java SDK或OpenAI兼容接口 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)； -- **安全实践**：API Key务必配置至环境变量，避免硬编码；Java SDK中`Generation`对象非线程安全，需自行管理同步 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md)； -- **响应解析**：`tongyi-intent-detect-v3`返回内容含XML标记（如``、），需用正则解析提取工具调用JSON [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 +- **地域限制**：`qwen-deep-research` 仅支持华北2（北京）地域，其他地域调用将失败； +- **SDK 限制**：`qwen-deep-research` 当前**不支持 Java SDK 与 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**，仅 Python DashScope SDK 可用； +- **域名迁移强制建议**：旧域名（如 `dashscope.aliyuncs.com`）虽仍可用，但新业务空间专属域名提供“卓越的性能和更高的稳定性”，生产环境务必迁移； +- **成本与配额**：`tongyi-intent-detect-v3` 提供开通后90天内100万 Token 免费额度；`farui-plus` 输入/输出成本为20元/百万 Token（文档2未列明输出成本，存在信息缺失）； +- **OCR 图像约束**：`qwen3.5-ocr` 对输入图像像素有硬性要求（`min_pixels`/`max_pixels`），超限将自动缩放，需在请求中显式配置； +- **GUI 模型行为约束**：`gui-plus-2026-02-26` 无终端或应用菜单访问权限，所有操作必须通过点击桌面图标启动应用，并需合理插入 `wait` 动作应对加载延迟。 ## 来源文档 +- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) - [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) - [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) -- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) - [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more.md b/skills/bailian-docs-llm-wiki/wiki/api/more.md index 20d025f8..ed06769a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more.md @@ -1,52 +1,64 @@ # more -`more` 是百炼平台面向高级用例提供的扩展能力集合，涵盖服务权限管理、知识库精准检索、临时凭证生成等关键功能。这些能力不直接参与模型推理主流程，但对构建安全、可控、可观察的企业级AI应用至关重要。开发者需结合具体场景按需启用，并严格遵循最小权限原则。 +`more` 是百炼平台面向开发者提供的扩展能力集合，涵盖临时凭证生成、服务关联角色管理、知识库高级检索过滤等关键功能。这些能力不直接参与模型推理，而是支撑安全调用、资源协同与精准检索等核心场景。本文档聚焦其技术细节与工程实践要点，适用于需要在可信/不可信环境集成百炼服务、构建复杂工作流或优化RAG效果的开发者。 ## 支持的模型/功能 -`more` 不对应特定模型，而是提供以下三类基础设施级功能： +`more` 并非模型名称，而是百炼平台中一组**基础设施级扩展能力**的统称，当前包含以下三类核心功能： -- **服务关联角色（SLR）管理**：为百炼各子功能（如工作流调用函数计算、知识库对接ADB-PG、数据同步访问OSS等）自动创建并托管云资源访问权限。详见 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 -- **知识库高级检索（SearchFilters）**：在 `Retrieve` 接口请求中嵌入结构化过滤条件，支持单值、多值、范围、模糊及标签查询，显著提升语义检索结果的相关性。该能力仅适用于已配置字段索引的数据查询型知识库。 -- **临时API Key生成**：通过后端服务调用 `/tokens` 接口，基于永久密钥签发短期有效的访问令牌，适用于前端直连等不可信环境。详见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 +- **临时API Key生成**：为浏览器、移动端等不可信客户端提供短期、可撤销的访问凭证，避免永久密钥泄露风险。该能力通过 `https://dashscope.aliyuncs.com/api/v1/tokens` 接口提供，详见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 +- **服务关联角色（SLR）管理**：百炼在启用特定功能（如函数计算节点、OSS数据导入、ADB-PG知识库存储）时，自动创建并绑定RAM服务关联角色，以最小权限原则访问其他云服务资源。完整角色列表及权限策略见 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- **知识库SearchFilters**：在调用 `Retrieve` 接口时，通过结构化过滤条件对语义检索结果进行后置精筛，显著提升结构化数据（如员工表、产品目录）的召回准确率。语法支持单值、多值、范围、模糊及标签查询，详见 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 -> **注意**：文档 1 中列出的 `AliyunServiceRoleForSFMTelemetry` 权限策略在末尾被截断（`"xtrace:Read*", "xtrace:Get*"` 后缺失完整内容），实际策略应以控制台或最新版RAM策略文档为准；同时，文档 2 中 `tag_query2()` 示例代码在末尾被截断，完整逻辑需参考SDK示例仓库。 +> **注意**：文档2中列出的 `AliyunServiceRoleForSFMAccessingMNS` 权限策略末尾存在截断（`"xtrace:Get*"` 后缺失闭合括号与完整JSON结构），实际部署应以RAM控制台中该角色绑定的最新系统策略为准；此为文档过时导致，非API行为变更。 ## 关键参数 -| 功能 | 参数名 | 类型 | 必填 | 说明 | 取值范围 | -|------|--------|------|------|------|----------| -| 临时API Key | `expire_in_seconds` | integer | 否 | 令牌有效期（秒） | `[1, 1800]`，默认 `60` | -| SearchFilters | `searchFilters` | array of object | 否 | 过滤条件数组，每个对象为一个AND分组 | 最多支持 5 个分组；每个分组内Key-Value对数量无硬限制，但总请求体大小 ≤ 1 MB | +| 功能 | 参数名 | 类型 | 必填 | 说明 | 示例 | +|------|--------|------|------|------|------| +| 临时API Key生成 | `expire_in_seconds` | Integer | 否 | TTL有效期（秒），取值范围 `[1, 1800]`，默认60秒 | `1800` | +| SearchFilters | `searchFilters` | Array of Object | 否 | 过滤条件数组，每个Object为一个AND子分组，支持 `eq/neq/gt/gte/lt/lte/like` 等操作符 | `[{"姓名": "张三"}, {"岗位": "技术员"}]` | +| SearchFilters（范围查询） | 字段值 | JSON String | 是 | 范围查询需序列化为JSON字符串，如 `{"gte": 20, "lte": 27}` | `{"gte": 20, "lte": 27}` | +| SearchFilters（模糊查询） | 字段值 | JSON String | 是 | 模糊查询需序列化为 `{"like": "技%员"}` 形式 | `{"like": "技%员"}` | ## 使用方式 -- **服务关联角色**：系统在首次启用对应功能（如发布含FC节点的工作流）时**自动创建**，无需手动调用API。角色名称与权限策略已固化，不可修改。删除前必须先解除所有依赖该角色的业务配置（如断开OSS连接、删除FC节点等），否则将导致功能异常。 -- **SearchFilters**：在 `RetrieveRequest` 请求体中直接传入 `searchFilters` 字段。例如： - ```json - { - "indexId": "o73yjlxxxx", - "query": "公司中姓名为张三的员工", - "searchFilters": [ - {"姓名": "张三"}, - {"岗位": "技术员", "年龄": {"gte": 20, "lte": 27}} - ] - } - ``` - 具体语法与字段类型约束请参考 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 -- **临时API Key**：向 `https://dashscope.aliyuncs.com/api/v1/tokens` 发起带 `Authorization: Bearer ` 的 POST 请求，可选添加 `?expire_in_seconds=N` 查询参数。响应中的 `token` 字符串可直接用于后续模型API调用的 `Authorization` 头。 +- **临时API Key**： + 在后端服务中，使用已配置的永久 `DASHSCOPE_API_KEY` 调用 `/api/v1/tokens` 接口（[生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)）。返回的 `token` 可直接用于后续模型请求的 `Authorization: Bearer ` 头。**切勿在前端硬编码或暴露永久密钥**。 + +- **服务关联角色**： + 角色由百炼在首次启用对应功能时**自动创建**，无需手动申请。开发者需确保RAM账号具备 `AliyunRAMFullAccess` 权限以查看角色，并在删除前按文档要求清理依赖资源（如先断开OSS连接、删除函数计算节点等）。具体操作请参考 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) 中各角色的“删除服务关联角色”章节。 + +- **SearchFilters**： + 在 `RetrieveRequest` 请求体中直接传入 `searchFilters` 字段（[知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)）。SDK调用示例中需注意： + - Python/Java SDK要求将范围、模糊等复杂查询**序列化为JSON字符串**再赋值（见文档3中 `json.dumps()` 用法）； + - 子分组间为AND逻辑，分组内字段为AND，无法改为OR； + - 标签查询（`tags` 字段）中，同一分组内多个标签为OR关系，不同分组间仍为AND。 ## 限制和注意事项 -- 所有服务关联角色均绑定特定百炼服务域名（如 `fc.sfm.aliyuncs.com`），**不可复用或跨服务授权**。手动修改其策略或删除角色将导致对应功能完全失效。 -- `SearchFilters` 仅对**数据查询型知识库**生效，文档型知识库不支持字段级过滤；多值查询需使用 `json.dumps(["val1","val2"])` 序列化为字符串传递；模糊查询 `like` 值中 `%` 为通配符。 -- 临时API Key 继承源密钥的全部权限（含模型白名单、知识库访问限制等），且**无法提前撤销**，仅能等待过期。生产环境务必严格控制 `expire_in_seconds` 时长，避免设置过长TTL。 -- 文档 1 中 `AliyunServiceRoleForSFMAccessingMNS` 明确声明“请勿修改、删除，或将其授予除服务关联角色之外的任何RAM身份”，此为强制安全要求，违反将导致数据同步中断且难以恢复。 +- **临时API Key**： + - 无法提前失效，仅能等待TTL过期； + - 继承生成者密钥的全部权限（含模型访问、知识库读写等），请严格管控生成密钥的权限粒度； + - 各地域Endpoint独立（北京/新加坡/弗吉尼亚），密钥不可跨地域复用。 + +- **服务关联角色**： + - 删除角色将导致对应功能完全不可用（如删除 `AliyunServiceRoleForSFMAccessFC` 后，工作流中函数计算节点将无法调用）； + - 部分角色（如 `AliyunServiceRoleForSFMAccessingMNS`）明确禁止手动修改或删除，违反将导致服务异常； + - 所有角色均需通过RAM控制台管理，百炼控制台不提供直接入口。 + +- **SearchFilters**： + - 仅对**数据查询型知识库**生效，文档型知识库不支持； + - 字段类型必须与知识库索引配置一致（如年龄字段需为 `double` 才能使用 `gte`）； + - 模糊查询（`like`）仅支持字符串字段，且 `%` 为通配符（`"技%员"` 匹配“技术员”“技师员”等）； + - 多值查询需将数组序列化为字符串（如 `["张三","李四"]` → `"[\\"张三\\",\\"李四\\"]"`），SDK已封装此逻辑。 + +> **注意**：文档3中Python示例 `multi_query()` 方法将 `names` 数组直接 `json.dumps()` 后赋值给 `search_filters`，但实际SDK（如 `alibabacloud_bailian20231229` v1.0.10+）已支持原生List传入，无需手动序列化。建议优先使用SDK最新版以避免兼容性问题。 ## 来源文档 +- [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md) -- [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md index 26ef989b..f6dc26e9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md @@ -1,61 +1,64 @@ # omni realtime api -Qwen-Omni-Realtime API 是一个基于 WebSocket 的实时多模态交互接口，支持语音输入、文本/音频输出、VAD 自动检测、工具调用与联网搜索（部分模型），适用于智能客服、虚拟助手等低延迟对话场景。其核心是双向事件流通信：客户端发送 `session.update`、`input_audio_buffer.append` 等事件，服务端返回 `session.created`、`response.audio.delta` 等事件。 +Qwen-Omni-Realtime API 是基于 WebSocket 的全模态实时交互接口，支持文本、音频、图像多模态输入与文本+音频双模态输出，适用于语音助手、智能客服等低延迟对话场景。其核心能力包括实时语音识别（ASR）、大模型流式推理、TTS 音频合成、工具调用（Function Calling）及可选的联网搜索，所有交互均通过事件驱动模型完成。 -## 支持的模型与功能 +## 支持的模型/功能 -- **支持模型**：`qwen3.5-omni-realtime`、`qwen3.5-omni-plus-realtime`、`qwen3.5-omni-flash-realtime`、`qwen3-omni-flash-realtime`、`qwen-omni-turbo-realtime`。各模型能力存在差异，详见 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 中的参数兼容性说明。 -- **核心模态**：默认支持 `["text", "audio"]` 输出；可设为 `["text"]` 仅输出文本。输入仅支持 `pcm` 格式音频（16 kHz 采样率）。 -- **语音活动检测（VAD）**：支持 `server_vad`（声学检测）和 `semantic_vad`（语义检测，**仅 `qwen3.5-omni-realtime` 支持**）[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 +- **模型系列**：当前支持 `qwen3.5-omni-realtime`、`qwen3.5-omni-plus-realtime`、`qwen3.5-omni-flash-realtime`、`qwen3-omni-flash-realtime` 和 `qwen-omni-turbo-realtime`。各模型在 VAD 类型、参数可调性、音色默认值等方面存在差异，详见 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 +- **多模态输入**：支持 PCM 音频（16 kHz）和 JPG/JPEG 图像（≤1080p，Base64 编码后 ≤256 KB），图像需在首次音频追加后发送。 +- **多模态输出**：支持 `["text"]` 或 `["text", "audio"]` 输出模态；音频输出固定为 24 kHz PCM，不可自定义采样率。 +- **语音活动检测（VAD）**：提供 `server_vad`（声学检测）和 `semantic_vad`（语义检测，仅 `qwen3.5-omni-realtime` 支持）两种模式，支持静默超时（`idle_timeout_ms`）主动引导对话。 - **高级功能**： - - 工具调用（`tools`）：所有支持模型均可配置，但仅 `qwen3.5-omni-realtime` 系列在文档中明确标注支持完整流程； - - 联网搜索（`enable_search`）：**仅 `qwen3.5-omni-realtime` 系列支持**，且与 `tools` 不兼容 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)； - - 声音复刻：需先调用独立声音复刻 API 创建音色，再在 `session.update` 中通过 `voice` 参数传入，**驱动模型必须与复刻时指定的 `target_model` 严格一致** [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)。 - -> **注意**：文档 2（服务端事件）中 `session.created` 示例显示 `model: "qwen3-omni-flash-realtime"`，而文档 1（客户端事件）中 `voice` 默认值表格将 `qwen3-omni-flash-realtime` 对应音色列为 `"Cherry"`，但文档 3（Python SDK）和文档 4（Java SDK）均将该模型写作 `"qwen3-omni-flash-realtime"`（无连字符），而文档 6（声音复刻）中 `target_model` 列表使用 `"qwen3.5-omni-flash-realtime"`。实际调用时请以控制台模型列表或最新 SDK 枚举为准，避免因命名不一致导致 `invalid_request_error`。 + - 工具调用（`tools`）：模型自主触发函数并返回参数，客户端执行后需回传结果并调用 `response.create`。 + - 联网搜索（`enable_search`）：仅 `qwen3.5-omni-realtime` 系列支持，与 `tools` 不兼容。 + > **注意**：文档 5 中提到的声音复刻功能（`qwen-voice-enrollment`）属于独立服务，需先创建音色再在 `session.update` 中通过 `voice` 参数引用，其模型绑定要求（如 `target_model` 必须与 Omni 模型一致）在 [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) 中有明确约束。 ## 关键参数 -所有参数均通过 `session.update` 客户端事件或 SDK 的 `update_session()` 方法设置： - -| 参数 | 类型 | 说明 | 兼容性 | -|------|------|------|--------| -| `modalities` | `["text"]` 或 `["text","audio"]` | 输出模态组合 | 全系列支持 | -| `voice` | `string` | 音色 ID，如 `"Chelsie"`、`"Tina"`；复刻音色需传入生成的 voice ID | 全系列支持 | -| `input_audio_format` / `output_audio_format` | `"pcm"` | 输入/输出音频格式，固定值 | 全系列支持 | -| `instructions` | `string` | 系统角色提示词 | 全系列支持 | -| `turn_detection.type` | `"server_vad"` 或 `"semantic_vad"` | VAD 类型 | `semantic_vad` 仅 `qwen3.5-omni-realtime` 支持 | -| `turn_detection.silence_duration_ms` | `integer [200, 6000]` | 静音触发响应阈值（毫秒） | 全系列支持 | -| `turn_detection.idle_timeout_ms` | `integer [5000, 30000]` | 静默超时主动引导时间 | **仅 `qwen3.5-omni-plus-realtime` 或 `qwen3.5-omni-flash-realtime` + `server_vad` 时生效** | -| `enable_search` | `boolean` | 启用联网搜索 | **仅 `qwen3.5-omni-realtime` 系列支持** | -| `tools` | `array` | 工具定义列表 | 全系列支持，但 `qwen3.5-omni-realtime` 文档最完整 | -| `temperature` / `top_p` / `top_k` | `float` / `float` / `integer` | 采样控制参数 | `qwen-omni-turbo` 系列**不支持修改** | -| `max_tokens` | `integer` | 最大输出 token 数 | `qwen-omni-turbo` 系列**不支持修改** | -| `smooth_output` | `boolean` or `null` | **仅 `qwen3-omni-flash-realtime` 系列支持**，控制口语化/书面化风格 | | - -> **注意**：`repetition_penalty` 和 `presence_penalty` 在文档 1 和文档 2 中默认值存在差异（如 `qwen3.5-omni-realtime` 的 `presence_penalty`，文档 1 写 `1.5`，文档 2 未明确，默认值以 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 为准。 +所有会话级配置均通过 `session.update` 事件或 SDK 的 `update_session` 方法设置，关键参数如下： + +| 参数 | 类型 | 说明 | 默认值/约束 | +|------|------|------|-------------| +| `modalities` | `array` | 输出模态，仅支持 `["text"]` 或 `["text","audio"]` | `["text","audio"]` | +| `voice` | `string` | TTS 音色名 | `qwen3.5`: `"Tina"`；`qwen3-flash`: `"Cherry"`；`qwen-turbo`: `"Chelsie"` | +| `input_audio_format` / `output_audio_format` | `string` | 固定为 `"pcm"`；输入需 16 kHz，输出为 24 kHz | — | +| `instructions` | `string` | 系统提示词，定义角色与行为边界 | — | +| `turn_detection.type` | `string` | `server_vad`（默认）或 `semantic_vad`（仅 `qwen3.5-omni-realtime`） | `server_vad` | +| `turn_detection.threshold` | `float` | VAD 灵敏度 [-1.0, 1.0] | `0.5` | +| `turn_detection.silence_duration_ms` | `int` | 静音触发阈值 [200, 6000] ms | `800` | +| `enable_search` | `boolean` | 启用联网搜索（仅 `qwen3.5-omni-realtime`） | `false` | +| `tools` | `array` | 工具定义列表，`type` 固定为 `"function"` | `[]` | +| `temperature` / `top_p` / `top_k` | `float`/`float`/`int` | 采样控制参数；`qwen-omni-turbo` 系列不支持修改 | 见 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 表格 | +| `max_tokens` | `int` | 响应最大 token 数（截断，不影响生成过程） | 模型最大输出长度 | +| `repetition_penalty` / `presence_penalty` | `float` | 重复惩罚参数；`qwen-omni-turbo` 系列不支持修改 | 见 [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 表格 | + +> **注意**：`smooth_output` 参数仅对 `qwen3-omni-flash-realtime` 生效，且文档 1 与文档 3、4 在默认值描述上存在不一致——文档 1 称 `true` 为默认值，而文档 3、4 明确标注 `null`（自动选择）为默认值。实际行为以 SDK 实现为准，建议显式设置。 ## 使用方式 -1. **建立连接**：使用 WebSocket 连接到地域专属域名（推荐 `wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime` 或 `wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime`），`{WorkspaceId}` 从控制台获取。 -2. **初始化会话**：连接后，服务端立即返回 `session.created` 事件。随后调用 `session.update` 设置初始配置（如 `modalities`, `voice`, `instructions`）。 -3. **输入数据**： - - **VAD 模式（推荐）**：持续发送 `input_audio_buffer.append`，服务端自动检测起止并提交；无需手动发 `commit` 或 `response.create` [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)。 - - **Manual 模式**：发送 `input_audio_buffer.append` → `input_audio_buffer.commit` → `response.create` 触发响应。 -4. **处理响应**：监听 `response.audio.delta`（流式音频）、`response.text.delta`（流式文本）、`response.done`（完成）等事件。 -5. **工具调用**：当收到 `response.function_call_arguments.done` 事件时，执行本地工具，再通过 `conversation.item.create` 回传结果，最后发 `response.create`（Manual 模式）或等待服务端自动响应（VAD 模式）。 +1. **建立连接**：使用 WSS 协议连接业务空间专属域名（推荐），格式为 `wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime`，其中 `{WorkspaceId}` 为控制台获取的业务空间 ID。 +2. **初始化会话**：连接后服务端立即返回 `session.created` 事件，含默认配置。随后调用 `session.update`（或 SDK `update_session`）覆盖默认参数。 +3. **输入处理**： + - **VAD 模式**（`enable_turn_detection=true`）：持续 `input_audio_buffer.append` 音频，服务端自动检测起止并提交；可选 `input_image_buffer.append` 图像。 + - **Manual 模式**（`enable_turn_detection=false`）：手动 `append` 后必须 `input_audio_buffer.commit` 提交。 +4. **触发响应**： + - VAD 模式下，语音停止后服务端自动触发 `response.create`。 + - Manual 模式或工具调用后，需显式发送 `response.create`。 +5. **处理响应**：监听 `response.content_part.added`（文本增量）、`response.audio.delta`（音频增量）、`response.done`（完成）等事件。 +6. **工具调用流程**：收到 `conversation.item.created`（`type="function_call"`）→ 执行本地函数 → 发送 `conversation.item.create`（`type="function_call_output"`）→ 发送 `response.create`。 + +SDK 封装了上述流程，Python 使用 `OmniRealtimeConversation` 类，Java 使用 `OmniRealtimeConversation` 类，二者均提供 `append_audio`、`commit`、`create_response` 等方法，详细用法见 [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 和 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md)。 ## 限制和注意事项 -- **音频限制**：输入音频必须为 16 kHz PCM；输出音频固定为 24 kHz PCM；单次 `append_audio` 数据量无明确上限，但 SDK 示例建议 ≤15 MiB。 -- **图片限制**：仅 JPG/JPEG；Base64 编码后 ≤256 KB；建议分辨率 480p/720p；发送频率 ≤1 张/秒。 -- **并发与超时**：单个 WebSocket 连接对应一个会话；`idle_timeout_ms` 仅在特定模型+VAD 组合下生效；`max_tokens` 截断响应但不影响生成过程。 -- **兼容性约束**： - - `tools` 与 `enable_search` **不可同时启用**； - - `qwen-omni-turbo` 系列模型**不支持修改** `temperature`、`top_p`、`top_k`、`max_tokens`、`repetition_penalty`、`presence_penalty`、`seed`； - - `semantic_vad` 仅 `qwen3.5-omni-realtime` 支持，其他模型设为该值将报错； - - 声音复刻音色必须与 Omni 调用模型严格匹配（如复刻时 `target_model="qwen3.5-omni-plus-realtime"`，则 Omni 调用时 `model` 和 `voice` 必须对应同一模型）[声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)。 -- **错误处理**：服务端返回 `error` 事件（含 `type`、`code`、`message`、`param`），需根据 `param` 字段定位问题参数。 +- **音频限制**：输入音频必须为 16 kHz PCM；单次 `append_audio` 数据量无硬限制，但缓冲区总大小建议 ≤15 MiB（文档 3、4 明确提及）。 +- **图像限制**：仅 JPG/JPEG；Base64 编码后 ≤256 KB；建议分辨率 480p/720p；发送频率 ≤1 张/秒。 +- **参数兼容性**： + - `tools` 与 `enable_search` 互斥，不可同时启用。 + - `qwen-omni-turbo-realtime` 系列模型**不支持修改** `temperature`、`top_p`、`top_k`、`max_tokens`、`repetition_penalty`、`presence_penalty`、`seed`（文档 1、3、4 均强调）。 +- **VAD 行为**：`semantic_vad` 仅 `qwen3.5-omni-realtime` 支持；`idle_timeout_ms` 仅在 `server_vad` + `qwen3.5-omni-plus-realtime` 或 `qwen3.5-omni-flash-realtime` 下生效。 +- **错误处理**：服务端返回 `error` 事件（如 `invalid_request_error`），需检查 `param` 字段定位问题，例如 `modalities` 值必须为 `["text"]` 或 `["text","audio"]`（文档 2 示例明确指出错误消息）。 +- **域名迁移**：旧域名（如 `dashscope.aliyuncs.com`）仍可用，但百炼官方强烈推荐迁移到业务空间专属域名以获得更高性能与稳定性（文档 3、4、5 均强调）。 ## 来源文档 @@ -63,7 +66,6 @@ Qwen-Omni-Realtime API 是一个基于 WebSocket 的实时多模态交互接口 - [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md) - [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) -- [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) - [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md index 98b04b94..baf91d80 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md @@ -1,65 +1,41 @@ # preparations -在调用阿里云百炼平台的模型或应用前，开发者需完成 API Key 获取、SDK/CLI 安装与配置、环境变量设置等基础准备。这些步骤是所有后续调用（文本生成、多模态理解、语音合成等）的前提，直接影响鉴权有效性、调用协议兼容性及安全性。本文档结构化梳理关键环节，聚焦可操作项，避免冗余说明。 +在调用阿里云百炼平台的模型或应用前，开发者需完成基础环境准备，包括获取并安全配置 API Key、安装必要的 SDK 或 CLI 工具、理解关键参数约束及常见错误处理机制。这些步骤是所有模型调用的前置依赖，直接影响服务可用性与安全性。 ## 支持的模型/功能 -百炼平台支持全模态能力调用，包括： -- **文本生成**：如 `qwen3.7-max`、`qwen3-235b-a22b-instruct-2507` 等大语言模型； -- **多模态理解与生成**：`qwen3-vl-plus`（视觉理解）、`qwen-image-2.0`（文生图）、`happyhorse-1.0-t2v`（文生视频）； -- **语音处理**：`cosyvoice-v3-flash`（TTS）、`paraformer-real-time`（ASR）； -- **向量与排序**：`text-embedding-v3`、`text-rerank-v3`。 - -所有模型均通过统一 API Key 鉴权，**无需为不同模型创建独立密钥**；权限由 API Key 所属业务空间决定，详见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) 中“API Key权限说明”章节。 - -> **注意**：文档 3 中 CLI 命令示例默认使用 `qwen3.7-max` 作为文本模型，但文档 4 的错误码明确指出部分思考模式模型（如 `qwen3-235b-a22b-thinking-2507`）**强制要求 `enable_thinking=true`**，且不支持非流式调用。实际选型需以[模型列表文档](https://help.aliyun.com/zh/model-studio/model-list)为准，不可仅依赖 CLI 默认值。 +百炼平台支持多模态模型（如 `qwen3-vl-plus`、`qwen-image-2.0`）、文本生成模型（如 `qwen3.7-max`）、语音合成（`cosyvoice-v3-flash`）、语音识别（`paraformer-real-time`）、向量嵌入（`text-embedding-v3`）及排序模型（`text-rerank-v3`）等。不同模型对输入格式、协议兼容性（OpenAI 兼容 / Anthropic 兼容）、输出模式（流式/非流式）有明确要求。例如，`qwen3-235b-a22b-thinking-2507` 等思考模式模型**仅支持流式调用**，且 `enable_thinking` 参数不可设为 `false`；而纯文本模型（如 `qwen3-max`）**不支持 `image_url` 等多模态 `content` 元素**，混用将触发 400 错误 [原文标题](../../raw/model-api-reference/preparations/error-code.md)。多模态能力需通过 `bl omni` 或 `bl vision describe` 等 CLI 命令或对应 SDK 接口调用。 ## 关键参数 -| 参数 | 说明 | 取值范围/格式 | 来源依据 | -|------|------|----------------|----------| -| `DASHSCOPE_API_KEY` | 鉴权凭证，必须配置为环境变量或显式传入 | `sk-ws-` 开头（新密钥）或 `sk-` 开头（旧密钥），长度固定 | [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) | -| `base_url` / `--base-url` | 服务端点地址，随地域和协议变化 | 如 `https://dashscope.aliyuncs.com/api/v1`（OpenAI 兼容）或 `https://dashscope.aliyuncs.com/anthropic/v1`（Anthropic 兼容） | [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) | -| `--region` | 地域标识 | `cn`（华北2）、`us`（弗吉尼亚）、`intl`（新加坡/东京等） | [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) | -| `enable_thinking` | 启用思考模式 | `true` 或 `false`，部分模型强制为 `true` | [错误码](../../raw/model-api-reference/preparations/error-code.md) | -| `stream` | 启用[流式输出](../concepts/streaming-output.md) | `true`（必需用于思考模式、Qwen-Omni 音频输出等） | [错误码](../../raw/model-api-reference/preparations/error-code.md) | +核心参数需严格遵循取值范围与类型约束： +- `temperature`: 必须在 `[0.0, 2.0)` 区间； +- `top_p`: 必须在 `(0.0, 1.0]` 区间； +- `max_tokens`: 上限由模型文档明确指定，超出将报错 `Range of max_tokens should be [1, xxx]`； +- `n`: 图像生成等场景中最大值为 `6`（CLI）或 `4`（HTTP API），超限触发 `Range of n should be [1, 4]`； +- `seed`: DashScope 协议下必须为 `[0, 9223372036854775807]` 内整数； +- `messages` 格式：纯文本模型要求 `content` 为字符串，多模态模型要求 `content` 数组元素为合法对象（`type` 仅限 `text`/`image_url`/`video_url` 等）[原文标题](../../raw/model-api-reference/preparations/error-code.md)； +- 结构化输出（`response_format={"type": "json_object"}`）时，提示词中**必须包含 `json` 关键词**，且 `enable_thinking` 必须为 `false`。 -## 使用方式 +> **注意**：文档 3 中 `bl image generate` 的 `--n` 参数默认值为 `1`，最大支持 `6`；而文档 4 的错误码说明中 `Range of n should be [1, 4]` 针对的是 HTTP API 的通用限制。实际使用时，请以目标接口（CLI vs HTTP）的文档为准——CLI 扩展了图像生成的并发上限，但标准 HTTP 接口仍遵循 `n ≤ 4` 规则 [原文标题](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 -### 1. 获取并配置 API Key -- 通过[控制台](https://bailian.console.aliyun.com/)创建 API Key，**主账号或具备 `API-Key` 权限的子账号**方可操作； -- 创建时选择 **全部权限**（快速上手）或 **自定义权限**（IP 白名单 + 模型范围）； -- **强烈建议配置为环境变量**：Linux/macOS 使用 `export DASHSCOPE_API_KEY="sk-ws-xxx"`，Windows 使用系统属性或 PowerShell 的 `[Environment]::SetEnvironmentVariable`； -- 美国（弗吉尼亚）地域不支持禁用/重置操作，且不显示完整明文密钥，需立即保存。 +## 使用方式 -### 2. 安装调用工具 -- **SDK 方式**： - - Python：`pip install -U dashscope`（原生）或 `pip install -U openai`（OpenAI 兼容）； - - Java/Node.js/Go：参考对应语言的 SDK 依赖声明（如 Maven/Gradle/GitHub）； -- **CLI 方式**： - - 要求 Node.js ≥ 22.12.0，仅支持 `npm install -g bailian-cli`； - - 认证推荐 `bl auth login --console`（浏览器 OAuth），备选 `bl auth login --api-key `； - - 支持 `--api-key` 临时传入、环境变量、配置文件三种鉴权方式，互不冲突。 +### API Key 获取与配置 +- **获取**：需主账号或具备 `管理员`/`API-Key` 权限的子账号，在[百炼控制台 API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建。新创建的 Key 统一以 `sk-ws` 开头，仅创建时可见明文，丢失需重置 [原文标题](../../raw/model-api-reference/preparations/get-api-key.md)。 +- **配置**：强烈建议通过环境变量 `DASHSCOPE_API_KEY` 设置（Linux/macOS/Windows 均支持永久或临时配置），避免代码硬编码。CLI 工具还支持 `bl auth login --api-key`、`bl config set` 或命令行 `--api-key` 临时传入等多种方式。 -### 3. 发起调用 -- 代码中：SDK 初始化时传入 `api_key` 和 `base_url`（如 `dashscope.ApiKeyAuth(api_key=..., base_url=...)`）； -- CLI 中：全局参数 `--api-key`、`--region`、`--base-url` 可覆盖配置； -- HTTP 请求：Header 中添加 `Authorization: Bearer sk-ws-xxx`，Body 指定 `model` 和 `messages`（或 `prompt`）。 +### SDK 与 CLI 安装 +- **SDK**：支持 DashScope 官方 SDK（Python/Java）及 OpenAI 兼容 SDK（Python/Node.js/Java/Go）。Python 环境需 `≥ 3.8`，Java 需 `≥ 8`，Go 需 `≥ 1.22` [原文标题](../../raw/model-api-reference/preparations/install-sdk.md)。 +- **CLI**：`bailian-cli` 仅支持 `npm install -g bailian-cli`（Node ≥ 22.12.0），不支持 `pnpm`/`yarn`。认证推荐 `bl auth login --console`（浏览器 OAuth），也可 `--api-key` 或环境变量方式。 ## 限制和注意事项 -- **密钥安全**：API Key 创建后**仅一次明文展示机会**（除美国地域外），关闭弹窗即不可恢复；禁止硬编码、日志打印、Git 提交；建议定期轮换。 -- **地域隔离**：API Key 与地域强绑定，华北2 创建的 Key 无法直接调用美国地域服务，需切换 `--region` 或创建对应地域 Key。 -- **参数强约束**： - - `temperature` 必须 ∈ [0.0, 2.0)，`top_p` ∈ (0.0, 1.0]，`n` ∈ [1, 4]（图像生成最多 6 张，但 `n` 参数上限为 4）； - - 思考模式（`enable_thinking=true`）**必须启用 `stream=true`**，且 `result_format` 固定为 `"message"`； - - 结构化输出（`response_format={"type": "json_object"}`）**与思考模式互斥**，需关闭 `enable_thinking`。 -- **输入限制**： - - `messages` 数组不能为空；纯文本模型禁止 `content` 为数组（含 `image_url` 等多模态元素），否则报错 `Unexpected item type in content`； - - 文件类调用（Qwen-Long）要求文件 ≤ 150 MB、≤ 15000 页、内容非空，且仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 格式。 -- **模型兼容性**： - - OpenAI SDK 调用需严格匹配百炼的[OpenAI 兼容接口规范](https://help.aliyun.com/zh/model-studio/compatibility-of-openai-with-dashscope)，如 `messages` 必须嵌套在 `input` 对象内（DashScope 协议）或平级（OpenAI 协议）； - - 不同 SDK 对 `seed` 等参数的校验逻辑可能差异（如 DashScope 协议要求 `seed ∈ [0, 9223372036854775807]`），应以[错误码文档](../../raw/model-api-reference/preparations/error-code.md)为准排障。 +- **地域与端点**：API Key 创建地域（如华北2、新加坡、美国弗吉尼亚）决定了默认 `base_url`，OpenAI 兼容与 Anthropic 兼容协议的端点不同，必须匹配所选协议。 +- **权限隔离**：API Key 权限由其**归属业务空间**决定，同一空间内 Key 权限一致；子业务空间 Key 仅可访问该空间已授权的模型与应用。 +- **安全红线**：API Key 明文禁止日志打印、代码提交、聊天记录留存；CLI 在 `auth status` 输出中仅显示脱敏字段（如 `masked: "sk-...xxx"`）。 +- **错误处理**：常见 400 错误（如 `Model not exist`、`InvalidParameter`）需核对模型 ID 大小写、参数范围及 `messages` 结构；`Arrearage` 类错误表明账户欠费，需充值后等待系统同步。 +- **文件限制**：Qwen-Long 模型支持 TXT/DOCX/PDF/EPUB/MOBI/MD，单文件 ≤ 150 MB 且 ≤ 15000 页；无效 URL 需以 `http://`/`https://`/`data:`/`file://` 开头，并注意 `X-DashScope-OssResourceResolve` Header 配置。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md index c4c0b4fd..3dd12bdb 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md @@ -1,55 +1,40 @@ # qwen api reference -Qwen 系列大语言模型通过百炼平台提供多种 API 接入方式，支持文本生成、工具调用、多轮对话等核心能力。开发者可根据现有技术栈（如 OpenAI 或 Anthropic 生态）或对功能完整性的需求，选择最适配的接口协议。所有接口均需通过 DashScope SDK 或 HTTP 直连调用，并依赖有效的 API Key 认证。 +Qwen 系列大语言模型通过百炼平台提供多种 API 接入方式，支持文本生成、工具调用、多轮对话等核心能力。开发者可根据技术栈兼容性、功能需求和运维复杂度选择合适接口。所有接口均需通过阿里云认证（AccessKey 或 STS [Token](../concepts/token.md)）调用，并遵循统一的计费与配额规则。 ## 支持的模型与功能 -当前 Qwen 系列支持以下主流接入协议： +当前支持的 Qwen 模型包括 `qwen-max`、`qwen-plus`、`qwen-turbo` 和 `qwen-vl`（多模态），具体能力因模型而异： -- **OpenAI 兼容 Chat Completions**：完全兼容 `openai>=1.0.0` 客户端，适用于快速迁移已有应用，但不支持原生工具调用（需自行封装）[原文标题](../../raw/model-api-reference/qwen-api-reference.md) -- **OpenAI 兼容-Responses**：在 Chat Completions 基础上增强，内置联网搜索、代码解释器和网页内容提取能力，自动维护对话历史，适合需要开箱即用增强功能的场景 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) -- **Anthropic 兼容 Messages**：支持 `tool_use` 和 `thinking` 模式，可直接声明工具 schema 并接收结构化 tool_result，但部分 Qwen 特有参数（如 `enable_search`）不可用 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) -- **DashScope 原生接口**：功能最全，支持全部模型参数（如 `top_p`, `stop`, `incremental_output`）、流式响应控制、自定义 stop token 及细粒度日志开关，是调试与高阶定制的首选。 +- **文本生成**：所有模型均支持基础 [prompt](../guides/prompt.md)-to-text 生成； +- **工具调用**：`qwen-max` 和 `qwen-plus` 支持[函数调用](../concepts/function-calling.md)（function calling）、联网搜索、代码解释器等扩展能力，需配合 [OpenAI兼容-Responses](../../raw/model-api-reference/qwen-api-reference.md) 或 [Anthropic兼容-Messages](../../raw/model-api-reference/qwen-api-reference.md) 使用； +- **多模态理解**：`qwen-vl` 仅通过 [DashScope](../../raw/model-api-reference/qwen-api-reference.md) 原生接口支持图像输入，不兼容 OpenAI/Anthropic 标准协议。 -> **注意**：原始文档中“OpenAI 兼容-Responses”被描述为“自动管理对话历史”，但实测中若未显式传入 `messages` 且未启用 `enable_session`，历史不会持久化；该行为与 DashScope 原生接口的 session 机制存在差异，建议以 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) 中的接口说明为准，并在生产环境显式管理上下文。 +> **注意**：文档中提及的“内置联网搜索”功能在 `qwen-turbo` 上默认不可用，实际支持情况以 [OpenAI兼容-Responses](../../raw/model-api-reference/qwen-api-reference.md) 的最新说明为准；若调用失败，请确认模型版本与接口组合是否匹配。 ## 关键参数 -| 参数名 | 类型 | 说明 | 适用接口 | -|--------|------|------|----------| -| `model` | string | 必填，如 `qwen-max`, `qwen-plus`, `qwen-turbo` | 全部 | -| `messages` | array | 对话消息列表，格式为 `[{role: "user", content: "..."}]` | Chat Completions / Responses / Anthropic Messages / DashScope | -| `tools` | array | 工具定义数组（OpenAI 格式或 Anthropic 格式） | Responses / Anthropic Messages / DashScope（需配合 `tool_choice`） | -| `stream` | boolean | 是否启用流式响应 | 全部（DashScope 支持更精细的 `incremental_output` 控制） | -| `max_tokens` | integer | 最大输出 token 数 | 全部 | -| `enable_search` | boolean | 是否启用联网搜索（仅 DashScope 和 Responses 支持） | DashScope / Responses | +| 参数 | 说明 | 必填 | 示例值 | +|------|------|------|--------| +| `model` | 模型标识符 | 是 | `"qwen-max"` | +| `messages` | 对话历史（OpenAI/Anthropic 格式）或 `input`（DashScope 格式） | 是 | `[{"role":"user","content":"你好"}]` | +| `temperature` | 控制输出随机性（0.0–2.0） | 否 | `0.7` | +| `top_p` | 核采样阈值（0.0–1.0） | 否 | `0.8` | +| `max_tokens` | 最大生成长度 | 否 | `1024` | +| `tools` / `tool_choice` | 工具定义与调用策略（仅部分模型+接口支持） | 否 | 见 [Anthropic兼容-Messages](../../raw/model-api-reference/qwen-api-reference.md) 文档 | ## 使用方式 -1. **认证**：通过环境变量 `DASHSCOPE_API_KEY` 或请求头 `Authorization: Bearer ` 认证 -2. **调用示例（DashScope 原生）**： - ```bash - curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "qwen-max", - "input": {"messages": [{"role":"user","content":"你好"}]}, - "parameters": {"max_tokens": 512} - }' - ``` -3. **SDK 调用（Python）**： - ```python - from dashscope import Generation - resp = Generation.call(model='qwen-max', messages=[{'role':'user','content':'你好'}]) - ``` +- **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**：使用标准 `openai` Python SDK，设置 `base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"` 并传入 DashScope API Key； +- **Anthropic 兼容接口**：需将 `Content-Type` 设为 `application/json`，并使用 `messages` 字段而非 `prompt`； +- **DashScope 原生接口**：推荐使用 `dashscope` SDK，支持更细粒度控制（如 `enable_search`、`enable_code_interpreter` 等布尔开关），详见 [DashScope](../../raw/model-api-reference/qwen-api-reference.md) 文档。 ## 限制和注意事项 -- 所有接口默认单次请求最大 `messages` 长度为 32768 tokens（含输入+输出），超限将返回 `400 Bad Request` -- `qwen-max` 和 `qwen-plus` 支持 32K 上下文，`qwen-turbo` 为 8K，实际可用长度受系统 [prompt](../guides/prompt.md) 占用影响 -- 流式响应中，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)返回 `delta.content` 字段，DashScope 原生接口返回 `output.text`（非增量）或 `output.choices[0].message.content`（增量模式需设 `incremental_output=true`） -- 工具调用结果必须由客户端解析并重新提交 `tool_result`，服务端不自动执行后续推理（Anthropic Messages 除外，其支持自动循环调用） +- 单次请求 `messages` 总 token 数上限为 32,768（`qwen-max`），其他模型略低； +- `qwen-vl` 图像输入仅支持 base64 编码或公网可访问 URL，不支持本地文件路径； +- 所有接口均不支持流式响应中的 `delta.tool_calls` 结构（仅返回完整 `tool_calls`），此行为与 OpenAI v1.0+ 不一致； +- 配额与计费按模型+输入/输出 token 分别统计，详细规则参见 [DashScope](../../raw/model-api-reference/qwen-api-reference.md) 官方说明。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md index 2ccd5b44..b7a9822e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md @@ -1,82 +1,88 @@ # toolkits and [frameworks](frameworks.md) -阿里云百炼平台提供多种 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)及配套工具链，支持开发者快速迁移现有应用或构建新场景。所有接口均基于统一的 `compatible-mode/v1` 协议层，通过调整 `base_url`、`api_key` 和 `model` 三个参数即可接入，无需重写业务逻辑。核心能力覆盖文本生成、视觉理解、向量嵌入、批量处理、会话管理与低代码集成（如 LangChain），适配从单次调用到大规模异步任务的全栈需求。 +阿里云百炼提供多种 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)及专用工具链，支持开发者快速迁移现有应用或构建新场景。所有接口均基于统一的 `compatible-mode/v1` 协议层，通过调整 `base_url`、`api_key` 和 `model` 三要素即可接入，无需重写核心逻辑。各接口在功能定位、模型支持和使用约束上存在明确分工，需按场景选型。 ## 支持的模型/功能 -百炼支持的 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)按功能划分为以下几类： +百炼支持的 OpenAI 兼容能力覆盖文本、视觉、向量、文件与对话管理等维度： -- **Chat Completions**：通用对话接口，支持 `qwen-plus`、`qwen3.7-plus`、`qwen-coder-turbo`、`deepseek-r1`、`kimi`、`glm` 等数十种文本与代码模型；也兼容多模态模型如 `qwen-vl-plus`、`qwen3-vl-plus` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 -- **Responses API**：面向智能体的增强型接口，内置联网搜索、网页抓取、代码解释器等工具，支持 `qwen3.7-max`、`qwen3.5-plus`、`qwen3-coder-next` 等新一代模型，显著简化复杂任务编排 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 -- **Completions**：专用于代码补全与内容续写，当前仅支持 `qwen-coder-turbo` 模型，支持前缀补全与“前缀+后缀”中间生成两种模式 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 -- **Vision**：图像理解专用接口，支持 `qwen-vl-plus`、`qwen3-vl-plus`、`QVQ`、`qwen-vl-ocr`，兼容 OpenAI 的 `image_url` 输入格式 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 -- **Embedding**：文本向量化接口，支持 `text-embedding-v4`（2048维）、`v3`、`v2`、`v1` 四代模型，支持 `dimensions` 参数动态指定维度，适用于检索增强（RAG）等场景 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 -- **Batch（文件输入）**：异步批量处理接口，支持 `qwen3.7-max`、`qwen3.5-omni-plus`、`qwen-vl-ocr` 等模型，单请求上下文最大支持 256K tokens，费用为实时调用的 50% [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md)。 -- **Conversations**：会话状态管理接口，支持跨设备/长时间中断的上下文持久化，配合 Responses API 实现自动历史注入，避免手动维护消息数组 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 +- **Chat Completions**：通用对话接口，支持 `qwen-plus`、`qwen3-*` 系列、`Qwen-VL`、`Qwen-Coder`、`DeepSeek`（三方直供）、`Kimi`、`GLM`、`MiniMax` 等模型，但 **Qwen-Audio 不支持该协议** [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **Responses API**：面向智能体的演进接口，内置联网搜索、网页抓取等工具，支持 `qwen3.7-plus`、`qwen3-coder-*` 等数十个 `qwen3-*` 模型，**不支持 `qwen-coder-turbo` 等旧版 coder 模型** [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 +- **Completions**：专用于代码/文本补全，**当前仅支持 `qwen-coder-turbo`**，且仅限北京地域 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 +- **Vision**：多模态理解接口，支持 `Qwen-VL`、`QVQ`、`Qwen-OCR`，其中 **QVQ 仅支持[流式输出](../concepts/streaming-output.md)** [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 +- **Embedding**：文本向量化接口，支持 `text-embedding-v1` 至 `v4`，**多模态 Embedding 模型（如 `qwen3-vl-embedding`）不兼容 OpenAI 接口** [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 +- **Files**：文件上传与管理，用途包括文档问答（`file-extract`）、批量推理（`batch`）和模型调优（`fine-tune`），单文件上限依用途而异（150 MB / 500 MB / 300 MB） [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 +- **Batch**：异步批量处理，支持两种模式： + - 文件输入（JSONL 格式）：适用于大规模任务，费用为实时调用的 50%； + - Batch Chat（同步阻塞）：单请求模式，保持实时 API 调用习惯，同样享 5 折优惠 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md)。 +- **Conversations**：会话状态管理，配合 Responses API 实现跨设备上下文延续，支持创建、查询、更新、删除会话及添加消息项 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 -> **注意**：文档 5（Batch 文件输入）与文档 7（Batch Chat）存在关键差异——前者为**异步文件提交模式**（需上传 JSONL 文件、轮询状态、下载结果），后者为**同步 HTTP 请求模式**（保持连接等待结果返回，单请求）。二者适用场景不同，不可混用；文档 7 明确声明“本接口仅支持提交单个请求”，而文档 5 支持千级并发请求批量处理。 +> **注意**：文档 6（Batch 文件输入）与文档 7（Batch Chat）对同一模型（如 `qwen-plus`）的适用性描述存在差异——前者明确列出 `qwen-plus` 在华北2（北京）可用，后者亦将其列入支持列表，但文档 7 的“适用范围”小节未注明地域限制，而文档 6 明确要求中国内地使用 `https://dashscope.aliyuncs.com`，国际使用 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`。实际使用时，**Batch Chat 必须使用专用域名 `https://batch.dashscope.aliyuncs.com`**，与常规 Batch 文件接口的域名不同，此为关键区别。 ## 关键参数 -所有 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)共享以下核心参数，行为与 OpenAI 官方一致： +所有 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)共用以下核心参数，部分接口有扩展： -| 参数 | 类型 | 说明 | -|------|------|------| -| `model` | string | 必填。模型名称，如 `"qwen3.7-plus"`、`"text-embedding-v4"`。注意：`qwen-audio` 不支持 OpenAI 协议，仅支持 DashScope 原生协议。 | -| `base_url` | string | 必填。服务端点，**必须使用业务空间专属域名**以获得最佳性能与稳定性：
• 北京：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
• 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`
• 弗吉尼亚：`https://dashscope-us.aliyuncs.com/compatible-mode/v1`
• 法兰克福：`https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1`
• 东京：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| `api_key` | string | 必填。阿里云百炼 API Key，**各地域 Key 不互通**，需按 `base_url` 所在地域分别获取并配置。 | -| `stream` | boolean | 可选。启用[流式输出](../concepts/streaming-output.md)（`true`），适用于长响应或前端实时渲染。 | -| `stream_options` | object | 可选。当 `stream=true` 时，设 `{"include_usage": true}` 可在最后一 chunk 返回 token 统计。 | -| `temperature` / `top_p` | float | 可选。互斥使用，控制生成多样性（`temperature ∈ [0, 2.0)`，`top_p ∈ (0, 1.0]`）。 | -| `max_tokens` | integer | 可选。限制响应最大 token 数，超限将截断（不影响模型内部生成过程）。 | - -> **注意**：`enable_thinking` 是 Batch 场景特有参数（见文档 5 和 7），用于显式开关思考模式（影响 token 计费），**必须作为 JSONL `body` 的顶层字段传入，不可置于 `extra_body` 中**。该参数在 Chat Completions 或 Responses 同步接口中无效。 +- `base_url`：必须配置，不同接口/地域有严格对应关系： + - Chat/Responses/Vision/Embedding/Conversations：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（北京）、`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡）等； + - Files/Batch（文件输入）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（中国内地）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（国际）； + - Batch Chat：**必须使用 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`**，与其他接口域名隔离。 +- `model`：模型名称需严格匹配支持列表，例如 `qwen3.7-plus`（Responses）、`qwen-coder-turbo`（Completions）、`text-embedding-v4`（Embedding）。 +- `api_key`：需与 `base_url` 所属地域一致（如北京地域 API Key 不能用于新加坡 endpoint）。 +- `stream`：布尔值，控制是否流式返回，默认 `false`；Vision 接口的 `QVQ` 模型强制流式。 +- `stream_options`：当 `stream=true` 时，设 `{"include_usage": true}` 可在最后一 chunk 返回 token 统计。 +- `enable_thinking`：仅 Batch 场景下有效，控制思考模式开关（`true`/`false`），需作为 JSONL `body` 的顶层字段，**不可置于 `extra_body` 内**。 +- `previous_response_id`（Responses API）：用于多轮对话，传入上一轮响应的顶层 `id`（UUID 格式），非 `output` 中 `msg_*` ID。 +- `purpose`（Files API）：必需字段，取值 `file-extract`、`batch` 或 `fine-tune`，决定文件用途及格式校验规则。 ## 使用方式 -### 1. SDK 调用（推荐） -安装对应 SDK 并初始化客户端： -```python -from openai import OpenAI -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" -) -``` -- **Chat**：`client.chat.completions.create(model=..., messages=[...])` -- **Responses**：`client.responses.create(model=..., input="...")` -- **Completions**：`client.completions.create(model=..., prompt="...")` -- **Embedding**：`client.embeddings.create(model=..., input="...")` -- **Batch（文件）**：先 `client.files.create(file=..., purpose="batch")`，再 `client.batches.create(input_file_id=..., endpoint="/v1/chat/completions")` -- **Conversations**：`client.conversations.create(items=[...])` → 获取 `id` 后用于后续 `responses.create(previous_response_id=...)` - -### 2. LangChain 集成 -- **OpenAI 兼容层**（`langchain_openai`）：仅支持部分模型（如 `qwen-plus`），依赖 `base_url` 指向百炼兼容端点 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)。 -- **DashScope 原生层**（`langchain-community` + `dashscope`）：支持全部百炼模型（含部署模型），使用 `ChatTongyi` 类，不依赖 OpenAI 协议。 - -### 3. HTTP 直连 -构造标准 OpenAI 格式请求： -```bash -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "qwen3.7-plus", - "messages": [{"role":"user","content":"你好"}] - }' -``` +### 基础调用流程 +1. **获取并配置凭证**：在百炼控制台获取对应地域的 API Key，并推荐配置至环境变量 `DASHSCOPE_API_KEY`。 +2. **初始化客户端**：使用 OpenAI SDK（Python/Node.js/Java/Go/C#）或 HTTP 客户端，设置 `base_url` 和 `api_key`。 +3. **构造请求**：按接口规范传入 `model`、输入内容（如 `messages`、`input`、`prompt`、`file`）及其他参数。 +4. **处理响应**：解析 JSON 结构，注意 `choices[0].message.content`（Chat）、`output_text`（Responses）、`data[0].embedding`（Embedding）等字段差异。 + +### 典型示例 +- **Chat Completions（非流式）**： + ```python + from openai import OpenAI + client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") + resp = client.chat.completions.create(model="qwen-plus", messages=[{"role":"user","content":"你好"}]) + print(resp.choices[0].message.content) + ``` +- **Responses API（多轮）**： + ```python + resp1 = client.responses.create(model="qwen3.7-plus", input="我的名字是张三") + resp2 = client.responses.create(model="qwen3.7-plus", input="你还记得我的名字吗？", + previous_response_id=resp1.id) # 注意传顶层 id + ``` +- **Files API（上传）**： + ```python + file_obj = client.files.create(file=Path("doc.pdf"), purpose="file-extract") # 用于 Qwen-Long/Qwen-Doc-Turbo + ``` +- **Batch Chat（同步）**： + ```python + client = OpenAI(..., base_url="https://batch.dashscope.aliyuncs.com/compatible-mode/v1").with_options(timeout=1800.0) + resp = client.chat.completions.create(model="qwen-plus", messages=[...]) # 阻塞等待完成 + ``` + +### LangChain 集成 +- **OpenAI 方式**：使用 `langchain_openai.ChatOpenAI`，仅支持 OpenAI 兼容模型（如 `qwen-plus`），`base_url` 设为 `https://dashscope.aliyuncs.com/compatible-mode/v1`。 +- **DashScope 方式**：使用 `langchain_community.chat_models.tongyi.ChatTongyi`，支持全部百炼文本模型，需安装 `dashscope` 包，`dashscope_api_key` 为凭证字段 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)。 ## 限制和注意事项 -- **地域与 Key 绑定**：API Key 与 `base_url` 所属地域强绑定（如北京 Key 不能用于新加坡 `base_url`），且各接口对地域支持不完全一致（例如 `completions` 接口仅支持北京地域）。 -- **域名迁移强制要求**：旧域名 `https://dashscope.aliyuncs.com` 和 `https://dashscope-intl.aliyuncs.com` 已不推荐使用，**所有新项目必须采用 `{WorkspaceId}.xxx.maas.aliyuncs.com` 专属域名**，否则可能遭遇性能下降或未来停服风险。 +- **地域与域名绑定**：API Key、`base_url`、模型可用性三者强绑定。例如，北京地域 API Key 无法用于 `dashscope-us.aliyuncs.com`；`qwen3.7-plus` 在 Responses API 中可用，但在 Completions API 中不可用。 +- **业务空间专属域名迁移**：北京、新加坡地域已启用 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 等新域名，**旧域名 `https://dashscope.aliyuncs.com` 将逐步停用**，文档 1、2、4、8 均强调此迁移要求。 - **模型能力差异**： - - `Qwen-Audio` 不支持 OpenAI 兼容协议（见文档 1）； - - `qwen3.5-omni-plus` 在 Batch 场景下不支持语音输出（文档 5 和 7）； - - `QVQ` 模型仅支持[流式输出](../concepts/streaming-output.md)（文档 4）。 -- **Batch 与 Conversations 的协同**：`previous_response_id`（Responses API）与 `conversation_id`（Conversations API）是两个独立的状态管理机制，前者用于单次响应链路，后者用于长期会话存储，不可混用。 -- **文件上传配额**：`purpose=file-extract`（文档分析）单文件上限 150 MB；`purpose=batch`（批量任务）单文件上限 500 MB；`purpose=fine-tune`（调优）单文件上限 300 MB（见文档 6）。 -- **错误处理**：所有接口遵循 OpenAI 错误格式（`{"error": {"code": "...", "message": "..."}}`），具体码表参考[统一错误码文档](https://help.aliyun.com/zh/model-studio/error-code)。 + - `Qwen-Audio` 不支持 OpenAI 兼容协议； + - `QVQ` 模型强制流式，无非流式选项； + - `qwen3.7-*` 系列模型在 Batch 场景默认开启思考模式，需显式设置 `enable_thinking=false` 关闭以控本。 +- **文件服务配额**：Files API 总存储上限 100 GB、最多 10,000 个文件，超限后上传失败，需主动清理。 +- **Batch 超时机制**：Batch Chat 默认等待 3600 秒（1 小时），超时断连；Batch 文件任务最长等待 24 小时，需轮询状态。 +- **安全实践**：**严禁硬编码 API Key**，务必通过环境变量或密钥管理服务注入；SDK 调用时优先使用 `os.getenv("DASHSCOPE_API_KEY")`。 ## 来源文档 @@ -84,8 +90,8 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ - [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) -- [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) +- [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md index 096fba21..8a6afa9a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md @@ -1,88 +1,47 @@ # vector and sort -`vector and sort` 是百炼平台提供的核心向量化与排序能力集合，涵盖文本、多模态内容的嵌入（Embedding）生成，以及基于语义相关性的精准重排序（Rerank）。该能力支撑[检索增强生成](../concepts/rag.md)（RAG）、跨模态搜索、聚类分析等关键AI应用，支持同步/异步调用、OpenAI兼容接口及多语言、多分辨率、多模态输入。 +百炼平台提供文本向量化（vector）、多模态向量化及文本排序（rerank）三大核心能力，覆盖语义搜索、RAG、跨模态检索等典型场景。所有服务均支持同步与异步调用模式，可通过 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 DashScope 原生 SDK 快速集成。开发者需根据数据规模、模态类型、延迟敏感度及精度要求选择合适模型与调用方式。 ## 支持的模型/功能 -### 文本向量模型 -- **同步接口**：支持 `qwen3.7-text-embedding`（最高128K token）、`text-embedding-v4`（最高8K token）、`text-embedding-v3/v2/v1` 等系列，提供灵活维度选择（如256–2560维）和多语种支持（最多201种）[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 -- **异步批处理接口**：仅支持 `text-embedding-async-v1/v2`，适用于超大批量文本（单次最多100,000行），但不支持动态维度配置，固定输出1536维向量[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **OpenAI兼容模式**：通过 `compatible-mode/v1/embeddings` endpoint 调用 `text-embedding-v4` 等模型，支持 `dimensions` 和 `encoding_format` 参数，便于生态迁移[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **通用文本向量模型**：支持 `qwen3.7-text-embedding`、`text-embedding-v4`、`v3`、`v2`、`v1` 等版本，适用于纯文本语义表征；其中 `qwen3.7-text-embedding` 支持最高 128,000 [Token](../concepts/token.md) 单条输入与 2560 维向量输出 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **批处理异步向量模型**：`text-embedding-async-v2` 支持单次 100,000 行文本批量处理，适用于大规模离线向量化任务 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **多模态向量模型**：`qwen3-vl-embedding`、`tongyi-embedding-vision-plus-2026-03-06` 等支持文本、图像、视频统一语义空间编码，提供独立向量与融合向量两种模式 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- **文本排序（Rerank）模型**：`qwen3-rerank`（纯文本）、`qwen3-vl-rerank`（多模态）、`gte-rerank-v2`（已进入下线过渡期）用于对召回结果进行精准重排序；注意 `gte-rerank` 模型将于 2026 年 05 月 30 日下线，应迁移至 `qwen3-rerank` [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 -### 多模态向量模型 -- 支持 `qwen3-vl-embedding`、`tongyi-embedding-vision-plus-2026-03-06` 等模型，统一文本/图像/视频向量空间，支持独立向量（各模态单独编码）与融合向量（跨模态联合编码）两种模式[原文标题](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 -- 关键参数如 `enable_fusion`（仅 `qwen3-vl-embedding`）、`res_level`（分辨率档位）、`max_video_frames`（视频帧采样上限）均需在 `parameters` 中显式指定。 - -### 排序（Rerank）模型 -- `qwen3-rerank`：纯文本排序，OpenAI兼容接口，支持 `instruct` 任务提示词，最大文档数500条[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 -- `qwen3-vl-rerank`：多模态排序，支持文本/图片/视频混合查询与文档，需使用 `input.query` 和 `input.documents` 结构化输入。 -- `gte-rerank-v2`：已进入下线倒计时（2026年5月30日），建议迁移到 `qwen3-rerank` 或 `qwen3-vl-rerank`[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -> **注意**：文档2中 `text-embedding-v4` 的“最大行数”为10，而文档1中 `text-embedding-async-v2` 的“单次请求文本最大行数”为100,000——二者属不同调用路径（同步 vs 异步），无矛盾；但文档2称 `qwen3.7-text-embedding` 支持“单行最长128,000 Token”，而文档1明确 `text-embedding-async-v2` 单行上限为2,048 Token，此为模型能力差异，非错误。 +> **注意**：文档 1 中 `text-embedding-v2` 的“单行最大 2,048 [Token](../concepts/token.md)”与文档 4 中 `qwen3-vl-rerank` 的“单条最大输入[Token](../concepts/token.md)：8,000（文本）”存在隐含矛盾——前者为向量模型输入限制，后者为排序模型输入限制，二者不可直接对比；但需注意 `qwen3-rerank` 的单条文档限制为 4,000 Token，而 `qwen3-vl-rerank` 文本类文档上限为 100 条 × 4,000 Token，实际使用中应以各模型自身文档为准。 ## 关键参数 -| 参数名 | 适用场景 | 说明 | 必选/可选 | -|--------|----------|------|-----------| -| `model` | 所有接口 | 模型名称，如 `text-embedding-v4`、`qwen3-vl-rerank` | 必选 | -| `input` | 同步/多模态/Rerank | 字符串、字符串数组、文件对象或 `{"contents": [...]}` 结构体；异步批处理仅支持 `url` 字段 | 必选 | -| `dimensions` | 同步文本模型 | 指定向量维度（如1024），仅 `qwen3.7-text-embedding`、`text-embedding-v3/v4` 支持；`multimodal-embedding-v1` 等固定维度模型不支持 | 可选 | -| `text_type` | 异步批处理 | `"query"` 或 `"document"`，影响向量表征优化方向 | 可选（默认 `"document"`） | -| `enable_fusion` | `qwen3-vl-embedding` | `true` 时返回融合向量，`false`（默认）时返回独立向量 | 可选 | -| `top_n` | Rerank | 返回前N个最相关结果，`qwen3-rerank` 直接置于顶层，`qwen3-vl-rerank` 需置于 `parameters` 内 | 可选 | -| `instruct` | `qwen3-rerank` / `qwen3-vl-rerank` | 任务指令（如 `"Retrieve semantically similar text."`），显著影响排序策略 | 可选 | +| 参数名 | 类型 | 说明 | 支持模型 | +|--------|------|------|----------| +| `model` | string | 必选，指定模型名称 | 全部 | +| `input` / `query` / `documents` | string / array / object | 必选，输入内容格式依模型而异：文本向量支持 `string`/`array`/`file`；多模态向量使用 `contents: [{text:"..."}, {image:"..."}]`；排序模型中 `qwen3-rerank` 要求 `query` 和 `documents` 同级，`qwen3-vl-rerank` 则需嵌套在 `input` 对象内 | [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)、[文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) | +| `dimensions` | integer | 可选，指定输出向量维度（如 1024、2048）；`text-embedding-v1/v2` 不支持该参数；`multimodal-embedding-v1` 固定 1024 维 | [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)、[Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) | +| `encoding_format` | string | 可选，仅支持 `"float"` | [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) | +| `top_n` | integer | 可选，排序模型返回前 N 个结果 | [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) | +| `instruct` | string | 可选，排序任务指令（如 `"Retrieve semantically similar text."`），影响相关性判断逻辑 | [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) | +| `enable_fusion` | boolean | 可选，仅 `qwen3-vl-embedding` 支持，启用后将 `contents` 中所有模态融合为单一向量 | [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) | ## 使用方式 -### 调用路径选择 -- **小批量实时向量化**（≤25条文本）：优先使用同步接口 `POST /compatible-mode/v1/embeddings`，延迟低、响应快。 -- **超大批量离线处理**（≥1000行）：必须使用异步批处理接口 `POST /api/v1/services/embeddings/text-embedding/text-embedding` + `GET /api/v1/tasks/{task_id}`，避免超时[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **多模态内容处理**：统一使用 `POST /api/v1/services/embeddings/multimodal-embedding/multimodal-embedding`，按 `contents` 数组构造输入。 -- **排序任务**：`qwen3-rerank` 用 OpenAI 兼容 `/compatible-api/v1/reranks`；其余 rerank 模型用 `/api/v1/services/rerank/text-rerank/text-rerank`。 - -### SDK 与 HTTP 差异 -- DashScope SDK 对参数进行了扁平化封装（如 `BatchTextEmbedding.call(..., url=..., text_type=...)`），无需手动构造 `input` 和 `parameters` 嵌套结构；HTTP 则严格要求 JSON 层级[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- OpenAI SDK 调用需设置 `base_url` 为 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，并传入 `DASHSCOPE_API_KEY` 作为 `api_key`。 - -### 多模态输入示例 -```json -{ - "model": "qwen3-vl-embedding", - "input": { - "contents": [ - {"text": "商品标题"}, - {"image": "https://example.com/1.jpg"}, - {"image": "https://example.com/2.jpg"}, - {"video": "https://example.com/demo.mp4"} - ] - }, - "parameters": { - "enable_fusion": true, - "dimension": 2048 - } -} -``` +- **同步调用（低延迟、小批量）**：适用于实时搜索、RAG 在线推理。使用 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)时，`base_url` 需配置为 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（文本向量）或 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks`（`qwen3-rerank`）；HTTP endpoint 为 `POST /embeddings` 或 `POST /reranks`。 +- **异步批处理（高吞吐、离线任务）**：适用于日志/商品库全量向量化。调用 `text-embedding-async-v2` 时需设置请求头 `X-DashScope-Async: enable`，并通过 `task_id` 轮询结果；文件需托管于公网可访问 URL（如 OSS），单文件 ≤ 200MB [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **多模态联合处理**：使用 `qwen3-vl-embedding` 或 `tongyi-embedding-vision-plus-2026-03-06` 时，通过 `contents` 数组传入混合模态对象；融合向量需确保所有模态在同一 `content` 对象内（如 `{"text":"...", "image":"..."}`），而非分散在多个数组元素中 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- **SDK 封装调用**：推荐使用 DashScope Python/Java SDK，自动处理认证、重试与响应解析。注意 SDK 参数扁平化（如 `top_n` 直接传参），而 HTTP 接口部分模型需嵌套在 `parameters` 或 `input` 内 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 ## 限制和注意事项 -- **Token 与尺寸限制**： - - 同步文本模型：`qwen3.7-text-embedding` 单行上限128,000 Token；`text-embedding-v4` 单行上限8,192 Token；异步批处理单行上限2,048 Token[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 - - 多模态模型：图片单张≤10 MB（`qwen3-vl-embedding`），视频≤50 MB；`tongyi-embedding-vision-plus` 图片≤3 MB[原文标题](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 - - Rerank：`qwen3-vl-rerank` 文本文档上限100条、图片上限40张、视频上限4个；总请求 Token 上限120,000[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -- **并发与配额**： - - 异步批处理：单用户并发运行中任务数上限3个，排队中+运行中总数上限50个[原文标题](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 - - 免费额度：各模型独立计算，如 `text-embedding-v2` 享50万Token免费额度，`qwen3-vl-embedding` 享100万Token，均限百炼开通后90天内有效。 - -- **关键注意事项**： - - HTTP 异步调用**必须**携带 `X-DashScope-Async: enable` 请求头，否则报错 `current user api does not support synchronous calls`。 - - `qwen2.5-vl-embedding` 仅支持融合向量，不支持 `enable_fusion` 参数（因其恒为 true）；`tongyi-embedding-vision-plus` 系列则不支持该参数，融合需将多模态字段置于同一 `content` 对象内。 - - `gte-rerank-v2` 已标记为下线模型，新项目请勿选用[原文标题](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 +- **Token 与行数限制严格区分**：`qwen3.7-text-embedding` 单条支持 128,000 Token 但最多 20 行；`text-embedding-v4` 单条仅 8,192 Token 且最多 10 行；`text-embedding-async-v2` 单次请求支持 100,000 行但每行限 2,048 Token —— 超限将被截断并导致语义失真。 +- **异步任务生命周期**：批处理任务 `task_id` 有效期为 24 小时，结果 URL 仅在此期间有效，需及时下载 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **模型兼容性风险**：`gte-rerank-v2` 已标记为下线模型，新项目禁止接入；`qwen2.5-vl-embedding` 仅支持融合向量且不支持 `multi_images`，与 `tongyi-embedding-vision-plus` 系列行为不一致，迁移时需重构输入结构。 +- **地域与 endpoint 绑定**：华北2（北京）地域的 `base_url` 与新加坡地域的 `base_url` 不同，且 `qwen3-rerank` 使用兼容模式 endpoint，而 `qwen3-vl-rerank` 使用原生 `/api/v1/services/...` endpoint，混用将导致 404 错误。 +- **免费额度时效性**：所有模型的免费额度（如 100 万 Token）均自百炼开通起 90 天内有效，过期未用完即作废。 ## 来源文档 -- [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) +- [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) - [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md index 889c11aa..76c0452a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md @@ -1,95 +1,96 @@ # video generation api -百炼平台的 Video Generation API 提供多种视频生成与编辑能力，包括文生视频（T2V）、图生视频（I2V）、参考生视频（R2V）、首尾帧生视频（KF2V）、视频编辑、风格重绘及数字人播报等。所有接口均采用异步调用模式，需通过 `task_id` 轮询获取结果，任务有效期为 24 小时。 +百炼平台的 Video Generation API 提供多种视频生成能力，包括文生视频（T2V）、图生视频（I2V）、参考生视频（R2V）、视频编辑、风格重绘及数字人播报等。所有接口均采用异步调用模式，需通过 `task_id` 轮询获取结果，任务 ID 有效期为 24 小时。开发者需确保模型、Endpoint URL 与 API Key 严格属于同一地域，跨地域调用将失败。 ## 支持的模型/功能 -API 支持以下主流视频生成模型及对应能力： +API 支持多系列模型，按能力可分为以下几类： -- **文生视频（T2V）**：`happyhorse-1.1-t2v`、`wan2.7-t2v-2026-06-12`、`pixverse/pixverse-c1-t2v`、`kling/kling-v3-video-generation`、`vidu/viduq3-turbo_text2video` -- **图生视频（I2V）**： - - 基于首帧：`happyhorse-1.1-i2v`、`wan2.7-i2v-2026-04-25`、`pixverse/pixverse-c1-it2v`、`vidu/viduq3-pro-fast_img2video` - - 基于首尾帧：`pixverse/pixverse-c1-kf2v`、`vidu/viduq3-turbo_start-end2video`、`wan2.2-kf2v-flash`（[万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)） -- **参考生视频（R2V）**：`happyhorse-1.1-r2v`、`wan2.7-r2v-2026-06-12`、`pixverse/pixverse-c1-r2v`、`vidu/viduq3-ad_reference2video` -- **视频编辑**：`happyhorse-1.0-video-edit`、`wan2.7-videoedit`、`wanx2.1-vace-plus`（[万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md)） -- **数字人与肖像动画**：`wan2.2-s2v`（说话/唱歌）、`emo-v1`（悦动人像）、`liveportrait`（灵动人像）、`videoretalk`（口型替换）、`animate-anyone-gen2`（舞蹈复刻） -- **专用功能**：`video-style-transform`（8种艺术风格重绘）、`emoji`（表情包模板驱动）、`wan2.2-animate-move`（图生动作） +- **文生视频（T2V）**：输入文本提示词生成视频，主流模型包括 `wan2.7-t2v-*`、`vidu/viduq3-*-text2video`、`kling/kling-v3-*-video-generation`、`pixverse/pixverse-*-t2v` 和 `happyhorse-1.1-t2v`。其中万相2.7支持自然语言分镜（如“第1个镜头[0-3秒] 全景…”），而可灵支持显式 `multi_shot` + `multi_prompt` 结构化分镜 [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)。 -> **注意**：`wan2.6` 及更早版本（如 `wan2.2`、`wanx2.1`）属于旧版协议，其 endpoint 路径为 `/api/v1/services/aigc/image2video/video-synthesis`，而 `wan2.7+`、`HappyHorse`、`PixVerse`、`Kling`、`Vidu` 等新模型统一使用 `/api/v1/services/aigc/video-generation/video-synthesis`。混用路径将导致 404 错误。 +- **图生视频（I2V）**：基于首帧图像生成视频，支持模型包括 `wan2.7-i2v-*`、`vidu/viduq3-*-img2video`、`pixverse/pixverse-*-it2v`、`happyhorse-1.1-i2v`；另有首尾帧生视频（KF2V）能力，由 `wan2.7-i2v-*`（新版）、`vidu/viduq3-*-start-end2video`、`pixverse/pixverse-*-kf2v` 及旧版 `wan2.2-kf2v-flash` 提供 [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md)。 + +- **参考生视频（R2V）**：融合多张参考图（人物/场景/物体）生成角色一致的视频，模型包括 `wan2.7-r2v-*`、`vidu/viduq3-*-reference2video`（分广告/短剧场景）、`pixverse/pixverse-*-r2v`、`happyhorse-1.1-r2v`。注意万相2.6旧版 R2V 使用 `reference_urls` 字段，而2.7新版统一使用 `media` 数组 [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md)。 + +- **视频编辑与重绘**：支持指令驱动编辑（如“转为黏土风格”）、局部替换、口型同步（VideoRetalk）、风格迁移（`video-style-transform`）及视频换人（`wan2.2-animate-mix`）。 + +- **数字人与表情包**：面向人像的轻量级生成，包括 `wan2.2-s2v`（说话/唱歌）、`emo-v1`（唱演）、`liveportrait`（播报）、`videoretalk`（口型替换）、`emoji`（表情包）和 `animate-anyone-gen2`（舞蹈）。 + +> **注意**：万相2.1–2.6 系列（如 `wan2.6-t2v`、`wan2.6-i2v-flash`）属旧版协议，其请求路径为 `/api/v1/services/aigc/video-generation/video-synthesis`，但部分旧版文档（如文档29）错误地指向 `/api/v1/services/aigc/image2video/video-synthesis`，实际应统一使用 `video-generation` 路径，以避免 404 错误。 ## 关键参数 -所有请求必须包含以下基础参数： - -- **`model`**（必选）：模型名称，严格区分大小写和版本后缀（如 `wan2.7-i2v-2026-04-25`）。 -- **`input`**（必选）： - - 文生视频：`{"prompt": "..."}` - - 图/参考生视频：`{"media": [{"type": "...", "url": "..."}], "prompt": "..."}`；部分旧模型（如 `wan2.2-kf2v-flash`）仍使用 `first_frame_url`/`last_frame_url` 字段。 -- **`parameters`**（可选）：常见字段包括： - - `resolution`（如 `"720P"`、`"1080P"`）或 `size`（如 `"1280*720"`） - - `duration`（秒数，通常支持 3–8 秒） - - `watermark`: `true`/`false`（默认 `true`） - - `audio`: `true`/`false`（仅部分模型支持音频生成） - - 多镜头控制：`wan2.7` 系列通过 `prompt` 内时间戳描述分镜；`wan2.6` 需显式设置 `"shot_type": "multi"` 和 `"prompt_extend": true`（[万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)） - -请求头必须包含： -- `X-DashScope-Async: enable`（强制异步） -- `Authorization: Bearer $DASHSCOPE_API_KEY` +所有请求必须包含以下基础字段： + +- `model`（必选）：模型标识符，如 `wan2.7-t2v-2026-06-12`、`vidu/viduq3-pro_text2video`。 +- `input`（必选）：包含 `prompt`（文本提示）及媒体资源（`media` 或 `video_url`/`image_url` 等），格式因模型而异。 +- `parameters`（可选）：控制输出质量与时长，常见参数包括： + - `resolution`：如 `"720P"`、`"1080P"`（部分模型也支持 `"540P"`、`"480P"`）； + - `duration`：视频时长（秒），通常范围为 3–8 秒； + - `size`：分辨率宽高（如 `"1280*720"`），与 `resolution` 互斥，优先级依模型而定； + - `watermark`：布尔值，启用/禁用水印（默认 `true`）； + - `audio`：布尔值，是否生成音频（仅部分 T2V/I2V 模型支持）； + - `prompt_extend`：旧版万相多镜头必需设为 `true`，新版万相2.7 已弃用该参数，改由 [prompt](../guides/prompt.md) 自然描述分镜。 + +请求头必须包含： - `Content-Type: application/json` +- `Authorization: Bearer $DASHSCOPE_API_KEY` +- `X-DashScope-Async: enable`（缺失将报错：“current user api does not [support](../guides/support.md) synchronous calls”） ## 使用方式 -1. **地域对齐**：模型、Endpoint URL 与 API Key 必须同属一个地域（如华北2北京、新加坡、美国弗吉尼亚等），跨地域调用必然失败。 -2. **Endpoint 选择**： - - 新业务空间推荐使用专属域名：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`（新加坡），性能与稳定性更优； - - 兼容旧域名：`https://dashscope.aliyuncs.com`（北京）、`https://dashscope-us.aliyuncs.com`（美国）、`https://dashscope-intl.aliyuncs.com`（国际）。 -3. **异步流程**： - - **步骤1（创建任务）**：`POST /api/v1/services/aigc/.../video-synthesis`，获取 `task_id`； - - **步骤2（轮询结果）**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`（或对应地域专属域名），直到 `status` 为 `"SUCCESS"`，响应中 `output.video_url` 即为生成视频地址。 -4. **SDK 支持**：DashScope SDK 已封装异步轮询逻辑，推荐开发者优先使用（[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)）。 +所有视频生成任务均遵循两步异步流程： + +1. **创建任务**：发送 `POST` 请求至对应 Endpoint，获取 `task_id`。Endpoint 因地域与模型类型略有差异： + - 华北2（北京）推荐使用业务空间专属域名：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis`； + - 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/...`； + - 美国/德国：`https://dashscope-us.aliyuncs.com/...` 或 `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/...`； + - 通用兼容地址（仍可用）：`https://dashscope.aliyuncs.com/...`（北京）、`https://dashscope-intl.aliyuncs.com/...`（国际）。 + +2. **轮询结果**：使用 `GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` 查询状态，直至 `status` 为 `"SUCCESS"`，响应中 `output.video_url` 即为生成视频地址。 + +> **注意**：`task_id` 有效期为 24 小时，超时后无法查询；请勿重复提交相同请求，否则可能触发限流或返回冗余任务。 ## 限制和注意事项 -- **地域隔离**：华北2（北京）与新加坡地域的 API Key、Endpoint、模型实例完全独立，不可混用；美国、德国等区域暂不支持业务空间专属域名。 -- **任务并发**：多数模型限流为 **1 个同时处理中任务**（如 `emo-v1`、`videoretalk`），排队任务需等待前序完成。 -- **输入规范**： - - 数字人类模型（`s2v`、`emo`、`liveportrait`）要求输入图片为正面清晰肖像，需先调用对应 `detect` 模型校验； - - 视频编辑/重绘类模型对输入视频分辨率、时长有隐式要求（如 `video-style-transform` 推荐 540P–720P，≤30秒）。 -- **过期模型**：`wan2.6` 及更早系列（如 `wan2.2`、`wanx2.1`）已标记为“推荐优先选用 wan2.7”，其文档明确提示为遗留接口（[万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md)），新项目应避免接入。 -- **错误处理**：缺失 `X-DashScope-Async` 请求头将返回 `current user api does not support synchronous calls`；`task_id` 超过 24 小时有效期查询将返回 `UNKNOWN` 状态。 +- **地域强绑定**：模型开通地域、API Key 所属地域、Endpoint 地域三者必须完全一致。例如，使用北京地域 API Key 调用新加坡模型将鉴权失败 [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md)。 +- **媒体资源要求**：图片需为公网可访问 URL（HTTPS），且尺寸建议 ≥ 512×512；视频时长建议 ≤ 10 秒；音频需为清晰人声 MP3/WAV。 +- **输入长度限制**：`prompt` 最长 5000 字符（Vidu），其他模型未明确说明但建议 ≤ 1000 字；过长内容将被自动截断。 +- **并发与限流**：多数模型 QPS 限制为 1–5，同时处理中任务数上限为 1–100（详见各模型资费文档）。例如 `liveportrait` 和 `videoretalk` 同一时刻仅允许 1 个运行中任务。 +- **废弃模型提醒**：万相2.1–2.6 系列（文档26–30）已标记为“旧版协议”，官方明确推荐迁移到万相2.7系列；爱诗（PixVerse）与可灵（Kling）为新上线主力模型，功能更全、性能更优。 ## 来源文档 - [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) -- [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) +- [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) -- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) +- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - [万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) - [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) -- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) -- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) -- [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) -- [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) +- [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) +- [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) +- [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) +- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) -- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) -- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) +- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) -- [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) +- [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) +- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md new file mode 100644 index 00000000..99fb9faa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md @@ -0,0 +1,69 @@ +# 应用开发框架对比：Managed Agents、Application Component 与 Toolkits + +## 对比目的与背景 + +在百炼平台构建 AI 应用时，开发者面临多种技术路径选择：是直接调用模型能力快速验证想法？还是构建可复用、可审计的智能体系统？抑或集成知识库与数据连接能力打造企业级应用？`Managed Agents`、`Application Component` 和 `Toolkits` 是百炼平台面向不同抽象层级提供的三类核心开发框架，分别聚焦于**智能体生命周期管理**、**数据与知识基础设施编排**、以及**标准化模型能力接入**。本页旨在从技术定位、能力边界、使用约束和工程实践角度进行客观对比，帮助开发者基于业务目标、团队能力与运维要求做出理性选型决策。 + +--- + +## 关键维度对比表 + +| 维度 | Managed Agents | Application Component | Toolkits（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)） | +|------|----------------|------------------------|------------------------------| +| **核心定位** | 托管式智能体运行时：统一管理会话、沙箱、工具链与事件流 | 数据与知识基础设施 API：聚焦数据连接、知识库构建、解析配置与元数据管理 | 标准化模型能力网关：提供 OpenAI 兼容协议，屏蔽底层模型差异，支持快速迁移与多模态调用 | +| **输入格式** | 结构化 JSON Event 消息数组（含 `role`/`type`/`content`），严格遵循 Session Schema；支持文本、图像 URL、文件引用（需预上传并审核通过） | 多样化资源操作请求：
• 文件：`AddFile` + `Parser` 指定解析器
• 知识库：`CreateIndex` + `SubmitIndexJob`
• 类目/连接器：ROA 风格参数（如 `CategoryId`, `IndexId`, `FileType`） | OpenAI 标准 REST 请求体：
• Chat：`messages: [{role, content}]`
• Vision：`messages` + `image_url` 或 `base64_image`
• Embedding：`input: string/array`
• Files：`file` + `purpose`（必填） | +| **输出格式** | SSE 流式事件（`message`, `session_status`, `tool_call` 等）或轮询历史事件；响应结构含 `event_id`, `session_id`, `output`（含 `text`, `tool_calls`, `files` 等） | ROA 风格 JSON 响应：
• 创建类：返回 `ResourceId`（如 `FileId`, `IndexId`）
• 查询类：返回完整资源对象（含状态、统计、元数据）
• 检索类（`Retrieve`）：返回 `chunks` 数组及 `score` | OpenAI 兼容 JSON：
• Chat/Responses：`choices[0].message.content` / `output_text`
• Embedding：`data[0].embedding`
• Vision：`choices[0].message.content` 或 `data[0].url`（QVQ 流式）
• Batch：异步 `id` + 回调通知 | +| **支持模型** | 仅百炼托管大模型（当前限 `qwen-plus` 等），模型 ID 必须为对象 `{"id": "qwen-plus"}`；不支持自定义/外部模型 | **不直接调用模型**；为模型调用提供数据支撑（如知识库检索结果作为 RAG 上下文） | 广泛支持：`qwen-plus`, `qwen3-*`, `Qwen-VL`, `QVQ`, `Qwen-OCR`, `text-embedding-v*`, `qwen-coder-turbo` 等；按接口能力隔离（如 Completions 仅支持 `qwen-coder-turbo`） | +| **API 端点** | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`（如 `ws_abc.cn-beijing.maas.aliyuncs.com`） | `bailian.{region}.aliyuncs.com`（公网）或 `bailian-vpc.{region}.aliyuncs.com`（VPC）；需显式指定 `WorkspaceId` 路径参数 | 多域名策略：
• Chat/Vision/Embedding：`https://{WorkspaceId}.{region}.maas.aliyuncs.com/compatible-mode/v1`
• Files/Batch（文件）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（中国内地）
• **Batch Chat（专用）：`https://batch.dashscope.aliyuncs.com/compatible-mode/v1`** | +| **认证方式** | Bearer [Token](../concepts/token.md)（`Authorization: Bearer `），API Key 与工作空间强绑定 | RAM AccessKey 签名（ROA），需最小权限策略（如 `AliyunBailianDataFullAccess`） | Bearer [Token](../concepts/token.md)（`Authorization: Bearer `），**API Key 必须与 endpoint 地域一致**（北京 Key 不能用于新加坡 endpoint） | +| **计费方式** | 按 Session 运行时长（秒）+ 工具调用次数 + 文件存储（GB/月）计费；沙箱资源消耗计入 Session 成本 | 按调用次数（QPS）+ 知识库存储（GB/月）+ 文件解析/切片处理量计费；无模型推理费用（仅为数据层） | 按模型调用 token 数（输入+输出）计费；Batch 享 5 折优惠；Files 上传免费，用途相关（如 `fine-tune` 有额外费用） | +| **典型场景** | • 可复用客服智能体（带审批流、多工具协同）
• 合规审计型业务助手（完整事件溯源、版本快照）
• 需沙箱隔离的代码执行/文件分析任务 | • 构建企业级知识库（PDF/Word/Excel 自动入库+切片）
• 管理多源数据连接（OSS、数据库类目）
• 定制化文档解析策略（如合同关键字段提取） | • 快速迁移 OpenAI 应用（零代码修改）
• 多模态应用（图文理解、OCR、向量化）
• 批量推理任务（日志分析、报告生成） | + +--- + +## 各方案适用场景建议 + +### ✅ 优先选用 **Managed Agents** +- 需要**端到端智能体生命周期管理**：如会话状态机（`idle`→`running`→`terminated`）、中断恢复、工具调用审批、事件流订阅。 +- 要求**强安全与合规控制**：沙箱环境隔离、Skill ZIP 包安全扫描、文件审核机制、版本锁定（禁止 `latest`）。 +- 构建**可复用、可归档、可审计**的智能体资产：Agent 版本化、Environment 复用、Session 快照追溯。 +- 场景示例：金融理财顾问（需风控工具调用+会话留痕）、HR 招聘助手（简历解析+面试问答+合规提示）。 + +### ✅ 优先选用 **Application Component** +- 核心需求是**数据与知识基础设施建设**：如将数百份 PDF 合同自动构建知识库、对接内部 OSS 存储、定制化解析规则。 +- 需要**精细化管理知识库元数据**：如按部门/项目分类管理 Index、动态追加文档、删除错误切片、监控索引质量。 +- 业务逻辑依赖**结构化数据连接**：如将 CRM 表格数据作为 RAG 上下文源（注意：`AddTable` 为受限功能，推荐控制台操作）。 +- 场景示例：法务合同审查系统（OSS 批量导入+法律条款切片+高精度检索）、产品文档中心（多格式解析+版本化更新）。 + +### ✅ 优先选用 **Toolkits（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)）** +- 追求**开发效率与生态兼容性**：已有 OpenAI SDK 代码，希望最小改动接入百炼模型。 +- 需要**灵活组合多模态能力**：如同时调用 `Chat`（对话）、`Vision`（图片理解）、`Embedding`（向量检索）构建多阶段流水线。 +- 执行**大规模批量任务**：如每日 10 万条用户反馈情感分析（Batch Chat 5 折）、千万级文本向量化（Embedding Batch）。 +- 场景示例：SaaS 客户支持[插件](../concepts/plugin.md)（OpenAI SDK 直接切换 endpoint）、电商商品图搜系统（Qwen-VL + Embedding）、AI 写作助手（qwen3.7-plus + qwen-coder-turbo 协同）。 + +--- + +## 技术选型参考指南（面向开发者） + +| 选型考量因素 | 推荐方案 | 说明 | +|--------------|----------|------| +| **是否需要模型推理以外的能力（如知识库、文件解析）？** | → 若需：**Application Component**
→ 若仅需模型调用：继续评估 | Application Component 是数据层基石，Toolkits/Managed Agents 均可消费其产出（如知识库检索结果传入 Agent 的 `input`）。 | +| **是否必须保证会话状态一致性与工具链可审计？** | → 是：**Managed Agents**
→ 否：考虑 Toolkits | Managed Agents 提供唯一具备完整状态机与事件溯源的框架；Toolkits 的 `Conversations` 仅提供轻量会话 ID 管理，无状态持久化保障。 | +| **团队是否已使用 OpenAI 生态（SDK、Prompt 工程规范）？** | → 是：**Toolkits**（首选）
→ 否：评估学习成本 | Toolkits 最小化迁移成本；Managed Agents 需理解 Agent/Environment/Session 三层抽象；Application Component 需熟悉 ROA 签名与资源生命周期。 | +| **是否涉及敏感数据处理（如客户隐私文件）？** | → 是：**Managed Agents**（沙箱隔离）或 **Application Component**（私有 VPC endpoint）
→ 否：均可 | Managed Agents 的 `cloud` 沙箱提供进程级隔离；Application Component 支持 VPC endpoint 避免公网传输；Toolkits 默认走公网（需确认合规策略）。 | +| **是否需要未来扩展自定义模型或外部服务集成？** | → 是：**Toolkits**（开放模型列表）或 **Application Component + 自研服务**
→ 否：Managed Agents 更省心 | Managed Agents 当前**不支持自定义模型**；Toolkits 持续扩展模型支持；Application Component 可通过连接器对接自研服务。 | +| **运维复杂度要求（CI/CD、监控、告警）？** | → 低：**Toolkits**（标准 HTTP 接口）
→ 中：**Application Component**（需管理 Index/Category 状态）
→ 高：**Managed Agents**（需监控 Session 状态机、Skill 安全扫描、文件审核） | Toolkits 接口行为最接近传统 REST；Managed Agents 引入更多异步状态（如 `session_status` 变更），需适配 SSE 或轮询。 | + +> **联合使用建议**：实际生产中三者常组合使用—— +> **Application Component** 构建知识库 → 输出 `IndexId` 与检索结果； +> **Toolkits**（`Embedding` + `Retrieve`）实现快速原型验证； +> **Managed Agents** 封装最终交付形态，将知识库检索结果、工具调用、用户会话统一编排为可发布智能体。 +> 此分层架构兼顾敏捷性、可维护性与企业级治理要求。 + +## 被对比主题页 + +- [managed agents api](../api/managed-agents-api.md) +- [application component api reference](../api/application-component-api-reference.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md deleted file mode 100644 index e7ec2ed1..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration.md +++ /dev/null @@ -1,81 +0,0 @@ -# 应用编排能力对比：托管智能体、应用组件与模型上下文协议 - -## 背景与目的 -在百炼平台构建复杂 AI 应用时，开发者需在不同抽象层级间进行技术选型：是直接调度原子能力（如知识检索、文件解析），还是封装为可复用的智能体实例？是通过标准化协议接入外部工具，还是在底层数据与模型之间建立结构化桥梁？本对比聚焦三大核心编排能力——**托管智能体（Managed Agents）**、**应用组件（Application Components）** 和 **模型上下文协议（Model Context Protocol, MCP）**，旨在帮助开发者清晰理解其定位差异、能力边界与协同关系，避免方案错配（例如用 Application Component API 实现会话状态管理，或用 MCP 直接替代知识库构建），从而做出高效、可维护、可扩展的技术决策。 - ---- - -## 关键维度对比 - -| 维度 | 托管智能体（Managed Agents） | 应用组件（Application Components） | 模型上下文协议（MCP） | -|------|------------------------------|-------------------------------------|------------------------| -| **核心定位** | **面向会话的智能体运行时服务**：提供带状态、带沙箱、事件驱动的 Agent 全生命周期托管能力 | **面向数据与知识的基础设施层**：提供知识库构建、文件/类目管理、Prompt 模板等基础能力的 OpenAPI 接口集合 | **面向工具调用的标准化协议层**：定义大模型与外部工具（搜索、地图、图表等）安全交互的通用通信契约，不处理模型推理本身 | -| **输入格式** | OpenAI-style message 数组（`[{"role":"user","content":[{"type":"text","text":"..."}]}]`），支持文本、图片、文件引用（需先上传并审核） | 多样化结构化输入：
• 文件上传：二进制流 + `AddFile` 元数据
• 知识库构建：`CreateIndex` + `SubmitIndexJob` JSON 配置
• Prompt 模板：含 `${variable}` 占位符的字符串 | 工具调用请求（Tool Call）JSON：
• 智能体场景：由模型自动生成，含 `tool.name` 与 `tool.input`
• 工作流场景：人工配置，支持变量引用（如 `上游节点.output`）
• 外部调用：符合 MCP Streamable HTTP 规范的 POST 请求体 | -| **输出格式** | SSE 流式事件（`message`, `tool_call`, `tool_result`, `session_status`），含完整会话状态变迁与中间结果 | 同步 REST 响应：
• 成功：标准 JSON（如 `{"IndexId": "idx_xxx", "Status": "CREATING"}`）
• 列表接口：分页结构（`NextToken`, `Items[]`）
• 检索结果：`Retrieve` 返回 `Chunks[]` 及相关性分数 | 工具执行结果（Tool Result）JSON：
• 格式由工具 `outputSchema` 定义（如天气服务返回 `{"temperature": 25.3, "condition": "sunny"}`）
• 支持流式响应（`streamableHttp` 协议下）或同步返回 | -| **支持模型** | **仅限百炼托管模型**（如 `qwen-plus`, `qwen-max`, `qwen3`），不支持自定义/外部模型接入；模型 ID 必须在创建 Agent 时显式指定且不可变更 | **不直接涉及模型调用**；为模型提供上下文支撑（如 RAG 检索结果、结构化 Prompt），可被任意百炼模型（包括千问系列、第三方模型）在应用层消费 | **不绑定模型**；作为工具通道服务于智能体/工作流中的任意百炼托管模型（推荐 Qwen-Max/Qwen3 提升工具调用准确率），**不可用于原始千问 API 调用** | -| **API 端点** | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`（REST + SSE） | `https://bailian.{region}.aliyuncs.com`（ROA 风格，如 `/bailian/2023-12-29/indexes`） | 无统一端点：
• 官方服务：`https://dashscope.aliyuncs.com/api/v1/mcps/{service}/{tool}`
• 自定义服务：由用户部署地址决定（如 FC 函数 URL） | -| **计费方式** | **按会话（Session）计费**：
• 基础费用：Agent 运行时长（秒） × 单价
• 附加费用：沙箱资源（CPU/GPU）、文件存储（≤20 MB/文件，30 天保留期）、Skill 执行 | **按操作与资源计费**：
• 文件上传/解析：按文件数与大小计费
• 知识库：按索引容量（GB）、构建时长（CU）、检索调用次数（QPS）计费
• Prompt 模板：免费 | **按工具调用计费**：
• 官方云服务（如 WebSearch）：29 元/千次 + QPS 限制
• 自定义服务：按调用时长（秒）计费（0.000156 元/秒），分“基础模式”与“极速模式” | -| **典型场景** | • 多轮对话客服机器人（需记忆用户偏好、调用订单系统）
• 自动化数据分析助手（上传 Excel → 解析 → 生成图表 → 解释结论）
• 内部 IT 支持 Agent（集成 Jira、Confluence 工具链） | • 构建企业专属知识库（PDF/Word/网页入库 + 分片优化）
• 管理客户资料类目体系与附件文件
• 统一维护销售话术、产品 FAQ 等 Prompt 模板 | • 智能体中自动触发天气查询、路径规划、联网搜索
• 工作流中串联“网页爬取 → 文本摘要 → PPT 生成”工具链
• 第三方应用（如 Cherry Studio）接入百炼工具生态 | - ---- - -## 适用场景建议 - -### ✅ 选择 **托管智能体** 当： -- 你需要一个**有状态、可中断、可审计的会话级执行单元**； -- 业务逻辑涉及**多步骤、条件分支、工具审批（如人工确认支付）**； -- 必须保障**沙箱隔离性**（如运行用户上传的 Python 脚本、调用敏感内部 API）； -- 团队希望**复用已验证的 Skill 包**（经安全扫描的 zip 工具组合）； -- 对**事件流实时性要求高**（如实时推送工具执行进度、用户消息确认）。 - -> ⚠️ 注意：若仅需单次问答（无状态）、或模型固定无需版本控制、或工具链简单无沙箱需求，则过度使用托管智能体将增加运维复杂度。 - -### ✅ 选择 **应用组件** 当： -- 你的核心诉求是**构建和管理知识资产**（RAG 知识库、文档中心、FAQ 库）； -- 需要**批量导入/更新/删除文件与类目**，并精细控制解析策略（如 PDF 用 DocMind、图片用 Qwen-VL）； -- 希望**标准化提示词工程**，实现跨应用复用与 A/B 测试（如不同销售话术模板）； -- 数据源来自内部系统（如 CRM 导出 CSV），需通过 API 自动化同步至百炼。 - -> ⚠️ 注意：应用组件不提供模型推理、会话管理或工具调用能力；它本质是“燃料供给系统”，需与智能体或工作流配合使用。 - -### ✅ 选择 **模型上下文协议（MCP）** 当： -- 你希望**解耦模型与工具**，让同一套智能体配置可灵活切换不同地图/搜索服务商； -- 需要**快速接入多个异构工具**（如同时用高德地图 + WebSearch + QuickChart），避免为每个工具单独开发适配器； -- 在**工作流中精确控制工具调用顺序与参数传递**（如将 OCR 结果作为搜索关键词）； -- 计划将百炼工具能力**嵌入第三方 IDE 或低代码平台**（通过 MCP SDK 标准化对接）。 - -> ⚠️ 注意：MCP 不解决知识库构建、文件管理、Prompt 版本控制等问题；它专注“调用什么工具”和“如何传参”，而非“从哪获取数据”或“如何组织会话”。 - ---- - -## 技术选型参考（面向开发者） - -| 你的问题 | 推荐方案 | 关键理由 | -|----------|----------|----------| -| “我需要一个能记住用户上句话、调用数据库查订单、再生成总结的聊天机器人” | ✅ **托管智能体** | 唯一支持会话状态机（`idle`→`running`→`idle`）、SSE 事件流、沙箱环境三者结合的方案 | -| “我要把公司 500 份产品手册 PDF 自动转成向量知识库，并支持按章节检索” | ✅ **应用组件** | `AddFile` + `CreateIndex` + `SubmitIndexJob` 是知识库构建的标准 API 流程，支持分片、监控与权限隔离 | -| “我的智能体有时需要查天气，有时需要搜新闻，能否不改代码就切换服务商？” | ✅ **MCP** | 通过控制台更换 MCP 服务绑定即可，模型调用逻辑（`tool.name`）完全不变，真正实现工具解耦 | -| “我想在 Python 脚本里批量上传 1000 个文件到百炼，并分类打标” | ✅ **应用组件** | `ApplyFileUploadLease` + `AddFile` + `AddCategory` 提供稳定、幂等的批量文件管理能力 | -| “我有一个自研的股票分析 API，想让百炼模型能调用它” | ✅ **MCP（自定义服务）** | 用 `streamableHttp` 类型接入，无需修改模型代码；KMS 加密密钥保障安全性；支持流式返回实时行情 | -| “我只需要调用一次千问 API 回答问题，不需要状态、不调用工具” | ❌ 三者均不适用 | 应直接使用 [DashScope ChatCompletion API](https://help.aliyun.com/zh/dashscope/developer-reference/quick-start) | - -### 🧩 协同使用最佳实践 -- **典型组合**：`应用组件`（构建知识库） → `MCP`（接入搜索工具） → `托管智能体`（封装为可对话的 Agent） -- **示例流程**： - 1. 用 Application Component API 将产品文档入库（`AddFile` → `SubmitIndexJob`）； - 2. 用 MCP 配置 WebSearch 服务，补充实时信息； - 3. 创建 Managed Agent，挂载该知识库（通过 RAG 插件）+ MCP 服务，设定系统提示词； - 4. 用户提问时，Agent 自动融合知识库检索结果与 WebSearch 结果生成回答。 - -> 💡 **关键提醒**：三者非互斥关系，而是分层协作——应用组件提供“数据燃料”，MCP 提供“工具插件”，托管智能体提供“运行引擎”。合理分层可显著提升系统可维护性与迭代效率。 - ---- -*最后更新：2025年4月* - -## 被对比主题页 - -- [managed agents api](../api/managed-agents-api.md) -- [application component api reference](../api/application-component-api-reference.md) -- [model context protocol](../guides/model-context-protocol.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md deleted file mode 100644 index 8b4f6c1a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation.md +++ /dev/null @@ -1,70 +0,0 @@ -# 图像、视频与3D生成能力对比 - -为帮助开发者快速理解百炼平台在多模态生成领域的技术边界与工程适配特性，本文系统对比图像生成（Image Generation）、视频生成（Video Generation）与3D生成（3D Generation）三大核心能力。对比聚焦于实际开发中高频关注的技术维度——包括调用模式、模型生态、输入输出约束、地域与计费策略等，旨在支撑产品规划、架构设计与模型选型决策。所有结论均基于当前（2024年Q3）百炼平台正式发布的API文档与运行时行为。 - -## 关键能力维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| **输入格式** | 文本（`prompt`）、图像URL（`image_url`）、局部掩码（`mask`）、风格参考图（`style_ref_url`）、涂鸦（`sketch`）等多模态组合；支持批量输入（如海报多文案） | 文本（T2V）、单图/首尾帧/参考图（I2V/KF2V/R2V）、视频片段（编辑类）、数字人肖像图（S2V/Emo）；`input` 结构统一为 `{"media": [...], "prompt": "..."}` 或纯 `{"prompt": ...}` | 文本（文生3D）、单张图像URL（单图生3D）、4张有序视角图数组（多图生3D，空位用 `{}` 占位）；三者互斥，不可混合 | -| **输出格式** | JPEG/PNG 图像（URL），支持水印控制；部分模型返回优化提示词（`prompt_extend`）、分割掩码（`instance_mask`）等辅助数据 | MP4 视频（URL），含可选音频轨道；部分模型返回分镜渲染图、关键帧序列；数字人模型额外输出 `.fbx` 或 `.glb` 动作文件 | GLB 模型文件（`pbr_model_url` 带PBR材质 / `base_model_url` 无贴图），及配套预览图 `rendered_image_url`；所有URL有效期仅2小时 | -| **支持模型（代表）** | `qwen-image-2.0-pro`（文字精度）、`wan2.7-image-pro`（4K高清）、`kling/kling-v3-*`（强风格）、`virtualmodel-v2`（电商虚拟模特）、`facechain-portrait-generation`（人物写真） | `wan2.7-t2v/i2v/r2v`（全链路新协议）、`kling/kling-v3-video-generation`（高动态）、`vidu/viduq3-*`（快节奏）、`emo-v1`（悦动人像）、`liveportrait`（灵动人像） | `Tripo/Tripo-H3.1`（高精度，≤200万面）、`Tripo/Tripo-P1.0`（快速，≤2万面）；仅Tripo官方模型，无第三方接入 | -| **API端点（标准路径）** | `/api/v1/services/aigc/image-generation/generation`（同步）
`/api/v1/services/aigc/image-generation/generation_async`（异步） | `/api/v1/services/aigc/video-generation/video-synthesis`（新模型统一路径）
`/api/v1/services/aigc/image2video/video-synthesis`（`wan2.6`及更早旧路径，已弃用） | `/api/v1/services/aigc/video-generation/3d-generation`（注意：路径含`video-generation`但属3D服务，为历史兼容命名） | -| **调用模式** | **混合模式**：
• 同步：`z-image-turbo`、`qwen-image-*`、`wan2.6-t2i` 等（响应 <15s）
• 异步：虚拟模特、背景生成、局部重绘等（需轮询 `task_id`） | **强制异步**：
所有模型均需 `X-DashScope-Async: enable`，创建任务后轮询 `GET /api/v1/tasks/{id}`；任务有效期24小时 | **强制异步**：
必须携带 `X-DashScope-Async: enable`；创建任务后轮询；`task_id` 有效期24小时，结果URL有效期仅2小时 | -| **计费方式** | 按**成功生成的图片张数**计费（非请求次数）；免费额度500张/90天（主账号+RAM共享）；单价因模型而异（例：`wanx-v1`: 0.16元/张，`wanx-style-repaint-v1`: 0.12元/张） | 按**成功生成的视频条数**计费；无公开免费额度说明，需按用量购买资源包或开通后付费；单价未在文档中统一公示，以控制台实时报价为准 | 按**成功生成的3D模型个数**计费；无免费额度；单价依模型版本区分（`H3.1` > `P1.0`），具体见控制台定价页；失败任务不扣费 | -| **典型场景** | • 电商素材生成（商品图、海报、模特图）
• 设计辅助（背景替换、风格迁移、AI试衣）
• 内容创作（插画、头像、锦书文字艺术）
• 工业应用（缺陷标注补全、图纸增强） | • 营销短视频（文生广告片、产品演示）
• 数字人播报（新闻、客服、培训）
• 影视预演（分镜动画、角色动作测试）
• 社交内容（表情包、GIF动图、AI舞蹈） | • 工业设计（概念建模、零部件快速原型）
• 游戏开发（低多边形资产生成）
• AR/VR内容生产（可交互3D对象）
• 电商3D展示（商品360°视图基础模型） | -| **地域可用性** | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；各区域API Key与Endpoint独立 | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；跨地域调用必然失败 | **仅华北2（北京）地域可用**；其他地域调用返回403或模型不可见错误 | -| **SDK支持** | DashScope SDK（Python/Java）完整封装同步/异步调用、自动重试、凭证管理 | DashScope SDK 封装异步轮询逻辑，推荐使用；避免手动实现长轮询 | DashScope SDK 支持，但需显式指定北京地域Endpoint；无专用3D模块，复用通用异步任务接口 | - -## 各方案适用场景建议 - -### ✅ 图像生成 —— 适合「高频、轻量、多样化」视觉内容生产 -- **首选场景**：需要快速产出大量静态图像的业务，如电商平台每日上新图生成、营销海报A/B测试、设计团队灵感草稿、个性化头像/证件照批量处理。 -- **技术优势**：同步调用降低延迟（<1s响应），支持精细控制（分辨率、水印、风格索引），垂直模型丰富（虚拟模特、鞋靴、海报专用）。 -- **规避风险**：避免用图像API生成含复杂运动/时间逻辑的内容（如“挥手动作”），此类需求应转向视频生成。 - -### ✅ 视频生成 —— 适合「动态表达、人机交互、时间序列」内容构建 -- **首选场景**：数字人驱动（企业IP形象播报）、短视频自动化生产（图文转视频）、影视工业预演（分镜动画）、社交互动内容（AI跳舞、口型同步）。 -- **技术优势**：原生支持多镜头描述（`wan2.7` [prompt](../guides/prompt.md)内时间戳）、首尾帧控制、参考图动作迁移；数字人模型提供人脸/语音/动作联合生成能力。 -- **规避风险**：勿用于生成超长视频（当前最大8秒）或高精度物理仿真（如流体、布料动力学）；3D空间一致性弱于专用3D生成。 - -### ✅ 3D生成 —— 适合「几何结构明确、需下游渲染/交互」的三维资产创建 -- **首选场景**：工业设计快速建模（如家具、小家电概念验证）、游戏美术管线中的基础网格生成、AR应用中轻量化3D商品模型、教育可视化教具制作。 -- **技术优势**：输出标准GLB格式（含PBR材质），可直接导入Unity/Unreal/Three.js；支持多视角输入提升几何准确性；`H3.1`版本达200万面，满足中等复杂度建模。 -- **规避风险**：不适用于生成无明确几何结构的抽象艺术（如“一团流动的光”）；不支持纹理编辑、UV展开等后期操作；仅北京地域可用，需提前规划部署架构。 - -## 面向开发者的选型参考指南 - -1. **评估输入复杂度** - - 若输入仅为文本或单图 → 优先评估图像生成（成本低、速度快）； - - 若需表达时间变化（动作、过渡、节奏）→ 必选视频生成； - - 若目标为可旋转、可光照、可碰撞的三维实体 → 唯一选择3D生成。 - -2. **检查地域与基础设施约束** - - 若业务已部署于新加坡或美国 → **排除3D生成**，并确认视频/图像模型在对应地域的可用性（部分垂直模型仅限北京）； - - 若需混合调用（如先图生图再图生视频）→ 确保所有服务使用**同一地域API Key与Endpoint**，避免跨域认证失败。 - -3. **权衡成本与质量要求** - - 追求极致性价比（千张级/日）→ 图像生成（有免费额度+单价透明）； - - 接受中等成本换取动态表现力 → 视频生成（按条计费，单条成本高于单图）； - - 愿为专业3D资产支付溢价 → 3D生成（`H3.1`单价显著高于`P1.0`，但面数与材质质量跃升）。 - -4. **验证端到端工作流可行性** - - 图像生成：检查是否需后续处理（如抠图→合成→视频），若链路过长，考虑直接使用视频生成的I2V； - - 视频生成：确认输入图是否满足数字人模型的正面肖像要求（需先调用`detect`接口校验）； - - 3D生成：验证输入图是否符合视角顺序（前/左/后/右），多图生3D对拍摄规范性要求高，建议先用单图模式快速验证。 - -5. **上线前必做事项** - - 所有能力均需配置 `X-DashScope-Async: enable` 请求头（视频/3D强制，图像部分模型强制）； - - 生产环境务必使用**业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），避免旧域名限流与延迟问题； - - 对异步任务实现健壮轮询（带指数退避、超时熔断、`task_id`有效期校验），禁止无限循环轮询。 - -> **最后提醒**：模型能力持续迭代，`wan2.7+`、`kling-v3`、`Tripo-H3.1` 等新版本已逐步替代旧模型（如 `wan2.2`、`wanx2.1`）。新项目开发请严格参照各能力文档顶部的「最新版API参考」链接，避免依赖已标记为“遗留接口”的旧路径与参数。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md new file mode 100644 index 00000000..f89e5b44 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md @@ -0,0 +1,65 @@ +# 图像生成与视频生成对比 + +为帮助开发者快速理解百炼平台中图像生成与视频生成两类能力的差异，明确技术选型边界与使用约束，本文从输入输出、模型支持、调用方式、计费策略等核心维度进行系统性对比。该对比基于当前（2024年Q3）平台正式发布的 API 能力与文档规范，适用于新项目接入、存量系统迁移及多模态方案架构设计。 + +| 维度 | 图像生成（Image Generation） | 视频生成（Video Generation） | +|------|------------------------------|------------------------------| +| **输入格式** | • 文生图：`input.prompt` 或 `input.messages`（含 text）
• 图生图/编辑：`input.messages` 数组（含 `{"text": "..."}` 和 `{"image": "url"}`），部分旧模型仍用 `input.ref_image`
• 图片 URL 需公网可访问、无中文路径、支持 HTTPS | • 文生视频（T2V）：`input.prompt` + 可选 `parameters.multi_shot`/分镜描述
• 图生视频（I2V）：`input.media`（首帧图 URL）或 `input.video_url`（短片）
• 参考生视频（R2V）：`input.media` 数组（多张参考图）
• 所有媒体 URL 必须 HTTPS、≥512×512（图）、≤10秒（视频） | +| **输出格式** | • 同步调用：直接返回 JSON，含 `output.results` 数组（每项含 `url`、`width`、`height`）
• 异步调用：轮询 `GET /api/v1/tasks/{task_id}`，响应含 `output.results`（单图或多图） | • 全部异步：轮询 `GET /api/v1/tasks/{task_id}`，响应含 `output.video_url`（H.264 MP4）、`output.duration`（秒）、`output.resolution`（如 `"1080P"`）
• 不返回帧序列或中间产物，仅最终视频文件 | +| **支持模型（主力）** | • 文生图：`qwen-image-2.0-pro`、`wan2.7-image-pro`、`z-image-turbo`、`vidu`
• 图像编辑：`qwen-image-edit-*`、`wanx-x-painting`、`virtualmodel-v2`
• 创意工具：`image-out-painting`、`wanx-background-generation-v2`、`aitryon-plus` | • 文生视频：`wan2.7-t2v-*`、`vidu/viduq3-*-text2video`、`kling/kling-v3-*-video-generation`、`pixverse/pixverse-*-t2v`
• 图/参考生视频：`wan2.7-i2v-*`、`vidu/viduq3-*-img2video`、`wan2.7-r2v-*`
• 数字人：`liveportrait`、`videoretalk`、`emo-v1` | +| **API 端点** | • **统一主入口**：
`POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation`（推荐）
• 历史路径（部分模型）：
`POST /api/v1/services/aigc/text2image/image-synthesis` 等 | • **全量异步专用入口**：
`POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis`
• 通用兼容地址（不推荐）：
`https://dashscope.aliyuncs.com/...`（北京）或 `https://dashscope-intl.aliyuncs.com/...`（国际） | +| **调用模式** | • **同步 & 异步混合**：
– 快速模型（如 `qwen-image-2.0-pro`、`wan2.6-t2i`）支持同步（<10s），可流式返回（需 `X-DashScope-Sse: enable`）
– 长耗时模型（如 `wanx-x-painting`、`image-out-painting`）强制异步 | • **强制异步**：
所有模型均需两步流程：① 创建任务获取 `task_id`；② 轮询 `GET /api/v1/tasks/{task_id}` 获取结果
• `task_id` 有效期严格为 **24 小时** | +| **计费方式** | • 免费额度：**500 张/账号/90天**（主账号与 RAM 子账号共享）
• 计费模型：
– 按成功生成图片计费（如 `wanx-v1`: 0.16元/张）
– 部分模型免费额度用尽即停用（无单价），如 `wanx-x-painting`、`shoemodel-v1`、`image-instance-segmentation` | • 免费额度：**100 秒视频生成时长/账号/90天**（按实际生成视频秒数累加）
• 计费模型：
– 按生成视频时长计费（如 `wan2.7-t2v`: 0.8元/秒，`vidu`: 1.2元/秒）
– 数字人模型按任务计费（如 `liveportrait`: 0.5元/次）
• 所有模型均无“免费额度外不可用”例外，超限后自动转计费 | +| **典型场景** | • 静态内容生产：电商主图、营销海报、AI头像、文字艺术（WordArt）、背景生成、商品试穿（鞋靴/服装）
• 图像增强：局部重绘、擦除补全、实例分割、风格迁移
• 快速原型：A/B测试图稿、UI素材生成、设计草图扩展 | • 动态内容生产：短视频广告、产品演示动画、分镜脚本可视化、数字人播报、虚拟主播口型同步
• 视频创作辅助：图转视频（I2V）、多图角色一致性视频（R2V）、自然语言分镜生成（万相2.7）
• 垂直应用：表情包生成（`emoji`）、唱演视频（`emo-v1`）、舞蹈驱动（`animate-anyone-gen2`） | + +## 各方案适用场景建议 + +### ✅ 推荐选择图像生成当： +- 业务需求聚焦于**静态视觉资产**，如电商平台的商品图、社交媒体封面、APP图标、个性化头像； +- 对**响应延迟敏感**（如实时交互式设计工具），且任务平均耗时 < 8 秒，可优先选用 `qwen-image-2.0-pro` 或 `wan2.7-image-pro` 同步接口； +- 需要**精细控制像素级输出**（如 4K 渲染、文字精准识别、局部编辑掩码），图像模型在空间保真度上显著优于视频模型首帧； +- 成本结构以**固定次数/张数**为主，且月用量稳定在数百张内，可充分复用免费额度。 + +### ✅ 推荐选择视频生成当： +- 核心目标是**动态表达与时间叙事**，如短视频营销、教学动画、数字人直播、AI分镜预演； +- 接受**异步工作流**（任务创建 → 轮询 → 下载），并能妥善管理 `task_id` 生命周期与失败重试逻辑； +- 需要**跨帧一致性能力**（人物/物体/风格在多帧中稳定呈现），R2V 与 I2V 模型专为此优化，图像模型无法替代； +- 业务具备**视频时长可预测性**（如统一生成 5 秒广告），便于成本建模；若需高频、短时（<3秒）视频，需注意部分模型最低时长限制（如 `kling` 最小 3 秒）。 + +### ⚠️ 需谨慎评估或组合使用的场景： +- **“动效化静态图”需求**（如将海报转为带缩放/平移的短视频）： + → 不建议直接调用视频生成，应先用图像生成产出高质量源图，再通过视频编辑模型（如 `video-style-transform` 或 `wan2.2-animate-mix`）添加运镜效果，兼顾质量与成本。 + +- **高并发实时图像/视频混合服务**（如用户上传图→生成图→生成对应视频）： + → 必须分离调用链路：图像生成走同步路径（低延迟），视频生成走异步路径（解耦阻塞）；同时注意地域强绑定——图像与视频模型若部署在不同地域（如图在北京、视频在新加坡），需分别配置 API Key 与 Endpoint。 + +- **需要帧级控制或导出中间帧**（如用于后期合成、AR叠加）： + → 当前两类 API 均**不提供帧序列下载**。若必须获取逐帧，需自行对生成视频做抽帧处理（注意版权与水印合规性），或联系平台申请定制化能力支持。 + +## 面向开发者的选型参考指南 + +1. **起步验证阶段**： + - 优先使用 `qwen-image-2.0-pro`（同步、北京/新加坡可用、免费额度覆盖）验证图像流程； + - 视频侧选用 `wan2.7-t2v-2026-06-12`（支持自然语言分镜、文档完善）+ `task_id` 轮询 SDK 封装，避免手动轮询。 + +2. **生产环境部署要点**： + - **域名与地域必须显式绑定**：禁用 `dashscope.aliyuncs.com` 通用域名，全部切换至业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），提升稳定性与性能； + - **错误处理标准化**：图像 API 需捕获 `429`（限流）、`400`（参数错误）；视频 API 必须处理 `401`（地域不匹配）、`404`（旧版路径错误，如误用 `/image2video/`）； + - **水印策略统一**：生产环境所有请求显式设置 `"watermark": false`，避免默认水印影响交付。 + +3. **模型升级路径建议**： + - 图像侧：逐步淘汰 `wan2.5-i2i-preview` 等旧版，迁移到 `wan2.7-image-pro`（4K）或 `qwen-image-2.0-pro`（文字渲染）； + - 视频侧：**立即停用万相2.1–2.6系列**（文档标记为“旧版协议”），全面切换至 `wan2.7-*` 或 `vidu/kling` 新主力模型，享受分镜解析、多镜头、音频生成等增强能力。 + +4. **成本监控关键指标**： + - 图像：监控 `total_images_generated` 与 `free_quota_remaining`（Dashboard 可查）； + - 视频：监控 `total_video_seconds_generated` 及各模型 `avg_duration_per_task`，警惕因 `duration` 参数设置过高导致意外超支。 + +> **最后提醒**：两类能力虽同属 AIGC，但底层计算范式、资源调度与 SLA 保障机制完全不同。切勿将图像 API 的同步思维套用于视频，亦不可期望视频模型输出单帧图像——尊重各自技术边界，方能构建稳健、可扩展的多模态应用。 + +## 被对比主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md index 87267d43..5ab605c6 100644 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md @@ -1,56 +1,57 @@ # [长期记忆](../concepts/long-term-memory.md)与知识库方案对比 -为帮助开发者在智能体（Agent）与 RAG 应用开发中做出精准技术选型，本文系统对比百炼平台两大核心上下文增强能力：**[长期记忆](../concepts/long-term-memory.md)（Long-Term Memory, 新版）** 与 **知识库（Knowledge Base）**。二者虽均基于向量检索与语义理解，但设计目标、数据来源、生命周期管理及集成范式存在本质差异。本对比聚焦实际工程落地维度，涵盖接口行为、模型依赖、计费逻辑与典型适用场景，旨在提供可操作的选型决策依据。 +为帮助开发者在百炼平台中科学选型，本文系统对比**[长期记忆](../concepts/long-term-memory.md)（新）**（含记忆库能力）与**知识库（RAG）** 两大核心数据增强方案。二者虽均支持语义检索与结构化管理，但设计目标、数据生命周期、适用角色和集成范式存在本质差异：**[长期记忆](../concepts/long-term-memory.md)聚焦“用户专属、动态演进、会话级上下文沉淀”，知识库专注“领域通用、静态注入、应用级知识赋能”**。本对比基于当前（2025年Q2）百炼平台正式版能力，面向智能体（Agent）、工作流及自定义应用的开发者提供技术决策依据。 ## 关键维度对比 -| 维度 | [长期记忆](../concepts/long-term-memory.md)（新） | 知识库 | -|------|----------------|---------| -| **核心定位** | 用户级、会话级**个性化上下文持久化**：捕获并结构化用户偏好、意图、习惯、关系等动态语义信息 | **领域/业务级静态知识注入**：为大模型提供私有、结构化或非结构化的外部事实性知识（文档、表格、音视频等） | -| **输入格式** | • `messages`：多轮对话数组（最多50条），自动提取记忆片段
• `custom_content`：纯文本（≤512字符），绕过提取直接写入
• 支持 `meta_data` 自定义元数据 | • 多源文件：PDF/DOCX/TXT/CSV/JSON/MP3/MP4 等（单文件 ≤150MB）
• 支持 API 批量上传或控制台导入
• 索引时自动切片（≤6000 Token/片）并抽取 `filename`/`date`/`author` 等元数据 | -| **输出格式** | • `SearchMemory` 返回结构化记忆片段列表，含 `id`、`content`、`score`、`meta_data`、`created_at`
• `GetUserProfile` 返回 JSON Schema 定义的结构化画像对象 | • 检索服务：返回带 `score`、`source`（文件名/页码）、`content`、`metadata` 的文本切片数组
• 问答服务：返回生成答案 + 引用溯源（高亮原文位置 + 文件链接） | -| **支持模型** | • **记忆提取**：由平台内置专用模型驱动（不暴露给用户选择）
• **检索排序**：默认向量模型 + 可选 `enable_rerank`（使用平台统一 rerank 模型）
• **不支持自定义模型替换** | • **检索模型**：支持指定向量模型（如 `qwen3-embedding`）
• **重排模型（Rerank）**：纯文本知识库仅支持 `qwen3-rerank`；多模态知识库支持 `qwen-vl-rerank` 等视觉专用模型
• **生成模型**：问答服务可自由绑定任意百炼平台支持的 LLM（Qwen 系列、DeepSeek、Llama3.1 等） | -| **API 端点** | • Base URL：`https://dashscope.aliyuncs.com/api/v2/apps/memory/`
• `POST /add`（新增）
• `POST /memory_nodes/search`（检索）
• `GET /memory_nodes?user_id=xxx`（分页查询）
• `DELETE /memory_nodes/{id}` / `PATCH /memory_nodes/{id}`（删/改） | • Base URL：`https://dashscope.aliyuncs.com/api/v2/knowledgebase/`（华北2地域）
• `POST /retrieval`（独立检索）
• `POST /qa`（问答服务）
• `POST /knowledgebases/{kb_id}/documents`（上传）
• 控制台创建后生成专属服务端点（含鉴权Token） | -| **计费方式** | • **按调用量计费**：
 – `AddMemory` / `SearchMemory` / `ListMemory` 等 API 调用按次计费（具体单价见控制台定价页）
 – 无知识库规格费、无向量模型 Token 费
• **无存储容量费**（记忆片段按账号配额管理） | • **双重计费**：
 – **规格费**：按知识库运行时长（标准版 0.03 元/小时）或 RCU（旗舰版 0.2 元/RCU/小时）
 – **模型费**：向量嵌入（Embedding）与 Rerank 模型按实际 Token 消耗计费（独立于规格费）
 – 问答服务中的 LLM 调用另计费 | -| **典型场景** | • 智能客服：记住用户历史投诉、设备型号、服务偏好
• 个人助手：持续跟踪日程提醒、饮食禁忌、旅行计划
• 教育 Agent：记录学生错题类型、薄弱知识点、学习节奏
• 游戏 NPC：维护玩家角色关系、阵营立场、任务进度 | • 企业知识问答：HR 政策、IT SOP、产品手册即时查询
• 法律/医疗辅助：基于法规条文、临床指南生成专业建议
• 客服工单处理：关联历史工单、解决方案库、产品变更日志
• 投研分析：从财报、研报 PDF 中提取关键财务指标与风险提示 | -| **数据生命周期** | • 默认无自动过期（需业务侧通过 `expire_time` 参数或定时任务清理）
• 控制台支持配置全局有效期（7/30/180天或永不过期）
• `UpdateMemory` 仅更新内容与 `meta_data`，不改变向量索引时间戳 | • 文档上传后即构建索引，无显式过期机制
• 更新知识需重新上传文件或调用 `update_document` API 触发增量索引
• 删除文档后，对应切片从向量库中移除（约1-5分钟生效） | -| **地域与权限** | • 全地域可用（与 DashScope API 一致）
• 仅需 `DASHSCOPE_API_KEY`（Bearer Token 认证） | • **仅限中国站华北2（北京）地域**
• 需子账号具备 `AliyunBailianDataFullAccess` 权限
• SDK 调用需配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` 等 AK/SK 环境变量 | -| **SDK 支持** | • Python：`agentscope-runtime>=1.1.5` 提供 `AddMemory`/`SearchMemory`/`ListMemory`/`DeleteMemory` 异步封装
• `UpdateMemory` 需直调 REST API
• OpenClaw 插件开箱即用（`autoCapture`/`autoRecall`） | • Python/Java SDK 提供完整生命周期管理（创建、上传、索引、检索、问答）
• 控制台生成的问答服务支持一键导出 SDK 调用示例
• 不提供 OpenClaw 原生插件 | +| 维度 | 长期记忆（新） | 知识库（RAG） | +|------|----------------|----------------| +| **核心定位** | 用户级长期上下文管理：自动提取、结构化建模、跨会话个性化召回 | 应用级知识增强：私有文档/多模态数据的语义索引与检索，提升大模型领域回答质量 | +| **输入格式** | • `messages`：结构化对话数组（最多50条，role/content 必填）
• `custom_content`：纯文本（≤512字符）
• 支持 `meta_data`（≤1 KB）作为辅助标签 | • 多模态文件：PDF/DOCX/TXT/CSV/XLSX/图片（≤20MB）、音视频（需转文字）
• 支持元数据抽取规则（创建时配置，不可变）
• 文本切片上限6,000 [Token](../concepts/token.md) | +| **输出格式** | • `SearchMemory` 返回结构化记忆节点列表（含 `id`, `content`, `score`, `meta_data`, `created_at`）
• `GetUserProfile` 返回 JSON Schema 校验的结构化画像对象 | • `Retrieve` 返回文本切片列表（含 `content`, `score`, `source_file_name`, `page_number`, `tags`）
• `Query`（问答服务）返回带引用溯源的自然语言答案 + 切片高亮片段 | +| **支持模型** | **不依赖特定大模型**：底层由平台统一记忆引擎处理；提取质量受输入对话结构影响，**无需开发者指定或切换模型** | • 预置模型：Qwen3/Qwen2.5/Qwen2/Long/Max/Turbo/Coder/Deep-Research/VL系列等
• 第三方模型：DeepSeek-R1/Llama3.1/Yi-Large 等（需模型已接入百炼）
• **必须显式指定模型 ID**（如 `qwen3.6-plus`）用于问答生成 | +| **API 端点** | • 统一 REST 基础路径：`https://dashscope.aliyuncs.com/api/v2/apps/memory/`
• 主要接口：`/add`, `/memory_nodes/search`, `/user_profile/get`, `/list`, `/delete` | • 分阶段 API：`/file/upload`, `/index/create`, `/index/job/submit`, `/retrieve`, `/query`
• 上层服务端点：`/knowledge-retrieval`, `/knowledge-qa`（支持多知识库联合） | +| **计费方式** | • **按调用次数计费**（QPM 限流严格）：
 - `AddMemory`: ≤120 QPM
 - `SearchMemory`: ≤300 QPM
 - 全账号总计 ≤3000 QPM
• 无存储容量费、无模型推理费 | • **双维度计费**（自2026年1月4日起）：
 - **规格费**：按知识库版本（标准版/旗舰版）按小时计费
 - **模型调用费**：按 [Token](../concepts/token.md) 计费（含 Rerank 排序 [Token](../concepts/token.md) + 问答生成 Token）
 - **关键成本因子**：`TopK` 值直接影响 Rerank Token 消耗（非仅最终返回数） | +| **典型场景** | • 智能客服中持续记录用户偏好（“不喜电话回访”“常用支付方式为支付宝”）
• 个人助理中管理日程提醒、健康习惯、购物清单等动态事项
• 教育 Agent 中跟踪学生错题类型、薄弱知识点、学习节奏变化 | • 企业内部知识问答（制度文档/产品手册/项目报告）
• 客服知识库（FAQ/工单案例/解决方案库）
• 多媒体内容检索（PPT 图文问答、会议录像字幕搜索）
• 行业垂类问答（医疗指南、法律条文、金融产品说明） | +| **数据生命周期** | • **无自动过期**（默认规则可配 7/30/180 天或永不过期）
• 全生命周期由开发者主动管理（`DeleteMemory`/`UpdateMemory`）
• `UpdateMemory` 对 `custom_content` 为全量覆盖 | • 文件上传即索引，**内容不可变**（修改需重新上传）
• 元数据抽取规则创建后不可修改
• 删除文件将触发异步索引重建 | +| **地域支持** | 全地域可用（与 DashScope API 一致） | **仅限华北2（北京）地域**（控制台与 API 均受限制） | +| **开发者控制粒度** | • 高：可精确控制每条记忆的 `user_id`/`memory_library_id`/`project_id`/`profile_schema`
• 支持细粒度分页（`page_num/page_size`）、相似度阈值（`min_score`）、召回数（`top_k`） | • 中高：可配置切片策略（推荐“智能切分”）、相似度阈值（0.01–1.0）、召回数（`max_retrieve_count`≤20）、标签过滤
• **元数据规则不可变**，需创建时一次性设定 | ## 各方案的适用场景建议 -### ✅ 优先选用 **长期记忆（新）** 当： -- 你需要**跨会话记住单个用户的行为特征与主观状态**（如“张三讨厌电话推销”、“李四每周三健身”）； -- 应用逻辑依赖**动态更新的用户画像**（年龄、职业、兴趣标签），且需与对话流深度耦合； -- 场景对**低延迟、高并发写入**有要求（如每轮对话结束自动存记忆），且无法接受知识库的文档上传/索引延迟； -- 数据敏感度高，**拒绝将用户对话原始内容上传至共享知识库**，要求严格按 `user_id` 隔离； -- 工程团队倾向轻量级集成，仅需几行 SDK 代码即可启用记忆能力，无需管理知识库规格与配额。 - -### ✅ 优先选用 **知识库** 当： -- 你的核心需求是**让大模型准确回答基于私有文档的问题**（如“最新版《员工手册》第5章关于年假的规定是什么？”）； -- 知识源为**批量、静态、结构化程度不一的业务文档**（合同模板、产品说明书、会议纪要），且需支持 PDF 表格识别、音视频转文字等多模态解析； -- 要求**答案可溯源、可审计**，必须明确标注引用来源（文件名+页码+段落）； -- 需要**混合检索能力**（向量 + 关键词）、**多知识库联合混排**（如同时查 HR 政策 + IT 流程 + 财务制度），并精细调控权重与标签过滤； -- 团队已具备文档治理流程，能接受知识库创建后**配置不可逆**（需重建），并愿意承担规格费与模型 Token 成本以换取更高精度与稳定性。 - -### ⚠️ 注意边界与组合策略 -- **不要混淆用途**：长期记忆 ≠ 用户文档存储空间；知识库 ≠ 用户偏好数据库。将用户聊天记录直接丢进知识库，既浪费成本又降低检索精度。 -- **推荐组合使用**：典型智能体架构中，**长期记忆负责“用户是谁、想要什么”**（个性化上下文），**知识库负责“世界是什么、规则是什么”**（领域知识）。二者通过不同 `user_id` 和 `kb_id` 隔离，再由 Agent 编排协同调用。 -- **性能兜底建议**：对高 QPS 场景（如百万级用户助手），长期记忆的 `SearchMemory`（300 QPM 限额）可能成为瓶颈，此时可结合本地缓存（如 Redis）暂存高频用户记忆；知识库则需根据并发量选择旗舰版 RCU 规格。 +### ✅ 选择「长期记忆（新）」当： +- 你的应用核心是**服务特定用户个体**，需在多次独立会话中保持对其偏好、状态、承诺事项的记忆（如：“上次说好下周三跟进合同”）； +- 数据天然以**对话形式产生**，且需从自然语言中自动提炼结构化事实（如从“我妈妈生日是1970年5月12日”中提取 `{"relation": "mother", "birthday": "1970-05-12"}`）； +- 你需要**轻量、低延迟、高频率**的增删查操作（如每轮对话后写入1条+每轮开始前检索3–5条），且不愿承担 Rerank 或大模型生成费用； +- 你使用 OpenClaw 等框架，希望**零代码接入自动捕获/自动召回**[插件](../concepts/plugin.md)，实现“记忆无感流转”。 + +### ✅ 选择「知识库（RAG）」当： +- 你的知识源是**静态、批量、多格式的文档集合**（如1000份PDF产品说明书），需为整个应用而非单个用户注入领域知识； +- 业务要求**强准确性、可溯源、防幻觉**，需通过引用原文片段佐证答案（如客服回复必须标注“依据《售后服务政策V3.2》第5.1条”）； +- 你需要处理**图片、表格、音视频等非纯文本数据**，并支持图文混合检索或视觉理解； +- 你已有成熟的数据管道（如每日同步CRM数据到OSS），需通过 API 自动化完成**文件上传→索引构建→服务发布**全流程； +- 你愿意为更高精度和更丰富能力（如多知识库联合检索、Query 改写、拒答控制）承担相应规格费与模型 Token 成本。 + +### ⚠️ 注意：二者可协同，非互斥 +- **典型协同模式**： + `用户当前会话 → 长期记忆召回其历史偏好（如“用户禁用语音播报”） → 知识库检索最新产品文档 → 大模型生成答案时，结合记忆约束（禁用语音）+ 知识依据（文档条款）生成合规响应` +- 技术实现：在 Agent 的 `tool_call` 或工作流中，**并行调用 `SearchMemory` 和 `KnowledgeRetrieval`**，将结果统一注入 LLM 提示词。 ## 面向开发者的选型参考 -| 选型问题 | 长期记忆（新） | 知识库 | 决策建议 | -|----------|----------------|---------|-----------| -| **我的数据是用户对话产生的个性化信息吗？** | 是 | 否 | ✔️ 选长期记忆 | -| **我的数据是公司内部的 PDF/Excel/音视频等业务资料吗？** | 否 | 是 | ✔️ 选知识库 | -| **我需要为每个用户单独隔离数据，且不能共享？** | 是（`user_id` 强隔离） | 否（知识库全局共享，靠权限控制访问） | ✔️ 选长期记忆 | -| **我需要答案附带原文出处，满足合规审计要求？** | 否（仅返回记忆内容） | 是（返回 `source` 字段与文件链接） | ✔️ 选知识库 | -| **我的应用部署在新加坡/法兰克福地域？** | 支持 | ❌ 不支持（仅华北2） | ✔️ 必须选长期记忆 | -| **我追求最低接入成本，希望 SDK 一行代码启用？** | `AddMemory().arun(...)` 即可 | 需先创建 KB、上传文档、触发索引，再调用检索 | ✔️ 选长期记忆 | -| **我需要支持视觉文档（扫描件/PPT）的 OCR 与理解？** | 不支持 | 支持（需创建“多模态知识库”，绑定 VL 模型） | ✔️ 选知识库 | - -> **最后建议**：首次集成时，请务必使用 cURL 或 Postman 验证基础 API 流程（而非直接依赖 SDK 封装），重点关注 `request_id` 日志排查；生产环境务必设置合理的 `min_score`（长期记忆 ≥0.5，知识库 ≥0.3）与 `top_k`（3–10），避免噪声干扰或召回不足。 +| 你的需求 | 推荐方案 | 关键理由 | +|----------|-----------|-----------| +| “我要做一个健身教练Bot，记住每个用户的运动目标、受伤史、每周训练反馈，并据此调整计划” | ✅ 长期记忆（新） | 用户专属、动态更新、对话驱动、低成本高频操作 | +| “我要搭建公司内部IT Helpdesk，让员工能问‘如何重置VPN密码’并返回准确步骤截图” | ✅ 知识库（RAG） | 多模态（图文）、静态知识、需引用溯源、强准确性要求 | +| “我的电商客服Bot既要懂商品参数（知识库），又要记用户本次投诉诉求（长期记忆）” | ✅ 两者协同 | 知识库提供通用产品信息，长期记忆保存本次会话的订单号、情绪标签、承诺时效 | +| “我需要在新加坡地域部署应用，且必须使用私有知识” | ❌ 知识库（不可用）→ ✅ 长期记忆（新） | 知识库地域限制为硬性约束，长期记忆无此限制 | +| “我每天新增10万条用户行为日志，需实时写入并支持语义搜索” | ⚠️ 谨慎评估 → 建议长期记忆 + 异步批处理 | `AddMemory` QPM 限流120，需拆分为多 `user_id` 并行或采用 `custom_content` 批量聚合；知识库不支持高频流式写入 | +| “我只有3个PDF文件，但要求极低延迟（<300ms）和零模型费用” | ✅ 长期记忆（新） | 可将PDF文本摘要后作为 `custom_content` 写入，用 `SearchMemory` 直接检索，规避Rerank与生成费用 | + +> **最后建议**： +> - **MVP 阶段优先试用长期记忆（新）**：API 简洁、上手快、成本可控，适合验证用户记忆价值； +> - **规模化知识服务必选知识库**：其多模态、高精度、生产级管控（拒答/溯源/审计日志）是长期记忆无法替代的； +> - **始终通过 `DASHSCOPE_API_KEY` 统一认证**，二者共享同一套密钥体系与配额管理，便于权限收敛。 ## 被对比主题页 diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md index 4390971f..371949d8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md @@ -1,51 +1,63 @@ -# 模型部署方案对比：高并发推理、生产部署与压缩优化 - -为帮助开发者在不同业务阶段（如流量洪峰应对、长期服务上线、成本敏感型部署）科学选型，本文系统对比百炼平台三大核心模型部署能力：**高并发推理（TPM 预留 & 快速模式）**、**生产级模型部署（[model production](../api/model-production.md)）** 和 **模型压缩（[model compression](../guides/model-compression.md)）**。三者定位互补——高并发推理聚焦 *运行时性能保障*，生产部署解决 *定制化模型落地闭环*，压缩优化则面向 *推理成本与资源效率平衡*。理解其差异是构建稳定、高效、可演进 AI 服务的关键前提。 - -## 关键维度对比 - -| 维度 | 高并发推理（TPM 预留 + 快速模式） | 生产部署（[model production](../api/model-production.md)） | 压缩优化（[model compression](../guides/model-compression.md)） | -|------|----------------------------------|------------------------------|------------------------------| -| **核心目标** | 保障高吞吐稳定性（TPM 预留）或极致首 token/流式延迟（快速模式） | 实现私有化微调模型的端到端上线与服务化 | 降低已训练模型的推理资源消耗与部署成本 | -| **输入格式** | 标准 OpenAI 兼容请求体（`messages`, `stream`, `temperature` 等）；快速模式需额外适配 `reasoning_content` 字段解析 | 微调：JSONL 训练数据集 URL；部署：`model_id` / `fine_tuned_model_id` + 实例规格等配置参数 | 微调成功且状态为 `SUCCEEDED` 的自定义模型 ID；可选校准数据集（最多 5 个已发布数据集） | -| **输出格式** | 标准 OpenAI 流式/非流式响应；快速模式返回含 `delta.reasoning_content` 和 `delta.content` 的双通道结构 | 微调任务输出 `fine_tuned_model_id`；部署后返回 `endpoint_url`（兼容 `/v1/chat/completions`） | 生成新模型 ID（源模型名 + 后缀），如 `my-qwen-ft-w8a8`，存于模型中心，可直接用于部署 | -| **支持模型** | **TPM 预留**：Qwen、GLM、DeepSeek、Kimi 等十余个主流基础模型（如 `qwen3.7-max-2026-05-20`, `deepseek-v4-pro`）；
**快速模式**：仅 `glm-5.2-fast-preview`（Preview 阶段，严格限定） | 支持基于 Qwen 系列等基础模型的全参微调（`full`）与 LoRA 微调（`lora`）；部署对象为微调产出模型或 `import_model` 导入的第三方模型 | **仅限百炼平台内微调产出的自定义模型**（如 `qwen3.5-flash-2026-02-23-finetuned-xxx`）；不支持基础模型、OSS 模型、第三方模型 | -| **API 端点** | **TPM 预留**：复用标准 MaaS 域名（如 `https://{workspace_id}.maas.aliyuncs.com/v1`），但 `model` 参数需替换为专属 TPM code（如 `tpm-qwen37max-xxx`）；
**快速模式**：必须使用专属地域域名（如 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），`model="glm-5.2-fast-preview"` | 微调：`POST /api/v1/fine_tuning_jobs`；部署：`POST /api/v1/deployments`；服务调用：`POST {endpoint_url}/v1/chat/completions` | 控制台操作为主（路径：模型 > 模型训练 > 模型压缩）；无公开 REST API，任务通过控制台提交与管理 | -| **计费方式** | **TPM 预留**：按预留容量（kTPM）预付费，超额部分自动降级为按量计费；缩容按 1.5 倍系数结算；
**快速模式**：Preview 阶段暂未明确独立计费规则，实际按底层资源消耗计费（建议监控） | 微调：按 GPU 小时计费；部署：按所选实例规格（如 `ecs.gn7i-c16g1.4xlarge`）的 MU 小时计费；版本管理不额外收费 | 压缩任务本身限时免费；压缩后模型的部署费用按 MU 规格单独计费（因规格降低而节省成本） | -| **典型场景** | - 大促期间客服机器人流量峰值保障（TPM 预留）
- 编程助手要求 <200ms 首 token 延迟（快速模式）
- Agent 多步推理中对 token 流速敏感的链路 | - 金融领域定制化报告生成模型上线
- 电商客服知识库问答模型迭代与灰度发布
- 将开源模型微调后封装为内部 SaaS 服务 | - 已验证效果的微调模型需降低 30%+ 推理成本
- 边缘侧或轻量级容器环境部署受限于显存/内存
- 快速验证不同量化精度（W4A4/W8A8）对业务指标的影响 | - -## 适用场景建议 - -- **选择高并发推理（TPM 预留）当**： - 你的模型已在生产环境稳定运行，但面临周期性流量高峰（如每日晚 8 点用户咨询激增），且 SLA 要求“99.9% 请求在 1s 内完成”，无法容忍公共资源池的随机限流。此时，TPM 预留是保障容量确定性的最优解，尤其适用于成熟业务线的稳态扩容。 - -- **选择快速模式（Fast mode）当**： - 你的应用对交互实时性极度敏感（如 IDE 内嵌代码补全、语音转文字后的即时意图分析），且能接受 Preview 阶段的技术不确定性。注意：仅 `glm-5.2-fast-preview` 可用，客户端需改造解析逻辑，**严禁用于支付、风控等强 SLA 场景**。 - -- **选择生产部署（[model production](../api/model-production.md)）当**： - 你需要将自有业务数据训练出的专属模型（如医疗问诊微调模型）长期、可靠、可回滚地上线。它提供完整的生命周期管理（训练→部署→版本→监控），是构建企业级 AI 应用的基石能力，适合从 PoC 迈向规模化落地的团队。 - -- **选择压缩优化（[model compression](../guides/model-compression.md)）当**： - 你的微调模型已通过业务验证，但部署成本过高（如需 2×A10 GPU），或目标环境资源受限（如单卡 24GB 显存）。通过 PTQ 量化可显著降低 MU 规格（如从 `gn7i-c16g1.4xlarge` 降至 `gn7i-c8g1.2xlarge`），在精度损失可控前提下实现成本优化，**必须在部署前执行，且不可逆**。 - -## 技术选型参考（面向开发者） - -| 你的关键诉求 | 推荐方案 | 关键动作提醒 | -|--------------|----------|--------------| -| “我的模型流量忽高忽低，怕高峰期被限流崩掉” | ✅ TPM 预留 | 计算真实 kTPM 需求（考虑长文本阶梯系数），使用专属 model code，监控“超额降级统计”避免隐性成本 | -| “用户抱怨补全太慢，首 token 要 1.2 秒，体验差” | ⚠️ 快速模式（仅限 GLM-5.2） | 确认业务能接受 Preview 风险；切换域名；解析 `reasoning_content`；压测排队延迟容忍度 | -| “我要用自己标注的 5000 条合同数据训练一个法律问答模型并上线” | ✅ 生产部署 | 优先选用 LoRA 微调（成本低、速度快）；规划 `endpoint_name`；部署后立即做 A/B 测试验证效果 | -| “这个微调好的模型效果不错，但部署要两台 A10，太贵了，能压小点吗？” | ✅ 压缩优化 | 在华北2地域操作；选 W8A8 模板作为起点；用历史测试集校准；部署后对比 accuracy & latency | -| “我想把 HuggingFace 上下载的 Llama3-8B-GGUF 模型直接部署” | ❌ 三者均不支持 | 百炼当前不支持直接导入 GGUF/AWQ 等外部量化格式；需先转换为百炼兼容格式或通过 `import_model` 流程验证 | -| “我需要同时跑 10 个不同版本的客服模型做灰度” | ✅ 生产部署（+ 版本管理） | 提工单申请提升部署实例配额（默认 5 个）；利用 `version_id` 精确路由流量 | -| “模型压缩后还能不能继续微调？” | ❌ 不可以 | 压缩不可逆！务必保留原始 `SUCCEEDED` 微调模型，所有后续迭代均从此开始 | - -> **重要提醒**：三类能力并非互斥，而是可组合使用——例如：对生产部署的 `qwen3.5-flash-finetuned-xxx` 模型执行压缩得到 `qwen3.5-flash-finetuned-xxx-w8a8`，再为其预留 TPM 并启用快速模式（若该模型未来支持）。但请注意：**快速模式当前仅对 `glm-5.2-fast-preview` 开放，不支持其他模型（含压缩后模型）**。技术演进请持续关注官方文档更新。 +# 模型部署方式对比：高并发推理、模型压缩与模型部署指南 + +## 背景与目的 + +在百炼平台的生产实践中，开发者常面临三类核心部署决策： +- **如何应对突发或持续的高并发请求**（如客服系统、实时搜索、AIGC 生成服务）； +- **如何降低自定义微调模型的部署成本与资源开销**（尤其在长周期、多实例场景下）； +- **如何选择最适配业务需求的模型服务化模式**（兼顾性能、弹性、可控性与成本）。 + +本文系统对比三大能力模块——**高并发推理（含快速模式与 TPM 预留）**、**模型压缩（量化优化）** 和 **模型部署（PTU / MU / [Token](../concepts/token.md) 用量）**，聚焦其技术定位、能力边界与协同关系，为开发者提供清晰、可落地的技术选型参考。需特别注意：三者并非互斥替代方案，而是分属不同层级的优化手段—— +✅ **模型压缩**作用于**模型资产层**（优化模型本身）； +✅ **高并发推理能力**作用于**服务调度与资源保障层**（优化请求处理路径）； +✅ **模型部署模式**作用于**服务交付层**（定义服务形态与计费契约）。 +合理组合使用（例如：对微调模型先压缩 → 再以 MU 模式部署 → 同时启用 TPM 预留保障关键流量），可实现性能、成本与稳定性的最优平衡。 + +--- + +## 关键维度对比表 + +| 维度 | 高并发推理 — 快速模式 | 高并发推理 — TPM 预留 | 模型压缩 | 模型部署 — PTU 模式 | 模型部署 — MU 模式 | 模型部署 — [Token](../concepts/token.md) 用量模式 | +|------|------------------------|--------------------------|------------|------------------------|------------------------|------------------------------| +| **核心目标** | 通过模型级加速提升单请求吞吐（TPS）与首 [Token](../concepts/token.md) 延迟 | 通过资源独占保障确定性吞吐（TPM）与服务稳定性 | 降低模型参数精度，减少部署所需计算资源（MU）与成本 | 提供预置容量的高并发、长上下文服务，支持前缀缓存与阶梯计费 | 提供资源独占、可定制规格的全功能模型服务（支持思考模式、PD 分离等） | 提供轻量、按调用计费的 LoRA 微调模型服务，适用于验证与低频场景 | +| **输入格式** | 标准 OpenAI 兼容 `messages`；流式时需解析 `delta.reasoning_content` / `delta.content` | 标准 OpenAI 兼容 `messages`；行为与标准模型一致 | 不直接接收推理输入；作用于模型文件（`.safetensors` 等） | 标准 OpenAI/Anthropic/DashScope 兼容格式；支持 `provisioned_tokens` 控制缓存 | 标准兼容格式；支持 `enable_thinking`, `max_context_length`, `rpm_limit` 等精细化控制 | 标准兼容格式；仅支持 LoRA 模型，`capacity` 字段必填但无效 | +| **输出格式** | 流式响应含 `reasoning_content`（思考过程）与 `content`（最终结果）双字段；非流式合并返回 | 与标准模型完全一致（无额外字段） | 无输出；生成新模型 ID（如 `my-qwen-ft-int4`） | 响应头含 `x-dashscope-ptu-overflow:true`（溢出时）；响应体含 `service_tier: "ptu-standard"` | 无专用额度标识；`service_tier` 通常不返回或为 `"default"` | 无专用额度标识；`service_tier` 通常不返回或为 `"default"` | +| **支持模型** | 仅 `glm-5.2-fast-preview`（北京/新加坡，Preview 阶段） | 多款主流模型：
• `GLM-5.2` / `GLM-5.1`
• `千问3.7-Max-2026-05-20`
• `DeepSeek-v4-Pro`
• `Kimi-K2.6` 等 | 仅百炼平台微调产出的**自定义模型**（如 `qwen3.5-flash-2026-02-23`）；不支持基础模型、第三方模型 | 主流预置模型：
• `glm-5.1`（200K 输入）
• `deepseek-v4-pro`
• `qwen3.7-plus-2026-05-26`（256K 输入）等 | 全部预置模型 + LoRA 微调模型（千问/GLM/DeepSeek/Kimi/MiniMax 等） | 仅 LoRA 微调模型：
• `qwen3-32b` / `qwen3-14b` 等
• **不支持全参微调、VL 模型** | +| **API 端点** | workspace 绑定域名：
`https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（北京）
或对应新加坡域名 | 全局兼容域名：
`https://dashscope.aliyuncs.com/compatible-mode/v1` | 无 API 端点；通过控制台或 API 提交压缩任务（异步） | 统一部署服务端点：
`/api/v1/services/{deployed_model}/completions` | 同上，统一端点 | 同上，统一端点 | +| **计费方式** | 按实际 token 用量计费；Preview 阶段策略可能调整 | 按购买 kTPM（千 tokens/分钟）预付费；超额部分自动降级为按量计费 | **压缩任务免费**（限时）；**压缩后模型部署仍按 MU 规格计费** | 按预购 `input_tpm` / `output_tpm` 计费；支持缓存折扣（如 `glm-5.1` 缓存命中按 20% 折算）；溢出自动转按量 | 按 `deploy_spec × capacity × 时长` 计费（如 `MU8*2`）；支持 PD 分离优化首 Token 延迟 | 按实际输入/输出 token 用量计费；无预付，无额度保障 | +| **典型场景** | • 对 `glm-5.2` 的极致低延迟需求（如实时对话机器人）
• 流量波动大但允许排队（超额度请求进入队列） | • 流量高度可预测的关键业务（如订单生成、风控决策）
• 需严格 SLA 保障（99.9% 可用性）
• 多模型混合部署需独立容量隔离 | • 自研 LoRA 微调模型需规模化部署（如 10+ 实例）
• 成本敏感型长期服务（节省约 50% MU 资源）
• 模型已验证精度达标，追求部署效率 | • 长文本处理（法律合同、代码分析）
• 高并发且流量平稳的 SaaS 服务
• 需利用前缀缓存优化重复提示（如客服知识库） | • 需要完全控制硬件规格与推理行为（如启用思考模式、自定义 context length）
• 对首 Token 延迟极度敏感（启用 PD 分离）
• 多租户隔离或安全合规要求强 | • A/B 测试与效果验证
• 内部工具、低频后台任务
• 快速原型上线，暂无稳定流量预期 | + +--- + +## 适用场景建议（面向开发者的技术选型指南） + +### ✅ 推荐组合策略（生产推荐） +| 业务需求 | 推荐方案 | 理由说明 | +|----------|-----------|----------| +| **高并发 + 长文本 + 成本可控** | `PTU 部署` + `前缀缓存` + `GLM/Qwen 长上下文模型` | PTU 提供确定性吞吐与长输入支持，缓存折扣显著降低高频重复提示成本；无需额外压缩或预留，开箱即用。 | +| **自研微调模型大规模上线** | `模型压缩` → `MU 部署`（压缩后模型） | 先量化降低 MU 占用（实测节省 56% 成本），再以 MU 模式部署获得资源独占与 PD 分离能力，兼顾成本与性能。 | +| **关键业务链路零容忍抖动** | `TPM 预留` + `PTU 部署`（同一模型） | TPM 保障核心流量基线容量，PTU 承担弹性峰值；二者可并行存在（TPM 用于主通道，PTU 用于备用通道），实现容量冗余。 | +| **极致低延迟对话服务（GLM 生态）** | `快速模式`（`glm-5.2-fast-preview`） | 当前唯一提供 `reasoning_content` 流式分离的加速方案，适合需实时展示思考过程的交互场景；注意其 Preview 属性与地域限制。 | + +### ⚠️ 需规避的常见误用 +- **❌ 混用 `glm-5.2-fast-preview` 与 TPM 预留**：二者技术路径冲突，`glm-5.2-fast-preview` 不支持 TPM 预留，强行配置将导致调用失败。 +- **❌ 对基础模型（如 `qwen3-72b`）执行模型压缩**：仅支持百炼微调产出的自定义模型，基础模型压缩入口不可见，操作无效。 +- **❌ 在 Token 用量模式下部署全参微调模型**：该模式明确不支持，导入会失败；应改用 MU 或 PTU 模式。 +- **❌ 为 MU 部署设置超出模型原生上限的 `max_context_length`**（如给 `qwen3-8b` 设 256K）：部署将成功但推理时触发截断或报错，需严格遵循基础模型文档上限。 + +### 🔧 开发者行动清单 +1. **评估流量特征**：若峰值/均值比 > 3，优先考虑 `TPM 预留` 或 `PTU`；若均值稳定，`MU` 更灵活。 +2. **检查模型来源**：是百炼微调模型？→ 可走 `模型压缩`；是官方预置模型？→ 直接选 `PTU/MU/Token`。 +3. **验证精度容忍度**：对压缩后模型务必在**业务真实测试集**上验证效果（尤其 NLU/NLG 任务），避免仅依赖 MU 数值。 +4. **统一 endpoint 管理**：所有部署模式均使用 `/services/{deployed_model}/completions`，通过 `model` 参数切换，便于灰度与路由。 +5. **监控溢出信号**：PTU 场景需监听响应头 `x-dashscope-ptu-overflow:true`，及时扩容或优化提示长度。 + +> 💡 **最后提醒**：百炼平台能力持续演进。`快速模式` 处于 Preview 阶段，`模型压缩` 仅限北京地域，`Token 用量` 不支持 VL 模型——请始终以控制台实时能力列表与最新文档为准，避免硬编码过期参数（如模型 ID、地域域名）。 ## 被对比主题页 - [model high speed inference](../guides/model-high-speed-inference.md) -- [model production](../api/model-production.md) - [model compression](../guides/model-compression.md) +- [model deployment 1](../guides/model-deployment-1.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md index 3cd824a8..05c5c14e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md @@ -1,52 +1,61 @@ -# 模型评估与监控体系对比 +# 模型评估与监控能力对比 -为帮助开发者在模型研发、上线与运维全生命周期中科学选型，本文系统对比百炼平台三大核心质量保障能力：**模型评测（Model Evaluation）**、**模型监控（Model Monitoring）** 和 **应用评测（Application Evaluation）**。三者定位互补：模型评测聚焦「结果质量」的离线、结构化打分；模型监控侧重「运行状态」的实时、可观测追踪；应用评测则面向「端到端智能体/工作流」的业务逻辑与交互效果评估。本对比旨在厘清能力边界、明确适用阶段与技术约束，避免功能误用或能力缺失。 +为帮助开发者在模型研发、上线及运维全生命周期中合理选用百炼平台的能力组件，本文对三大核心可观测性能力——**模型评测（Model Evaluation）**、**模型监控（Model Monitoring）** 和 **应用评测（Application Evaluation）** 进行系统性对比分析。三者定位互补： +- **模型评测**聚焦于*离线、定量、多维度的能力归因*，服务于模型选型与调优验证； +- **模型监控**侧重于*在线、实时、指标化的服务健康度观测*，保障生产稳定性与成本可控； +- **应用评测**专用于*端到端智能体/工作流级的质量闭环验证*，覆盖 RAG、Agent 编排等复杂链路的输出可信度评估。 + +本对比基于当前（2024年Q3）百炼平台正式功能，面向技术选型决策者提供客观、可落地的参考依据。 ## 关键维度对比 | 维度 | 模型评测（Model Evaluation） | 模型监控（Model Monitoring） | 应用评测（Application Evaluation） | -|------|------------------------------|-------------------------------|-------------------------------------| -| **核心目标** | 量化模型推理输出质量（语义正确性、事实一致性、指令遵循度等） | 实时观测模型服务稳定性、性能、成本与异常行为 | 评估智能体/工作流级应用的端到端业务效果（如问答准确率、任务完成率、RAG链路健壮性） | -| **输入格式** | 必须提供三元组：`Prompt` + `Output`（模型生成）+ `Completion`（参考答案）；支持 `.xls`/`.xlsx`/`.jsonl` 格式评测数据集 | 无需显式输入——自动采集所有调用请求的原始 `input` 与 `output`（北京地域可审计），指标基于 API 调用日志聚合 | 支持多种结构化输入：
• 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）
• 新版：按「智能体」「工作流」「自定义」类型自动生成参数模板（支持 `query`/`response`/`history` 等字段映射） | -| **输出格式** | 结构化评分结果：
• 综合得分（各维度平均）
• 维度级通过率/分布图
• 逐样本明细（含 AI 评分、规则分数、人工标签） | 多粒度时序指标：
• 控制台卡片（小时级延迟）
• Prometheus 指标（分钟级，需高级监控）
• 审计日志（北京地域，含原始 I/O 与 Token 用量） | 分层报告输出：
• 总体正确率 / Pass率
• BadCase 归因（检索失败/切片错误/重排偏差等）
• 多评估器并行结果（LLM/Code/历史模型）
• 标签筛选后的细分统计 | -| **支持模型** | **仅文本生成类模型**（预置及调优后模型），不支持多模态、语音、结构化输出模型 | **全量支持**：覆盖控制台所有可选模型（含千问系列、开源快照、三方模型及全部调优模型） | **受限支持**：
• 自动评测：仅 `qwen-max` 和 `qwen-plus` 可作为裁判模型
• 应用关联：支持任意已发布智能体/工作流（不限底层模型） | -| **API 端点** | 无独立公开 API；通过控制台创建评测任务触发，结果通过 `/api/v1/evaluation/tasks/{id}` 查询（需权限） | 提供标准 Prometheus HTTP API：
`GET {endpoint}/api/v1/query_range`
支持 `model_call_count`、`model_usage` 等 20+ 指标查询 | 无标准化 REST API；评测任务通过控制台发起，结果数据可通过 `/api/v1/application-evaluation/tasks/{id}` 获取（内部接口，文档未开放） | -| **计费方式** | • 被评测模型推理费用（使用评测数据集时）
• 裁判模型费用（仅大模型评估维度）
• 规则/人工评估零额外费用 | • 无单独监控费用
• 所有监控数据采集免费
• **但调用本身产生常规模型费用**（Token 计费） | • LLM 评估器调用：按 Token 计费（`qwen-max`/`qwen-plus`）
• 评测集生成：按 Token 计费
• Code 评估器：执行脚本免费（不调用模型） | -| **典型场景** | • A/B 测试新 Prompt 效果
• 验证模型微调前后性能提升
• 选型决策：对比多个候选模型在特定任务上的综合得分
• 合规审计：人工复核高风险输出 | • 生产环境告警（失败率突增、P99 延时超标）
• 成本治理：识别高 Token 消耗应用/用户
• 性能优化：分析首 Token 延时瓶颈
• 安全审计：追踪内容安全拦截事件 | • RAG 应用迭代：定位检索/切片/重排环节短板
• 智能体上线前验收测试
• 多版本横向对比（同一评测集验证不同 workflow 配置）
• 客户反馈归因：将 BadCase 关联至具体评估器标签 | +|------|------------------------------|------------------------------|------------------------------------| +| **核心目标** | 量化模型文本生成能力（准确性、一致性、安全性等），支持横向对比与归因分析 | 实时观测模型调用行为、性能瓶颈、成本消耗与异常事件，保障服务 SLA | 评估智能体/工作流整体输出质量（含知识检索、逻辑推理、格式合规等），支持链路级问题定位 | +| **输入格式** | `EvaluationSet`（JSONL，含 `prompt` + `completion`）或已含 `output` 的推理结果集（CSV/JSONL） | 无显式“输入数据集”；自动采集所有调用请求（HTTP/SDK）原始参数与上下文（北京地域部分模型支持请求/响应内容） | 自动评测：知识库 → 自动生成 `query`+`referenceAnswer` JSONL；手动/新版：XLS/XLSX（含 `Prompt`/`Completion`/`SessionId`）或自定义结构化表（JSONL/CSV） | +| **输出格式** | 结构化评分报告（综合得分、通过率、分布直方图）、逐条明细（含 LLM 判定理由、规则匹配结果、人工标签） | 多维时间序列指标（Prometheus 格式）、可视化看板（RPM/TPM/延迟/P99/失败率）、告警通知（短信/钉钉/Webhook）、原始日志（北京地域启用后） | 自动评测：总正确率、BadCase 归因分类（如“检索失效”“幻觉”“格式错误”）、调优建议；新版任务：各评估器评分明细、人工标签统计、交叉分析图表 | +| **支持模型类型** | **仅文本生成类模型**（预置 & 调优模型），不支持多模态、语音、向量模型 | **全模态支持**：LLM（qwen-plus/max）、视觉、语音、全模态、向量模型（所有公开及调优模型） | **仅智能体（Agent）与工作流（Workflow）应用**（需已发布）；底层依赖 `qwen-max`/`qwen-plus` 执行 LLM 评估，但不直接评测基础模型 | +| **API / SDK 支持** | ❌ **不提供公开 API/SDK**；仅控制台操作（自动化需通过 PAI Judge Model API 替代） | ✅ **完整 Prometheus API 支持**（HTTP 接口 + Basic Auth）；支持 Grafana 集成、自建系统对接；控制台提供 RESTful 告警管理接口 | ✅ **新版评测任务支持 OpenAPI**（创建/查询/停止任务、获取结果）；旧版自动评测暂无 API；手动评测无 API | +| **计费方式** | - 被评测模型推理费（使用评测数据集时按输入/输出 [Token](../concepts/token.md) 计费）
- 裁判模型费（大模型评估维度，按 [Token](../concepts/token.md) 计费）
- 规则/人工评估：零模型费用 | - **无额外评测费用**；所有监控数据采集免费
- 仅基础调用本身产生模型费用（与是否开启监控无关）
- 日志审计（北京）按存储量计费 | - 自动评测/LLM 评估器：调用 `qwen-max`/`qwen-plus`，按 [Token](../concepts/token.md) 计费
- Code 评估器：无额外调用成本
- 评测集存储：免费（≤20MB/个） | +| **典型场景** | • 新模型上线前能力基线测试（vs C-Eval/GSM8K）
• Prompt 工程效果验证（A/B 测试）
• 微调模型 vs 原始模型能力衰减分析
• 安全合规性人工抽检 | • 生产环境服务稳定性巡检（延迟突增、失败率飙升）
• 成本优化（识别高 Token 消耗调用、限流根因）
• 故障快速定位（结合日志追踪首 Token 延迟、内容安全拦截）
• 多模型路由策略效果验证 | • RAG 知识库更新后问答质量回归测试
• Agent 工作流编排逻辑正确性验证（如“订机票→查天气→发邮件”链路）
• 多应用横向对比（相同知识库下不同 Agent 设计优劣）
• 用户反馈 BadCase 的深度归因分析 | +| **地域限制** | • 自定义评测：全地域支持
• 基线评测（C-Eval 等）：**仅北京地域可用** | • 普通监控：全地域
• 高级监控（分钟级）、告警、日志审计：**仅北京、新加坡支持**；弗吉尼亚支持分钟级监控但不支持告警/日志 | • 自动评测：**仅北京地域支持**（依赖知识库服务）
• 新版评测任务/手动评测：全地域支持（评测集上传与任务执行） | +| **数据时效性** | 离线批处理：任务完成后即时生成结果（耗时取决于数据量与裁判模型负载） | • 普通监控：约 1 小时延迟
• 高级监控：≤5 分钟延迟（指标）；日志存在分钟级延迟需手动刷新 | • 自动评测：任务执行中实时显示进度，完成后即时出结果
• 新版/手动评测：人工标注提交后即时更新统计 | ## 各方案适用场景建议 -- **选择模型评测（Model Evaluation）当**： - ✅ 需要对**单次模型输出质量进行深度语义评判**（如“回答是否完整覆盖问题要点”“是否存在幻觉”）； - ✅ 有明确参考答案（Completion），且任务具备可定义的评分维度（如“事实准确性”“指令遵循度”）； - ✅ 处于模型开发/调优阶段，需高频、小批量验证效果； - ❌ 不适用于无参考答案的开放域任务，或需实时响应的线上服务保障。 - -- **选择模型监控（Model Monitoring）当**： - ✅ 需要**7×24 小时守护生产服务稳定性**（如设置“失败率 >5%”告警）； - ✅ 关注**性能基线变化**（如 TPM 下降 30%）、**成本异常**（某 API Key 单日 Token 消耗翻倍）； - ✅ 运维团队需对接 Grafana/Prometheus 构建统一可观测平台； - ❌ 不适用于评估“回答好不好”，仅能回答“调用快不快、成不成、花多少钱”。 - -- **选择应用评测（Application Evaluation）当**： - ✅ 评估对象是**封装了 Prompt、工具、知识库的智能体或工作流**，而非裸模型； - ✅ 需要**归因分析 BadCase 根源**（例如：“80% 错误源于知识切片过短”，而非“模型答错了”）； - ✅ 业务逻辑复杂，需混合 LLM 语义判断 + Code 规则校验（如“返回 JSON 必须含 `status:success` 字段”）； - ❌ 不适用于纯文本生成模型的基础能力 benchmark（此时应选模型评测）；旧版评测集已逐步淘汰，新建项目务必使用新版「智能体/工作流」评测集。 - -## 技术选型参考（面向开发者） - -| 你的需求 | 推荐方案 | 关键理由 | 注意事项 | -|----------|----------|----------|----------| -| **快速验证一个新 Prompt 在 GSM8K 上的效果** | 模型评测（基线评测） | 直接调用预置 C-Eval/GSM8K 数据集，5 分钟获取基准分，无需准备数据 | 仅北京地域可用；结果不可下载，仅作快速参考 | -| **上线后发现客服机器人响应变慢，需定位是模型还是网络问题** | 模型监控（高级监控 + 首Token延时指标） | 分钟级查看 `model_first_token_duration`，结合 `model_call_duration_p99` 对比，若首Token延时高而总延时低，说明网络或前置服务瓶颈 | 需开通高级监控，且模型部署在北京/新加坡/弗吉尼亚 | -| **知识库问答应用上线后客户投诉“答案不相关”，需找出是检索不准还是模型理解错** | 应用评测（新版 + 多评估器） | 创建「检索相关性」Code 评估器 + 「答案事实性」LLM 评估器，用同一评测集并行跑分，BadCase 自动归因到具体环节 | 必须使用新版评测集；需为知识库配置切片策略并发布应用 | -| **为合规要求，每月人工抽检 100 条金融咨询回答是否含违规表述** | 模型评测（人工评估维度） | 创建 Pass/Fail 标签，上传待检数据集，分配标注人员，结果自动统计通过率并留痕 | 人工评估无裁判模型费用；标注完成后才标记“评测完成” | -| **构建企业级 AI 运维看板，集成模型调用量、成本、错误率与业务指标** | 模型监控（Prometheus API + 自建 Grafana） | 所有指标通过标准 HTTP API 拉取，可与现有监控栈无缝融合，支持 `workspace_id`/`apikey_id` 等多维下钻 | 高级监控需手动开启；认证使用 AccessKey，注意密钥安全 | - -> **重要提醒**：三者非互斥关系，而是**协同闭环**—— -> 🔹 模型评测发现“事实性得分低” → 触发应用评测深入归因 → 定位到知识切片问题 → 优化切片策略 → 用模型监控确认线上 P99 延时未恶化 → 再用模型评测验证修复效果。 -> 开发者应根据所处阶段（开发/测试/上线/运维）和关注焦点（质量/性能/成本/归因），组合使用这三类能力，构建完整的 AI 质量保障体系。 +| 场景描述 | 推荐方案 | 关键理由 | +|----------|----------|----------| +| **新训练的文本生成模型（如 Qwen2-7B-Chat）需验证其数学推理、中文理解能力是否达到 SOTA 水平** | ✅ 模型评测（基线评测） | 直接复用平台预置 C-Eval/GSM8K 数据集，5 分钟内获得标准化分数，支持与历史模型横向对比；无需构建应用或部署服务。 | +| **线上客服对话机器人（基于 qwen-plus 的智能体）近 2 小时失败率从 0.5% 升至 12%，需快速定位是模型超时、知识库未命中还是限流导致** | ✅ 模型监控（高级监控 + 日志审计） | 查看 `model_call_duration_p99`、`model_usage{usage_type="input_tokens"}` 及 `error_code="429"` 指标趋势；在北京地域启用日志审计后，直接查看失败请求的原始 [prompt](../guides/prompt.md) 与 error message。 | +| **完成一轮 RAG 知识库更新后，需系统性验证 200 个典型用户问题的回答质量，并生成“检索失效占比”“答案幻觉率”等归因报告** | ✅ 应用评测（自动评测） | 知识库自动触发评测集生成 → 调用 `qwen-max` 打分 → 输出结构化归因（如“73% BadCase 因切片不完整”），直接指导知识库切分策略优化。 | +| **需对多个自研微调模型（text2sql、摘要生成）进行 A/B 测试，且要求支持 BLEU/ROUGE 规则评估 + LLM 语义评估混合打分** | ✅ 模型评测（自定义评测） | 创建混合维度（如 `SQL准确性-BLEU` + `SQL准确性-LLM评分`），上传统一评测数据集，一次任务输出双维度结果，避免跨工具数据拼接。 | +| **企业需将百炼模型调用指标接入自建运维平台（如 Zabbix），实现与数据库、API 网关指标统一告警** | ✅ 模型监控（Prometheus API） | 通过标准 PromQL 查询 `model_call_count{model="qwen-plus", workspace_id="xxx"}`，与现有监控栈无缝集成，无需定制开发。 | +| **客户成功团队需定期抽检销售助手智能体的回答，由业务专家人工标注“专业度”“合规性”两个维度（5 分制）** | ✅ 应用评测（新版评测任务 + 人工标签） | 创建含 `professional_score`/`compliance_score` 标签的评测集，关联 LLM 评估器（初筛）+ 人工标注流程，结果自动聚合统计，支持导出 Excel 报告。 | + +## 技术选型参考指南（面向开发者） + +- **优先选择模型监控，当您需要：** + ✅ 实时感知服务健康状态（SLA、延迟、错误） + ✅ 追踪成本消耗并优化预算（Token/图像/视频用量下钻） + ✅ 构建自动化运维体系（Prometheus/Grafana/告警联动） + ⚠️ 注意：若需分钟级洞察或日志审计，请确认地域支持（北京/新加坡）。 + +- **优先选择模型评测，当您需要：** + ✅ 对基础模型（非应用层）做能力量化（如“该模型在 BBH 上得分 68.2，比上一版提升 3.1”） + ✅ 验证 Prompt/LoRA 微调效果（控制变量法，固定数据集与裁判模型） + ✅ 满足合规审计要求（生成可追溯、可复现的评分报告） + ⚠️ 注意：仅支持文本生成模型；基线评测不可下载结果，建议关键任务使用自定义评测。 + +- **优先选择应用评测，当您需要：** + ✅ 评估端到端智能体/工作流输出（而非单个模型） + ✅ 深度归因链路问题（如 RAG 中“检索→重排→生成”各环节贡献度） + ✅ 混合自动化与人工评审（LLM 初筛 + 专家终审） + ⚠️ 注意：自动评测强依赖知识库与北京地域；新版评测任务更灵活，推荐新项目首选。 + +> **组合使用建议**： +> - **研发阶段**：用 *模型评测* 验证基础模型能力 → 用 *应用评测* 验证智能体封装效果 → 上线后用 *模型监控* 持续守护。 +> - **故障排查**：先看 *模型监控* 发现异常指标（如 `first_token_duration` 突增）→ 再用 *应用评测* 复现 BadCase 并归因 → 最后用 *模型评测* 隔离是否为模型自身退化。 +> - **成本治理**：通过 *模型监控* 识别高消耗调用 → 提取样本用 *模型评测* 分析低分原因（如 Prompt 过长导致输出冗余）→ 优化后再次 *模型评测* 验证改进效果。 ## 被对比主题页 diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md index 95adc3cb..c86d8b89 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md @@ -1,47 +1,51 @@ # 函数调用 -函数调用（Function Calling）是百炼平台中模型主动识别用户意图、生成结构化工具调用请求，并交由外部系统执行的能力。它使大模型能突破自身知识与能力边界，安全、可控地接入实时搜索、代码执行、图像生成、OCR解析、GUI操作等外部服务，实现“思考→规划→调用→整合”的闭环推理。 +函数调用（Function Calling）是百炼平台中大模型主动识别用户意图、自主触发外部工具并结构化传递参数的核心能力。它使模型不仅能生成文本，还能在推理过程中决策是否需要调用特定工具（如计算器、搜索、代码执行、图像生成等），并将结果无缝整合进最终响应。 ## 在百炼平台的不同场景中，这个概念如何使用 -函数调用并非单一接口，而是贯穿多类模型与交互范式的统一能力机制，具体体现为以下三种典型模式： +函数调用在百炼平台中并非单一接口能力，而是贯穿多个技术路径的横切机制，具体体现为： -- **通用模型的自主工具调用**：`qwen3.7-plus`、`qwen3.6-flash`、`qwen3.5-omni-plus-realtime` 等主流模型在启用 `tools` 参数后，可基于用户输入自动决策是否调用、调用哪个工具、传入哪些参数，并返回标准化的 `tool_calls` 响应（含 `tool_id` 和 `arguments`）。该过程完全由模型内部推理完成，开发者只需提供工具定义（名称、描述、参数 schema），无需编写调度逻辑。 +- **Omni Realtime API（实时语音交互场景）**： + 在 `qwen3.5-omni-realtime` 等实时多模态模型中，函数调用通过 `tools` 字段声明支持的工具列表（`type: "function"`），模型在流式推理过程中可自主触发工具调用，并返回标准化的 `tool_calls` 事件；客户端需执行对应逻辑后，将结果通过 `response.create` 回传，继续后续对话流。该模式严格依赖 WebSocket 事件驱动，适用于低延迟语音助手、智能客服等场景。 -- **专用意图模型的显式决策**：`tongyi-intent-detect-v3` 是专为函数调用设计的轻量级模型，不生成自然语言回复，而是直接输出结构化意图标签（如 `"search"`、`"calculate"`）或完整工具调用指令（`INTENT_MODE` 模式）。适用于需强确定性、低延迟的路由/分发场景，常作为智能体前置网关。 +- **[插件](plugin.md)（Plug-in）体系（智能体/工作流场景）**： + [插件](plugin.md)本质即函数调用的工程化封装。开发者定义工具（Tool）后，模型（如 `qwen-plus`、`qwen-max`）根据用户输入自动规划调用序列；在智能体（Agent）中由模型自主决策，在工作流（Workflow）中可显式编排节点顺序。所有[插件](plugin.md)均需配置 `tool_id`、输入/输出 Schema 和鉴权方式，调用过程对上层应用透明。 -- **垂直领域模型的内嵌工具链**：`gui-plus-2026-02-26` 通过 `computer_use` 工具实现 GUI 自动化；`qwen3.5-ocr` 在图文混合输入下自动触发结构化解析；`qwen-deep-research` 在研究流程中隐式调用检索与报告生成子模块。这些模型将函数调用深度集成至业务逻辑，对外表现为端到端能力，而非显式 `tool_calls` 字段。 +- **[OpenAI 兼容接口](openai-compatible-interface.md)（Chat Completions / Responses API）**： + 通过标准 `tools` + `tool_choice` 字段声明函数集合与调用策略（如 `"auto"`、`"required"` 或指定 `{"type": "function", "function": {"name": "xxx"}}`），模型返回 `tool_calls` 数组（含 `id`、`function.name`、`function.arguments`）；开发者解析后同步执行并构造 `tool_message` 回传。此方式兼容主流 SDK，适合快速迁移或通用对话应用。 -> ⚠️ 注意：并非所有模型均支持函数调用。例如 `qwen-long`（10M上下文）明确不支持；`qwen3.7-max` 不支持结构化输出，因而无法返回合规的 `tool_calls`；`qwen-omni-turbo-realtime` 系列虽支持 `tools` 参数，但文档未确认其完整调用流程，建议优先选用 `qwen3.5-omni-realtime` 系列。 +- **自定义模型集成（DashScope 原生调用）**： + 对支持函数调用的模型（如 `qwen3.7-plus`），可通过 DashScope 原生 `/api/v1/services/aigc/text-generation/generation` 接口，以 `tools` 参数注入工具定义，配合 `enable_search=false` 等约束使用。该路径更灵活，但需自行处理协议细节与错误重试。 + +> ⚠️ 注意：函数调用能力**不跨模型通用**——必须选用明确支持该能力的模型（如 Omni Realtime 系列、qwen-plus/max、qwen3.7-plus 等），旧版模型（如 `qwen-turbo` 部分版本）或专用模型（如 `qwen-coder-turbo`）可能不支持。 ## 关键参数和配置 | 参数 | 类型 | 说明 | 必填 | 示例 | |------|------|------|------|------| -| `tools` | `array` | 工具定义列表，每个元素包含 `tool_id`（字符串）、`description`（功能描述）、`parameters`（JSON Schema，定义必选/可选字段及类型） | 是（启用函数调用时） | `[{"tool_id": "calculator", "description": "执行数学计算", "parameters": {"type": "object", "properties": {"expression": {"type": "string"}}, "required": ["expression"]}}]` | -| `tool_choice` | `string` 或 `object` | 控制调用策略：
`"auto"`（默认，模型自主决定）
`"none"`（禁用调用）
`{"type": "function", "function": {"name": "xxx"}}`（强制指定工具） | 否 | `"auto"` | -| `enable_search` | `boolean` | **仅 `qwen3.5-omni-realtime` 系列支持**，启用内置联网搜索（与 `tools` 互斥） | 否 | `true` | -| `result_format` | `string` | 必须设为 `"message"`（推荐），确保响应中包含 `tool_calls` 字段；设为 `"text"` 将丢失结构化调用信息 | 是（推荐） | `"message"` | +| `tools` | `array` | 工具定义列表，每个元素为 `{ "type": "function", "function": { "name", "description", "parameters" } }` | 是 | `[{ "type": "function", "function": { "name": "calculator", "description": "计算数学表达式", "parameters": { "type": "object", "properties": { "expression": { "type": "string" } }, "required": ["expression"] } } }]` | +| `tool_choice` | `string` 或 `object` | 控制调用策略：
• `"auto"`（默认，模型自主决定）
• `"none"`（禁用）
• `{"type": "function", "function": {"name": "xxx"}}`（强制调用指定函数） | 否（默认 `auto`） | `"auto"` 或 `{"type":"function","function":{"name":"quark_search"}}` | +| `tool_id`（插件专用） | `string` | 插件市场中工具的唯一标识符，用于控制台绑定或 API 引用 | 是（插件场景） | `"quark_search"`、`"code_interpreter"` | +| `input parameters` | `object` | 实际传入工具的参数对象，字段名与 `tools[].function.parameters.properties` 严格一致；Object 类型子属性**不可为空** | 是（调用时） | `{"expression": "sqrt(144) + 2 * 3"}` | +| `biz_params`（插件专用） | `object` | 业务系统透传参数（非 LLM 识别），需在插件配置中设为“业务透传”模式 | 否 | `{"user_id": "u123", "session_id": "s456"}` | -- **工具 ID 命名规范**：必须全局唯一、语义清晰（如 `quark_search`, `code_interpreter`），避免空格/特殊字符；官方插件 ID 可在控制台插件详情页复制。 -- **参数 Schema 要求**：`parameters` 必须为合法 JSON Schema，`required` 数组需准确声明必填字段；`Object` 类型参数仅支持 `POST` 请求，`GET` 请求中禁止使用。 -- **响应解析要点**：成功调用后，模型响应 `message` 中 `role` 为 `"assistant"`，`content` 为空或为中间思考，`tool_calls` 数组包含调用详情；后续需开发者自行执行工具并以 `tool_result` 角色提交结果，继续对话。 +- **Schema 规范**：`parameters` 必须符合 OpenAPI 3.0 的 JSON Schema 子集（支持 `string`/`number`/`boolean`/`object`/`array` 及 `required` 字段），嵌套 `object` 中所有属性均为必填。 +- **安全要求**：自定义工具 URL 必须为 HTTPS，响应头需含 `Access-Control-Allow-Origin: *` 或明确允许百炼域名；鉴权参数（如 `api_key`）应置于 Header（`Authorization: Bearer xxx`）或 Query（需在插件配置中声明参数名）。 +- **调用限制**：单次对话最多触发 10 次工具调用（含重复调用同一工具），总次数受应用配额约束。 ## 面向开发者，简洁实用 -- ✅ **快速验证**：用 `qwen3.7-plus` + `calculator` 工具，发送 `"123 * 456 = ?"`，观察是否返回 `tool_calls`。 -- ✅ **调试技巧**：若模型未触发调用，检查 `description` 是否足够清晰、`parameters.required` 是否遗漏关键字段、`messages` 中是否提供足够上下文。 -- ✅ **生产建议**： - - 对高可靠性场景（如金融计算），优先使用 `tongyi-intent-detect-v3` 做意图路由，再交由专用工具执行； - - 实时语音对话中，`qwen3.5-omni-plus-realtime` 支持流式 `tool_calls` 事件，可边听边规划； - - 自定义插件务必完成在线调试并发布为“已发布”状态，否则调用失败且错误码不直观（常见 `130040`）。 -- ❌ **避坑提醒**：`tools` 与 `enable_search` 不能同时启用；`qwen-long` 等超长上下文模型不支持该能力；流式响应（`stream=true`）中 `tool_calls` 仅在最终 chunk 返回，勿在中间 chunk 解析。 +- ✅ **快速验证**：优先选用 `qwen-plus` 或 `qwen3.7-plus` 模型 + [OpenAI 兼容接口](openai-compatible-interface.md)，用 `tools` + `tool_choice="auto"` 启动最小可行测试。 +- ✅ **调试技巧**：开启 `stream=false` 获取完整响应，检查 `choices[0].message.tool_calls` 是否存在；若无调用，检查 `tools` Schema 是否匹配用户 query 语义，或尝试 `tool_choice="required"` 强制触发。 +- ✅ **错误定位**：常见失败原因包括——`tools` 中 `name` 与实际工具 ID 不一致、`parameters` 字段缺失或类型错误、自定义工具返回非 JSON 或 HTTP 状态码非 200。 +- ✅ **生产建议**：工具执行超时应设为 ≤10s；回传结果需精简（避免大文本），关键字段用 `output parameters` 显式声明，便于模型提取摘要。 ## 关联主题页 -- [model experience](../guides/model-experience.md) - [omni realtime api](../api/omni-realtime-api.md) - [plug in](../guides/plug-in.md) -- [more models](../api/more-models.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [more about models](../api/more-about-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md index f41f9f1e..a2f84bc5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md @@ -1,48 +1,51 @@ # 长期记忆 -长期记忆是百炼平台提供的结构化、持久化上下文管理能力，用于跨会话、跨对话地存储和检索用户意图、偏好、事件、计划等语义化信息，突破大模型单次推理的上下文窗口限制，支撑个性化、连贯的智能体体验。 +长期记忆是百炼平台提供的结构化、持久化用户上下文管理能力，用于突破大模型单次会话的上下文窗口限制，实现跨会话、跨请求的用户偏好、关键事件与结构化属性的自动提取、语义检索与全生命周期管理。 -## 在百炼平台的不同场景中如何使用 +## 在百炼平台的不同场景中，这个概念如何使用 -- **智能体（Agent）应用**：当前新版智能体（Agent 2.0）**不原生支持长期记忆**，仅提供短期记忆（最多30轮对话历史）。如需长期记忆能力，需通过 SDK 或 API 主动调用 `AddMemory` / `SearchMemory`，在 Agent 工具链中集成记忆读写逻辑（例如：在 `system_prompt` 中提示“请先检索用户历史偏好”，再调用 `SearchMemory` 工具注入上下文）。 +长期记忆不是被动存储，而是主动参与智能体决策闭环的核心组件，其使用方式因应用范式而异： -- **工作流（Workflow）应用**：可通过节点间传递 `user_id`，在关键节点（如“初始化”或“响应生成”前）调用 `SearchMemory` 注入个性化上下文；也可在用户输入处理节点后调用 `AddMemory` 持久化新信息。配合会话变量（`historyList`）实现短期+长期双层记忆协同。 +- **智能体（Agent 2.0）应用**：作为“有状态智能体”的基石，通过 `memory_search` 等运行时工具在 `before_agent_start` 钩子中自动注入相关记忆，或由 Agent 主动调用 `AddMemory` / `SearchMemory` 实现个性化响应（如“您上周提到要学习 Python，需要我推荐课程吗？”）。OpenClaw [插件](plugin.md)可实现零代码接入，自动捕获对话中的事实并召回历史上下文。 -- **高代码应用**：完全由开发者自主控制。推荐在 Python 应用中使用 `agentscope-runtime` SDK 封装的异步工具类（`AddMemory`, `SearchMemory`, `ListMemory`, `DeleteMemory`），结合业务逻辑实现记忆生命周期管理（如注册回调、触发更新、设置过期策略）。 +- **工作流（Workflow）应用**：在节点编排中显式调用 `memory_store` 或 `memory_search` 工具，将流程中间结果写入长期记忆，或在条件分支前检索用户画像字段（如 `profile_schema="health_habits"`），驱动差异化路径执行。 -- **OpenClaw 等框架集成**：通过官方插件 `modelstudio-memory-for-openclaw` 开箱启用全自动机制：`autoCapture`（对话结束自动提取并写入）、`autoRecall`（对话开始前按 `user_id` 自动检索 Top-K 记忆），并暴露 `memory_search` / `memory_store` 工具供 Agent 主动调用。 +- **高代码应用**：通过 `agentscope-runtime` SDK（≥1.1.5）直接集成，以编程方式控制记忆生命周期。例如，在 Python 函数中解析用户上传的体检报告后，调用 `AddMemory(user_id=uid, custom_content=summary, meta_data={"category": "health"})`；后续查询时传入 `meta_data={"category": "health"}` 进行精准过滤。 -- **用户画像构建**：需预先调用 `CreateProfileSchema` 定义结构化字段（如 `"age": "整数，用户年龄"`），并在 `AddMemory` 请求中传入 `profile_schema` ID，平台将自动从对话中抽取并聚合属性，后续可通过 `GetUserProfile` 获取完整画像。 +- **RAG 增强场景**：与知识库检索正交协同——知识库提供通用领域知识，长期记忆提供专属用户事实（如“张三对青霉素过敏”），二者可在提示词中融合注入，显著提升回答准确性与个性化程度。 + +- **Managed Agents 沙箱环境**：虽沙箱本身无状态，但可通过 `SearchMemory` 在会话启动时拉取用户历史配置（如“默认导出格式为 CSV”），再写入沙箱文件系统，实现“有状态行为”的模拟。 ## 关键参数和配置 -| 参数名 | 类型 | 必填 | 说明 | -|--------|------|------|------| -| `user_id` | string | 是 | 记忆归属主键（≤64 字符），用于严格隔离不同用户数据空间，所有接口均需传入。 | -| `memory_library_id` | string | 否 | 目标记忆库 ID（≤32 字符）；不传则使用账号默认记忆库（不可删除）。 | -| `project_id` | string | 否 | 记忆片段提取规则 ID；不传则使用对应记忆库的默认规则（控制台可配置有效期：7/30/180 天或永不过期）。 | -| `profile_schema` | string | 否 | 用户画像 Schema ID；仅当需触发结构化属性抽取时必填。 | -| `messages` / `custom_content` | array / string | 二选一 | `messages`: 对话数组（最多50条），用于自动提取语义记忆；`custom_content`: 最多512字符纯文本，绕过提取直接写入。 | -| `meta_data` | object | 否 | 自定义键值对（如 `{"category": "preference", "source": "onboarding"}`），支持后续按字段过滤或业务标记。 | -| `top_k` | integer | 否（`SearchMemory` 默认10，OpenClaw插件默认5） | 检索返回的最大记忆条数（1–100）。 | -| `min_score` | double | 否（默认0.3，控制台推荐0.5–0.7） | 向量相似度阈值 [0,1]，低于此值的结果被过滤。 | -| `expire_time` | integer (Unix timestamp) | 否 | 秒级时间戳，显式指定记忆过期时间；优先级高于 `project_id` 规则中的默认有效期。 | +| 参数名 | 类型 | 必填 | 说明 | 推荐值 | +|--------|------|------|------|--------| +| `user_id` | string | 是 | 用户唯一标识符（≤64 字符），所有操作均以此为隔离边界，**不可为空或重复复用** | `"u_abc123"` | +| `messages` / `custom_content` | array / string | 互斥 | `messages`：自动提取（最多 50 条，一问一答计 2 条）；`custom_content`：直接写入文本（≤512 字符） | 优先用 `messages` 提升提取质量 | +| `memory_library_id` | string | 否 | 显式指定记忆库 ID（≤32 字符），未传则使用默认库；可在控制台“记忆库”页获取 | 生产环境建议显式指定 | +| `project_id` | string | 否 | 记忆片段规则 ID，控制提取逻辑（如仅提取提醒类内容）；未传则使用默认规则 | 规则需在控制台预先配置 | +| `profile_schema` | string | 否 | 用户画像模板 ID，触发结构化字段抽取（如 `{"name":"string","age":"integer"}`） | 模板需先调用 `CreateProfileSchema` 创建 | +| `top_k` | integer | 否 | `SearchMemory` 返回最大条数（1–100），影响召回精度与延迟 | 3–10（默认 10） | +| `min_score` | double | 否 | 相似度阈值（0.0–1.0），低于此值的结果被过滤；**统一使用浮点值，非整数** | 0.5–0.7（默认 0.3） | +| `meta_data` | object | 否 | 自定义元数据（≤1 KB），支持后续按字段过滤（如 `{"source": "wechat", "priority": "high"}`） | 用于分类、权限或业务路由 | -> ⚠️ 注意：`UpdateMemory` 仅更新内容与 `meta_data`，不改变向量索引时间点；`timestamp` 元字段为秒级 Unix 时间戳（非毫秒）。 +> ⚠️ 注意：`profile_schema` 参数实际为**可选**，传入无效 ID 将静默忽略，不报错；`expire_at` 字段或控制台规则配置决定记忆有效期（支持 7/30/180 天或永不过期），**平台不提供自动过期清理，需开发者主动调用 `DeleteMemory` 管理生命周期**。 -## 面向开发者的实用建议 +## 面向开发者，简洁实用 -- **首选 SDK**：安装 `pip install agentscope-runtime>=1.1.5`，直接使用 `AddMemory`, `SearchMemory`, `ListMemory`, `DeleteMemory` 异步工具类，避免手动构造 HTTP 请求与认证头。 -- **调试先行**：首次集成务必用 cURL 验证基础流程（如 `curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add -H "Authorization: Bearer $DASHSCOPE_API_KEY" -d '{"user_id":"u123","messages":[...]}')`，再迁移到 SDK。 -- **限流应对**：阿里云账号级总限流 3000 QPM（`AddMemory` ≤120 QPM，`SearchMemory` ≤300 QPM），超限返回 `429`，需实现指数退避重试。 -- **时效性管理**：长期记忆**无自动失效机制**，业务侧必须主动维护生命周期 —— 建议在关键业务节点（如用户注销、偏好变更）调用 `DeleteMemory` 或设置 `expire_time`。 -- **错误排查**：所有 API 响应含 `request_id`，是定位问题的关键标识；结合控制台「API 调用日志」与文档中的错误码表快速诊断。 +- **认证**：所有 API 请求必须携带 `Authorization: Bearer $DASHSCOPE_API_KEY` 和 `Content-Type: application/json`。 +- **SDK 优先**：安装 `agentscope-runtime>=1.1.5`，直接使用封装类（如 `AddMemory`, `SearchMemory`），避免手写 HTTP 请求。 +- **输入优化**：确保 `messages` 中 `role` 明确为 `"user"`/`"assistant"`，内容语义清晰（避免模糊指代），可显著提升自动提取准确率。 +- **检索技巧**：搜索时优先使用自然语言查询（`query` 字段），而非 `messages` 数组；若需高精度匹配，结合 `meta_data` 过滤 + `min_score=0.65`。 +- **限流应对**：账号级总 QPM ≤3000（`AddMemory` ≤120，`SearchMemory` ≤300），突发流量建议加本地缓存或队列削峰。 +- **调试建议**：首次集成时，先用 `ListMemory` 查看已写入内容，确认 `user_id` 和 `meta_data` 是否符合预期；再测试 `SearchMemory` 的召回效果。 ## 关联主题页 - [long term memory new](../api/long-term-memory-new.md) - [memory library overview](../guides/memory-library-overview.md) -- [llm application](../guides/llm-application.md) - [managed agents](../guides/managed-agents.md) +- [application support](../guides/application-support.md) +- [llm application](../guides/llm-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-context-protocol.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-context-protocol.md new file mode 100644 index 00000000..4bfa6cd9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/model-context-protocol.md @@ -0,0 +1,50 @@ +# 模型上下文协议（MCP） + +模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化能力接入协议，用于在大语言模型与外部工具服务之间建立安全、可互操作、声明式的信息通道。它基于开源 MCP 标准（[modelcontextprotocol.io](https://modelcontextprotocol.io/)）实现，并升级为 Streamable HTTP 协议，屏蔽底层通信细节，使开发者无需为每个工具单独编写适配逻辑即可集成能力。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +MCP 是百炼平台中**工具能力的统一抽象层**，不直接绑定模型，而是作为智能体和工作流的“能力底座”被调用： + +- **智能体应用（Agent）**：在智能体编辑页的「MCP 服务」区域一键添加最多 5 个已开通的 MCP 服务（如 `Amap Maps`、`WebSearch`）。大模型根据用户自然语言指令自动推理并动态调用，无需显式指定工具名或参数结构——提示词中建议明确提及服务名称（例如：“请调用 WebSearch MCP 获取最新 AI 会议信息”），以提升调用准确率。 + +- **工作流应用（Workflow）**：在画布中拖入「MCP 节点」，手动绑定一个具体 MCP 工具（如 `maps_weather`）。需前置一个大模型节点（如 `qwen-plus`）将用户输入解析为结构化参数（如 `{"city": "杭州"}`），再通过参数映射传递至 MCP 节点执行。适用于需精确控制调用时机、顺序与输入输出的编排场景。 + +- **Managed Agents（托管智能体）**：MCP 服务可作为 `tools` 显式注入到 Managed Agent 的运行环境中，与沙箱内 `bash`、`read` 等内置工具协同使用，支撑多步、有状态的复杂任务（如“先查天气，再规划路线，最后生成行程图”）。 + +> ⚠️ 注意：MCP **不支持**通过 DashScope SDK 直接调用千问 API（如 `dashscope.Generation.call`）时注入；也不支持在 Assistant API 的 `tools` 字段中以 OpenAI 格式传入 MCP 工具。它仅限百炼平台内原生智能体/工作流/Managed Agents 场景使用。 + +## 关键参数和配置 + +| 参数类别 | 字段/配置项 | 说明 | 开发者须知 | +|----------|-------------|------|-----------| +| **协议类型** | `type`（必填） | 取值为 `stdio` / `sse` / `streamableHttp`，必须与服务端点严格匹配：
• `sse` → 对应 `/sse` 端点
• `streamableHttp` → 对应 `/mcp` 端点
配置错误将返回 `11200058` 错误码 | 部署自定义 MCP Server 时，务必在服务配置中声明正确 `type`，并在百炼控制台选择对应协议 | +| **部署模式** | `基础模式` / `极速模式` | • 基础模式：按调用时长计费（0.000156 元/秒），无部署费
• 极速模式：额外收取部署费（0.000036 元/秒），适合高频、低延迟场景 | 高频调用推荐极速模式；首次调试建议用基础模式快速验证 | +| **安全凭证** | KMS 加密凭据 | 敏感参数（如 `AMAP_MAPS_API_KEY`、`FIRECRAWL_API_KEY`）**必须**通过百炼控制台的 KMS 凭据管理功能加密后引用，禁止明文填写 | 控制台配置 MCP 服务时，所有带锁图标（🔒）的字段均需关联 KMS 凭据 | +| **外部调用地址** | `mcp_url` | 格式为 `https://dashscope.aliyuncs.com/api/v1/mcps/{service_name}/mcp`（如 `WebSearch`）
外部 SDK 集成时需配合 `DASHSCOPE_API_KEY` 使用 | 外部调用仅支持 `streamableHttp` 协议；`/mcp` 后缀不可省略 | + +## 面向开发者，简洁实用 + +- ✅ **开通即用**：前往 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market)，点击「立即开通」→ 自动完成服务注册与权限授权。 +- ✅ **自定义三路径**： + - 快速本地试跑：`npx mcp-server-stdio` 或 `uvx mcp-server-sse` 启动标准 MCP Server； + - 封装现有 API：通过「AI 网关」导入 RESTful 接口，自动转换为 MCP 工具； + - 对接云产品：从阿里云 OpenAPI 导入（如 OSS 文件上传、ECS 实例管理），一键暴露为 MCP 工具。 +- ✅ **调试技巧**： + - 提示词中避免模糊指令（❌“帮我查一下” → ✅“调用 WebSearch MCP 搜索‘2024 Qwen 最新论文’”）； + - 工作流中 MCP 节点的输入参数，优先使用上游节点的 `output.xxx` 引用，而非硬编码； + - 自定义 MCP Server 日志需输出到 `stdout`，便于函数计算（FC）日志排查。 +- ❌ **禁止事项**： + - 不得访问本地文件、硬件设备或未打通网络的私有数据库； + - 自定义服务托管于 FC，无固定出口 IP，访问云数据库等资源需配置白名单或 VPC； + - 不支持在非百炼平台环境（如本地 Python 脚本直连千问 API）中启用 MCP。 + +## 关联主题页 + +- [model context protocol](../guides/model-context-protocol.md) +- [plug in](../guides/plug-in.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [managed agents](../guides/managed-agents.md) +- [application support](../guides/application-support.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md deleted file mode 100644 index 0af3e708..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md +++ /dev/null @@ -1,80 +0,0 @@ -# 多模态能力 - -多模态能力指百炼平台模型对文本、图像、视频、音频等多种数据类型进行联合理解、生成与推理的能力，支持跨模态信息融合与协同处理，是构建智能体、内容创作、工业质检等复杂AI应用的核心基础。 - -## 在百炼平台的不同场景中，这个概念如何使用 - -多模态能力并非单一模型特性，而是贯穿多个能力域的底层技术范式，在以下典型场景中体现为具体能力组合： - -- **视觉理解与推理**：`qwen3.7-plus`、`qwen3.5-omni-plus` 等全模态大模型可同时接收文本指令 + 多张图像/视频片段，执行OCR识别、图表解析、视频事件定位、结构化输出（如JSON）及Function Calling（例如调用天气API后结合截图分析出行建议）。单请求最多支持2048张图片或64段视频，输入按统一Token规则计费（图像Token ≈ h×w/(32×32)+2）。 - -- **音视频端到端处理**：`qwen3.5-omni-plus-realtime` 支持语音输入→文本理解→工具调用→语音合成全流程，无需拆解ASR/TTS模块；S2S模型可直接处理带背景音的语音流，并在响应中保留语调、节奏等声学特征。 - -- **生成类多模态协同**： - - 图像生成中，`wan2.7-image-pro` 接收文本+参考图+风格图三重输入，实现精准可控的图文混合生成； - - 视频生成中，`wan2.7-t2v-2026-06-12` 支持“[prompt](../guides/prompt.md) + 自定义音频文件注入”，实现音画同步生成； - - 3D生成中，`Tripo/Tripo-H3.1` 允许文生3D、单图生3D或多图（前/左/后/右）联合重建，输入模态决定几何重建精度与纹理生成策略。 - -- **跨模态检索与重排序**：`qwen3-rerank` 可对图文混合结果集（如含标题、缩略图、描述的搜索项）进行联合语义重排序；`text-embedding-v4` 虽为文本模型，但其向量空间经多模态对齐训练，可与图像Embedding模型（如`qwen-vl-embed`）共用相似度计算，支撑跨模态检索。 - -> ⚠️ 注意：并非所有模型均具备完整多模态能力。例如 `qwen-long`（10M上下文）专注长文本处理，**不支持图像/视频输入**；`qwen3.7-max` **不支持结构化输出与Function Calling**，即使输入含多模态数据，也仅作单向理解，无法触发工具链。选型时请以[model experience](model-experience.md)中各模型的能力矩阵为准。 - -## 关键参数和配置 - -多模态能力的启用与控制依赖以下关键参数（均置于`parameters`对象内，部分需配合特定请求头）： - -| 参数 | 类型 | 说明 | 典型值/约束 | 适用模型示例 | -|------|------|------|-------------|--------------| -| `max_image_count` / `max_video_count` | integer | 单请求最大媒体数量 | `qwen3.7-plus`: 2048 / 64；`qwen3.5-omni-plus`: 256 / 512 | 全模态理解模型 | -| `enable_thinking` | boolean | 启用分步推理模式（Chain-of-Thought），提升复杂多模态任务准确率 | `true` / `false`（默认`false`） | `qwen3.7-plus`, `qwen3.6-flash` | -| `tool_choice` | string / object | 控制工具调用策略（自动/指定/禁用），影响多模态意图识别后的动作决策 | `"auto"`, `"required"`, `{"type": "function", "function": {"name": "get_weather"}}` | 支持Function Calling的全模态模型 | -| `X-DashScope-Async` | request header | **强制异步调用标识**，所有耗时多模态生成任务（视频/3D）必须设置为`"enable"` | `"enable"`（必填） | `happyhorse-1.1-t2v`, `Tripo/Tripo-H3.1` | -| `audio` | boolean | 视频生成中是否启用音频轨道合成 | `true`（默认）/ `false` | `wan2.7-t2v-2026-06-12`, `happyhorse-1.1-t2v` | -| `texture_quality` / `geometry_quality` | string | 3D生成中贴图精细度与网格面数控制，直接影响多图输入的重建保真度 | `"standard"` / `"detailed"`；`"standard"` / `"ultra"` | `Tripo/Tripo-H3.1` | - -> ✅ 实用提示： -> - 多模态输入必须通过`input`字段统一组织（非`messages`），格式为`{"messages": [...]}`（文本为主）或`{"media": [...], "prompt": "..."}`（媒体为主）； -> - 混合输入时，**图像/视频URL必须为公网可访问的HTTPS链接**，且需提前校验可用性（3D生成要求JPEG/PNG，分辨率20–6000像素）； -> - 所有异步多模态任务（视频/3D）的`task_id`有效期严格为24小时，结果URL（如`pbr_model_url`）仅保留2小时，请及时下载。 - -## 面向开发者，简洁实用 - -- **快速验证**：用`curl`测试`qwen3.7-plus`的图文理解能力： - ```bash - curl -X POST https://dashscope.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "qwen3.7-plus", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"type": "text", "text": "这张图里有哪些商品？价格分别是多少？"}, - {"type": "image_url", "image_url": {"url": "https://example.com/product.jpg"}} - ] - } - ] - }, - "parameters": {"max_output_tokens": 1024} - }' - ``` - -- **避坑指南**： - - ❌ 不要对`qwen-long`传入图片——会返回`400 Bad Request`； - - ❌ 不要省略`X-DashScope-Async: enable`调用视频/3D接口——会报错`current user api does not support synchronous calls`； - - ✅ 优先使用业务空间专属域名（如`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），多模态请求延迟降低30%+； - - ✅ 生产环境务必显式设置`max_output_tokens`，避免长输出导致超时或计费激增。 - -多模态能力的本质是“统一接口、混合输入、协同输出”。开发者只需关注业务需求的数据组合（文+图？音+视？文+图+3D？），平台将自动调度最适配的模型与计算资源——你负责定义“做什么”，我们负责实现“怎么做”。 - -## 关联主题页 - -- [model experience](../guides/model-experience.md) -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) -- [qwen api reference](../api/qwen-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md deleted file mode 100644 index e6a43738..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md +++ /dev/null @@ -1,82 +0,0 @@ -# OpenAI 兼容接口 - -OpenAI 兼容接口是百炼平台提供的一组标准化 API 协议层，严格遵循 OpenAI REST API 的路径、请求/响应结构、参数命名与语义规范（如 `/v1/chat/completions`），使开发者能直接复用 OpenAI SDK（如 `openai>=1.0.0`）或现有代码逻辑调用千问（Qwen）及第三方模型，实现零改造迁移。 - -## 在百炼平台的不同场景中，这个概念如何使用 - -OpenAI 兼容接口不是单一接口，而是一套按能力分层的协议集合，覆盖多种模型类型与任务场景： - -- **通用对话生成**：通过 `Chat Completions` 接口（`POST /compatible-mode/v1/chat/completions`）调用 `qwen3.7-plus`、`deepseek-r1`、`kimi-k2.7-code` 等文本/代码模型，支持标准 `messages` 格式、流式响应（`stream=true`）和基础采样参数（`temperature`, `top_p`, `max_tokens`）。 - -- **智能体增强能力**：`Responses API`（同路径但启用特定模型如 `qwen3.7-max`）在兼容基础上内置联网搜索、网页提取、代码解释器等工具链，自动处理工具调用循环，无需客户端手动解析 `tool_calls` 并重发 `tool_result`。 - -- **多模态理解**：`Vision` 接口（`/v1/chat/completions`）支持 OpenAI 格式的 `image_url` 输入，兼容 `qwen-vl-plus`、`qwen3-vl-plus` 等视觉语言模型，可混合文本与图像消息。 - -- **向量化与排序**：`Embedding`（`/v1/embeddings`）和 `Rerank`（`/v1/rerank`）接口完全对齐 OpenAI Embedding/Rerank 规范，支持 `dimensions`、`input` 数组、`top_n` 等关键参数，无缝接入 RAG 流水线。 - -- **批量异步处理**：`Batch` 接口（`/v1/batch`）提供 OpenAI 风格的异步提交能力，适用于高吞吐文本处理，单次支持 256K tokens 上下文。 - -- **会话状态管理**：`Conversations` 接口（`/v1/conversations`）配合 Responses 使用，实现跨请求的上下文持久化，避免手动拼接 `messages`。 - -> ⚠️ 注意：并非所有百炼模型都支持 OpenAI 协议——例如 `qwen-audio`、`wan2.6-t2i`（文生图）仅提供 DashScope 原生接口；Qwen3 系列部分高级能力（如 `enable_thinking`）需通过原生 SDK 或 `extra_body` 显式传递，OpenAI 兼容接口默认不透传。 - -## 关键参数和配置 - -| 参数 | 类型 | 说明 | 必填 | -|------|------|------|------| -| `model` | string | 模型 ID，如 `qwen3.7-plus`、`text-embedding-v4`、`qwen3-rerank`。**必须与 Base URL 所在地域匹配**（如 `qwen3.7-plus-us` 仅限美国地域）。 | 是 | -| `base_url` | string | 服务端点，**必须使用业务空间专属域名**以保障性能与稳定性：
• 北京：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
• 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`
• 弗吉尼亚：`https://dashscope-us.aliyuncs.com/compatible-mode/v1`
• 法兰克福/东京：同北京格式，替换对应地域代码 | 是 | -| `api_key` | string | 百炼 API Key，**严格按地域与计费方案隔离**（Token Plan、Coding Plan、按量计费 Key 不互通）。需通过 [API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建并匹配 `base_url` 地域。 | 是 | -| `messages` | array | 对话消息列表，格式为 `[{"role": "user", "content": "..."}]`。`Responses API` 可自动注入历史，但生产环境建议显式传入以确保可控性。 | Chat Completions / Responses / Vision 等需对话场景下必填 | -| `stream` | boolean | 启用流式响应（`true`）。返回 `data: {...}` SSE 格式，每 chunk 含 `delta.content`。 | 否 | -| `stream_options` | object | 当 `stream=true` 时，设 `{"include_usage": true}` 可在末尾 chunk 返回 token 统计（`usage` 字段）。 | 否 | -| `temperature` / `top_p` | float | 控制生成随机性，二者互斥。`temperature ∈ [0, 2.0)`，`top_p ∈ (0, 1.0]`。 | 否 | -| `max_tokens` | integer | 最大输出 token 数，影响响应长度与计费。不同模型有硬上限（如 `qwen3.7-plus` 支持 32K 上下文，实际可用受系统 [prompt](../guides/prompt.md) 占用）。 | 否 | - -## 面向开发者，简洁实用 - -- ✅ **快速上手**：只需三步——获取 API Key → 构造 `base_url`（含 WorkspaceId）→ 用 OpenAI SDK 调用，无需修改业务代码。 -- ✅ **灵活切换**：同一套 SDK 可无缝切换 Qwen、DeepSeek、Kimi 等模型，仅需改 `model` 参数。 -- ✅ **生产就绪**：推荐使用业务空间专属 `base_url`（而非公共域名），获得更低延迟、更高并发与独立配额。 -- ❌ **避坑提示**： - - 不要混用地域 Key 与 URL（如北京 Key + 美国 URL → 401）； - - `qwen-audio`、`wan2.6-t2i` 等模型**不支持** OpenAI 协议，请查文档确认模型兼容性； - - 工具调用（`tools`）在 Chat Completions 中需客户端自行解析并重发，`Responses API` 才支持自动执行； - - `enable_search`、`enable_thinking` 等 Qwen 特有参数**不在 OpenAI 协议内**，需通过 DashScope SDK 或 `extra_body` 传递。 - -示例（Python）： -```python -from openai import OpenAI - -client = OpenAI( - api_key="sk-xxx", # 替换为你的百炼 API Key - base_url="https://your-workspace-id.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" -) - -# 标准对话 -resp = client.chat.completions.create( - model="qwen3.7-plus", - messages=[{"role": "user", "content": "用 Python 写一个快速排序"}] -) -print(resp.choices[0].message.content) - -# 流式响应 -for chunk in client.chat.completions.create( - model="qwen3.7-plus", - messages=[{"role": "user", "content": "讲个笑话"}], - stream=True, - stream_options={"include_usage": True} -): - if chunk.choices[0].delta.content: - print(chunk.choices[0].delta.content, end="") -``` - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [get started with models](../guides/get-started-with-models.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [vector and sort](../api/vector-and-sort.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md new file mode 100644 index 00000000..666ae79b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md @@ -0,0 +1,74 @@ +# OpenAI 兼容接口 + +OpenAI 兼容接口是百炼平台提供的一组标准化 REST API，严格遵循 OpenAI 的请求/响应协议（如 `/v1/chat/completions`、`/v1/embeddings` 等路径），支持使用标准 `openai` SDK（Python、Node.js 等）或通用 HTTP 客户端快速接入，无需修改业务逻辑即可调用百炼托管的多种大模型与能力。 + +## 在百炼平台的不同场景中如何使用 + +OpenAI 兼容接口不是单一接口，而是一套按能力分层、按场景隔离的协议族，开发者需根据目标功能选择对应子接口： + +- **通用对话（Chat Completions）**：适用于多轮文本交互，支持 `qwen-plus`、`qwen3.7-plus`、`Qwen-VL`、`DeepSeek`、`Kimi`、`GLM` 等数十个模型；不支持 `Qwen-Audio`。 +- **智能体响应（Responses API）**：面向 Agent 场景的增强型接口，内置联网搜索、网页抓取等工具调用能力，仅支持 `qwen3-*` 系列模型（如 `qwen3.7-plus`），**不兼容旧版 `qwen-coder-turbo` 等模型**。 +- **代码补全（Completions）**：专用于代码生成与补全，当前**仅支持 `qwen-coder-turbo`**，且仅限华北2（北京）地域。 +- **多模态理解（Vision）**：支持图像输入的 `Qwen-VL`、`QVQ`、`Qwen-OCR`，其中 `QVQ` 仅支持[流式输出](streaming-output.md)。 +- **文本向量化（Embedding）**：支持 `text-embedding-v1` 至 `v4`，但**多模态 Embedding 模型（如 `qwen3-vl-embedding`）不兼容该协议**。 +- **文件管理（Files）**：用于文档上传、批量推理（Batch）、模型微调（Fine-tune）等，单文件上限依用途为 150 MB / 500 MB / 300 MB。 +- **异步批量处理（Batch）**：含两种模式： + - *文件输入*（JSONL 格式）：适用于大规模任务，费用为实时调用的 50%； + - *Batch Chat*（同步阻塞）：保持实时 API 调用习惯，同样享 5 折优惠，**必须使用专用域名 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`**。 +- **会话管理（Conversations）**：配合 Responses API 实现跨设备上下文延续，支持创建、查询、更新、删除会话及添加消息项。 +- **应用调用（Application Call）**：通过 `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` 同步或异步调用已发布的智能体/工作流应用，支持 `stream=true` [流式输出](streaming-output.md)或 `background=true` 异步触发。 + +> ⚠️ 注意：不同子接口的 `base_url`、地域、API Key 和模型支持范围均严格隔离，混用将导致 401 或 404 错误。 + +## 关键参数和配置 + +所有 OpenAI 兼容接口共用以下核心配置项，必须正确设置： + +| 参数 | 说明 | 必填 | 示例值 | +|------|------|------|--------| +| `base_url` | 接口根地址，**按子接口和地域严格区分**：
• Chat/Responses/Vision/Embedding/Conversations：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（新加坡）
• Files/Batch（文件输入）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（中国内地）
• Batch Chat：**必须为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`** | ✅ | `https://dashscope.aliyuncs.com/compatible-mode/v1` | +| `api_key` | DashScope API Key，**必须与 `base_url` 所属地域一致**（如北京 Key 不可用于新加坡 endpoint） | ✅ | `sk-xxx` | +| `model` | 模型 ID，**必须从对应子接口的支持列表中精确选取**（大小写敏感，不可拼写错误） | ✅ | `"qwen3.7-plus"`、`"text-embedding-v4"`、`"qwen-coder-turbo"` | +| `stream` | 布尔值，控制是否启用流式响应（`true`/`false`），部分接口（如 Batch Chat）不支持 | ❌ | `true` | + +常用请求体参数（以 `/v1/chat/completions` 为例）： +- `messages`: OpenAI 标准格式数组，如 `[{"role":"user","content":"你好"}]` +- `temperature`: 控制随机性（0.0–2.0），默认 `0.7` +- `top_p`: 核采样阈值（0.0–1.0），默认 `1.0` +- `max_tokens`: 最大生成 token 数，建议显式设置以防超限 +- `tools` / `tool_choice`: 仅 Responses API 及部分模型支持，用于[函数调用](function-calling.md) + +## 面向开发者：简洁实用指南 + +1. **选对接口**：先明确需求——是通用对话？还是调用智能体？需要嵌入向量？还是批量处理？再查对应子接口文档，确认模型支持与地域限制。 +2. **配对三要素**：`base_url` + `api_key` + `model` 必须同地域、同协议、同能力域，缺一不可。 +3. **用标准 SDK**：推荐 `openai==1.40.0+`（Python）或 `@openai/openai-node`（Node.js），只需设置 `base_url` 和 `api_key`，其余调用方式与 OpenAI 完全一致： + ```python + from openai import OpenAI + client = OpenAI( + api_key="sk-xxx", + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1" + ) + response = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你好"}] + ) + print(response.choices[0].message.content) + ``` +4. **避坑提示**： + - `qwen-vl` 图像输入仅支持 base64 或公网 URL，不支持本地路径； + - 所有接口**不返回 `delta.tool_calls`**（仅返回完整 `tool_calls`），需按 `finish_reason: "tool_calls"` 解析； + - `stream=true` 时，`usage` 字段仅在末尾 chunk 中出现； + - 工作流应用调用**仅支持华北2（北京）地域**，智能体应用建议保持地域一致。 + +如遇 401（认证失败）、404（模型不支持）或 422（参数错误），请优先核对 `base_url` 域名、`api_key` 地域归属、`model` 名称拼写及子接口适用范围。 + +## 关联主题页 + +- [qwen api reference](../api/qwen-api-reference.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [application call](../api/application-call.md) +- [bailian application calling](../guides/bailian-application-calling.md) +- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md b/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md index 2800f69d..fddfdf6d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md @@ -1,56 +1,57 @@ -# 插件机制 +# 插件 -插件机制是百炼平台提供的核心能力扩展框架，通过标准化接口将外部工具（API、服务或计算能力）安全、可控地集成到大模型推理链路中，使模型在保持语言理解能力的同时，具备实时搜索、代码执行、图像生成、专业计算等超越纯文本推理的增强能力。 +插件是百炼平台用于扩展大模型能力的核心机制，通过将外部工具（如 HTTP API）集成到推理链路中，弥补大模型在实时搜索、精确计算、代码执行、图像生成等场景的固有局限。它以“工具”为最小可调用单元，支持官方预置、三方市场及完全自定义三种来源，既可由大模型自主规划调用，也可在工作流中显式编排执行。 -## 在百炼平台的不同场景中，这个概念如何使用 +## 在百炼平台的不同场景中如何使用 -插件机制并非单一技术实现，而是贯穿多个能力层的统一抽象，其具体形态和使用方式因场景而异： +- **智能体应用（Agent 2.0）**：在控制台「应用编排」→「MCP 区块」中添加已发布的插件（或其转换的 MCP 服务）。官方插件仅限与同业务空间内的智能体关联；自定义插件需先发布为 MCP 服务。大模型根据用户输入自动识别意图、选择工具并组织参数，完成“思考-执行-反思”闭环。 + +- **工作流应用（Workflow）**：将插件作为独立节点拖入画布，与其他节点（如大模型、条件判断）连接。执行顺序和输入参数由开发者显式编排，不依赖模型决策，适用于流程确定、结果可控的自动化任务（如订单状态查询+通知发送）。 -- **智能体应用（Agent）**：插件作为“可调用工具”被注入模型上下文。大模型基于用户输入语义自主规划是否调用、调用哪个插件及传入参数（如 `calculator` 计算 `237 × 48`），整个过程无需开发者编写调度逻辑。官方插件（如 `quark_search`）、三方插件（云市场服务）和自定义插件均可在此模式下启用。 +- **Assistant API 调用**：在请求 payload 的 `tools` 字段中声明工具列表（含 `type: "function"`、`function.name`、`function.description` 和 `function.parameters`），并通过 `tool_choice` 控制策略（如 `"auto"` 或指定 `{"type": "function", "function": {"name": "calculator"}}`）。这是最轻量、最灵活的集成方式，适合已有 OpenAI 兼容架构的快速迁移。 -- **工作流应用（Workflow）**：插件以显式节点形式存在，开发者手动拖拽、配置输入/输出连接与参数映射（如将上一节点提取的地址传给 `amap_weather` 工具）。此时插件不依赖模型决策，适用于确定性、多步骤、需精确控制的业务编排。 +- **Managed Agents（托管智能体）**：插件需以 MCP 协议服务形式接入，作为沙箱内可调用的外部能力。适用于需长时运行、多步交互、文件读写与命令执行的复杂任务（如数据分析报告生成），工具调用与沙箱环境深度协同。 -- **Assistant API 调用**：通过 `tools` 数组声明插件 ID 及结构化描述（含 `name`、`description`、`parameters` JSON Schema），由 SDK 或平台自动完成 function calling 的请求构造、响应解析与结果注入，实现与 OpenAI 兼容的工具调用范式。 +- **高代码应用**：通过 SDK 或 MCP Client 直接调用已注册插件，支持在 Python 逻辑中混合模型推理与工具调用，实现高度定制化业务编排（如风控规则引擎 + 实时征信 API 调用）。 -- **MCP（Model Context Protocol）服务**：作为插件机制的协议升级形态，MCP 提供更严格的标准化通信契约（如 `streamableHttp` 协议、JSON Schema 输入校验、KMS 加密凭据管理），支持跨平台工具复用与统一治理。所有 MCP 服务（包括官方 Amap Maps、WebSearch 及自定义部署服务）在百炼中均以“插件”身份被发现、授权和调用。 - -> ⚠️ 注意：`Skill`（技能）虽常被类比为“插件”，但其本质不同——Skill 是预打包的、无网络调用的本地计算能力（如 CSV 清洗、PDF 解析），由百炼调度引擎基于 `description` 语义匹配触发，不涉及 HTTP 请求或外部服务授权，因此**不属于插件机制范畴**。 +> ⚠️ 注意：所有插件调用均要求主账号或 RAM 子账号已授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`；RAM 用户还需额外授予 `ram:CreateServiceLinkedRole` 权限。 ## 关键参数和配置 -以下参数在自定义插件或 MCP 服务配置中必须准确设置，直接影响调用成功率与安全性： - -| 参数 | 说明 | 开发提示 | -|------|------|----------| -| **`tool_id`（工具 ID）** | 插件内唯一标识具体工具的字符串（如 `text_to_image`），用于 Assistant API 的 `tools` 声明或工作流节点选择。可在控制台插件详情页复制。 | 避免使用空格、中文或特殊符号；同一插件内不可重复。 | -| **`url` + `path`** | 插件服务根地址（如 `https://api.example.com`）与工具路径（如 `/v1/generate`），拼接后构成完整调用 URL。MCP 中对应 `url` 字段（`type=streamableHttp` 时必填）。 | 确保 URL 可公网访问且 HTTPS 启用；路径区分大小写。 | -| **`inputSchema`（JSON Schema）** | 定义输入参数结构，直接影响模型参数提取准确性。必须包含 `type`、`properties`，推荐使用 `required` 明确必填项。 | 示例：`{"type":"object","properties":{"prompt":{"type":"string"}},"required":["prompt"]}` | -| **鉴权方式** | 支持 `Header`（如 `Authorization: Bearer `）或 `Query`（如 `?api_key=xxx`）；MCP 强制要求 `Authorization` header 为 `Bearer `。 | 敏感密钥（如地图 API Key）**必须通过 KMS 加密存储**，禁止明文配置。 | -| **`output_parameters`（输出字段）** | 指定 API 返回 JSON 中哪些顶层字段供模型读取（如 `{"image_url": "string"}`），需扁平、非嵌套、非空。 | 避免返回大体积二进制或原始 HTML；仅保留模型生成回复所需的最小数据集。 | +| 参数 | 必填 | 说明 | 示例值 | +|------|------|------|--------| +| `tool_id` | 是 | 工具唯一标识符，用于模型识别与路由 | `"quark_search"`, `"text_to_image"` | +| `input.parameters` | 是（自定义插件） | 输入参数结构，需严格匹配 API 契约：
• 类型必须明确（`string`/`number`/`object`）
• `object` 类型子字段**不可为空**
• 鉴权参数若在 Header，需指定 `Type`（如 `"bearer"`）；若在 Query，需在插件配置中声明参数名 | `{ "query": "杭州天气", "region": "hangzhou" }` | +| `input.pass_mode` | 是 | 传参方式：
• `"llm_recognition"`：由大模型从用户 query 中抽取
• `"biz_pass_through"`：由上游系统通过 `biz_params` 主动注入 | `"llm_recognition"` | +| `output.parameters` | 是 | 输出字段定义，所有字段均为必填，描述需精简准确，便于模型提取关键信息 | `[{"name": "result", "description": "搜索摘要", "type": "string"}]` | +| `plugins`（API 请求） | 否（但启用插件时需） | Assistant API 中显式启用的插件 ID 列表 | `["calculator", "generate_qrcode"]` | -## 面向开发者，简洁实用 +- **URL 与协议要求（自定义插件）**： + - 必须为 HTTPS 协议； + - 响应头需包含 `Access-Control-Allow-Origin: *` 或明确允许百炼域名； + - 工具路径必须以 `/` 开头（如 `/v1/search`），与插件基础 URL 拼接后构成合法完整地址。 -- **快速起步**：优先使用控制台「插件市场」添加官方插件（如 `code_interpreter`），无需配置即可在智能体中测试；确认 `AliyunServiceRoleForSFMAccessCloudAPI` 角色已授权（主账号一键授权，RAM 用户需 `ram:CreateServiceLinkedRole` 权限）。 +- **模型兼容性**：仅以下模型支持插件调用： + `qwen-turbo`, `qwen-plus`, `qwen-max`, `qwen-vl-max`, `qwen-vl-plus`。 + 推荐优先选用 `qwen-plus` 或 `qwen-max` 进行开发验证。 -- **自定义插件上线三步**： - 1. 在控制台创建插件 → 填写 `tool_id`、`url`、`path`； - 2. 定义 `inputSchema` 和 `output_parameters`（用 JSON Schema 校验器验证）； - 3. **在线调试通过并发布为“已发布”状态**（草稿/未启用 = 调用失败）。 +## 面向开发者：简洁实用提示 -- **避坑指南**： - - 所有插件调用均**只透传 `Authorization` header**，其他自定义 header 会被平台剥离； - - `Object` 类型输入参数在 `GET` 请求中不被支持（仅 `POST` 允许）； - - 智能体最多添加 10 个工具（含 Skill），MCP 服务上限为 5 个； - - 实际模型兼容性以控制台运行结果为准，文档列表可能滞后（如 `qwen2.5` 系列需实测）。 +- ✅ **快速起步**：直接使用官方插件（如 `calculator`、`text_to_image`），无需配置，控制台一键启用即可测试。 +- ✅ **调试必做**：自定义插件发布前，务必使用控制台「在线调试」功能验证连通性、参数解析与响应格式。 +- ✅ **参数安全**：`object` 类型输入中，所有子字段必须提供默认值或明确标记 `required`；空字段将触发错误码 `130022`。 +- ✅ **鉴权简化**：仅支持透传 `Authorization` header；其他自定义 header 将被忽略，请勿依赖。 +- ❌ **禁止行为**：`code_interpreter` 插件禁用网络访问（`requests` 不可用）和本地文件上传；`quark_search` / `github_search` 仅返回元信息，不抓取网页正文或源码。 +- 📉 **调用限制**：单次对话最多调用 10 个工具（含重复调用），且受应用配额约束；删除插件将级联删除其下所有工具，已关联应用立即失效。 -- **调试技巧**：开启 `stream=True` 查看模型思考过程（含 tool call 步骤）；检查返回错误码（如 `130040` = 参数描述缺失，`11200054` = MCP 协议解析失败），对照文档定位问题。 +> 提示：插件本质是“可被语言模型理解并调度的标准化 API”。设计时请遵循 OpenAPI 3.0 规范，用清晰的 `description` 和最小必要参数降低模型幻觉风险。 ## 关联主题页 - [plug in](../guides/plug-in.md) -- [skill](../guides/skill.md) -- [model context protocol](../guides/model-context-protocol.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) +- [managed agents](../guides/managed-agents.md) +- [llm application](../guides/llm-application.md) - [application support](../guides/application-support.md) -- [more about models](../api/more-about-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md index 2a278f6b..ed8ee2d3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md @@ -1,49 +1,40 @@ # 检索增强生成 -检索增强生成（Retrieval-Augmented Generation，RAG）是一种将大语言模型（LLM）的生成能力与外部知识源的精准检索能力相结合的技术范式。它通过在模型推理前动态召回相关上下文片段，并将其注入提示词（Prompt），使模型能在私有、领域专属或时效性强的知识基础上生成更准确、可溯源、抗幻觉的回答。 +检索增强生成（Retrieval-Augmented Generation，RAG）是一种将大语言模型（LLM）的生成能力与外部知识源的精准检索能力相结合的技术范式。它通过在模型推理前动态检索相关文档片段，并将其作为上下文注入提示（[prompt](../guides/prompt.md)），显著提升回答的事实准确性、领域专业性与可控性，避免幻觉，是百炼平台支撑私有知识问答、智能客服、企业知识助手等生产级应用的核心技术底座。 ## 在百炼平台的不同场景中，这个概念如何使用 -在百炼平台中，RAG 不是单一接口，而是贯穿多个能力层的协同工作模式，核心围绕**知识库**这一基础设施展开，支持三种主流集成路径： +在百炼平台中，“检索增强生成”并非单一接口，而是贯穿多个能力层级的横切架构模式，开发者可根据需求选择不同抽象层级的实现方式： -- **应用内嵌 RAG（推荐用于生产级智能体/工作流）** - 在智能体或工作流应用配置页中直接绑定已创建的知识库，设置“相似度阈值”“权重”和调用策略（如“必定调用”）。工作流中拖入“知识库节点”，配置 `TopK` 和输入变量（如 `query`），再连接至大模型节点；模型提示词中通过 `{result}` 引用召回内容。该方式支持多轮对话改写、Agentic 规划搜索及引用溯源。 +- **基础检索层（Knowledge API）**：直接调用 `/api/v1/indices/knowledge/search` 接口，执行跨知识库的语义检索，返回原始文本切片（chunk）。适用于需完全自控 RAG 流程（如自定义 Query 改写、混合检索、多路召回融合）的高级场景。 +- **端到端问答层（Knowledge QA）**：调用 `/api/v2/apps/knowledge/chat` 接口，平台自动完成“检索 → 上下文组装 → 大模型生成 → 引用标注”全流程，支持流式 SSE 响应与三阶段（规划→工具调用→生成）输出，适合快速构建生产就绪的问答服务。 +- **应用集成层（Application + Knowledge Base）**：在智能体或工作流应用中绑定已发布的知识库，并配置“必定调用”或“按需调用”策略；RAG 行为由 `application support` 统一调度，与[插件](plugin.md)、[流式输出](streaming-output.md)等能力无缝协同。 +- **框架集成层（LlamaIndex / Spring AI Alibaba）**：通过 SDK 封装复用百炼云端知识库与模型能力。例如 LlamaIndex 使用 `DashScopeCloudIndex` 构建索引，Spring AI Alibaba 使用 `DashScopeDocumentRetriever` 直接检索，均无需管理向量存储与嵌入模型。 +- **低代码渠道层（AppFlow）**：在网站、企业微信、钉钉等渠道嵌入 AI 助手时，后台自动启用 RAG 能力——只需上传文档、创建知识库并绑定至应用，即可零代码启用私有知识增强。 -- **独立服务形态（适合快速验证与 API 集成）** - 通过控制台“知识检索”或“知识问答”标签页发布统一服务：可跨最多 15 个知识库联合检索，配置混合检索（向量+关键词）、Rerank 模型（如 `qwen3-rerank`）、拒答与防泄漏策略。发布后通过标准化 HTTP API 调用： - - 检索：`POST /api/v1/indices/knowledge/search` → 返回结构化切片； - - 问答：`POST /api/v2/apps/knowledge/chat` → 默认 SSE 流式响应，含 planning、tool calling、generation 三阶段输出。 - -- **框架集成（面向开发者快速构建）** - - **LlamaIndex**：调用 `DashScopeCloudIndex.from_documents()` 构建云端知识库，通过 `SimilarityPostprocessor` + `DashScopeRerank` 控制召回与重排，最终 `query_engine.query()` 触发端到端 RAG。 - - **Spring AI Alibaba**：使用 `DashScopeDocumentRetriever` 按 `INDEX_NAME` 检索上下文，并自动注入提示词交由 `qwen-max` 等模型生成；或通过 `DashScopeAgent` 调用已发布的智能体应用（隐式封装 RAG 逻辑）。 - -> ⚠️ 注意：所有 RAG 路径均依赖知识库预置——知识库必须部署在华北2（北京）地域，且需完成文档上传、解析、向量化与索引构建；不支持裸知识库 ID 直接调用，问答接口必须传入已绑定知识库的 `app_id`。 +所有场景均依赖同一套知识库基础设施（切片、向量化、重排、元数据过滤），确保行为一致与效果可复现。 ## 关键参数和配置 -| 参数 | 所属层级 | 类型 | 说明 | 典型值 | 备注 | -|------|----------|------|------|--------|------| -| `retrieval_top_k` / `top_k` | 应用层 / API 层 | integer | 初始召回切片数量（向量/关键词双路） | `3`–`10` | 过高增加 Token 开销，过低影响召回完整性；最大支持 100 | -| `similarity_threshold` | 应用层 / 知识库层 | float (0.01–1.0) | Rerank 后过滤阈值，仅保留得分高于此值的切片 | `0.4`–`0.6` | 值过高易漏召，过低引入噪声；纯文本知识库专用 | -| `max_retrieved_chunks` | 应用层 | integer | 最终传递给大模型的上下文切片总数 | `1`–`20` | 控制 Prompt 长度与成本，建议 ≤10 | -| `weight` | 多知识库混排 | float | 知识库在联合检索中的相对优先级 | `1.0`, `2.0` | 权重越高，同 Query 下该库切片排序越靠前 | -| `tags` | 知识库层 | string array | 单文件最多 32 个标签，用于精准范围过滤 | `["finance", "2024Q3"]` | 支持 `AND` 语义匹配，提升高干扰场景精度 | -| `metadata` 字段 | 索引层 | key-value | 在切片索引时注入结构化信息（如 `filename`, `date`, `author`） | — | 实现“先过滤、再检索”，降低误召率 | +RAG 效果高度依赖以下关键参数，需根据业务目标权衡精度、延迟与成本： -> ✅ 提示：参数生效位置不同——`similarity_threshold` 和 `weight` 在知识库或应用配置页设置；`top_k` 和 `stream` 在 API 请求 Body 或 Header 中指定；`chunk_size`/`chunk_overlap` 仅在知识库创建时一次性配置，不可修改。 +| 参数 | 作用域 | 说明 | 推荐值 | 注意事项 | +|------|--------|------|--------|----------| +| `top_k` | 检索/API/框架 | 单次检索返回的最大文本切片数 | `3–10`（问答）、`5–20`（调试） | 知识 API 默认 `5`，最大 `20`；LlamaIndex 对应 `similarity_top_k`；Spring AI 对应 `topK` | +| `score_threshold` | 应用/API/框架 | 重排后相似度阈值（0.01–1.0），低于此值的切片被过滤 | `0.3–0.6` | 过高易漏召，过低引入噪声；需结合命中测试调优 | +| `enable_rerank` | 应用/API | 是否启用百炼内置重排模型（提升相关性排序质量） | `true`（默认） | 启用后增加少量延迟与 [Token](token.md) 消耗，但显著改善结果质量 | +| `tags` | 检索/API/控制台 | 按业务标签（如 `product_v2`, `faq_2024`）精准筛选知识库文件 | `["product_v2"]` | 最多支持 32 个标签，创建知识库时需预先配置 | +| `max_retrieve_count` | 知识库配置 | Rerank 阶段输入的最大候选切片数（影响 [Token](token.md) 消耗） | `20–100` | 初检 TopK 可设更高，但最终送入 Rerank 的数量受此限制 | -## 面向开发者，简洁实用 +> ⚠️ 注意：`workspaceId` 是所有 RAG 请求的必需前置——必须用于构造专属 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`），不可复用 DashScope 公共域名；且仅华北2（北京）地域可用。 -- **起步最快**：控制台创建知识库 → 上传 PDF/DOCX/TXT → 发布“知识问答”服务 → 调用 `/api/v2/apps/knowledge/chat`，只需 `workspaceId` + `API Key` + `app_id` + `messages`。 -- **调试关键**：开启 SLS 日志监控，关注 `data.nodes[]` 字段确认召回质量；流式响应需按 `event: chunk` 解析，末尾 `event: done` 包含完整结果与 `docReferences`。 -- **避坑指南**： - - 域名必须含 `workspaceId`，API Key 必须归属该 workspace，否则 `401`； - - 知识库类型（文档/图片/表格）决定可用模型：纯文本仅支持 `qwen3-rerank`，多模态知识库才可用 `VL-Max`； - - 文件上传后需等待解析完成（1–6 分钟），未就绪时检索返回空； - - 免费额度覆盖全部 RAG 场景，但向量模型与 Rerank 模型按 Token 单独计费，非包含在知识库规格费中。 +## 面向开发者，简洁实用 -RAG 的本质是“让模型知道它该知道的”。在百炼，你只需聚焦业务知识——平台负责高效检索、可信增强、稳定生成。 +- ✅ **快速上手**：控制台创建知识库 → 上传 PDF/DOCX → 发布 → 绑定至智能体应用 → 开启“必定调用”，5 分钟启用 RAG。 +- ✅ **调试优先**：先用 `/api/v1/indices/knowledge/search` 查看原始检索结果，验证切片质量与 `score_threshold` 设置是否合理。 +- ✅ **效果优化**：启用“智能切分”策略（优于固定长度）、配置 Meta 信息抽取（如 `file_name`）、添加业务标签，三者组合可大幅提升定向召回率。 +- ✅ **成本控制**：`max_retrieve_count` 和 `top_k` 直接影响 [Token](token.md) 消耗；生产环境建议 `top_k=5` + `enable_rerank=true`，平衡效果与开销。 +- ❌ **避坑提醒**：知识库未发布则不可检索；`workspaceId` 与 OpenAPI 的 `project_id` 无映射关系；SSE 流式响应需客户端正确解析 `text/event-stream`。 ## 关联主题页 @@ -51,6 +42,6 @@ RAG 的本质是“让模型知道它该知道的”。在百炼，你只需聚 - [knowledge base](../guides/knowledge-base.md) - [frameworks](../api/frameworks.md) - [application use cases](../guides/application-use-cases.md) -- [data connection overview](../guides/data-connection-overview.md) +- [application support](../guides/application-support.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md index efba1ac1..3be77ebf 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md @@ -1,63 +1,54 @@ # 流式输出 -流式输出（Streaming Output）是百炼平台提供的一种实时响应机制，允许模型在生成结果的过程中，将输出内容分块、逐步推送至客户端，而非等待全部内容生成完毕后一次性返回。该机制显著降低端到端延迟，提升用户交互体验，尤其适用于对话类、语音合成、长文本生成等对实时性敏感的场景。 +流式输出（Streaming Output）是指模型在生成响应过程中，将结果以增量方式分块、实时返回给客户端，而非等待整个响应完成后再一次性返回。这种方式显著降低端到端延迟，提升用户体验，尤其适用于语音助手、实时对话、长文本生成等对响应速度敏感的场景。 -## 在百炼平台的不同场景中，这个概念如何使用 +## 在百炼平台的不同场景中如何使用 -流式输出在百炼平台中并非统一协议，而是根据接口类型和底层能力采用不同传输机制，开发者需按场景适配解析方式： +流式输出在百炼平台中并非统一开关，而是按接入协议和模型能力分层支持，需结合具体接口显式启用： -- **HTTP SSE（Server-Sent Events）流式** - 主要用于 `knowledge/chat`（知识问答）和 `application-call/responses`（OpenAI 兼容 Responses API）等 REST 接口。服务端以 `text/event-stream` MIME 类型响应，每条消息格式为： - ``` - event: chunk - data: {"output":{"text":"你好"}} - - event: chunk - data: {"output":{"text":"，很高兴为您服务。"}} - - event: done - data: {"output":{"text":"，很高兴为您服务。"},"usage":{...}} - ``` - 客户端需监听 `chunk` 事件持续拼接文本，并在收到 `done` 事件后处理最终结果与统计信息。 +- **Omni Realtime API（WebSocket）**：原生支持流式文本与音频输出。服务端通过 `text.delta` 和 `audio.delta` 事件持续推送增量内容（如逐字文本、连续 PCM 音频帧），无需额外参数；`modalities: ["text", "audio"]` 即默认启用双模态流式输出。 +- **Qwen API（HTTP/OpenAI 兼容）**：需显式设置 `stream=True`（OpenAI 标准）或 `stream=true`（DashScope 原生）。支持文本 token 级别流式返回（`delta.content`），但**不支持 `delta.tool_calls` 结构**——工具调用结果始终在流结束时以完整 `tool_calls` 字段一次性返回。 +- **Application Support（Assistant/Agent API）**：需**同时设置 `stream=True` 和 `incremental_output=True`** 才能启用真正的增量流式（即仅返回新增 token，非全量重传），否则 `stream=True` 仅返回标准 OpenAI-style delta 流（含重复历史）。 +- **Vision API（Qwen-VL/QVQ）**：部分模型（如 QVQ）**仅支持流式输出**，必须设置 `stream=True`，否则请求将被拒绝。 +- **Batch Chat / Files / Embedding 等异步或非交互接口**：**不支持流式输出**，仅提供最终结果。 -- **WebSocket 实时流式** - 专用于 `Qwen-Omni-Realtime` 系列 API。通过双向 WebSocket 连接，服务端主动推送结构化事件（如 `response.text.delta`、`response.audio.delta`），支持文本、音频、工具调用状态等多模态增量输出，适用于语音助手、实时字幕等低延迟场景。 - -- **DashScope 原生 HTTP 流式** - 在 `Generation.call` 等原生接口中，通过 `stream=true` 启用，返回标准 SSE 格式；若需更细粒度控制（如仅增量返回新 token），可配合 `incremental_output=true` 参数，此时 `output.text` 字段为本次增量内容，而非累计全文。 - -- **非流式回退兼容** - 所有支持流式的接口均提供 `stream=false` 选项（默认值因接口而异），此时返回单次完整 JSON 响应，结构与流式末尾 `event: done` 的 `data` 字段一致，便于快速调试或轻量集成。 - -> ⚠️ 注意：流式能力依赖服务端配置。例如工作流应用需在「结束节点」显式开启“流式输出”开关并重新发布；Omni Realtime 模型需在 `session.update` 中设置 `modalities: ["text", "audio"]` 才能触发音频流。 +> ⚠️ 注意：流式能力与模型强绑定。例如 `qwen-omni-turbo-realtime` 支持低延迟音频流，而 `qwen-turbo`（HTTP 文本模型）虽支持 `stream=True`，但无音频流能力；`qwen-vl` 在 OpenAI Vision 接口下支持流式，但在 DashScope 原生接口中不支持。 ## 关键参数和配置 -| 参数 | 类型 | 说明 | 所属接口/场景 | -|------|------|------|----------------| -| `stream` | `boolean` | 全局开关，启用流式传输（SSE 或 WebSocket）。设为 `true` 时，响应头含 `Content-Type: text/event-stream`（HTTP）或建立 WebSocket 连接（Realtime）。 | 所有支持流式的接口（`knowledge/chat`, `responses`, `Generation`, `Application.call` 等） | -| `incremental_output` | `boolean` | **仅 DashScope 原生接口有效**。当 `stream=true` 时，若设为 `true`，则 `output.text` 返回本次增量内容；若为 `false`（默认），则返回当前累计全文。 | `dashscope.Generation` / 原生 `/generation` 接口 | -| `modalities` | `string[]` | **仅 Omni Realtime API 有效**。指定输出模态组合，如 `["text"]` 或 `["text","audio"]`，决定是否触发对应流式事件。 | `qwen3.5-omni-realtime` 等 WebSocket 接口 | -| `session_id`（流式上下文） | `string` | 在支持会话的流式调用中（如 `Application.call`），`session_id` 用于关联多轮流式响应，确保上下文连续性。注意：`Responses API` 异步模式（`background=true`）不支持流式。 | `Application.call`, `responses` 同步流式 | - -## 面向开发者，简洁实用 - -- ✅ **首选 SDK 调用**：Python 使用 `dashscope` SDK 的 `stream=True` 参数（如 `Generation.call(..., stream=True)`），SDK 自动处理 SSE 解析与事件分发，避免手动解析 `event:` 行。 -- ✅ **HTTP 调试建议**：用 `curl -N` 或浏览器 DevTools 的 Network → EventStream 查看原始流数据；生产环境务必设置超时（如 `timeout=120s`），防止连接挂起。 -- ✅ **错误处理要点**：流式请求失败时，可能已部分接收数据。请检查 HTTP 状态码（如 `401`, `429`）及首个 `event: error` 消息，不要仅依赖 `event: done`。 -- ❌ **避坑提示**： - - 工作流应用未在结束节点开启“流式输出” → 即使传 `stream=true` 也返回非流式响应； - - Omni Realtime 使用 `qwen-omni-turbo-realtime` 模型 → 不支持 `temperature`/`max_tokens` 等参数，但流式功能正常； - - `knowledge/chat` 接口 `stream=false` 时，响应结构与流式 `done` 事件 payload 完全一致，可复用同一解析逻辑。 - -流式输出是构建高性能 AI 应用的关键能力。正确启用并解析它，能让您的产品获得接近本地响应的流畅体验。 +| 参数 | 类型 | 说明 | 是否必需 | 备注 | +|------|------|------|----------|------| +| `stream` | `boolean` | 启用流式传输协议（HTTP chunked encoding / WebSocket event stream） | 是（流式场景） | 所有支持流式的接口均需设为 `true` | +| `incremental_output` | `boolean` | 启用增量式流式（仅返回本次新增内容） | 仅 Assistant/Agent API 需要 | 设为 `true` 可避免前端重复渲染；默认 `false` | +| `modalities` | `array` | Omni Realtime 中指定输出模态，决定流式内容类型 | 是（Omni Realtime） | 必须包含 `"text"`；添加 `"audio"` 则启用音频流（24 kHz PCM） | +| `output_audio_format` | `string` | Omni Realtime 中固定为 `"pcm"`，不可更改 | — | 音频流格式已固化，无需配置采样率或编码 | + +- **不推荐依赖的隐式行为**： + - 不要假设 `stream=True` 自动启用增量输出（仅 Assistant API 需配 `incremental_output`）； + - 不要尝试在不支持流式的接口（如 Completions、Embedding）中设置 `stream`，将导致 400 错误； + - `smooth_output`（Omni Realtime）影响音频流平滑度，但**不影响文本流行为**，属音频后处理参数，非流式开关。 + +## 面向开发者：简洁实用建议 + +- ✅ **首选 WebSocket 实时流**：对语音助手、实时客服等场景，直接使用 Omni Realtime API（`wss://.../realtime`），天然低延迟、双模态、事件驱动，无需手动解析 chunk。 +- ✅ **HTTP 场景统一用 `stream=True`**：[OpenAI 兼容接口](openai-compatible-interface.md)（Chat Completions / Responses）和 DashScope 原生接口均支持，返回格式一致（`{"delta": {"content": "..."}}`）。 +- ✅ **前端处理要点**： + - 监听 `data:` 行（HTTP）或 `message` 事件（WebSocket），拼接 `delta.content`； + - 检查 `finish_reason` 字段判断流是否结束（`"stop"` / `"length"` / `"tool_calls"`）； + - 对 Assistant API，务必检查 `incremental_output` 是否生效，避免重复渲染。 +- ❌ **避免踩坑**： + - 不要在 `tools` 调用场景中期待 `delta.tool_calls` —— 工具参数始终在流末尾完整返回； + - 不要跨地域混用 `api_key` 和 `base_url`（如北京 key 调用新加坡 endpoint），会导致流式连接失败； + - `max_tokens` 仅控制截断长度，**不影响流式生成过程**，流仍会持续直到自然结束或超时。 + +流式输出是百炼平台实现“实时 AI”的基础设施能力。正确配置参数、匹配接口协议、理解各模型限制，即可高效构建响应迅捷、体验流畅的 AI 应用。 ## 关联主题页 -- [knowledge](../api/knowledge.md) -- [application call](../api/application-call.md) - [omni realtime api](../api/omni-realtime-api.md) - [qwen api reference](../api/qwen-api-reference.md) -- [bailian application calling](../guides/bailian-application-calling.md) +- [model experience](../guides/model-experience.md) +- [application support](../guides/application-support.md) +- [toolkits and frameworks](../api/toolkits-and-frameworks.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md index af5d22b3..2ecf6286 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md @@ -1,51 +1,42 @@ -# Token 计量与管理 +# Token -Token 计量与管理是百炼平台对大模型调用资源消耗进行标准化度量、实时追踪、精准计费与精细化治理的核心机制。它以 **Token** 为最小计量单元，覆盖输入、输出、缓存等全链路消耗，并统一映射到 Credits（Token Plan）或按量账单（Pay-as-you-go），支撑成本控制、用量分析与性能优化。 +Token 是百炼平台中用于计量模型调用资源消耗的核心计费与观测单位，代表模型处理输入（[prompt](../guides/prompt.md)）和生成输出（completion）过程中所消耗的文本单元。在百炼体系中，Token 不仅是 Credits 抵扣、用量监控和成本分析的基础粒度，也是性能诊断（如首 Token 延时）、模型评测（裁判模型计费）及多模态资源折算（图像/音频等按规则换算为等效 Token）的统一标尺。 ## 在百炼平台的不同场景中，这个概念如何使用 -- **Token Plan 订阅服务**：以 Credits 为计费单位，按实际消耗的 `input_tokens`、`output_tokens` 和 `cache_tokens` 动态抵扣；模型白名单严格限定可计量范围，非白名单模型调用不计入 Credits，可能触发按量扣费。 -- **模型监控（Model Monitoring）**：提供分钟级/小时级 `model_usage` 指标（含 `input_tokens`/`output_tokens`/`total_tokens` 等 `usage_type` 维度），支持按 `model`、`apikey_id`、`workspace_id` 等标签过滤，用于成本归因与异常排查（北京地域支持单次请求级 Token 查看）。 -- **应用观测（Application Monitoring）**：在 Span 级别精确统计 `Input Tokens` 与 `Output Tokens`，关联至具体节点（如 `LLM`、`EMBEDDING`），支持按链路深度、节点类型、状态筛选，是智能体/工作流成本拆解与性能瓶颈定位的关键依据。 -- **模型评测（Model Evaluation）**：评测任务执行时，被评测模型推理和裁判模型评分均产生 Token 消耗——前者计入被评测模型用量，后者计入裁判模型用量，二者独立计量、分别计费。 -- **应用支持（Application Support）**：插件调用、RAG 检索、[流式输出](streaming-output.md)等能力本身不额外计 Token，但其触发的模型调用（如 LLM 生成响应、Embedding 向量化）仍遵循标准 Token 计量规则；`incremental_output=True` 不改变总 Token 数，仅影响传输方式。 +- **计费与配额**：Token Plan 团队版以 Credits 统一抵扣，1 Credit = 1,000 Tokens（文本模型），图像生成按张数折算为等效 Token（如 `qwen-image-2.0` 每张 ≈ 500 Tokens），抵扣顺序为「坐席月度额度 → 共享用量包 → 暂停服务」。 +- **可观测性**： + - 应用观测（Application Monitoring）中，LLM 节点的「Token 总量」= 输入 Token + 输出 Token；Embedding 节点仅统计向量化输入的 Token 量； + - 模型监控（Model Monitoring）提供细粒度 `input_tokens` / `output_tokens` / `cache_tokens` 等 `usage_type` 指标，支持按业务空间、模型、API Key 下钻分析； + - 所有 Token 消耗均分钟级同步（高级监控）或小时级聚合（基础监控）。 +- **模型评测**：大模型评估类维度需调用裁判模型（如 `qwen3.7-max`），其评分过程产生的输入与输出 Token 单独计费，费用计入评测任务账单；规则评估与人工评估不产生 Token 费用。 +- **API 调用约束**：`max_tokens` 参数直接限制输出长度上限（单位：Token），超限将触发 `Range of max_tokens should be [1, xxx]` 错误；多模态模型（如 `qwen3-vl-plus`）的 `messages.content` 中每张图片、每段视频均按预设规则折算为 Token 并计入总限额。 ## 关键参数和配置 -| 参数 | 说明 | 注意事项 | -|------|------|----------| -| `input_tokens` | 模型接收到的 Prompt、上下文、工具描述、图片 Base64 编码等输入内容所占 Token 数 | 图片按分辨率折算（如 `qwen-image-2.0` 使用固定 token 开销 + 可变视觉 token）；系统自动去除冗余空格与换行，但不压缩语义 | -| `output_tokens` | 模型实际生成的文本或结构化响应（含 function call 参数）所占 Token 数 | 流式响应中累计计数，`incremental_output` 不影响总量；截断（`max_tokens`）会限制此值上限 | -| `cache_tokens` | 模型缓存 Prompt 或历史对话产生的额外开销（如 KV Cache 预分配） | 当前仅部分模型（如 `qwen3.7-plus`）在启用缓存优化时显式计量，多数场景隐含在 `input_tokens` 中 | -| `total_tokens` | `input_tokens + output_tokens`（缓存通常不单独计入） | 计费与监控口径统一以此为准 | -| `usage_type` | Prometheus 指标 `model_usage` 的关键 label | 必须显式指定 `input_tokens`/`output_tokens`/`total_tokens` 才能正确聚合 | - -> ⚠️ 重要约束： -> - Token 计量基于模型实际 tokenizer 行为，**不接受客户端预估**；开发者不可自行计算并传入 `token_count` 参数。 -> - 所有计量均发生在服务端，调用返回的 `usage` 字段（如 OpenAI 兼容 API 中的 `"usage": {"prompt_tokens": ..., "completion_tokens": ...}`）为唯一可信来源。 -> - 图像生成（`multimodal-generation` API）与视觉理解（多模态输入）的 Token 计算逻辑与纯文本模型不同，需查阅对应模型文档确认细则。 +- `max_tokens`：必填整数，指定最大输出 Token 数，取值范围由模型文档明确限定（如 `qwen3.6-plus` 为 `[1, 8192]`），不可设为 0 或负数。 +- `input_tokens` / `output_tokens`：只读指标，由平台自动统计，用于监控、告警与账单结算，开发者无需手动传入。 +- Token 折算规则（非 API 参数，但影响用量）： + - 文本：UTF-8 编码下，中文字符约 1–2 Token/字，英文单词平均 1.3 Token/词； + - 图像：`qwen-image-2.0` 默认 500 Tokens/张，`qwen-image-2.0-pro` 为 1,200 Tokens/张； + - 音频/视频：按时长与模型规格折算（如 `cosyvoice-v3-flash` 每秒语音 ≈ 15 Tokens）； + - 缓存 Token（`cache_tokens`）：启用 KV Cache 时复用历史计算结果，按实际节省量计为负 Token，降低总消耗。 ## 面向开发者，简洁实用 -- ✅ **必查返回值**：每次成功调用后，务必解析响应中的 `usage` 字段，用于本地日志记录、预算预警或用量上报。 -- ✅ **善用监控工具**：在北京地域部署关键应用时，开启模型监控的「高级监控」并配置告警，当 `model_usage{usage_type="total_tokens"}` 异常飙升时快速定位问题模型或恶意请求。 -- ✅ **成本优化实践**： - - 对长上下文场景，优先启用 `cache_tokens` 支持的模型（查看模型文档支持列表）； - - 评测任务中，用规则评估替代大模型评估可规避裁判模型 Token 费用； - - Token Plan 用户应定期检查控制台「用量分析」，识别高消耗模型/成员，及时调整分配策略。 -- ❌ **避免踩坑**： - - 不要复用通用 API Key（`sk-`）调用 Token Plan 模型——将导致 401 错误或意外按量扣费； - - 不要尝试通过修改 `max_tokens` 或 [prompt](../guides/prompt.md) 格式“欺骗”Token 计量——平台按真实 tokenizer 输出计费； - - 不要依赖前端渲染逻辑（如 Markdown 解析）估算 Token——实际消耗由模型侧 tokenizer 决定。 - -Token 计量是百炼平台资源治理的基石。理解它，就是掌握成本、性能与合规的主动权。 +- ✅ **务必检查 `max_tokens` 上限**：调用前查阅对应模型文档，避免因超限返回 400 错误。 +- ✅ **用环境变量管理 API Key**：设置 `DASHSCOPE_API_KEY=sk-ws-xxx`，杜绝硬编码与日志泄露。 +- ✅ **监控 Token 用量防超支**：在控制台开启「高级监控」，配置 `model_usage{usage_type="total_tokens"}` 告警，阈值建议设为月度预算的 80%。 +- ✅ **图像/多模态调用需显式声明**：文本模型（如 `qwen3.6-plus`）不支持 `image_url`；必须使用 `qwen-image-2.0` 等专用模型 ID，并通过 `/multimodal-generation` endpoint 调用。 +- ❌ **不要跨地域混用 Key 与 Base URL**：Token Plan 专属 Key 仅支持华北2（北京）地域，且必须搭配其指定 Base URL（如 `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。 +- ❌ **勿在自动化脚本中滥用 Token Plan**：该套餐仅限交互式工具（Cursor/Claude Code 等）使用，批量调用将触发封禁。 ## 关联主题页 - [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) +- [preparations](../api/preparations.md) - [application monitoring](../guides/application-monitoring.md) +- [model monitoring](../guides/model-monitoring.md) - [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [application support](../guides/application-support.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md index c409420b..c1549c2d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md @@ -1,47 +1,48 @@ # application evaluation -应用评测是百炼平台用于系统化评估智能体/工作流应用输出质量的核心能力，支持自动与手动两种评测范式。自动评测基于大模型与知识库自动生成评测集并完成端到端评分，适用于快速迭代与横向对比；手动评测则依赖人工构建评测集与标注，适用于高精度、强主观性或需深度归因的场景。两类评测均围绕评测集、评测任务、评估器与标签四大核心组件展开，形成可配置、可复用、可追溯的质量保障闭环。 +application evaluation 是百炼平台用于系统化评估智能体/工作流应用输出质量的核心能力，支持自动与手动两种评测范式。它通过评测集驱动、多维度评估器协同、人工标签补充的混合机制，实现从数据构建、任务执行到归因分析的完整闭环，适用于模型迭代、知识库更新、Prompt调优等关键场景的质量验证。 ## 支持的模型/功能 -- **自动评测**：仅支持 `qwen-max` 和 `qwen-plus` 两种模型用于评测集生成与最终评分，不支持其他模型（如 `qwen-turbo` 或 `qwen2` 系列）[原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。该限制同样适用于评测规则配置阶段。 -- **评估器类型**：新版评测体系支持 LLM 评估器（调用大模型进行语义评分）、Code 评估器（执行 Python 脚本进行规则校验）及基于历史评测任务自动生成的 LLM 评估器 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。LLM 评估器默认限时免费，但实际调用仍产生 Token 费用。 -- **评测集类型**：当前存在两套并行体系： - - 旧版仅支持 **对话分析**（`.xls`/`.xlsx`）和 **知识问答**（`.jsonl`）两类，分别用于手动评测与自动评测 [原文标题](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)； - - 新版扩展为 **智能体**、**工作流** 和 **自定义** 三类，支持按应用出入参结构自动生成模板，并引入版本管理与表结构编辑能力 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)。 -> **注意**：文档 3 与文档 4 对评测集类型的定义存在明显差异——前者限定为“对话分析”和“知识问答”，后者升级为“智能体/工作流/自定义”。这反映平台已从单一 RAG 场景向通用应用评测演进，**旧版类型已逐步被新版覆盖，新建评测应优先采用新版评测集**。 +- **自动评测**：基于知识库自动生成评测集，调用大模型（当前仅支持 `qwen-max` 和 `qwen-plus`）完成端到端评分与归因分析，适用于单应用深度诊断或多应用横向对比 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **手动评测**：依赖人工构建的结构化评测集（XLS/XLSX 格式），由人工对模型输出进行打标（如“较差/一般/较好”），适合需强主观判断或高置信度校验的场景 [原文标题](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)。 +- **新版评测体系**：支持**智能体**、**工作流**、**自定义**三类评测集，配合可插拔的**评估器**（LLM 或 Code 类型）与**标签管理**，实现灵活的多维度自动+人工混合评测 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 +> **注意**：旧版自动评测（文档 1）与新版评测任务（文档 5）在架构上存在显著差异：前者为封闭式流程（知识库→生成评测集→固定模型打分），后者为开放式框架（评测集+评估器+标签自由组合）。两者共存但不兼容，新版不支持旧版的“RAG归因分析”能力，旧版亦无法使用新版的 Code 评估器或布尔值标签等功能。 ## 关键参数 -- **评测集字段映射**：所有评估器（尤其是预置模板）对输入字段有明确要求。例如，“问答相关性”评估器必需 `query` 和 `response` 字段；若评测集字段名为 `Prompt`/`Completion`，必须在参数映射中显式绑定 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 -- **分类采样数**：自动评测中，需为每种任务类型（事实型、教程型等）单独设置采样数量，直接影响评测覆盖面与 Token 消耗 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 -- **评估器评分范围与阈值**：LLM/Code 评估器均需配置 `评分范围`（如 `0-1` 或 `1-5`）和 `通过阈值`（如 `0.8` 或 `4`），二者共同决定 Pass/Fail 判定逻辑，且必须在 Prompt 中保持语义一致 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 -- **标签类型约束**：标签创建时需指定类型（分类/布尔值/数字/文本），不同类型对应不同筛选条件与标注方式，影响后续指标统计维度 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)。 +- **评测集类型**： + - `知识问答`（JSONL）：用于自动评测，含 `query`、`referenceAnswer`、`fineKeywords`、`coarseKeywords`、`queryType` 字段 [原文标题](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)； + - `对话分析`（XLS/XLSX）：用于手动评测，含 `Prompt`、`Completion`、`SessionId` 字段； + - `智能体/工作流/自定义`：新版评测集类型，字段结构由应用出入参或用户自定义决定。 +- **评估器参数**： + - LLM 评估器需配置 `模型`、`Prompt`、`评分范围`（如 0–1 或 1–5）、`通过阈值`； + - Code 评估器需定义 `入参`（如 `query`, `response`）、`Python 执行函数`（返回数值评分）； + - 所有评估器必须完成**字段映射**（如将评测集的 `question` 字段映射至评估器变量 `query`），否则任务无法创建。 +- **标签类型**：支持分类（多选枚举）、布尔值（True/False）、数字（1–5 分）、文本（自由输入）四类，用于人工标注与统计分析 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)。 ## 使用方式 -1. **准备数据基础**： - - 创建评测集：可选择自动生成（仅限知识问答类型，依赖知识库）或手动上传（支持 `.xls`/`.xlsx`/`.jsonl` 格式，单文件 ≤20MB）[原文标题](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)； - - 发布评测集：草稿状态不可用于评测，必须点击“发布”使其生效； - - （可选）创建标签与评估器：按业务需求定义多维标注体系与自动化评分规则。 - -2. **发起评测任务**： - - **自动评测**：进入控制台自动评测页面 → 选择已发布且配置知识库的智能体应用 → 选择知识库 → 生成或选用评测集 → 设置采样数与评测模型 → 发起任务 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)； - - **手动评测**：上传评测集 → 进入手动评测页 → 选择应用与已发布评测集 → 配置评测维度 → 开始评测 → 人工打标（较差/一般/较好 或 1–5 分）→ 提交结果； - - **新版评测任务**：支持“不关联应用”（纯人工标注）、“智能体”或“工作流”关联模式，并可同时添加最多 10 个评估器与任意标签 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 - -3. **分析与迭代**： - - 查看报告：自动评测提供总正确率、BadCase 归因（模型理解/重排/检索/切片/未获取知识）、RAG 各类型得分；手动评测提供人工标注汇总； - - 使用标签筛选 BadCase，结合评估器结果定位问题根因； - - 基于归因建议优化 Prompt、知识库切分策略或检索配置，发布新版本后复用同一评测集验证效果。 +1. **准备评测数据**： + - 自动评测：确保目标智能体已**发布**、**配置知识库**、**开通应用观测**； + - 手动评测：下载模板，按 `Prompt`/`Completion`/`SessionId` 填写 XLS/XLSX 文件并上传发布； + - 新版评测：创建评测集时选择类型（智能体/工作流/自定义），下载模板填写后上传，**必须发布**才可用于任务。 +2. **创建评测任务**： + - 旧版自动评测：在控制台依次完成「选择应用→选择知识库→生成评测集→配置采样数与模型→发起评测」； + - 新版评测任务：在任务创建页选择「评测集+版本」、「关联应用类型（智能体/工作流/不关联）」、「添加评估器（≤10个）并完成字段映射」、「配置标签」； + - 手动评测：在「手动评测」页面选择已发布应用与已发布评测集，进入标注流程。 +3. **执行与分析**： + - 自动评测结果直接生成总正确率、BadCase 归因（如“检索无效”“切片不完整”）及调优建议； + - 新版任务支持「数据明细」（查看每条评估器评分与人工标签）与「指标统计」（综合得分、各评估器通过率柱状图）； + - 手动评测需逐条点击「标注」，选择评价等级后保存。 ## 限制和注意事项 -- **权限与前提**：自动评测要求子账号具备 `管理员` 或 `应用评测-操作` 权限，且目标应用必须已发布、配置知识库、并加入应用观测列表 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 -- **数量限制**：单次自动评测最多支持 8 个应用横向对比；单个评测任务最多添加 10 个评估器；单次上传评测集文件不超过 10 个，单文件 ≤20MB。 -- **Token 消耗**：所有调用大模型的操作（评测集生成、自动评分、LLM 评估器）均产生 Token 费用，预估消耗仅为参考，实际以账单为准；试运行也会消耗少量 Token [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 -- **评测失败处理**：自动评测中失败用例不计入正确率计算；手动评测中未完成打标的条目不影响已完成部分的统计。 -- **兼容性提示**：新版评测任务（文档 5/6/7）与旧版自动/手动评测（文档 1/2/3）共存，但二者数据模型与流程不互通。**新建项目应统一使用新版体系**，旧版功能仅维持兼容，不再新增特性。 +- **应用限制**：自动评测仅支持**已发布的智能体应用**，且多应用横向评测时所有应用必须共享至少一个知识库 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)；新版评测任务中，已发布的评测集若被任务引用则不可删除。 +- **模型与计费**：自动评测与 LLM 评估器均调用 `qwen-max`/`qwen-plus`，产生 [Token](../concepts/token.md) 费用；Code 评估器无额外调用成本；预估 [Token](../concepts/token.md) 消耗为参考值，实际以账单为准 [原文标题](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **配置不可变性**：评测任务创建后，其关联的评测集、应用、评估器映射关系**不可修改**；如需调整，必须新建任务 [原文标题](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 +- **文件约束**：评测集上传支持 `.xls`/`.xlsx`（≤20MB/个，单次≤10个）或 `.jsonl`（知识问答专用）；新版自定义评测集支持任意表结构，但创建后类型不可更改。 +- **权限要求**：子账号需具备 `管理员` 或 `应用评测-操作` 权限才能使用自动评测功能。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md index 28d20d23..6c605f32 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md @@ -1,61 +1,70 @@ # application monitoring -应用观测（Application Monitoring）是阿里云百炼平台提供的端到端可观测能力，用于追踪和分析智能体、工作流及高代码类应用的内部执行链路。它支持查看调用延时、Token 消耗、模型思考过程及各节点状态，并提供分钟级指标聚合与原始 Span 数据导出能力。该功能基于 OpenTelemetry 构建，需依赖可观测链路服务，**当前不提供 API 接口** [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +应用观测（Application Monitoring）是阿里云百炼平台提供的端到端可观测能力，用于追踪应用内部调用链路、分析模型响应延时、查看推理过程及 [Token](../concepts/token.md) 消耗等关键指标。该功能基于 OpenTelemetry 架构实现，数据同步频率为分钟级，适用于调试、性能优化与真实场景评测。> **注意：应用观测目前暂无 API 接口**，所有操作需通过控制台完成，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 ## 支持的模型/功能 -- **支持的应用类型**：智能体应用、工作流应用、高代码应用（但高代码应用仅上报 `CHAIN` 根节点，**不支持内部链路追踪**） -- **核心可观测维度**： - - 调用链路（Root Span / All Span / Model Span 三种视图） - - 延时（含平均首 Token 耗时、平均调用时长） - - Token 统计（输入/输出/总量） - - 状态（正常/错误，含错误类型细分） - - 节点类型与嵌套关系（如 `LLM`、`RETRIEVER`、`EMBEDDING`、`GUARDRAIL` 等） -- **扩展能力**： - - 数据标注（布尔值、分类、数字、文本四类标签） - - 批量导出（JSONL / Excel） - - 添加 Span 到评测集（支持字段映射与导入策略配置） -- **不支持场景**：通过 Assistant API 创建的智能体应用 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)；[长期记忆](../concepts/long-term-memory.md)中的检索过程；高代码应用内部节点 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +- **支持的应用类型**：智能体应用、工作流应用、高代码应用（但高代码应用仅上报 `CHAIN` 根节点，[不支持内部链路追踪](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **不支持的应用**：通过 Assistant API 创建的智能体应用（[原文明确说明](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **可观测节点类型**： + - 通用节点：`CHAIN`（根节点，名称如 `AgentApp`/`WorkflowApp`/`FullCodeApp`）、`LLM`、`RETRIEVER`（含 `TextRetriever`/`VectorRetriever`）、`EMBEDDING`、`RERANKER`、`REWRITER`、`GUARDRAIL`、`TOOL`； + - 工作流专属节点：`START`、`END`、`API`、`CLASSIFIER`、`TEXT_CONVERTER`、`SCRIPT`、`CONDITION`、`FUNCTION_COMPUTE`、`APP_FLOW`； + - 智能体专属节点：`AGENT`； +- **附加能力**：Span 数据导出（JSONL/Excel）、添加至评测集、多维度标签标注（布尔/分类/数字/文本）、交互式展开追踪链路。 + +> **注意**：文档中关于 `FullCodeApp` 的说明存在潜在矛盾——正文称“不支持追踪其内部调用链路”，但附录又将其列为 `CHAIN` 类型节点。实际行为以 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) 中“高代码应用”章节为准：仅上报根节点，无嵌套子节点。 ## 关键参数 -| 参数 | 说明 | 备注 | +| 参数 | 说明 | 来源 | |------|------|------| -| `Request ID` / `Trace ID` / `Span ID` | 用于精准定位单次调用或子链路 | 可在节点详情页点击「查看 ID」获取 | -| `Span Name` | 节点逻辑名称（如 `AgentApp`, `TextRetriever`, `LLM`） | 支持模糊匹配筛选 | -| `Status` | `normal` 或 `error`，错误时可进一步区分类型（如 `GuardrailBlocked`, `LLMTimeout`） | — | -| `Latency (ms)` | 节点执行耗时（含网络与模型推理时间） | `LLM` 节点延时包含流式响应全过程 | -| `Input Tokens` / `Output Tokens` | Embedding 或 LLM 调用的 Token 数量 | 定义见 [附录](#f0ed9407canlv)（原文档） | -| `Label` | 用户自定义标注字段（类型强约束） | 与评测系统共享标签管理 | - -> **注意**：`TextRetriever` 和 `VectorRetriever` 默认返回 100 个切片，且**暂不支持调整数量**；此限制在文档中被多次强调，属设计约束而非临时限制。 +| **Request ID / Trace ID / Span ID** | 用于精确检索单次调用链路，可在节点详情页点击“查看 ID”获取 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **延时（ms）** | LLM 节点延时包含完整输出过程；平均首 [Token](../concepts/token.md) 耗时专用于流式调用场景 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **[Token](../concepts/token.md) 总量** | = 输入 Token + 输出 Token；Embedding 节点 Token 量仅统计向量化输入 | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **状态** | `正常` 或 `错误`（含细分错误类型） | [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) | ## 使用方式 -### 前置配置（仅首次使用需执行） -1. 使用主账号（或已授权子账号）进入 [应用观测配置](https://bailian.console.aliyun.com/tab=app?tab=app#/app-observe) 页面； -2. 授权 `AliyunServiceRoleForOpenTelemetry` 服务关联角色； -3. 开通可观测链路 OpenTelemetry 服务并初始化 LogStore。 +1. **前提配置**（主账号或已授权子账号操作）： + - 授权可观测链路 OpenTelemetry 服务角色权限； + - 开通 OpenTelemetry 服务； + - 初始化 LogStore 存储（开通后通常分钟级生效）； + > 子账号需额外配置 `AliyunBailianFullAccess`、页面权限及 `ram:CreateServiceLinkedRole` 策略，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 + +2. **启用观测**： + - 进入 [应用观测](https://bailian.console.aliyun.com/tab=app?tab=app#/app-observe)，点击“选择被观测的应用” → “添加”； + - **仅已发布且归属当前业务空间的应用可见**；未发布应用需先通过“管理应用” → “发布”。 -> 子账号需额外配置 `CreateServiceLinkedRole` 权限策略，详见 [常见问题](#cd0f1152d50hj)（原文档锚点）。 +3. **数据查看与筛选**： + - 支持三种 Span 展示模式：`Root Span`（默认）、`All Span`、`Model Span`； + - 过滤器支持按状态、Span Name、输入/输出关键词、延时、Token 量、标签等条件组合筛选； + - 可按 Request ID/Trace ID/Span ID 搜索，时间范围最长 30 天。 -### 日常操作流程 -1. **添加应用**：在应用观测列表中点击「添加」，仅支持已发布且归属当前业务空间的应用； -2. **查看数据**： - - 在 Span 列表页切换筛选模式（Root/All/Model Span）； - - 使用过滤器按 `Status`、`Span Name`、`Input`、`Output`、`Latency`、`Tokens` 或 `Label` 组合筛选； - - 单击节点名称展开详情，查看原始请求/响应、标注记录、子节点等； -3. **导出与复用**： - - 点击「导出数据」下载 JSONL 或 Excel； - - 选中 Span 后点击「添加到评测集」，完成字段映射与导入策略配置。 +4. **高级操作**： + - **导出数据**：Trace 列表页右上角支持 JSONL/Excel 导出； + - **添加到评测集**：支持批量 Span 导入，字段映射最多 50 个； + - **数据标注**：与评测系统共享标签体系，支持四类标注类型并实时保存。 ## 限制和注意事项 -- **无 API 支持**：应用观测为纯控制台功能，不开放 SDK 或 RESTful 接口 [应用观测 (raw/application-user-guide/application-monitoring/application-observation.md)](../../raw/application-user-guide/application-monitoring/application-observation.md)； -- **数据延迟**：指标同步频率为**分钟级**，不适用于实时告警场景； -- **存储计费**：功能本身免费，但底层 OpenTelemetry 存储费用需单独承担； -- **高代码应用限制**：即使开启观测，也仅上报 `FullCodeApp` 根节点，无法观测其内部[函数调用](../concepts/function-calling.md)或自定义逻辑——若需细粒度追踪，必须在代码中集成 `AgentScope-AI` 的 Tracing 模块并部署时启用 `--telemetry enable` 参数； -- **权限要求**：子账号开通需满足三重权限（`AliyunBailianFullAccess` + 页面写入权限 + `CreateServiceLinkedRole` 策略），缺一不可。 +- **功能限制**： + - 无公开 API，无法程序化接入； + - 不支持[长期记忆](../concepts/long-term-memory.md)（Long-term Memory）中的检索过程观测； + - 高代码应用无法观测内部节点，仅上报 `FullCodeApp` 根节点； + - TextRetriever / VectorRetriever 默认返回 100 个切片，**不支持数量调整**。 + +- **配置与权限**： + - 必须使用主账号首次开通，或确保子账号已获完整权限（含创建服务关联角色）； + - 应用观测本身免费，但底层 OpenTelemetry 存储与计算按量计费，费用独立于百炼资源包。 + +- **数据时效性**： + - 指标同步延迟约 1–3 分钟； + - 监控统计支持按分钟/小时/天聚合，最长回溯 30 天； + - 关闭观测后历史数据停止同步，重新开启仅采集新增数据。 + +- **开发适配（高代码应用）**： + - 需在代码中集成 AgentScope-AI 的 `Tracing` 模块； + - 部署时必须添加 `--telemetry enable` 启动参数，否则无数据上报。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md index 04de291f..38a762eb 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md @@ -1,57 +1,46 @@ # application permission management -百炼平台的权限管理以“业务空间”为最小单元，支持跨地域、多角色的精细化控制，覆盖模型调用/调优/部署、用户页面访问、API Key 管理及 OpenAPI 接口调用等核心场景。权限策略严格遵循阿里云 RAM 体系，需结合控制台操作与 RAM 策略协同配置。详细设计逻辑请参见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +百炼平台的权限管理以“业务空间”为最小管理单元，支持跨地域、多角色的精细化控制，覆盖模型调用/调优/部署、用户页面访问、API Key 管理及 OpenAPI 接口调用等核心场景。权限策略严格遵循阿里云 RAM 体系，需结合主账号与 RAM 用户角色协同配置。所有权限生效均依赖业务空间归属关系，且 API Key 权限继承自其所属空间而非用户控制台权限 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 ## 支持的模型/功能 -- **模型级管控**：支持对单个模型在指定业务空间内独立设置： - - 调用权限（含控制台 & API） - - 调优（训练）权限 - - 部署权限 -- **资源维度隔离**：业务空间按地域物理隔离，同一地域内可创建多个业务空间，但**单个业务空间不能跨地域存在**（详见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 -- **角色能力矩阵**： - | 功能 | 超级管理员 | 业务空间管理员 | 普通用户 | - |---|---|---|---| - | 模型调用 & 限流 | ✅ | ❌ | ❌ | - | 模型调优 | ✅ | ❌ | ❌ | - | 模型部署 | ✅ | ❌ | ❌ | - | 用户管理 | ✅ | ✅ | ❌ | - | 页面权限管理 | ✅ | ✅ | ❌ | - | API Key 管理 | ✅ | ✅ | ❌ | - | OpenAPI 接口权限 | ❌（仅主账号可开通） | ❌ | ❌ | - -> **注意**：文档中多次强调“默认业务空间无法设置模型调用/调优/部署限制”，但未明确说明该限制是否适用于所有地域。实际配置时请以控制台实时提示为准，避免依赖默认空间进行生产环境权限隔离 —— 此点与 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) 中“应用于生产环境”章节推荐的按环境划分空间策略存在隐含冲突。 +- **模型级控制**：支持对单个模型设置调用（含控制台 & API）、调优（训练）和直接部署三类开关，仅在**非默认业务空间**中可配置；默认业务空间对所有模型开放全部能力 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +- **角色分级**： + - **超级管理员**：拥有 `AliyunBailianFullAccess` 策略，可跨地域、跨空间管理模型限流、用户、API Key 及空间生命周期； + - **业务空间管理员**：仅管理指定空间内的用户权限、页面可见性及模型可用性； + - **普通用户**：仅能使用被显式授权的页面与资源，无管理能力。 +- **细粒度页面权限**：通过控制台「权限管理」页签为 RAM 用户分配具体菜单项（如“模型体验-操作”“批量推理-操作”），但该设置**不影响 API Key 的调用能力** [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +- **OpenAPI 接口权限**：RAM 用户默认无权调用应用、知识库、Prompt 工程等 OpenAPI，需主账号在 RAM 控制台额外授予 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess` 策略。 ## 关键参数 -- **业务空间 ID（Workspace ID）**：API 调用必需参数，用于标识资源归属空间，获取方式见 [获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 -- **API Key 归属约束**：单个 API Key 仅绑定**一个地域 + 一个业务空间 + 一个 RAM 用户**，不可迁移；其可用模型与限流策略完全继承自归属业务空间的配置（[API-Key 权限](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 -- **限流粒度**：支持 QPM（每分钟请求数）和 Token 限流两种模式，均在业务空间维度配置。 -- **OpenAPI 权限策略**：必须由阿里云主账号在 RAM 控制台显式授予 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess`，RAM 用户默认无权调用应用、知识库、Prompt 工程等核心 OpenAPI（[OpenAPI 接口权限](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 +| 参数 | 说明 | 约束 | +|------|------|------| +| `workspace_id` | 业务空间唯一标识，API 调用必需参数 | 必须与 API Key 所属空间一致；获取方式见 [获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id) | +| `qpm_limit` / `token_limit` | 模型级请求/[Token](../concepts/token.md) 限流值（QPM、TPM） | 仅在非默认业务空间中可设置；全局配额按空间比例分配（如生产环境占 60%） | +| `api_key` | 绑定至单一地域+单一业务空间+单一用户的凭证 | 不可跨空间/跨用户迁移；华北2（北京）新创建的 API Key 默认归属主账号（自 2026-03-25 起） | +| `ip_whitelist` | API Key 的 IP 访问白名单 | 仅华北2（北京）地域支持 | -## 使用方式 - -1. **角色初始化**： - - 超级管理员：主账号或拥有 `AliyunBailianFullAccess` 策略的 RAM 用户，通过全局管理菜单（[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) / [新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/business_management) / [弗吉尼亚](https://modelstudio.console.aliyun.com/us-east-1?tab=globalset#/efm/business_management)）统一配置。 - - 业务空间管理员：由超级管理员或同空间管理员在控制台「权限管理」页签中为 RAM 用户授予「管理员」角色。 +> **注意**：文档中多次强调“默认业务空间无法设置模型调用/调优/部署限制”，但未明确说明该限制是否适用于所有地域。实际配置时请以控制台界面为准——若某地域默认空间页面中缺失限流开关，则视为不可配置。 -2. **模型权限开通（必需前置步骤）**： - - 超级管理员需先在全局管理菜单中为业务空间启用目标模型的**调用、调优或部署权限**（默认业务空间自动全开，但不支持限流）。 - -3. **用户权限分配**： - - 控制台操作：在业务空间「权限管理」页签中，为 RAM 用户勾选对应功能模块权限（如「模型体验-操作」「模型调优-操作」「批量推理-操作」等）。 - - API 调用：为用户在目标业务空间创建 API Key，Key 的能力范围由该空间模型权限决定，**不受用户控制台权限影响**。 +## 使用方式 -4. **OpenAPI 授权**： - - 主账号登录 RAM 控制台 → 找到目标 RAM 用户 → 添加 `AliyunBailianDataFullAccess`（读写）或 `AliyunBailianDataReadOnlyAccess`（只读）系统策略。 +1. **初始化空间**：超级管理员通过全局管理菜单（[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)｜[新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/business_management)｜[弗吉尼亚](https://modelstudio.console.aliyun.com/us-east-1?tab=globalset#/efm/business_management)）创建非默认业务空间，并为该空间开通目标模型的调用、调优或部署权限。 +2. **分配角色**： + - 超级管理员：在 RAM 控制台为 RAM 用户附加 `AliyunBailianFullAccess`； + - 业务空间管理员：在百炼控制台「权限管理」页签中为用户勾选「管理员」角色。 +3. **配置模型权限**： + - 控制台调用：为目标用户分配「模型体验-操作」「批量推理-操作」等页面权限； + - API 调用：为用户在对应空间创建 API Key（自动继承空间级模型权限）。 +4. **启用 OpenAPI**：主账号在 RAM 控制台为 RAM 用户绑定 `AliyunBailianDataFullAccess` 或只读策略。 ## 限制和注意事项 -- **地域强绑定**：业务空间与地域一一对应，API Key、模型限流、用户权限均不可跨地域复用。 -- **默认空间限制**：默认业务空间无法配置模型调用/调优/部署限制，且不支持限流，**严禁用于生产环境**（[权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) 明确建议按环境或业务线新建独立空间）。 -- **API Key 生命周期**：RAM 用户被移出业务空间后，其 API Key **立即失效**（重新加入后恢复）；若在 RAM 控制台删除该用户，则 Key **永久失效**。 -- **账单与预付费权限**：RAM 用户需额外授予 `AliyunBSSReadOnlyAccess`（查看账单）或 `AliyunBSSOrderAccess`（购买预付费）策略，且这些权限作用于**全部阿里云产品**，非百炼专属，授权需谨慎。 -- **IP 白名单支持范围**：仅华北2（北京）地域的 API Key 支持设置 IP 访问白名单。 +- **地域隔离**：业务空间严格绑定单一地域，跨地域资源不可共享；即使同名空间（如 `project-prod-workspace`）在不同地域也互不关联。 +- **API Key 绑定刚性**：一个 API Key 仅归属一个地域、一个业务空间、一个用户，删除用户或将其移出空间将导致其 API Key 失效（重新加入后恢复）。 +- **权限继承逻辑**：API Key 的模型调用能力完全由其所属业务空间的模型开关与限流策略决定，**不受用户控制台页面权限影响**；例如用户无「模型体验」权限但仍可通过 API Key 调用已开通模型。 +- **账单与预付费权限**：RAM 用户需单独授予 `AliyunBSSReadOnlyAccess`（查看账单）或 `AliyunBSSOrderAccess`（购买预付费）策略，且该授权作用于**全阿里云产品**，非百炼专属。 +- **默认空间例外**：所有限流与模型开关功能在默认业务空间中不可用，建议生产环境务必使用自建非默认空间 [原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md index 0df1e08f..3cace675 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md @@ -1,51 +1,55 @@ # application publishing and sharing -百炼平台支持将已发布的智能体应用（Agent 1.0）或工作流应用以多种方式对外共享与集成，包括生成可访问的 UI 应用、发布为跨平台机器人（钉钉/微信）、封装为可复用组件、以及接入音视频实时互动场景。所有发布行为均需基于已上线的应用，并受 Agent 版本、权限空间和计费模型约束。 +百炼平台支持将智能体（Agent 1.0）和工作流应用以多种方式发布与共享，包括作为可复用组件接入其他AI应用、生成网页UI界面、集成至钉钉/微信等第三方平台，以及启用音视频实时互动能力。所有发布行为均需在统一业务空间下完成，且不同发布渠道对应用版本（Agent 1.0 vs Agent 2.0）有明确兼容性要求。 ## 支持的模型/功能 -- **仅限 Agent 1.0**：魔笔分享渠道、钉钉机器人、微信公众号、组件发布、音视频实时互动等功能**全部仅支持 Agent 1.0 智能体应用**；Agent 2.0 应用不支持上述任何发布渠道，仅可通过 API 调用 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 -- **UI 应用支持范围更广**：UI 设计器支持集成**智能体应用（Agent 1.0/2.0）和工作流应用**，但前提是二者与 UI 所属业务空间一致 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 -- **组件来源多样**：智能体应用和工作流应用均可发布为组件，且组件可在智能体或工作流中被引用 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **组件化能力**：智能体或工作流应用可发布为标准化组件，供其他智能体或工作流调用，实现功能复用。组件支持预设系统参数（如 `query`、`imageList`），并可通过别名、描述、可见性、传参方式（业务透传 / 模型识别）精细控制接入逻辑 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **UI应用**：通过可视化UI设计器构建网页界面，支持拖放式布局、多端适配（PC/H5）、权限管理（匿名访问、OIDC/OAuth 2.0）、数据库与文件存储集成，并一键发布至开发或生产环境 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **第三方平台集成**：支持将Agent 1.0应用发布至钉钉机器人、微信公众号，需配置API Key、平台凭证（Client ID/Secret、AppID、卡片模板ID等）及回调地址；也支持音视频实时互动（H5/APP扫码体验或SDK集成） [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **模型兼容性**：组件节点、UI应用及第三方渠道均依赖百炼托管模型（如千问-Max-Latest）或MCP服务（如Amap Maps、QuickChart）；音视频互动仅支持图文对话类应用（智能体/工作流），不支持纯语音/视频原生模型。 -> **注意**：文档 1 明确限定“分享渠道均为 Agent 1.0 功能”，而文档 3 在“准备工作”中指出 UI 设计器支持“智能体应用或工作流应用”，未限定 Agent 版本；结合控制台实际能力，UI 集成对 Agent 2.0 的支持是例外情形，但组件发布、钉钉/微信等渠道严格不兼容 Agent 2.0。 +> **注意**：文档3明确指出“分享渠道（魔笔分享渠道、钉钉、微信、组件、音视频实时互动）均为 **Agent 1.0** 智能体应用的功能。**Agent 2.0** 智能体应用仅支持通过 API 调用，不支持上述分享渠道”，而文档1未提及此限制。开发者在选择发布方式前，必须确认目标应用为Agent 1.0版本，否则将无法配置对应渠道。 ## 关键参数 -| 参数 | 说明 | 约束 | -|------|------|------| -| `API Key` | 用于身份认证与调用鉴权，必须与应用、UI 同属一个业务空间 | 缺失时需在发布流程中创建或管理；钉钉/微信/音视频配置均依赖此密钥 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | -| `query` / `imageList` | 组件预设系统参数：`query`（String，必填）传递用户文本输入；`imageList`（Array，非必填）传递图像公网地址 | 预设参数不可删除，无需显式定义；若组件不处理图像，应将 `imageList` 设置为“是否可见 = 否” [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) | -| `传参方式`（业务透传 / 模型识别） | 决定参数值由调用方提供（业务透传）还是由大模型从上下文推断（模型识别） | **工作流中模型识别无效**：即使配置为“模型识别”，仍需上游节点明确传入值 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | +| 参数名 | 类型 | 必填 | 用途 | 说明 | +|--------|------|------|------|------| +| `query` | String | 是 | 用户输入文本指令 | 组件默认入参，用于传递自然语言查询（如“查询杭州天气”）；在智能体中启用“模型识别”时由大模型自动填充 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) | +| `imageList` | Array | 否 | 图像公网URL列表 | 仅当组件使用图像理解模型时生效；非图像场景需设置“是否可见=否”隐藏该参数 | +| `biz_param` | Object | 否（按需） | 业务透传参数容器 | API调用时传入，用于显式提供`query`等参数值；测试时可在“入参变量配置”中手动填写 | +| API Key | String | 是（UI/第三方渠道必需） | 鉴权凭证 | 必须与应用、UI设计器位于同一业务空间；未正确配置将导致“无法选择API Key”错误 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) | +| 回调地址 / 模板ID / Client ID | String | 是（钉钉/微信必需） | 平台对接凭证 | 钉钉需卡片模板ID + Client ID/Secret；微信需AppID；均需在对应开放平台创建应用后获取 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | ## 使用方式 -1. **UI 应用发布** - 进入应用「发布渠道」页签 → 选择「UI 应用」→ 创建后跳转至 UI 设计器 → 编辑并发布至开发/生产环境。开发环境链接有效期 24 小时，生产环境需订阅付费套餐并绑定域名 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +1. **发布为组件** + - 在应用编辑页点击「发布应用」→ 勾选「发布应用组件」，或进入「组件管理」面板创建； + - 配置组件名称、描述、参数别名、传参方式（业务透传/模型识别）及可见性； + - 接入智能体：在技能配置中选择组件，大模型根据描述+上下文自动触发； + - 接入工作流：拖入「组件节点」，手动连接上游节点输出至`query`等参数。 -2. **钉钉/微信机器人** - - 钉钉：需在钉钉开放平台创建应用，获取 `Client ID`/`Client Secret` 和 AI 卡片 `Template ID`，并在百炼配置回调地址；授权 SLR 及 API-KEY 传输为必要前置步骤 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 - - 微信：需在微信公众号后台获取 `AppID`，完成开发者授权；发布后生成客服二维码供扫码体验。 +2. **发布为UI应用** + - 方式一：从已有应用发布 → 进入发布渠道 → 选择「UI应用」→ 自动填充基础信息； + - 方式二：新建UI → 选模板（如企业AI知识库Lite）→ 配置API Key、智能体、数据库映射 → 拖放组件编辑 → 发布至开发/生产环境； + - 开发环境链接24小时失效，生产环境需订阅付费套餐并绑定自定义域名 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 -3. **组件发布与引用** - - 发布：在应用「发布渠道」→「组件」→ 填写名称、描述、参数别名及传参方式 → 确定发布。 - - 引用：智能体中作为技能添加；工作流中拖入「组件节点」并绑定输入（如 `系统变量/query`）→ 输出可直接接入下游节点 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 - -4. **音视频实时互动** - 仅支持图文类应用（智能体/工作流），需配置 API Key → 生成临时体验二维码（24 小时有效）→ 发布后开通智能媒体服务并授权 SLR → 可选 H5/APP 扫码或 SDK 集成 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +3. **分享至第三方平台** + - **钉钉/微信**：在应用「发布平台」页签授权计算巢AppFlow → 配置平台凭证 → 获取回调地址/二维码 → 在钉钉群@机器人或微信扫码使用； + - **音视频互动**：在「AI实时互动」页签配置API Key → 生成临时体验二维码（24小时有效）→ 发布后支持H5扫码或SDK集成。 ## 限制和注意事项 -- **Agent 版本硬性限制**：除 UI 集成外，所有发布渠道（魔笔、钉钉、微信、组件、音视频）均**不支持 Agent 2.0**；尝试对 Agent 2.0 应用执行相关操作将失败或无响应。 -- **嵌套与多级调用风险**：组件间禁止 A→B→A 的循环调用（导致死循环），也应避免 A→B→C 的三级以上链式调用（易超时） [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 -- **环境与权限隔离**：UI 应用、API Key、智能体/工作流必须归属同一业务空间，否则无法关联或发布 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 -- **计费责任归属**：所有通过分享链接产生的模型调用、存储、带宽等费用，均由应用创建者 UID 账号承担，与访问者无关 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 -- **生产环境成本**：UI 应用发布至生产环境需订阅团队版及以上套餐；开发环境免费但链接 24 小时失效 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **版本限制**：仅Agent 1.0支持UI设计器、钉钉/微信发布、组件化及音视频互动；Agent 2.0仅支持API调用，此差异已在文档3中明确，但文档1未警示，开发者务必核验应用版本 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **嵌套与多级调用**：禁止A调用B、B再调用A（循环嵌套），会导致无限递归；A→B→C等三级以上调用易超时，应尽量扁平化设计 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **参数约束**：工作流中即使设置参数为“模型识别”，也不会自动推断值，必须通过上游节点显式传入；智能体中“模型识别”依赖参数描述质量，描述模糊将导致填充失败。 +- **环境与计费**：UI开发环境免费但24小时失效；生产环境需付费套餐；模型调用、文件存储（1GB免费）、数据库（0.3GB免费）均按量计费 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 +- **业务空间隔离**：API Key、应用、UI设计器必须归属同一业务空间，否则无法关联资源或出现配置项不可见问题。 ## 来源文档 -- [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) - [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) +- [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md index e00e62de..fc7cbc9b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md @@ -1,42 +1,45 @@ # application [support](support.md) -`application support` 指百炼平台为构建和运行 AI 应用（含智能体、RAG 应用、插件集成等）所提供的核心能力支持体系，涵盖模型调用、插件扩展、知识检索增强、[流式输出](../concepts/streaming-output.md)及数据管理等关键环节。开发者需结合服务协议与技术规范进行开发与部署。相关法律约束和合规要求详见 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 +`application support` 指百炼平台为构建和运行 AI 应用（如智能体、RAG 应用、[插件](../concepts/plugin.md)集成应用等）所提供的核心能力支持体系，涵盖模型调用、[插件](../concepts/plugin.md)扩展、知识检索增强、[流式输出](../concepts/streaming-output.md)等关键功能。开发者可通过 Assistant API 或 Agent 框架接入，需关注参数配置、协议约束及服务边界。所有能力均受 [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=5176.28197581.0.0.16e829a4HTC9FE) 约束，具体条款详见 [原文标题](../../raw/application-user-guide/application-support/application-related-agreements.md)。 ## 支持的模型与功能 -- **插件能力**：官方提供六类内置插件：Python代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub搜索；其中部分插件需申请开通。 -- **自定义插件**：支持通过 API 注册自定义插件，大模型可解析其参数定义并调用（参见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第3条）。 -- **RAG（知识检索增强）**：支持多知识库并行检索，按配置策略（如相似度得分）选取 topN 片段后融合生成，适用于问答、客服、教育等场景（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第5条）。 -- **流式与增量输出**：支持 `stream=True` 实现流式响应；进一步启用 `incremental_output=True` 可获得真正增量式 token 输出（非全量重传），适用于前端实时渲染（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第8条）。 +- **内置[插件](../concepts/plugin.md)**：当前官方支持 6 类插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索。部分插件需申请开通 [原文标题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **自定义插件**：支持通过符合 OpenAPI 规范的 HTTP 接口注册；大模型可理解插件描述及参数结构，并据此生成调用逻辑；但**仅支持透传 `Authorization` header**，其他自定义 header 将被忽略（见文档 1 第 10 条）。 +- **RAG（知识检索增强）**：支持多知识库并行检索，按配置权重与相似度得分聚合结果后选取 topN 片段；适用于问答、客服、教育等场景 [原文标题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **[流式输出](../concepts/streaming-output.md)**：支持增量式流式响应，需同时设置 `stream=True` 和 `incremental_output=True`（文档 1 第 8 条）。 -> **注意**：文档2中第4条称“Assistant API 可提供各种类，方便调优”，但未明确具体类名或接口契约；当前 SDK 与 OpenAI 兼容 API 中实际暴露的是 `assistant` 类型资源（非 `Assistant` 类），该描述易引发歧义，建议以 [API 参考文档](https://help.aliyun.com/zh/model-studio/developer-reference) 为准。 +> **注意**：文档 1 中“Agent 和 Assistant API 的最大区别”（第 4 条）表述模糊且缺乏技术细节，实际差异应以最新版 [Assistant API 文档](https://help.aliyun.com/zh/model-studio/assistant-api-overview) 为准；该 FAQ 条目已过时，不建议作为架构选型依据。 ## 关键参数 | 参数 | 类型 | 说明 | |------|------|------| | `stream` | bool | 启用[流式输出](../concepts/streaming-output.md)（逐 token 返回） | -| `incremental_output` | bool | 在 `stream=True` 基础上启用增量式输出（仅返回新增 token，非累计内容） | -| `MD5`（文件上传） | string | 文件完整性校验值，必填（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第3条） | -| `authorization`（插件调用） | header | 插件 HTTP 请求中唯一支持透传的 header；其他自定义 header 将被丢弃（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第10条） | +| `incremental_output` | bool | 启用增量式流式输出（仅返回新增内容，非全量重传） | +| `knowledge_retrieval` | object | 控制 RAG 行为，含 `top_k`、`score_threshold`、`enable_rerank` 等子字段 | +| `plugins` | list | 指定启用的插件 ID 列表（如 `["python_interpreter", "qrcode_generator"]`） | ## 使用方式 -- **插件调用**：注册插件时需声明 `name`、`description`、`parameters`（JSON Schema 格式），系统自动注入至模型上下文；调用时模型生成结构化 function call 请求，平台负责路由与执行。 -- **RAG 应用测试**：若检索结果不准确，可通过回复下方“问题反馈”按钮提交，或复制 `RequestId` 提交工单（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第6条）。 -- **Markdown 渲染**：模型输出中的 `**text**` 等标记需由前端自行解析并渲染为加粗等样式（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第7条）。 -- **备案与合作**：接入通义千问模型并上架应用市场/小程序前，须完成 [应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)，并提交工单申请合作协议（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第11条）。 +- **调用入口**：统一通过 `/v1/applications/{app_id}/chat` 接口发起请求（RESTful）或使用 SDK 封装的 `assistant.chat()` 方法。 +- **插件配置**：在应用控制台中绑定插件，或在 API 请求中显式声明 `plugins` 参数；自定义插件需提前在「插件管理」中完成注册与鉴权配置。 +- **RAG 配置**：上传文件至知识库（仅支持小写后缀 `.pdf`, `.doc`, `.docx`；见文档 1 数据管理第 1 条），并在应用中关联知识库 ID；空行会导致结构化数据截断（文档 1 第 4 条）。 +- **错误处理**：RAG 结果不准确时，可通过界面反馈按钮提交问题，或复制 `RequestId` 提交工单 [原文标题](../../raw/application-user-guide/application-support/application-faq.md)。 ## 限制和注意事项 -- **文件上传**：仅支持 `.pdf`（小写后缀）、`.doc`、`.docx`；结构化数据导入时，空行将导致后续行被截断（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第1、4条）。 -- **知识库容量**：单业务空间上限为 10 万个文档；超限时需提交工单申请扩容（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第2条）。 -- **协议约束**：所有应用必须遵守 [阿里云百炼服务协议](../../raw/application-user-guide/application-support/application-related-agreements.md) 及 [开源模型协议条款说明](../../raw/application-user-guide/application-support/application-related-agreements.md)，尤其注意数据使用、模型输出责任归属等条款。 -- **插件安全限制**：自定义插件无法透传除 `Authorization` 外的任何 HTTP header，服务端会主动剥离其余 header 字段（[常见问题](../../raw/application-user-guide/application-support/application-faq.md) 第10条）。 +- **文件上传**：单业务空间上限 10 万个文档；超限时需提交工单申请扩容（文档 1 第 2 条）。 +- **MD5 校验**：上传接口必填 `Content-MD5` 头，用于验证文件完整性（文档 1 第 3 条）。 +- **协议约束**： + - 自定义插件不收费，但 [prompt](prompt.md) 优化、API 调用及测试窗使用将计费； + - 所有应用上线前须完成 [应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)，并签署通义千问合作协议（文档 1 应用备案章节）； + - 开源模型使用须遵守 [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms)（见 [原文标题](../../raw/application-user-guide/application-support/application-related-agreements.md)）。 +- **渲染提示**：模型输出中的 `**text**` 为 Markdown 加粗语法，需前端自行解析渲染（文档 1 第 7 条）。 ## 来源文档 -- [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) - [常见问题](../../raw/application-user-guide/application-support/application-faq.md) +- [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md index 203965d9..84b6d259 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md @@ -1,56 +1,47 @@ # application [use cases](use-cases.md) -百炼平台支持多种典型业务场景下的 AI 应用快速落地，核心模式为“大模型应用（LLM） + 知识增强（RAG） + 多端集成”。所有方案均基于统一的百炼应用作为推理后端，通过 AppFlow 实现零代码连接主流企业通讯与内容平台（如网站、企业微信、钉钉、微信公众号），并支持本地化知识库部署。开发者可复用同一套 Prompt 工程、知识库配置和评测流程，显著降低多渠道 AI 助手的构建与维护成本。 +阿里云百炼平台支持将大模型能力快速集成至多种主流企业级通信与网站渠道，构建面向客户、员工或私域用户的 AI 助手。典型场景包括在网站、企业微信、微信公众号、钉钉等平台嵌入 RAG 增强的智能问答服务，全程无需编码，依托 AppFlow 实现低代码连接，结合百炼应用配置与知识库管理完成端到端交付。所有方案均兼容新用户免费额度，适用于快速验证与轻量级生产部署。 ## 支持的模型/功能 -- **基础模型**：默认推荐 `qwen-plus`（即文档中提及的“千问-Plus”或“Qwen3.5-Plus”），在效果、速度与成本间取得平衡；也可按需切换为 `qwen-max`（高精度）、`qwen-turbo`（低延迟）或 `qwen-flash`（超低成本）。> **注意**：文档 1 明确指定模型为 `Qwen3.5-Plus`，而文档 2、3、4 均写为“千问-Plus”，二者实际为同一模型的不同命名；当前控制台显示名称以 `qwen-plus` 为准，建议开发者以控制台实际可选模型列表为准 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 -- **核心能力**： - - 智能体（Agent）应用：支持角色设定（Prompt）、工具调用（如知识库检索）、多轮对话管理； - - RAG 增强：通过知识库实现私有领域问答，支持 PDF/DOCX/TXT/Excel 等格式上传与向量化； - - 多模态支持：文档 2、4、5 均明确列出 `.png`, `.jpg`, `.jpeg`, `.bmp`, `.gif` 等图片格式支持，适用于产品图谱、说明书图像理解等场景。 +- **核心模型**：推荐使用 `Qwen3.5-Plus`（见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)）或 `千问-Plus`（见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)），该模型在效果、速度与成本间取得平衡；亦支持 `qwen-max`（高精度）、`qwen-turbo`（低延迟）等变体，适用于不同响应 SLA 要求。 +- **RAG 增强能力**：所有用例均依赖百炼知识库实现私有知识注入，支持 PDF/DOCX/TXT/Excel 等格式上传（见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)），并提供“必定调用”“按需调用”等知识引用策略。 +- **本地化 RAG 选项**：对于需完全控制文档切分、嵌入模型与向量存储的场景，可采用 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 方案，支持自定义切分逻辑、本地部署嵌入模型（如 GTE-Chinese-Large）及 Gradio API 对接。 + +> **注意**：文档 1 明确指定模型为 `Qwen3.5-Plus`，而文档 2、3、5 均使用 `千问-Plus`。二者为不同代际模型，`Qwen3.5-Plus` 是更新版本，具备更强推理与多轮对话能力；若需一致性建议优先选用 `Qwen3.5-Plus`，除非业务明确要求兼容旧版 `千问-Plus` 接口行为。 ## 关键参数 -| 参数类别 | 参数名 | 说明 | 可配置位置 | -|----------|--------|------|------------| -| **模型层** | `temperature` | 控制生成随机性，值域通常为 0.0–1.0 | 百炼应用配置页、[基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 的 Gradio 界面 | -| | `max_tokens` | 限制模型输出最大 token 数 | 同上 | -| | `top_p` / `top_k` | 影响采样多样性 | 百炼应用高级设置（部分模型支持） | -| **RAG 层** | `retrieval_top_k` | 召回片段数（如“召回 3 个最相关段落”） | [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 的 Gradio 界面；百炼知识库引用配置中对应“相似度阈值”与“权重” | -| | `similarity_threshold` | 过滤低相关性召回结果的阈值（0–1） | 百炼应用配置页 > 知识库 > “相似度阈值”字段 | -| | `chunk_size` / `chunk_overlap` | 文档切分粒度（影响检索精度） | 百炼知识库创建时的“索引设置”；本地 RAG 应用中可自定义切分逻辑 | +- **百炼应用 ID 与 API Key**：所有集成方案必需，用于 AppFlow 或客户端调用百炼推理服务（见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) 第 1.2 节）。 +- **平台凭证**： + - 企业微信：需 `企业 ID`、`AgentId`、`Secret`（见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) 第 2.2 节）； + - 微信公众号：需 `AppID` 及管理员扫码授权（见 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) 第 2 节）； + - 钉钉：需 `Client ID` 与 `Client Secret`（见 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) 第 2.2 节）； + - 网站嵌入：仅需前端脚本，无后端凭证（见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) 第 3.2 节）。 +- **RAG 参数**（本地 RAG 场景）：包括 `召回片段数`、`相似度阈值`、`温度`、`最大回复长度` 等，可在 `chat.py` 中直接调整（见 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 第三节）。 ## 使用方式 -1. **统一后端：创建百炼应用** - 所有场景均始于百炼控制台的[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 创建**智能体应用** → 配置模型、Prompt 与知识库。应用发布后获得唯一 `AppID` 和调用所需的 `API Key`。 - -2. **前端集成：通过 AppFlow 连接目标平台** - - **网站嵌入**：使用 AppFlow 创建 AI 助手 → 关联百炼应用 → 生成悬浮挂件脚本 → 插入 HTML 即可 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 - - **企业微信/钉钉/微信公众号**：使用 AppFlow 预置模板（如“企业微信自建应用大模型自动回复”）→ 分别配置平台凭证（企业 ID/AgentId/Secret 或 Client ID/Secret 或 AppID）与百炼凭证 → 获取 Webhook URL → 在对应平台后台完成消息接收配置。 - -3. **知识增强：配置知识库（可选但推荐）** - - 上传文件至百炼[数据中心](https://bailian.console.aliyun.com/?tab=app#/data-center) 或[数据连接](https://bailian.console.aliyun.com/cn-beijing?tab=app#/connector/list)； - - 在[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面创建标准版知识库； - - 在应用配置页启用知识库，设置调用方式（如“必定调用”）及相似度阈值。 +1. **创建百炼应用**：统一通过百炼控制台 → 应用管理 → 创建智能体应用，配置 Prompt（如 `"你叫小助，可以帮助用户解答产品选购、使用等方面的问题。"`）并发布。 +2. **配置知识库**：上传文档 → 创建知识库 → 在应用配置中绑定知识库并设为“必定调用”（各文档均采用此流程，细节略有差异：文档 1 使用“数据连接”页签，文档 2/3/5 使用“文件”或“知识库”页签）。 +3. **集成至目标平台**： + - **网站**：通过 AppFlow 创建 AI 助手 → 获取悬浮挂件脚本 → 插入 HTML（见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)）； + - **企业微信/钉钉/微信公众号**：使用 AppFlow 预置模板 → 配置平台凭证与百炼凭证 → 获取 Webhook URL → 在对应平台后台完成消息接收配置（如企业微信的“API接收消息”、钉钉的“HTTP模式机器人”、公众号的“服务器配置”）。 +4. **验证与日志**：各平台均支持直接对话测试；如需审计，可通过 AppFlow 添加 SLS 日志节点记录对话（见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) 第 4 节）。 ## 限制和注意事项 -- **免费额度与计费**：新用户享有百炼免费额度，覆盖教程全部操作；额度耗尽后按 token 计费，具体见 [新用户免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota) [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 -- **文件限制**：单文档最大 100 MB 或 1000 页，单图片最大 20 MB，最多上传 200 个文件；知识库创建过程需等待解析（通常 1–6 分钟）。 -- **平台特异性约束**： - - 微信公众号：未认证订阅号仅支持被动回复（5 秒超时限制），建议完成认证或选用 `qwen-turbo` 模型提速 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)； - - 企业微信：配置 API 接收消息时需通过域名主体校验，若无自有备案域名，需通过 AppFlow 的 Nginx 代理或计算巢实例解决 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)； - - 钉钉：机器人消息接收模式**必须选择 HTTP 模式**，Stream 模式不兼容 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md)。 -- **本地 RAG 场景**：适用于需完全私有化部署、灵活控制文档切分与嵌入模型的场景，但需自行维护 Python 环境（3.8–3.12）及依赖，且不直接集成百炼控制台的统一监控与评测能力 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 +- **认证要求**：微信公众号未认证时仅支持被动回复（5 秒超时限制），建议完成认证以启用主动消息能力（见 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) 第 2 节说明）。 +- **可信 IP 与域名**：企业微信/钉钉要求配置可信 IP 白名单；若使用 AppFlow Webhook，需通过计算巢 Nginx 代理或自有域名解析解决主体校验问题（见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) “常见问题”章节）。 +- **文件限制**：云端知识库单文档上限为 100 MB 或 1000 页；本地 RAG 方案亦不建议上传超 100 MB 文件（见 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) 第三节）。 +- **模型兼容性**：`Qwen3.5-Plus` 为文档 1 所指定，其余文档未更新至该版本，开发者需自行确认控制台可用模型列表，避免因模型下线导致配置失败。 ## 来源文档 - [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) -- [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) +- [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md index e9119569..a43c4c3c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md @@ -1,6 +1,6 @@ # bailian [application call](../api/application-call.md)ing -百炼应用调用是指通过 DashScope SDK 或标准 HTTP API，将已发布的百炼智能体应用或工作流应用集成到自有业务系统中。调用过程统一使用 `POST /api/v1/apps/{app_id}/completion` 接口，支持单轮/多轮对话及插件参数透传，适用于各类 AI 增强场景。 +百炼应用调用是指通过 DashScope SDK 或标准 HTTP API，将已发布的百炼智能体应用或工作流应用集成至第三方业务系统。调用过程统一使用 `POST /api/v1/apps/{app_id}/completion` 接口，支持单轮/多轮对话、自定义[插件](../concepts/plugin.md)参数透传等核心能力，适用于各类 AI 增强型业务场景。 ## 支持的模型/功能 @@ -8,66 +8,75 @@ - [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)（即单智能体应用） - [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)（原“智能体编排应用”，已由工作流应用替代） - **核心能力**： - - 单轮文本生成（`prompt` 输入 → `output.text` 输出） + - 单轮文本生成（`prompt` 输入） - 多轮对话（通过 `session_id` 或显式 `messages` 数组管理上下文） - - 自定义插件参数透传（需在应用内配置插件并启用“业务透传”参数模式） -- **底层模型**：实际执行模型由应用发布时绑定的模型决定（如 `qwen-max`、`qwen-plus`），调用方无需指定；响应中 `usage.models[].model_id` 字段可查实际使用的模型。 + - 自定义[插件](../concepts/plugin.md)参数透传（需在应用中配置[插件](../concepts/plugin.md)节点并启用业务透传） +- **底层模型**：实际执行由应用绑定的模型（如 `qwen-max`、`qwen-plus` 等）完成，调用方无需指定模型 ID；模型信息在响应 `usage.models[].model_id` 中返回。 -> **注意**：文档 2 明确声明“本文档仅适用于华北2（北京）地域”，而文档 1 和文档 3 未限定地域。若跨地域调用失败，请优先确认应用所在地域与 API Endpoint 是否匹配（当前仅北京地域支持工作流应用调用）。 +> **注意**：文档2明确声明“本文档仅适用于华北2（北京）地域”，而文档1和文档3未限定地域。若跨地域调用失败，请优先确认应用部署地域与 API Endpoint 是否匹配（当前所有示例均指向 `dashscope.aliyuncs.com`，该域名默认路由至北京地域）。 ## 关键参数 | 参数名 | 类型 | 必填 | 说明 | |--------|------|------|------| -| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面获取 | -| `prompt` | string | 否（多轮对话时可省略） | 当前轮次的用户输入指令；若使用 `messages` 则此字段被忽略 | -| `biz_params` | object | 否 | 用于传递自定义插件参数，结构为 `{ "user_defined_params": { "": { "": } } }`；详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) | -| `session_id` | string | 否 | 启用云端会话管理时使用，有效期 1 小时，最多 50 轮 | -| `messages` | array | 否 | 替代 `prompt` 的推荐方式，格式同 OpenAI：`[{ "role": "user", "content": "..." }, { "role": "assistant", "content": "..." }]`；若同时传 `session_id` 和 `messages`，以 `messages` 为准 | +| `app_id` | string | ✅ | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面获取 | +| `prompt` | string | ⚠️（见下文） | 单轮请求时必需；若使用 `messages` 进行多轮对话，则此项可省略 | +| `biz_params` | object | ❌ | 用于传递自定义插件参数，结构为 `{ "user_defined_params": { "": { "": } } }`。详见 [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) | +| `session_id` | string | ❌ | 启用云端会话管理时提供，有效期 1 小时，最多支持 50 轮对话 | +| `messages` | array | ❌ | 替代 `prompt` 的多轮对话方式，格式同 OpenAI-style `[{ "role": "user/system/assistant", "content": "..." }]`；若同时传 `session_id` 和 `messages`，以 `messages` 为准 | ## 使用方式 ### 1. 准备工作 -- 获取 API Key：前往[密钥管理](https://bailian.console.aliyun.com/?tab=model#/api-key)创建并配置为环境变量 `DASHSCOPE_API_KEY`（**强烈推荐**，避免硬编码） -- 获取 `app_id`：在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)中复制目标应用 ID -- 安装 SDK（可选）：Python 执行 `pip install -U dashscope`；Java/Node.js 等参见对应语言示例 - -### 2. 发起调用 -- **SDK 方式（推荐）**： - ```python - from dashscope import Application - response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id="YOUR_APP_ID", - prompt="你是谁？", - biz_params={"user_defined_params": {"plugin_abc": {"query": "test"}}} - ) - print(response.output.text) - ``` -- **HTTP 方式（通用）**： - ```bash - curl -X POST https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "input": { - "prompt": "你是谁？", - "biz_params": { - "user_defined_params": { - "plugin_abc": {"query": "test"} - } - } - } - }' - ``` +- 获取 API Key：前往 [密钥管理](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建并复制。 +- 获取 `app_id`：在 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) 页面对应应用卡片上复制。 +- （推荐）配置环境变量：`export DASHSCOPE_API_KEY=sk-xxx`，避免代码硬编码。 + +### 2. 调用方式（任选其一） +- **DashScope SDK**（Python/Java/Node.js/C#/Go 等）：封装了认证、序列化与错误处理，推荐生产环境使用。SDK 版本要求：Python ≥ 1.14.0（插件参数）、Java ≥ 2.12.0（多轮对话支持）。 +- **HTTP API**：直接调用 `POST https://dashscope.aliyuncs.com/api/v1/apps/{app_id}/completion`，需手动设置 `Authorization: Bearer ` 请求头。 + +### 3. 示例（Python SDK） +```python +from dashscope import Application +import os + +# 单轮调用（含插件参数） +biz_params = { + "user_defined_params": { + "your_plugin_code": {"article_index": 2} + } +} +response = Application.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + app_id="YOUR_APP_ID", + prompt="寝室公约内容", + biz_params=biz_params +) + +# 多轮调用（显式 messages） +messages = [ + {"role": "user", "content": "你是谁？"}, + {"role": "assistant", "content": "我是通义千问。"}, + {"role": "user", "content": "今天天气如何？"} +] +response = Application.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + app_id="YOUR_APP_ID", + messages=messages # 注意：此时不传 prompt +) +``` ## 限制和注意事项 -- **地域限制**：工作流应用调用仅支持华北2（北京）地域；智能体应用无明确地域限制，但建议与应用部署地域一致以降低延迟。 -- **会话管理**：`session_id` 有效期为 1 小时且最多承载 50 轮对话；生产环境推荐自行维护 `messages` 数组以获得完全控制权。 -- **插件参数**：必须在插件工具配置中将参数“传参方式”设为 **业务透传**，否则 `biz_params` 中的参数不会生效。 -- **错误处理**：所有调用均返回标准 HTTP 状态码（如 `401 Unauthorized`、`404 Not Found`）及 `request_id`，用于问题定位；错误码详情请参考[开发者参考文档](https://help.aliyun.com/zh/model-studio/developer-reference/error-code)。 -- **安全要求**：API Key **严禁硬编码**于源码或前端代码中；务必通过环境变量或密钥管理服务注入。 +- **地域限制**：工作流应用调用[仅支持华北2（北京）地域](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)，智能体应用无明确地域限制，但建议保持应用与调用端地域一致。 +- **会话管理**：`session_id` 由服务端生成并返回于响应中（`output.session_id`），客户端需自行保存并在后续请求中复用；若使用 `messages`，则完全由客户端维护上下文。 +- **插件参数**：必须满足以下条件才能生效： + - 插件工具的输入参数“传参方式”必须设为 **业务透传**； + - 应用内已关联该插件且已发布； + - `biz_params.user_defined_params.` 中的 `plugin_code` 必须与控制台插件卡片显示的 ID 完全一致。 +- **错误处理**：所有调用均需检查 `response.status_code`（HTTP）或 `response.status_code`（SDK），非 `200` 时解析 `message` 和 `request_id` 用于排查，参考 [错误码文档](https://help.aliyun.com/zh/model-studio/developer-reference/error-code)。 +- **安全实践**：严禁在代码中硬编码 `DASHSCOPE_API_KEY`；务必通过环境变量或密钥管理服务注入。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md index 84ae13e8..e68b139e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md @@ -1,49 +1,52 @@ # data connection overview -数据连接是阿里云百炼平台统一管理外部数据源的核心能力，为应用提供安全、可控的数据接入通道。它支持结构化与非结构化数据的接入，并通过平台托管或流处理两种模式实现数据访问，是构建知识增强型智能体（Agent）和 RAG 应用的基础组件。所有连接器均需在业务空间内创建并绑定至具体应用，其配置直接影响后续检索与调用行为。 +数据连接是阿里云百炼平台统一管理外部数据源的核心机制，为应用提供安全、可控的数据接入能力。它支持将企业自有数据库、文档系统、对象存储等异构数据源接入百炼环境，并在对话或智能体执行过程中实时检索与引用。所有连接器均通过统一控制台创建与配置，权限由RAM策略集中管控。 ## 支持的模型/功能 -数据连接器按数据访问方式分为两类： +数据连接器按数据访问模式分为两类： -- **平台托管类**：适用于静态文件与表格数据，包括 - - `文件`：支持 PDF、Word、Markdown 等非结构化文档，依赖[文档理解](https://help.aliyun.com/zh/document-mind/product-overview/overview-of-document-understanding#9a4f5fb91fpps)能力进行解析（详见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“导入文件”章节）； - - `表格`：支持 CSV、Excel 等结构化数据，支持自定义表头与字段类型（如 `image_url`），但表结构一旦确定不可修改。 +- **平台托管型**：适用于非结构化与结构化静态数据，包括： + - `文件`：支持 PDF、Word、Markdown 等格式，依赖[文档理解](https://help.aliyun.com/zh/document-mind/product-overview/overview-of-document-understanding#9a4f5fb91fpps)能力进行向量化（详见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“导入文件”章节）； + - `表格`：支持 CSV、Excel（XLS/XLSX），支持自定义表头与字段类型（如 `image_url` 字段触发图片向量索引生成）。 -- **流处理类**：适用于实时数据库与在线服务，包括 - - `MySQL`、`PostgreSQL`、`PolarDB-X 2.0`：仅通过 **DMS 导入数据源** 方式创建的连接器支持执行 SQL 查询；自定义方式创建的连接器仅支持元数据同步，不支持直接查询（该限制在 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各数据库连接器说明中反复强调）； - - `语雀`：对接语雀知识库，依赖个人访问 Token，**仅支持公网版本语雀**； - - `OSS`：访问对象存储中的文件，需开通向量检索服务方可使用 `searchOSSFile` 和 `searchOSSFileByFileName` 工具（参见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) “OSS连接器”章节）。 +- **流处理型**：适用于实时查询动态数据，包括： + - `MySQL`、`PostgreSQL`、`PolarDB-X 2.0`：仅通过 **DMS 导入数据源** 方式创建的连接器支持 SQL 查询执行（[原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) 明确指出“创建自定义数据源方式不支持直接执行SQL”）； + - `语雀`：对接语雀知识库，需公网版 [Token](../concepts/token.md)； + - `OSS`：访问对象存储中文件，依赖向量检索服务实现 `searchOSSFile` 等工具调用（参见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) “OSS连接器”说明）。 -> **注意**：`MySQL` 与 `PostgreSQL` 连接器均要求数据库账号具备高权限（如 REPLICATION 或 Superuser），且 PostgreSQL 必须将 `wal_level` 设置为 `logical`；而 PolarDB-X 2.0 **仅支持私网连接**，不支持公网，且不兼容自建实例——这些关键差异已在原始文档中明确区分，开发者需严格遵循。 +> **注意**：`PolarDB-X 2.0` 连接器**仅支持私网接入**，且不支持自建实例；而 `MySQL` 和 `PostgreSQL` 均支持公网/私网双模式，但后者要求 `wal_level=logical` —— 此配置差异在原始文档中被明确列出，无矛盾。 ## 关键参数 -| 参数类别 | 关键字段 | 说明 | -|----------|----------|------| -| **通用** | 连接器名称、描述 | 名称需唯一且易识别；描述影响智能体调用准确度，建议明确数据内容与用途 | -| **文件/表格** | 存储位置（平台存储 / 自有 OSS） | 平台存储提供免费额度（文件连接器限 200,000 文件 / 1 TB，表格连接器限 1 TB）；自有 OSS 需添加 `bailian-connector-access` 标签（值为 `ReadAndWrite`） | -| **数据库类** | 数据库地址、端口、用户名、密码、dbName（PostgreSQL/PolarDB-X 必填） | MySQL 默认端口 3306，PostgreSQL 默认 5432；PolarDB-X 仅支持私网，且数据库地址/端口由实例自动填充 | -| **语雀/OSS** | Tenant access token（语雀）、Bucket 选择（OSS） | 语雀 Token 需从 [语雀开放 API](https://www.yuque.com/yuque/developer/api) 获取；OSS Bucket 需添加 `bailian-datahub-access` 标签（值为 `read`），且**不支持归档/冷归档存储类型** | +| 参数类别 | 关键项 | 说明 | +|----------|--------|------| +| **通用** | 连接器名称、描述 | 名称需唯一可识别；描述影响智能体调用准确度，建议注明数据范围与用途 | +| **平台托管** | 存储位置（平台存储 / 自有OSS） | 平台存储提供限时免费额度（文件：200,000个/1TB；表格：1TB）；自有OSS需添加 `bailian-connector-access` 标签（值 `ReadAndWrite`） | +| **流处理（DB类）** | 数据库地址、端口、用户名、密码、dbName（PostgreSQL/PolarDB-X 必填） | MySQL 默认端口 3306，PostgreSQL 默认 5432；PolarDB-X 仅支持私网，且数据库地址/端口自动填充 | +| **语雀/OSS** | Tenant access token（语雀）、Bucket 名称（OSS） | 语雀 [Token](../concepts/token.md) 需从[语雀开放 API](https://www.yuque.com/yuque/developer/api) 获取；OSS Bucket 需添加 `bailian-datahub-access` 标签（值 `read`），并开通向量检索服务 | ## 使用方式 -1. **创建连接器**：进入 [数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list) 页面 → 单击 **创建连接器** → 选择类型 → 填写基本信息与连接参数 → （可选）点击 **开始检测** 或 **连接检测** 验证连通性 → 确认创建。 -2. **导入数据**： - - 文件连接器：进入详情页 → 选择类目 → **导入数据** → 本地上传 → 配置解析方式（默认/自定义）与标签 → 确认； - - 表格连接器：进入详情页 → 在 **数据表管理** 下新建或选择数据表 → 上传 Excel 或自定义表头 → 确保列名与类型严格匹配； - - 数据库/OSS/语雀类连接器无需手动导入，数据实时访问。 -3. **绑定应用**：在应用配置中显式关联已创建的数据连接器，方可启用对应检索能力（如 `searchFile`、`querySQL` 等工具）。 +1. **创建连接器**：进入 [数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list) 控制台 → 单击“创建连接器” → 选择类型 → 填写基本信息与连接参数 → （可选）点击“开始检测”验证连通性 → 确认提交。 +2. **导入数据**： + - 文件/表格连接器：进入连接器详情页 → 选择类目或数据表 → 上传本地文件或配置表结构 → 设置解析方式（如大模型文档解析）与标签 → 提交。 + - DB 类连接器：无需手动导入，数据保留在原库，应用通过 SQL 工具（如 `queryMySQL`）实时查询。 +3. **在应用中调用**：在智能体或 API 调用中，通过预置工具（如 `searchFile`、`queryPostgreSQL`）指定连接器 ID 与查询条件，平台自动路由至对应数据源。 ## 限制和注意事项 -- **权限约束**：RAM 用户需主账号授予 `AliyunBaiLianFullAccess` 或最小化权限策略（含 `bailian:ListConnectors`、`bailian:CreateConnector` 等动作），详见 [权限管理](https://help.aliyun.com/zh/model-studio/application-permission-management-overview)。 -- **网络与白名单**：MySQL 公网连接需将百炼服务 IP 段加入数据库白名单；PostgreSQL 自建实例需配置 `pg_hba.conf` 允许 `100.64.0.0/16` 访问；PolarDB-X 仅支持私网，必须同地域部署。 -- **解析与时效性**：文件导入后生成独立副本，**仅支持查看最近 90 天内导入的文件**；高峰时段解析可能延迟数小时，偶现超时，建议错峰操作。 -- **功能边界**： - - `MySQL`/`PostgreSQL`/`PolarDB-X 2.0` 的 SQL 执行能力**仅对 DMS 导入方式生效**，自定义方式不支持（原始文档多次强调，勿混淆）； - - OSS 连接器若未开通向量检索服务，则 `searchOSSFile` 等工具不可用； - - 文件连接器**不支持直接导入 JSON/CSV/YAML**，需先转为 XLSX/XLS 格式。 +- **权限前提**：必须由主账号或已授权 RAM 用户操作；涉及 SLR 授权（如 DMS、DTS、PolarDB-X 角色）时需显式同意（见 [原文标题](../../raw/application-user-guide/data-connection-overview/data-connection.md) “前置条件”与各连接器章节）。 +- **存储限制**： + - 文件连接器：仅可查看最近 90 天内导入的文件（文件副本独立存储，不与源同步）； + - OSS 连接器：**不支持归档、冷归档、深度冷归档类型 Bucket**；开启 Referer 防盗链时，须将 `*.console.aliyun.com` 加入白名单。 +- **功能限制**： + - MySQL/PostgreSQL/PolarDB-X 的 SQL 执行能力**严格绑定 DMS 导入方式**，自定义数据源方式仅支持元数据同步，不可查； + - 表格连接器中 `image_url` 字段要求链接**公开可访问**，否则图片抓取失败； + - 文件导入暂不支持 JSON/CSV/YAML 格式，需转为 XLSX/XLS 后再上传。 +- **网络与配置**： + - PostgreSQL 自建实例需额外配置 `listen_addresses` 允许 `100.64.0.0/16` 网段访问； + - PolarDB-X 2.0 连接器**仅支持私网**，且必须与实例同地域。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md index bf395d88..4564e2df 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md @@ -1,42 +1,66 @@ # fine tuning -fine tuning 是阿里云百炼平台提供的模型定制化能力，允许开发者基于自有数据对预训练模型进行增量训练，以提升其在特定任务、领域或风格上的表现。该能力覆盖文本生成、视觉理解、图像/视频生成及语音合成等多模态模型，支持 SFT（监督微调）、CPT（持续预训练）和 DPO（直接偏好优化）等多种训练范式。所有 fine tuning 任务当前均仅支持华北2（北京）地域，且需使用该地域的 API Key [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +fine tuning（微调）是阿里云百炼平台提供的核心模型优化能力，允许开发者基于自有数据对预训练大模型进行定制化训练，从而在特定任务、领域或风格上显著提升效果。它适用于图像生成、视频生成、文本生成、语音合成等多种模态，支持高效微调（LoRA）与全参微调两种模式，兼顾效果与成本。微调后的模型可独立部署为在线服务，直接用于生产环境。 -## 支持的模型与功能 +## 支持的模型/功能 -- **文本生成**：支持 Qwen 系列全量模型（如 `qwen3-8b`, `qwen2.5-7b-instruct`）及千问-VL 视觉语言模型，提供 CPT、SFT（含高效 LoRA 和全参）、DPO 三种训练方式。具体支持矩阵详见 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -- **图像生成**：仅支持万相系列模型（`wan2.7-image-pro`, `wan2.7-image`），采用 SFT-LoRA 高效微调，适用于文生图（t2i）和图生图（i2i）场景 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -- **视频生成**：支持万相图生视频模型（`wan2.7-i2v`, `wan2.2-kf2v-flash` 等），同样基于 SFT-LoRA，支持基于首帧或首尾帧的特效/动作定制 [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 -- **语音合成**：仅支持 `cosyvoice-v3-flash` 模型，通过 SFT-LoRA 进行单发音人音色定制，产物为独立部署的专属音色模型，不支持多音色切换 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +百炼平台支持多模态、多场景的 fine tuning，覆盖主流业务需求： -> **注意**：文档 4 中称“阿里云百炼推荐您如果**模型支持全参训练，请优先选择全参训练**”，但文档 1、2、7 均明确限定图像、视频、语音类模型**仅支持 `efficient_sft`（LoRA）**，且文档 3 的支持矩阵中，`wan*` 和 `cosyvoice*` 系列未列出任何全参训练选项。因此，对非文本生成模型，全参训练不可用，该推荐不适用。 +- **图像生成**：支持 `wan2.7-image-pro` 和 `wan2.7-image` 模型，通过 SFT-LoRA 微调实现人物形象、IP 风格、特效（如“末日废土红黑机甲”）的稳定复现 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **视频生成**：支持 `wan2.7-i2v`、`wan2.5-i2v-preview`、`wan2.2-i2v-flash`（首帧驱动）及 `wan2.2-kf2v-flash`（首尾帧驱动）等模型，用于定制动作、转场与特效（如“金钱雨”“时尚杂志”） [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 +- **文本生成**：覆盖 Qwen 系列全量模型（如 `qwen3-8b`、`qwen3-32b`、`qwen2.5-72b-instruct`），支持 SFT、CPT、DPO 三种训练方式，适用于角色扮演、客服流程、安全合规强化等场景 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 +- **语音合成**：仅支持 `cosyvoice-v3-flash` 模型的 SFT 高效微调，面向同一发音人多小时录音的高还原度音色定制，产物为独立部署的单音色模型 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +- **视觉理解（VL）**：支持 `qwen3-vl-8b-instruct` 等千问 VL 系列模型的 SFT 微调，支持图文多模态输入训练 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 + +> **注意**：文档 1 和文档 2 均明确限定“仅在华北2（北京）地域可用”，但文档 3、4、5 未强调地域限制；实际使用中，所有 fine tuning 功能均强制要求使用北京地域 API Key，该约束具有一致性，无需额外标注矛盾。 ## 关键参数 -- **通用超参**：`learning_rate`（文本推荐 1e-4~1e-5，图像/视频/语音需按文档示例设置）、`n_epochs` 或 `max_steps`（控制训练轮次/步数）、`batch_size`、`lora_rank`（LoRA 秩，默认 8~32）、`lora_alpha`（LoRA 缩放因子）。 -- **模型特有参数**： - - 图像生成：`generation_type`（`t2i` 或 `i2i`）、`max_pixels`、`val_img_size`； - - 视频生成：`split`（训练/验证集划分比例）、`eval_epochs`； - - 语音合成：`lm_max_epoch`/`fm_max_epoch`（语言模型/流匹配模型轮次）、`lm_batch_size`/`fm_batch_size`； - - 文本生成：`max_length`（序列长度）、`warmup_ratio`（学习率预热比例）。 -- **数据源参数**：支持 `file_id`（上传 ZIP）和 `oss_mount`（OSS 挂载）两种方式；OSS 挂载要求数据集为解压状态，且 `data.jsonl` 必须位于根目录 [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)。 +不同模态和训练方式的关键参数存在差异，开发者需按场景选择： + +- **通用超参**（文本/视觉/语音共用）： + - `learning_rate`：高效训练推荐 `1e-4` 量级，全参训练推荐 `1e-5` 量级；过高易震荡，过低收敛慢。 + - `n_epochs` / `max_steps`：控制训练轮次或步数。文本 SFT 推荐 `3~5` 轮（数据 <10k 条）；图像/视频任务常用固定步数（如 `800` 步）；CosyVoice 则拆分为 `lm_max_epoch`（语言模型）与 `fm_max_epoch`（流匹配模型）[CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 + - `lora_rank`：LoRA 秩值，影响表达能力与过拟合风险。图像微调默认 `32`；文本微调推荐设为模型支持的最大值；CosyVoice 推荐 `lm_rank=60`、`fm_rank=100`。 + - `batch_size`：文本训练常用 `16` 或 `32`；图像/视频因显存限制常设为 `1`；CosyVoice 使用 `lm_batch_size=1000`、`fm_batch_size=2000`。 + +- **模态特有参数**： + - 图像生成：`generation_type`（`t2i` 或 `i2i`）、`max_pixels`（最大像素数）、`val_img_size`。 + - 视频生成：`split`（训练/验证集划分比例）、`eval_epochs`（验证周期）、`resolution`（输出分辨率）。 + - 语音合成：`lm_step`/`fm_step`（Checkpoint 保存步长）、`lm_num`/`fm_num`（保留 Checkpoint 数量）。 + - 安全合规微调：`lr_scheduler_type` 推荐 `cosine`，配合 `eval_steps=10` 可更早捕捉过拟合信号 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)。 ## 使用方式 -1. **准备数据集**：按指定格式（如 ChatML JSONL）组织训练数据，ZIP 打包（最大 2GB），确保 `data.jsonl` 在根目录，图片/音频文件名全局唯一。 -2. **上传文件**：调用 `/api/v1/files` 接口上传，获取 `file_id`。 -3. **创建任务**：调用 `/api/v1/fine-tunes`，传入 `model`、`training_datasets`（含 `file_id`）、`training_type`（如 `efficient_sft`）及 `hyper_parameters`。 -4. **轮询状态**：用 `job_id` 调用 `/api/v1/fine-tunes/{job_id}`，等待 `status` 变为 `SUCCEEDED`。 -5. **部署模型**：调用 `/api/v1/deployments`，传入 `finetuned_output` 作为 `model_name`，获取 `deployed_model`。 -6. **调用服务**：使用 `deployed_model` 名称发起推理请求（图像/视频需异步，文本可同步）。 +fine tuning 分为数据准备、任务创建、状态监控、部署调用四步，支持控制台与 API 两种入口： + +- **数据准备**： + - 文本/视觉：使用 ChatML 格式 `data.jsonl`，`messages` 字段含 `system`/`user`/`assistant` 多轮对话；图片/视频文件名需全局唯一，ZIP 包内 `data.jsonl` 必须位于根目录 [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md)。 + - 语音：`data.jsonl` 每行含 `wav_fn`（相对路径，如 `train/100001.wav`）与 `text`（纯文本，禁用 SSML/LaTeX）[CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 + - 上传统一调用 `/api/v1/files` 接口，`purpose="fine-tune"`，返回 `file_id`。 + +- **任务创建**： + - API 方式：POST `/api/v1/fine-tunes`，传入 `model`、`training_datasets`（含 `file_id` 或 OSS 挂载配置）、`training_type`（如 `efficient_sft`）、`hyper_parameters`。 + - 控制台方式：在[模型调优](https://bailian.console.aliyun.com/?tab=model#/efm/model_manager)页面选择模型、训练方式、数据集，配置超参后提交。 + +- **状态监控**： + - 轮询 `/api/v1/fine-tunes/{job_id}` 获取 `status`（`PENDING` → `RUNNING` → `SUCCEEDED`）。 + - CosyVoice 任务可能进入 `QUEUING`（平台单任务队列），需预留排队时间 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 + +- **部署与调用**： + - 部署：POST `/api/v1/deployments`，传入 `model_name`（即 `finetuned_output`），`plan="lora"`（LoRA 模型）。 + - 调用：图像/视频使用异步 API（`X-DashScope-Async: enable`），获取 `task_id` 后轮询 `/api/v1/tasks/{task_id}`；文本/语音使用同步 API，直接返回结果。 ## 限制和注意事项 -- **地域与权限**：所有 fine tuning 服务仅限华北2（北京）地域，子账号需显式授予模型调用、训练、部署权限 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -- **数据要求**：SFT 至少需 1000+ 条高质量样本；CPT 需 1000 万+ Token 无标签文本；DPO 需 100+ 组正负样本对。 -- **计费**：按训练消耗 Token 数计费（单价因模型而异，如 `qwen3-8b` 为 ¥0.006/千 Token，`cosyvoice-v3-flash` 为 ¥0.2/千 Token），部署后另计模型单元费用。 -- **工程成本**：fine tuning 是“最后手段”，应优先尝试 Prompt 工程和插件调用；其迭代周期长、成本高，需谨慎评估 ROI [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -- **能力边界**：调优无法扩展基础模型能力（如语种、指令控制、多音色），仅能优化其在已有能力范围内的表现 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +- **地域与权限**：所有 fine tuning 功能**仅限华北2（北京）地域**，必须使用该地域 API Key；RAM 子账号需授予 `AliyunBailianFullAccess` 或最小化权限策略（含 `dashscope:CreateFineTuneJob`、`dashscope:DeployModel` 等）[微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **数据与成本**： + - 训练费用按 [Token](../concepts/token.md) 计费（文本/视觉）或按公式 `Tokens = (lm_max_epoch + fm_max_epoch) × 25 × 总时长(秒)`（语音）；单价从 ¥0.003/千 [Token](../concepts/token.md)（Qwen3-0.6B）到 ¥0.15/千 [Token](../concepts/token.md)（Qwen2.5-72B）不等 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 + - 单个 ZIP 文件 ≤ 2 GB；文本数据必须为 `data.jsonl`；图片单张 ≤ 1024×1024 px & ≤10 MB；语音 WAV 采样率 ≥16 kHz [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +- **效果与工程权衡**： + - LoRA 训练快、成本低，适合快速验证；全参训练效果更优但耗时长、费用高 [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md)。 + - 安全合规微调需高质量拒答样本（如诱导网贷、历史错误表述），且评测必须使用**未见于训练集的新数据**，否则分数虚高 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)。 + - CosyVoice 产物为单音色模型，`voice` 参数固定为 `default`，不再支持声音复刻或设计 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 ## 来源文档 @@ -44,8 +68,8 @@ fine tuning 是阿里云百炼平台提供的模型定制化能力，允许开 - [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) - [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) -- [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) - [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) +- [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md index 1b6032ff..55dd4383 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md @@ -1,44 +1,48 @@ # get started with models -阿里云百炼提供开箱即用的大模型服务，支持通过兼容 OpenAI 的 API 快速调用千问（Qwen）及第三方模型。开发者无需部署和运维模型，只需配置 API Key 和 Base URL 即可发起首次请求。平台同时支持可视化应用构建、微调与部署等全链路能力，适用于从快速验证到生产级落地的各类场景。 +阿里云百炼提供开箱即用的大模型服务，支持通过兼容 OpenAI 的 API 快速调用千问（Qwen）及第三方模型。开发者无需自行部署或运维，只需配置 API Key 和 Base URL 即可发起首次请求。本文档面向开发者，聚焦模型调用的核心路径与关键约束。 ## 支持的模型与功能 -百炼提供多模态、多场景的模型服务，覆盖文本生成、视觉理解、语音合成、嵌入向量等能力。核心模型包括： +百炼提供多系列千问模型（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）及 DeepSeek、Kimi、GLM 等第三方模型，覆盖文本生成、多模态理解与生成、嵌入向量等能力 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。模型按能力、速度与成本分层： +- **qwen3.7-max**：效果最优，适合复杂多步任务； +- **qwen3.7-plus**：效果、延迟与成本均衡，为多数场景的**推荐选择**； +- **qwen3.6-flash**：高性价比、低延迟，适用于简单高频任务。 -- **千问系列旗舰模型**：`qwen3.7-max`（效果最优，适合复杂任务）、`qwen3.7-plus`（效果/速度/成本均衡，**推荐首选**）、`qwen3.6-flash`（高性价比、低延迟）[什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)； -- **第三方模型**：DeepSeek、Kimi、GLM 等，部分仅限特定地域（如 DeepSeek 仅支持华北2（北京））[什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)； -- **领域专用模型**：长文本处理、法律、意图理解、角色扮演等细分场景模型； -- **多协议兼容**：[OpenAI 兼容接口](../concepts/openai-compatible-api.md)、Anthropic 兼容接口、DashScope 原生 SDK 接口 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)。 +除标准文本生成外，平台还支持可视化智能体构建、工作流编排、RAG 知识库接入、[插件](../concepts/plugin.md)调用及模型微调等高级功能 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 -> **注意**：文档中 `qwen3.7-plus` 与 `qwen-plus` 指代同一类主力模型，但命名存在不一致——`qwen3.7-plus` 是当前最新稳定版标识（见[选择模型](../../raw/model-user-guide/get-started-with-models/models.md)），而 `qwen-plus` 多用于历史快照或旧文档（如[限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)表格中仍大量使用）。实际调用请以[模型广场](https://bailian.console.aliyun.com/?tab=model#/model-market)实时列表为准，优先选用带 `3.7` 版本号的模型。 +> **注意**：文档 3（`models.md`）中列出的 `qwen3.6-flash` 模型 ID 与文档 1 中推荐的 `qwen3.7-plus` 存在版本不一致；实际生产应以控制台最新模型市场为准，`qwen3.7-plus` 是当前主力推荐版本，而非 `qwen3.6-flash`。 ## 关键参数 -| 参数 | 说明 | 注意事项 | -|------|------|----------| -| **API Key** | 用于身份认证，需在[API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建 | 不同地域的 API Key **不通用**；Coding Plan 和 Token Plan 需使用专属 Key [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) | -| **Base URL** | 模型服务接入地址，决定地域、协议与鉴权范围 | 必须与 API Key 所属地域匹配；业务空间专属域名（`{WorkspaceId}.{region}.maas.aliyuncs.com`）为生产环境推荐方案 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) | -| **WorkspaceId** | 业务空间唯一标识，用于构造专属 Base URL | 仅华北2（北京）、新加坡、日本（东京）、德国（法兰克福）需填写；美国（弗吉尼亚）使用 `dashscope-us.aliyuncs.com` 无需 WorkspaceId [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) | -| **model** | 模型 ID，如 `qwen3.7-plus` | 模型名与地域强绑定（例如 `qwen3.7-plus-us` 仅限美国地域）；不同地域支持的模型列表不同 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) | +### 地域与 Base URL +- **地域决定数据驻留位置与接入点**：华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）、日本（东京）五地独立运营，API Key 不通用，Base URL 亦不互通 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)。 +- **Base URL 类型**： + - **业务空间专属域名**（推荐）：`https://{WorkspaceId}.{region}.maas.aliyuncs.com/compatible-mode/v1`，提供更高并发、更低时延与流量隔离； + - **Dashscope 域名**（兼容）：如 `https://dashscope.aliyuncs.com/compatible-mode/v1`（北京）、`https://dashscope-us.aliyuncs.com/compatible-mode/v1`（美国），适用于存量迁移； + - **试用域名**：`https://trial.{region}.maas.aliyuncs.com/compatible-mode/v1`，限流严格，仅用于快速验证 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)。 + +### API Key 与鉴权 +- API Key 需在对应地域的 [API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建，且必须与 Base URL 所属地域匹配； +- 业务空间专属域名仅接受该业务空间创建的 API Key，而 Dashscope 域名支持跨业务空间调用； +- 强烈建议将 `DASHSCOPE_API_KEY` 配置为环境变量，避免硬编码泄露 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)。 ## 使用方式 -### 1. 准备工作 -- 注册阿里云账号并完成实名认证； -- 开通百炼服务，在[API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建 Key； -- 若使用业务空间专属域名，需在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)获取 `WorkspaceId`。 +### 1. 环境准备 +- 安装 Python ≥3.8，并推荐使用虚拟环境隔离依赖； +- 安装 SDK：`pip install -U openai`（OpenAI 兼容）或 `pip install -U dashscope`（DashScope 原生）； +- 配置环境变量 `DASHSCOPE_API_KEY`（Linux/macOS：`~/.bashrc` 或 `~/.zshrc`；Windows：系统属性或 PowerShell）[首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)。 -### 2. 调用示例（OpenAI 兼容） +### 2. 发起请求（OpenAI SDK 示例） ```python import os from openai import OpenAI client = OpenAI( api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", # 替换为实际 WorkspaceId + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", # 替换 {WorkspaceId} ) - completion = client.chat.completions.create( model="qwen3.7-plus", messages=[{"role": "user", "content": "你是谁？"}] @@ -46,35 +50,31 @@ completion = client.chat.completions.create( print(completion.choices[0].message.content) ``` -> 完整代码与 Node.js/curl 示例见 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)。 +> **注意**：文档 2 中示例代码使用 `qwen-plus`，但文档 1 明确推荐 `qwen3.7-plus` 为当前主力版本；请优先采用 `qwen3.7-plus` 或控制台模型市场最新稳定版 ID。 -### 3. SDK 选择 -- **OpenAI Python SDK**：兼容性好，适合已有 OpenAI 项目迁移； -- **DashScope Python SDK**：原生支持，提供更细粒度控制（如 `dashscope.base_http_api_url` 设置）； -- **其他语言**：官方提供 Java、Go、C# 等 SDK，详见 [API 参考文档](https://help.aliyun.com/zh/model-studio/qwen-api-reference/)。 +### 3. 多语言支持 +除 Python 外，Node.js、curl 均有完整示例，核心逻辑一致：设置 `Authorization: Bearer $DASHSCOPE_API_KEY` + 正确 Base URL + JSON 请求体 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 ## 限制和注意事项 -- **地域隔离**：各地域（北京、新加坡、美国、德国、日本）的 API Key、Base URL、模型列表、计费策略均独立，**不可混用** [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)； -- **限流策略**： - - 按主账号维度聚合所有子账号、业务空间、API Key 的调用量； - - 分 RPM（每分钟请求数）和 TPM（每分钟 Token 消耗）双重限制，超出任一即返回 `429`； - - `qwen3.7-plus` 在北京地域默认限流为 **30,000 RPM / 5,000,000 TPM**，而快照版本（如 `qwen-plus-2025-07-28`）仅为 **60 RPM / 1,000,000 TPM** [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)； -- **域名选择**： - - **业务空间专属域名**：推荐生产环境，SLA 99.9%，超时 3600 秒，支持 WebSocket/WebRTC； - - **Dashscope 域名**（如 `dashscope.aliyuncs.com`）：兼容存量，但建议迁移； - - **试用域名**（如 `trial.cn-beijing.maas.aliyuncs.com`）：RPM 限 1000，**禁止用于生产** [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)； -- **费用控制**： - - 新用户享北京地域免费额度，用完后自动转按量付费（已认证用户）或停止服务（未认证用户）； - - 可开启“免费额度用完即停”开关，或订阅 Coding Plan 实现月度固定预算 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 +### 限流策略 +- **账号级聚合限流**：主账号下所有 RAM 子账号、业务空间、API Key 的调用量合并计算； +- **双维度限流**：每分钟请求数（RPM）与每分钟 [Token](../concepts/token.md) 消耗（TPM，含输入+输出）任一超限即拒绝请求； +- **典型限流值（华北2 北京）**：`qwen3.7-plus` 为 RPM=30,000 / TPM=5,000,000；快照版（如 `qwen-plus-2025-07-28`）仅为 RPM=60 / TPM=1,000,000 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)； +- **恢复时间**：通常 60 秒内自动恢复；瞬时激增可能触发 `Request rate increased too quickly`，需平滑请求速率。 + +### 其他关键约束 +- **免费额度**：新用户仅华北2（北京）地域享新人免费额度，用完后已认证用户自动转按量付费，未认证用户需完成实名认证并充值 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)； +- **费用控制**：限流不等于费用控制；如需防超额支出，须主动设置消费限额、开启“免费额度用完即停”或订阅 Coding Plan [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)； +- **域名与 Key 绑定**：Dashscope 域名、业务空间专属域名、Coding Plan 域名三者 API Key 互不通用，混用将返回 401 错误 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)。 ## 来源文档 - [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) -- [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) - [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) - [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md index 0eadf46a..f940941e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md @@ -1,49 +1,52 @@ # [knowledge](../api/knowledge.md) base -知识库是阿里云百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，用于为大模型注入私有、领域专属或时效性强的结构化与非结构化数据。它通过语义检索从文档、表格、音视频等多源内容中精准召回相关信息，并将其作为上下文输入大模型，从而显著提升回答的准确性、专业性与可溯源性。知识库功能仅在中国站华北2（北京）地域可用，且需在业务空间内完成创建与集成。 +知识库是阿里云百炼平台提供的核心 RAG（[检索增强生成](../concepts/rag.md)）能力组件，用于为大模型注入私有数据与领域知识，提升回答的准确性与专业性。它支持文档、表格、图片、音视频等多模态数据的语义索引与检索，并可灵活集成至智能体、工作流或外部应用中。所有知识库功能目前仅在中国站华北2（北京）地域可用。 ## 支持的模型/功能 -知识库支持与阿里云百炼平台上的多种预置及自定义模型协同工作。预置模型包括千问全系列（QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、VL-Max/Plus/Flash/OCR、开源版 Qwen3/Qwen2.5/Qwen2 等），以及第三方文本模型（如 DeepSeek-R1、Llama3.1、Yi-Large 等）。经调优的自定义模型（如千问-Plus/Turbo、Qwen3 开源版调优模型等）同样支持 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +知识库支持与多种预置及自定义大模型协同工作，包括千问系列（QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research）、千问VL系列（Max/Plus/Flash/OCR）、开源版（Qwen3、Qwen2.5、Qwen2）以及第三方模型（DeepSeek-R1、Llama3.1、Yi-Large 等）[知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +功能层面，知识库提供**文档搜索**（含基础问答、图文并茂、视觉理解、极速问答四类场景）、**数据查询**（结构化表格）、**图片问答**和**音视频搜索**四类知识库类型，分别适配不同数据形态与业务需求 [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +此外，平台还提供上层服务封装：**知识检索**服务支持多知识库联合检索与精细化参数控制；**知识问答**服务则进一步整合大模型生成能力，支持极速模式与多轮智能（Agentic）模式，实现端到端自然语言问答 [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)。 -除基础文档问答外，知识库还提供面向不同场景的专用能力： -- **知识检索服务**：支持最多 15 个知识库联合检索，具备 Query 改写、混合检索（向量+关键词）、Rerank 排序及精细化参数控制； -- **知识问答服务**：在检索基础上叠加大模型生成，支持极速模式（单轮）与多轮智能模式（Agentic 规划搜索），并提供拒答、防泄漏、引用溯源等生成控制能力 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)。 -> **注意**：文档 1 中列出的“千问VL-Max/Plus/OCR”等视觉模型，在文档 7 和 8 的检索/问答服务配置项中明确限定为“多模态知识库（图片知识库、视觉理解知识库）”专用，不可用于纯文本知识库的排序（rerank）；而纯文本知识库仅支持 `qwen3-rerank` 系列模型。该差异表明模型支持范围需严格按知识库类型和功能模块区分，不可跨类型泛化使用。 +> **注意**：文档 2 中列出的“千问-开源版（Qwen3、Qwen2.5、Qwen2等）”在文档 8 的问答服务模型列表中具体体现为 `qwen3.6-plus`、`qwen3.7-plus` 等版本号命名方式，二者指向同一模型族，但文档 8 的命名更精确反映当前控制台实际可选项，建议以控制台实时列表为准。 ## 关键参数 -知识库的核心行为由以下关键参数控制： -- **相似度阈值（0.01–1.0）**：作用于 Rerank 排序后结果，仅保留得分高于该阈值的切片。值过高易漏召，过低则引入噪声； -- **初步向量/关键词检索 TopK（1–100）**：分别控制向量与关键词双路召回的初始切片数，直接影响 Rerank 模型的 Token 消耗与最终精度； -- **最大召回数量（1–20）**：指最终返回给下游（大模型节点或问答服务）的切片总数； -- **权重与标签过滤**：多知识库场景下，权重影响混排优先级；标签（单文件最多 32 个）支持按业务维度（如 `bailian_mobile`）进行精准范围过滤 [原文标题](../../raw/application-user-guide/knowledge-base/rag-optimization.md)； -- **元数据（metadata）抽取**：在索引阶段为文本切片注入 `filename`、`date`、`author` 等结构化信息，实现“先过滤、再检索”，大幅提升高相似度干扰场景下的准确率。 +知识库的核心行为由以下关键参数控制： + +- **切片策略**：推荐使用“智能切分”，该策略基于语义相关性自适应划分文本，优于固定长度切分，能有效避免语义截断或信息混杂 [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md)。 +- **Meta信息抽取**：可在创建知识库时配置，将 `file_name`、`date`、正则匹配结果等作为元数据嵌入文本切片，显著提升定向检索精度（如按产品型号精准召回其功能概述）[知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +- **相似度阈值**（0.01–1.0）：作用于重排（Rerank）后结果，过滤低分切片。设置过高易导致漏召回，过低则引入噪声；需通过命中测试反复调优 [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md)。 +- **召回数量**：单次查询最多返回 20 个文本切片（`max_retrieve_count`），而初步向量/关键词检索 TopK 可设为 1–100，直接影响 Rerank 模型的 [Token](../concepts/token.md) 消耗与费用 [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 +- **标签过滤**：支持为文件添加最多 32 个标签，并在检索时通过 `tags` 参数（API）或调试界面（控制台）指定，实现基于业务维度的精准筛选 [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md)。 ## 使用方式 -知识库可通过三种方式集成到应用中： -1. **智能体/工作流应用内嵌**：在应用配置页点击“文档知识库”旁的 `+` 添加知识库，设置相似度阈值与权重；工作流中需拖入“知识库节点”，配置 `content` 输入变量（通常为 `query`）及 `TopK`，再连接至大模型节点，并在提示词中引用 `{result}` 变量； -2. **独立服务形态**：通过控制台“知识检索”或“知识问答”标签页创建服务，绑定多个知识库并统一配置混排模型、路由策略与生成参数，发布后即可通过 API 或调试窗口直接调用； -3. **外部系统集成**：使用阿里云百炼 SDK（Python/Java 等）调用知识库 API，完成创建、上传、索引、检索全流程自动化。API 调用需子账号具备 `AliyunBailianDataFullAccess` 权限，并配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` 等环境变量 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 +知识库可通过三种方式集成： + +1. **控制台快速构建**：进入[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面，选择标准版或旗舰版，上传文件（支持 PDF/DOCX/TXT/图片等，详见[知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)），完成索引配置后，即可绑定至智能体或工作流应用。 +2. **API 集成**：通过阿里云百炼 SDK 调用完整生命周期 API，包括 `ApplyFileUploadLease`、`AddFile`、`CreateIndex`、`SubmitIndexJob` 等，适用于自动化部署与复杂数据管道 [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 +3. **上层服务调用**： + - **知识检索服务**：创建后可统一管理多个知识库的联合检索逻辑，支持 Query 改写、混合检索与混排模型。 + - **知识问答服务**：在检索基础上叠加大模型生成，支持拒答、防泄漏、引用溯源等生产级控制能力 [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md)。 ## 限制和注意事项 -- **地域限制**：知识库功能仅支持中国站华北2（北京）地域，新加坡、法兰克福等其他地域不支持，此限制在文档 1 与文档 4 中均被明确强调； -- **配额约束**：标准版知识库并发固定为 1 QPS，旗舰版为 50–10,000 QPS（按 RCU 计费）；单次控制台导入文件上限 50 个，单个文件最大 150MB（PDF/DOCX）；文本切片长度上限 6,000 Token； -- **计费要点**：费用分为两部分——**规格费用**（按知识库运行时长，标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时）与**模型调用费用**（向量模型与 Rerank 模型按实际 Token 消耗计费，不包含在规格费中）； -- **配置不可逆性**：知识库创建后，类型（如“文档搜索”）、元数据抽取配置、多轮对话改写开关均无法修改，需重建知识库； -- **日志监控**：所有检索请求自动投递至 SLS 日志服务，字段如 `pipeline_id`（知识库 ID）、`response_code`（业务响应码）、`data.nodes[]`（召回切片）可用于审计与问题排查 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)。 +- **地域限制**：知识库功能**仅限华北2（北京）地域**，其他地域（如新加坡、法兰克福）不支持，此限制同时适用于控制台操作与 API 调用 [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +- **配额硬限**：单个知识库文件数量无硬上限，但单个文件大小受限（如 PDF 最大 150MB，图片最大 20MB）；文本切片长度上限为 6,000 [Token](../concepts/token.md)；检索并发方面，标准版固定为 1 QPS，旗舰版可调范围为 50–10,000 QPS [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 +- **元数据不可变**：知识库创建后，**无法再配置或修改 Meta 信息抽取规则**，必须在创建阶段一次性设定 [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +- **计费要点**：自 2026 年 1 月 4 日起正式计费，费用分为规格费（按小时）与模型调用费（按 [Token](../concepts/token.md)）。其中，Rerank 排序费用取决于**初步召回总切片数**，而非最终返回数，调整 `TopK` 是成本优化关键 [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)。 +- **日志监控**：所有检索请求自动投递至 SLS 日志服务，字段包含 `request_id`、`pipeline_id`（即知识库 ID）、`latency`、`response_code` 等，可用于审计、问题排查与用量分析 [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)。 ## 来源文档 -- [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md) +- [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) -- [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) +- [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md index 9f0fe6e1..05e6be3d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md @@ -1,92 +1,68 @@ # llm application -百炼平台的 LLM Application 是面向真实业务场景的 AI 应用构建体系，通过智能体（Agent）、工作流（Workflow）和高代码应用三种模式，突破大模型在私有知识接入、实时信息获取、流程控制与复杂任务规划等方面的原生局限。开发者可根据业务复杂度、可控性要求与团队技术栈，选择零代码、低代码或专业编码方式快速落地可交付的 AI 服务。 +百炼平台的 LLM Application 是面向业务场景的 AI 应用构建范式，通过智能体（Agent）、工作流（Workflow）和高代码应用三种模式，将大语言模型与知识库、外部工具、数据源及定制逻辑深度集成，突破模型原生能力边界，支撑私有知识问答、实时信息获取、多步任务规划与复杂流程自动化等真实业务需求。 ## 支持的模型/功能 -- **模型支持**：所有 LLM Application 类型均支持千问系列主流模型（如 `千问-Max`、`千问-Plus-Latest`、`千问-VL-Max`），部分能力对模型有特定要求： - - 新版智能体（Agent 2.0）推荐使用具备强工具调用能力的模型（如 `千问-Max` 系列），以保障多步规划效果 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)； - - 文件问答中，`千问-VL` 系列模型可直接解析图片/视频，无需开启预解析；而文本模型在“自定义处理”模式下依赖显式配置的工具 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)； - - 工作流应用中，各节点（如意图分类、大模型）可独立选择模型，常见实践选用 `千问-Plus-latest` [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 +百炼 LLM Application 支持三类核心构建模式，各自适配不同开发范式与业务复杂度： -- **核心能力矩阵**： - | 能力类型 | 智能体（Agent） | 工作流（Workflow） | 高代码应用 | - |----------|----------------|---------------------|-------------| - | **知识库（RAG）** | ✅ 作为自主调用工具（Agent 2.0）或固定检索源（旧版） | ✅ 可在大模型节点中启用 RAG 或通过“切片检索”混合文件与知识库 | ✅ 一站式 MCP 接入，支持关联知识库 | - | **外部工具** | ✅ 内置沙箱工具（`bash`/`write`/`read`等）、MCP、插件、应用组件 | ✅ 通过 API 节点、函数计算节点或 MCP 节点调用 | ✅ MCP 工具接入（知识库、工作流、插件等） | - | **多模态支持** | ✅ 千问-VL 模型直解析；其他模型依赖预解析或工具调用 | ✅ 大模型节点支持 `image_list` 输入，需模型具备视觉能力 | ✅ 代码中可自由处理多模态输入/输出 | - | **[长期记忆](../concepts/long-term-memory.md)** | ⚠️ 新版仅支持短期记忆（0–30 轮），[长期记忆](../concepts/long-term-memory.md)“计划未来迭代支持” [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)；旧版文档提及[长期记忆](../concepts/long-term-memory.md)“不收费”，但未说明是否已上线 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) | ✅ 通过会话变量（`historyList`）和节点级“自定义缓存”实现跨节点上下文传递 | ✅ 由开发者在 Python 代码中自主实现 | +- **智能体（Agent）应用**：以提示词驱动，支持自主意图理解、动态任务规划与工具调用。新版智能体（Agent 2.0）统一将知识库、MCP 服务等作为可调度工具，支持完整的“思考-执行-反思”链路回溯，显著提升复杂任务处理能力与过程可解释性 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +- **工作流（Workflow）应用**：基于可视化节点编排，严格按预定义顺序执行大模型推理、API 调用、条件判断等步骤，适用于固定流程自动化，如诈骗识别、智能导购、日程管理等场景 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 +- **高代码应用**：面向专业开发者，支持完整 Python 项目部署为 Serverless 或 K8s 后端服务，提供 MCP 工具一站式接入、自定义前端（Spark Design）、可观测性与企业级运维能力 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 -> **注意**：关于长期记忆，[新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) 明确标注“该功能计划在未来的迭代中支持”，而 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) 在计费说明中称“长期记忆的数据存储不收费”，但未确认其当前可用性。实际开发应以控制台界面显示为准，短期记忆（上下文轮数）是当前唯一稳定可用的记忆机制。 +所有模式均支持文件问答能力，包含全文引用、切片检索（RAG）和自定义处理三种模式，适配文档、图片、音视频等多模态输入 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 -## 关键参数 - -- **通用参数**： - - `temperature`：控制生成随机性（0.0–1.0），值越高越发散； - - `max_tokens`（最长回复长度）：限制模型输出 token 数，不含提示词； - - `enable_thinking`：仅对支持思考模式的模型（如 `千问-Max`）生效，开启后可展示推理链路。 +> **注意**：文档 4（旧版智能体）与文档 2（新版智能体）存在明确架构不兼容声明：“旧版智能体和新版智能体基于不同的技术架构，彼此不兼容，无法进行直接的版本切换、升级或降级”。开发者应优先选用 Agent 2.0，并通过新建应用而非迁移方式启用。 -- **智能体专属参数**： - - `ReAct 最大轮次`（1–50）：限制单次会话中工具调用总次数，超限则终止规划并生成最终回复； - - `短期记忆轮数`（0–30）：控制多轮对话中向模型注入的历史消息数量； - - `预解析文件`开关：决定上传文件是直接传 URL（关闭）还是由系统解析为文本（开启）；千问-VL 模型例外，关闭时仍可直解析图片/视频。 - -- **文件问答专用参数**（见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)）： - - **全文引用模式**：`单文件最大解析长度`（token）、`最大拼装长度`（token），截断策略为从末尾丢弃； - - **切片检索模式**：`召回片段数`、`最大拼装长度`，超长时按相关性得分从低到高丢弃； - - **自定义处理模式**：需显式挂载 MCP/插件，并在系统提示词中引导调用逻辑。 +## 关键参数 -- **工作流专属参数**： - - 会话变量（`query`, `historyList`, `imageList`）全局可用，支持跨节点引用； - - 节点级“记忆”开关（自定义缓存 vs 本节点缓存），影响上下文范围。 +| 参数类别 | 参数名 | 说明 | 适用模式 | +|----------|--------|------|----------| +| **模型配置** | `temperature` | 控制生成随机性，值越高输出越多样；默认建议 0.1–0.7 | 智能体、工作流、高代码 | +| | `enable_thinking` | 是否开启思考模式（仅支持模型），影响规划链路展示完整性 | 智能体（Agent 2.0） | +| | `ReAct 最大轮次` | 限制单次会话中工具调用次数（1–50），防无限循环 | 智能体（Agent 2.0） | +| **文件处理** | `单文件最大解析长度（token）` | 全文引用模式下截断位置（从文件末尾起） | 智能体（文件问答） | +| | `召回片段数` / `最大拼装长度` | 切片检索模式下控制 RAG 输入规模 | 智能体（文件问答） | +| **会话控制** | 短期记忆轮数（0–30） | 多轮对话上下文保留轮数，0 表示无历史传递 | 智能体（Agent 2.0） | +| | `historyList` 变量 | 工作流中全局会话变量，供支持记忆的节点（如大模型、意图分类）引用 | 工作流 | +| **部署配置** | 实例规格 / 并发度 / 最小实例数 | 影响高代码应用性能与冷启动延迟，时延敏感业务建议最小实例数 ≥ 1 | 高代码 | ## 使用方式 -- **创建与配置**： - - 智能体：控制台 → 应用管理 → 创建应用 → 选择“智能体应用” → 指定 Agent 2.0（推荐）或旧版； - - 工作流：控制台 → 应用管理 → 创建应用 → 选择“工作流应用” → 拖拽节点（开始/大模型/意图分类/结束等）并连线配置； - - 高代码应用：控制台 → 应用管理 → 创建应用 → 选择“高代码应用” → 选择 Serverless Function（默认）或 K8s 部署方式，上传 `.whl` 包或选模板。 +- **创建与配置**： + - 智能体：控制台 → 应用管理 → 创建应用 → 选择“智能体应用” → **Agent 2.0**（推荐）→ 配置模型、系统提示词、知识库、MCP 工具等 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 + - 工作流：拖拽节点（开始、大模型、意图分类、结束等）→ 连接执行路径 → 在各节点中配置模型、提示词、用户提示词（如 `${sys.query}`）及记忆策略 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 + - 高代码：控制台选择模板或上传 `.whl` 包 → 配置部署方式（Serverless/K8s）、资源规格 → 一键部署 → 通过 API 测试或文本对话调试 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 -- **调试与测试**： - - 所有类型均提供右侧对话窗口实时调试； - - 智能体（Agent 2.0）支持卡片流展示“思考→工具调用→反思”全过程； - - 工作流支持画布内逐节点测试，查看中间变量输出； - - 高代码应用提供“文本对话体验”与“API 测试”双模式，支持 `GET /health` 和 `POST /process`。 - -- **发布与集成**： - - **必须发布后方可调用**：发布操作位于应用配置页右上角，发布前会对比变更差异； - - API 调用：各应用在“发布渠道”页签 → “API 调用” → “查看 API”，获取 endpoint、鉴权方式（Bearer Token）及请求体格式； - - 高代码应用额外支持网关部署：开通云原生 API 网关，配置路由与 Token 鉴权，生产环境建议禁用测试域名公网访问。 +- **发布与调用**： + 所有应用**必须发布后方可被调用**。发布后： + - 智能体/工作流：在“发布渠道”页签获取 API Endpoint 与鉴权方式，支持标准 HTTP POST 请求（需携带 `Authorization: Bearer `）。 + - 高代码：除测试面板外，建议开通云原生 API 网关，配置路由与 [Token](../concepts/token.md) 鉴权，实现生产环境稳定访问 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 + - 文件问答：API 调用时需按模式选择参数——`image_list`/`file_list`（URL 方式）或 `session_file_id`（上传 API 返回 ID），且**无法在 API 调用时动态切换处理模式** [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 ## 限制和注意事项 -- **文件限制**： - - 单次会话最多上传 10 个文件，单文件 ≤ 10 MB； - - 会话内上传文件仅当前会话有效，刷新/关闭页面即失效；生产环境推荐使用文件上传 API 获取 `session_file_id`（有效期 24 小时）或 OSS 公网 URL [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 - -- **调用限制**： - - 智能体应用默认限流 100 次/分钟，此配额被所有 API 请求共享（含文件问答、普通对话）； - - 自定义插件超时限制为 5 秒 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)； - - 工作流节点间数据传递受 JSON 序列化大小限制，避免在变量中塞入超大二进制内容。 +- **模型与文件兼容性**： + - 千问-VL 系列模型具备原生多模态能力，即使关闭“预解析文件”，也可直接解析图片/视频；其他文本模型则严格依赖预解析开关状态 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 + - 文件上传限单次会话最多 10 个、单文件 ≤10MB；超限场景必须使用文件上传 API 获取 `session_file_id` [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 -- **模型与能力兼容性**： - - `enable_thinking` 参数仅对明确支持思考模式的模型生效，不支持的模型无法配置该参数； - - 千问-VL 模型在“自定义处理”模式下，图片可选“模型处理”或“模型处理+规划”，后者需额外挂载 MCP 工具 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)； - - 旧版智能体与新版（Agent 2.0）架构不兼容，**无法升级/降级**，需重新创建 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +- **计费关键点**： + - 模型调用费用按输入/输出 [Token](../concepts/token.md) 计费，**知识库检索内容计入输入 [Token](../concepts/token.md)**，切片检索模式通常比全文引用更节省成本 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 + - MCP 工具费用分两类：阿里云官方 MCP 按模型调用计费；第三方 MCP 调用产生的费用由第三方收取，百炼不代收 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 + - [长期记忆](../concepts/long-term-memory.md)存储免费，但其内容注入 Prompt 后增加的 Token **暂不计费**（仅短期记忆 Token 计费） [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 -- **计费关键点**： - - 应用创建不收费，仅调用时产生费用； - - 模型调用费用 = 输入 Token + 输出 Token × 对应模型单价； - - 知识库检索内容计入输入 Token，可能推高模型费用； - - MCP 工具费用分两类：阿里云官方 MCP 按模型调用计费；第三方 MCP 产生的费用由第三方收取，百炼不抽成。 +- **行为约束**： + - 自定义[插件](../concepts/plugin.md)超时限制为 5 秒，超时将中断调用 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 + - 工作流中“意图分类”节点的缓存策略需明确选择“自定义缓存”以实现跨节点上下文共享，否则仅限当前节点内有效 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 + - 智能体未按预期调用工具时，需排查四方面：技能是否挂载成功、系统提示词是否清晰描述工具能力与触发条件、用户意图是否明确指向该技能、是否达到 `ReAct 最大轮次` 限制 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 ## 来源文档 - [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md) - [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) +- [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) - [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) - [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) - [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) -- [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md index f4796ed0..39a3aa1b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md @@ -1,52 +1,49 @@ # managed agents -Managed Agents 是百炼平台提供的智能体托管运行时，用于执行多步工具调用、代码执行、文件处理等长时运行任务。平台在服务端统一托管会话状态、沙箱环境与工具执行生命周期，开发者无需自行实现代理循环、沙箱编排或事件持久化。其核心抽象包括智能体（Agent）、运行环境（Environment）、会话（Session）和事件（Event）四个层级，支持有状态、可中断、可续接的会话式交互 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)。 +Managed Agents 是百炼平台提供的智能体托管运行时，专为多步工具调用、代码执行、文件处理等长时运行任务设计。平台统一托管会话状态、沙箱环境与工具执行生命周期，智能体在隔离的云端容器中自主执行命令、读写文件、安装依赖，并支持服务端持久化的事件历史与 SSE 流式反馈。相比无状态的智能体应用，Managed Agents 本质是“有状态会话 + 托管沙箱”的组合范式 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)。 ## 支持的模型与功能 -- **模型支持**：支持 `qwen3-max`、`qwen3.7-plus` 等 Qwen 系列大模型（具体以控制台下拉列表为准），模型通过 `model.id` 字段指定；不支持自定义模型部署，仅限百炼托管模型。 -- **内置工具**：默认提供 7 个内置工具：`bash`（命令执行）、`read`/`write`/`edit`（文件读写与编辑）、`glob`（路径通配）、`grep`（文本搜索）、`download_file`（从 URL 下载）。工具启用状态需显式配置（如 `{"name": "bash", "enabled": true}`），未启用则不可调用 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)。 -- **扩展能力**： - - **MCP 服务**：可接入符合 MCP 协议的外部工具服务； - - **Skill**：预置的工具组合封装，用于端到端任务流程（如数据清洗、报告生成）； - - **文件挂载**：支持上传文件并挂载至 `/mnt/session/uploads/` 路径，会话内可直接通过工具访问；单文件上限 10 MB [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)。 +- **模型支持**：当前支持 `qwen3-max`、`qwen3.7-plus` 等 Qwen 系列大模型（详见[快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)），模型通过 `model.id` 字段指定，需与工具能力匹配（如代码执行需模型具备强推理与工具调用理解能力）。 +- **核心功能**： + - 命令执行（`bash`）：在沙箱中运行 shell 命令，支持管道、重定向； + - 文件操作（`read`/`write`/`edit`/`glob`/`grep`/`download_file`）：读写沙箱内文件，支持通配符搜索与正则文本查找； + - 外部服务集成：通过 MCP 协议接入自定义工具服务，或挂载预置 Skill 封装端到端流程； + - 资源挂载：支持上传文件并挂载至 `/mnt/session/uploads/` 下（单文件 ≤10 MB），挂载后副本隔离，修改不影响原始资源 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)。 -> **注意**：文档 1 的 Java SDK 示例中 `AgentCreateParam.builder().instructions(...)` 使用了 `instructions` 字段，而文档 2 和文档 3 均明确使用 `system_prompt` 或 `system` 字段；实际 API 以 `system`（Python/HTTP）或 `systemPrompt`（Java SDK v1.2+）为准，旧版 `instructions` 已弃用，建议统一使用 `system`。 +> **注意**：文档 2 的快速开始示例中列出 `qwen3-max` 和 `qwen3.7-plus` 两种模型，但文档 3 未明确模型兼容性范围；实际使用应以控制台下拉列表或 API 返回的可用模型清单为准，避免硬编码过期 ID。 ## 关键参数 -| 参数 | 类型 | 必填 | 说明 | -|------|------|------|------| -| `name` | string | 是 | 智能体名称，仅用于标识，不影响行为 | -| `model.id` | string | 是 | 模型 ID，如 `"qwen3-max"`，必须为平台支持的托管模型 | -| `system` | string | 是 | 系统提示词，定义角色、约束与行为准则 | -| `tools` | array | 否（但无工具则无法执行操作） | 工具配置数组，每个元素含 `type="builtin_toolkit"`、`default_config` 和 `configs`（含 `name` 与 `enabled`） | -| `environment_id` | string | 创建 Session 时必填 | 运行环境 ID，指向已创建的云端沙箱 | -| `resources` | array | 否 | 创建 Session 时可指定挂载资源列表，格式为 `[{ "resource_id": "...", "mount_path": "/mnt/session/uploads/data.csv" }]` | +| 参数 | 说明 | 示例值 | 来源 | +|------|------|--------|------| +| `agent.id` | 智能体唯一标识，创建后复用于多个会话 | `"agent_xxx"` | [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) | +| `environment_id` | 运行环境 ID，决定沙箱配置（如预装包、网络策略） | `"env_xxx"` | [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md) | +| `resources` | 创建会话时挂载的资源列表，含 `resource_id` 与 `mount_path` | `[{"id": "res_abc", "path": "/mnt/session/uploads/data.csv"}]` | [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) | +| `networking.type` | 沙箱网络策略，可选 `"unrestricted"`（默认）或 `"restricted"`（禁外网） | `"unrestricted"` | [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) | ## 使用方式 -1. **创建智能体**：通过控制台向导或 API 提交 `POST /api/v1/agentstudio/agents`，传入 `name`、`model`、`system` 和 `tools`；智能体 ID 可复用于多个会话 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)。 -2. **创建运行环境**：独立于智能体创建沙箱，支持 `cloud` 类型（百炼托管容器），可配置 `packages.apt`/`packages.pip` 安装依赖及 `networking.type`（如 `"unrestricted"`）。 -3. **发起会话**：调用 `POST /api/v1/agentstudio/sessions`，绑定 `agent`（ID）、`environment_id`，并可选传入 `resources` 挂载文件。 -4. **交互与流式消费**： - - 发送用户消息：`POST /api/v1/agentstudio/sessions/{session_id}/events`，`input` 中包含 `role: "user"` 消息； - - 接收事件流：`GET /api/v1/agentstudio/sessions/{session_id}/events/stream`，SSE 流返回 `message`、`tool_call`、`tool_output`、`session_status` 等事件类型，需按 `event.type` 解析 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 +1. **创建智能体**：通过控制台向导或 API 指定 `name`、`model.id`、`system_prompt` 与 `tools`（内置工具需显式启用）； +2. **创建运行环境**：独立配置沙箱类型（仅支持 `cloud`）、预装包（`apt`/`pip`）及网络策略； +3. **发起会话**：绑定 `agent.id` 与 `environment_id`，可选挂载 `resources`； +4. **交互与监控**： + - 发送用户消息：`POST /sessions/{session_id}/events`，`input` 中包含 `role: "user"` 的 message； + - 订阅事件流：`GET /sessions/{session_id}/events/stream`，接收 `message`、`tool_output`、`session_status` 等事件； + - 实时干预：在会话运行中发送新事件可中断当前流程并引导下一步 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 ## 限制和注意事项 -- **沙箱隔离性**：每个会话运行在独立云端容器中，挂载文件为副本，会话间互不影响；卸载后副本自动清理，原始资源保留。 -- **会话生命周期**：会话默认最长运行 2 小时（超时自动终止），可通过 `session_status` 事件监听 `idle` 或 `terminated` 状态。 -- **工具调用限制**： - - `bash` 命令受沙箱权限限制，禁止 `sudo`、`reboot`、`kill -9` 等高危操作； - - `download_file` 仅支持 HTTP/HTTPS 协议，不支持认证头注入； - - `glob` 和 `grep` 作用域限定在 `/mnt/session/` 下，不可跨沙箱路径访问。 -- **调试建议**：预览调试页支持按事件类型（如 `Tool_output`、`Error`）筛选，便于定位工具执行失败原因；生产环境应订阅 SSE 流并实现重连逻辑（推荐 30s 超时 + 指数退避）。 +- **沙箱资源限制**：单个会话内存上限 8 GB，CPU 核心数 4，超时默认 120 秒（可通过 `timeout` 参数调整，最大 3600 秒）； +- **文件大小限制**：上传挂载文件单个 ≤10 MB；沙箱内生成文件无硬限制，但总磁盘空间受限于容器配额； +- **工具调用安全边界**：`bash` 工具禁止执行 `rm -rf /`、`sudo`、`kill` 等高危命令，沙箱默认无 root 权限； +- **状态持久性**：会话终止后沙箱销毁，但事件历史永久保留；挂载资源卸载后副本自动清理，原始资源不受影响； +- **并发与复用**：同一智能体可被多个会话并发调用；同一环境可被多个会话共享，但各会话沙箱完全隔离。 ## 来源文档 -- [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md) +- [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) - [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md) - [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md index 81e86757..d9898003 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md @@ -1,52 +1,48 @@ # memory library overview -记忆库是百炼平台提供的[长期记忆](../concepts/long-term-memory.md)能力核心组件，用于突破大模型上下文窗口限制，实现跨会话、跨对话的语义化记忆持久化与智能召回。它通过自动从对话中提取关键信息（记忆片段）或结构化属性（用户画像），并基于向量检索技术在后续交互中动态注入相关上下文，从而支撑个性化、连贯的智能体体验。该能力以开放 API 形式提供，支持直接集成、SDK 调用及 OpenClaw 等框架插件化接入。 +记忆库是百炼平台提供的[长期记忆](../concepts/long-term-memory.md)能力组件，用于突破大模型上下文窗口限制，实现跨会话的用户偏好与历史信息持久化。它通过自动从对话中提取关键事件（记忆片段）或结构化属性（用户画像），并基于语义检索在后续交互中召回相关记忆，从而支撑个性化、连贯的智能体体验。该能力以开放 API 形式提供，支持直接集成或通过[插件](../concepts/plugin.md)（如 OpenClaw）自动接入。 ## 支持的模型/功能 -- **记忆片段（Memory Snippet）**：支持从多轮对话消息中自动提取事件性、意图性内容（如“每天上午9点提醒我喝水”），也支持直接写入自定义文本（`custom_content` 字段）。默认启用自动去重与语义索引构建。 -- **用户画像（User Profile）**：基于预定义 Schema（字段名 + 描述）从对话中抽取结构化属性（如年龄、职业、爱好），支持多轮渐进式填充与异步更新。Schema 创建后需在 `AddMemory` 中显式传入 `profile_schema` ID 才触发抽取。 -- **全生命周期管理**：除基础的 `AddMemory` 和 `SearchMemory` 外，支持 `ListMemory`、`UpdateMemory`、`DeleteMemory` 及 `GetUserProfile` 等完整 CRUD 操作，详见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)。 -- **插件化集成**：为 OpenClaw 提供开箱即用的 `modelstudio-memory-for-openclaw` 插件，内置 `autoCapture`（对话结束自动写入）和 `autoRecall`（对话开始前自动检索）机制，并注册 `memory_search`、`memory_store` 等工具供 Agent 主动调用 —— 具体配置方式见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +- **记忆片段**：从对话消息中自动提取关键事件（如“每天上午9点提醒我喝水”），支持自定义内容写入、语义检索、动态更新与去重。适用于大多数[长期记忆](../concepts/long-term-memory.md)场景。 +- **用户画像**：基于预定义模板（`CreateProfileSchema`）从对话中抽取结构化属性（如年龄、职业、爱好），支持多轮渐进式填充与 `GetUserProfile` 查询。适用于需固定字段的业务场景。 +- **自动捕获与召回**：在 OpenClaw 等框架中，可通过[插件](../concepts/plugin.md)生命周期钩子（`agent_end`/`before_agent_start`）实现无感的记忆写入与注入，详见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +- **工具集成**：除核心 API 外，还提供 `memory_search`、`memory_store`、`memory_list`、`memory_forget` 等运行时工具，供 Agent 主动调用。 -> **注意**：文档 2 声称“生成的记忆片段与用户画像暂无失效日期”，但文档 1 明确指出默认记忆片段规则有效期为 180 天，且控制台支持配置 7/30/180 天或永不过期。实际行为以控制台配置及 `AddMemory` 请求中 `expire_time` 参数为准，文档 2 的表述已过时。 +> **注意**：文档 3 声称“生成的记忆片段与用户画像暂无失效日期”，但文档 1 明确指出默认记忆片段规则有效期为 180 天，且可在控制台配置为 7/30/180 天或永不过期。实际行为以控制台配置及 `AddMemory` 请求中显式指定的 `expire_at` 或规则设置为准，建议以 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) 中的规则配置为准。 ## 关键参数 | 参数 | 类型 | 是否必填 | 说明 | |------|------|----------|------| -| `user_id` | string | 是 | 记忆隔离的主键，不同 `user_id` 数据完全隔离；OpenClaw 插件中为必填项，SDK/API 中亦为必需字段。 | -| `messages` | array | 否（与 `custom_content` 二选一） | 对话消息数组，用于自动提取记忆片段；格式为 `[{role: "user"/"assistant", content: "..."}]`。 | -| `custom_content` | string | 否（与 `messages` 二选一） | 直接写入的原始文本内容，绕过自动提取逻辑。 | -| `memory_library_id` | string | 否 | 指定目标记忆库 ID；不填则使用默认记忆库（每个账号自带一个，不可删除）。 | -| `project_id` | string | 否 | 指定记忆片段规则 ID；不填则使用该记忆库下默认规则。 | -| `profile_schema` | string | 否 | 用户画像 Schema ID；仅当需触发画像抽取时必填。 | -| `meta_data` | object | 否 | 自定义元数据键值对，用于分类、过滤或业务标记（如 `"location_name": "北京"`）。 | -| `top_k` | number | 否（默认 5） | `SearchMemory` 返回的最大记忆条数；OpenClaw 插件中默认为 5，建议设为 3–10 平衡效果与性能。 | -| `min_score` / `similarity_threshold` | number (0.0–1.0) | 否（默认 0.0 / 0.5） | 检索相似度阈值；文档 1 控制台推荐 0.5–0.7，文档 3 CLI 默认为 0（即无阈值），实际应按业务精度要求调整。 | +| `user_id` | string | 是 | 用户唯一标识，用于隔离不同用户的记忆空间；同一 `user_id` 共享命名空间。 | +| `memory_library_id` | string | 否 | 记忆库 ID；不填则使用默认记忆库（见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)）。 | +| `project_id` | string | 否 | 记忆片段规则 ID；不填则使用默认规则（见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)）。 | +| `profile_schema` | string | 否 | 用户画像模板 ID；用于触发结构化属性抽取（见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)）。 | +| `meta_data` | object | 否 | 自定义元数据，用于分类管理（如 `"location_name": "北京"`），支持后续按字段过滤。 | +| `top_k` | number | 否（默认 5） | `SearchMemory` 返回的最大记忆条数，推荐值 3–10（见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)）。 | +| `min_score` / `similarity_threshold` | number | 否（默认 0 / 0.5–0.7） | 相似度阈值（0.0–1.0），用于过滤低相关性结果；文档 2 使用 `minScore`（0–100 整数），文档 1 和 3 使用浮点阈值，实际 API 接受浮点值，建议统一使用 0.5–0.7 区间。 | ## 使用方式 -1. **API 直接调用**：配置 `DASHSCOPE_API_KEY` 环境变量后，通过 HTTP 请求调用标准 REST API（如 `POST /api/v2/apps/memory/add`）。所有接口均支持 cURL 与 Python SDK（`agentscope-runtime`）两种方式，示例详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 -2. **SDK 集成**：安装 `pip install agentscope-runtime`，使用封装好的工具类（如 `AddMemory`, `SearchMemory`, `CreateProfileSchema`）进行异步调用，避免手动构造请求体与处理认证头。 -3. **OpenClaw 插件**：通过 `openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw` 安装，并在 `~/.openclaw/openclaw.json` 中配置 `apiKey` 和 `userId` 即可启用全自动捕获与召回；也可通过 CLI（如 `openclaw modelstudio-memory search "用户偏好"`）或 Agent 工具调用进行手动干预。 +1. **准备环境**：设置 `DASHSCOPE_API_KEY` 环境变量（获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)）。 +2. **写入记忆**：调用 `AddMemory`，传入 `messages`（对话历史）或 `custom_content`（直接写入文本），指定 `user_id` 及可选参数（如 `profile_schema`）。示例见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)。 +3. **检索记忆**：调用 `SearchMemory`，传入 `user_id` 和自然语言查询（`query` 字段）或 `messages` 数组，返回语义匹配的记忆列表。 +4. **管理记忆**：使用 `ListMemory` 分页查看、`UpdateMemory` 修改内容、`DeleteMemory` 删除特定记忆节点（见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)）。 +5. **[插件](../concepts/plugin.md)集成（OpenClaw）**：安装 `@modelstudio/modelstudio-memory-for-openclaw` 插件，配置 `apiKey` 和 `userId`，启用 `autoCapture`/`autoRecall` 即可实现全自动记忆流转（见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)）。 ## 限制和注意事项 -- **速率限制（阿里云账号级别）**： - - 所有 API 总计 ≤ 3000 QPM - - `AddMemory` ≤ 120 QPM - - `SearchMemory` ≤ 300 QPM - 超限将返回 `429 Too Many Requests`，需自行实现重试退避逻辑。 -- **延迟特性**：`SearchMemory` 端到端延迟约 200–500ms，`AddMemory` 约 500–1000ms；OpenClaw 插件中 `autoCapture` 为异步执行，不影响主响应流。 -- **默认记忆库约束**：每个账号自带一个默认记忆库，不可删除，但可编辑名称、描述及规则；新建记忆库最多支持 50 条记忆片段规则 + 50 条用户画像规则。 -- **画像提取时效性**：调用 `AddMemory` 写入含画像信息的对话后，需等待约 3 秒再调用 `GetUserProfile` 获取结果，因系统需异步完成抽取与存储。 -- **兼容性说明**：OpenClaw 插件不支持阿里云百炼 Coding Plan 的 API Key，仅接受标准 DashScope API Key —— 此限制在 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) 中明确标注。 +- **配额限制**：阿里云账号级别限流，总计 ≤3000 QPM；其中 `AddMemory` ≤120 QPM，`SearchMemory` ≤300 QPM（见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) 和 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)）。 +- **延迟特性**：`AddMemory` 端到端延迟约 500–1000ms，`SearchMemory` 约 200–500ms；自动捕获为异步执行，不影响主链路响应速度。 +- **默认记忆库**：每个账号自带一个不可删除的默认记忆库，已预置一条有效期 180 天的“默认项目”规则（见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)）。 +- **用户画像提取**：单次对话难以覆盖全部字段，建议通过多轮对话渐进收集；画像字段名应语义唯一（如避免同时定义“年龄”“岁数”），描述需具体（见 [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)）。 +- **API Key 兼容性**：仅支持百炼平台标准 API Key，不支持 Coding Plan 的 API Key（见 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)）。 ## 来源文档 - [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) -- [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) +- [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md index 9061852e..cbf69bc5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md @@ -1,42 +1,44 @@ # model compression -模型压缩是百炼平台提供的量化能力，用于将全精度微调模型转换为低精度版本，在保持推理能力的前提下显著降低部署所需的 MU 规格与成本。该功能属于模型生产链路中的可选环节，位于[模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)之后、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-1/#3bc53b23c7shc)之前。**压缩不可逆**，产出模型不支持继续微调或二次压缩。 +模型压缩是百炼平台提供的量化优化能力，通过降低模型参数精度（如 FP16 → INT4/INT8），在保持推理能力基本不变的前提下显著减少部署所需的 MU 资源与成本。该功能仅作用于百炼平台微调产出的自定义模型，属于模型生产链路中可选但关键的一环：[模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634) → 模型压缩 → [模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-1/#3bc53b23c7shc)。**压缩不可逆**，压缩后模型既不支持继续微调，也不支持二次压缩。 ## 支持的模型与功能 -- **支持模型类型**：仅限通过百炼平台完成微调训练的自定义模型（即“微调产出模型”），不支持基础模型（如原始 Qwen）、第三方模型或 OSS 托管模型。 -- **当前支持系列**：Qwen 系列（例如 `qwen3.5-flash-2026-02-23`），具体以控制台实时展示为准。详见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中的“支持压缩的模型”表格。 -- **功能范围**：当前仅实现**后训练量化（PTQ）**，不包含结构剪枝、知识蒸馏等其他压缩技术。该限定在 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“功能概述”中已明确说明。 - -> **注意**：文档中提及“压缩不可逆”且“不支持二次压缩”，但未说明是否支持对同一源模型多次创建不同模板的压缩任务——实际支持，只要源模型状态为 `SUCCEEDED` 即可重复提交任务。此行为与文档中“切换源模型会自动清空已选的量化模板”逻辑一致，无矛盾。 +- **支持模型类型**：仅限百炼平台微调训练产出的自定义模型（如 `qwen3.5-flash-2026-02-23`），不支持基础模型、第三方模型或未完成微调的中间模型。 +- **支持的量化方法**：当前仅提供**量化（Quantization）**，不包含结构剪枝、知识蒸馏等其他压缩技术。详见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中“功能概述”章节。 +- **地域限制**：仅华北2（北京）地域可用。 +- **核心价值**：以 `qwen3.5-flash-2026-02-23` 为例，部署规格可从 `MU1*2`（¥108/小时）降至 `MU8*1`（¥47/小时），成本节省约 56%。该数据来自 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的实测示例，实际效果因模型与模板而异。 ## 关键参数 | 参数 | 是否必填 | 说明 | |------|----------|------| -| **任务名称** | 是 | ≤50 字符；建议含模型简称、量化方式、版本号（如 `qwen35-flash-w8a8-v1`） | -| **量化产出模型名后缀** | 是 | 仅小写字母+数字，≤8 位；将拼接至源模型名后（如源模型 `my-qwen-ft` + 后缀 `w4a4` → `my-qwen-ft-w4a4`） | -| **量化模板** | 是 | 卡片式选择；模板名中 MU 编号越大，部署规格越小、成本越低，但潜在精度损失可能增加。须先选源模型才可选模板。 | -| **校准数据** | 条件必填 | 仅当所选模板需校准输入时显示；最多选 5 个已在[数据管理](https://help.aliyun.com/zh/model-studio/manage-data/#9d2f7039bfo1a)中发布并启用的数据集；不支持 OSS 挂载数据集。 | +| **任务名称** | 是 | ≤50 字符；建议含模型简称、量化方式、版本号（如 `qwen35-flash-int4-v1`） | +| **量化产出模型名后缀** | 是 | 仅小写字母+数字，≤8 位；将拼接至源模型名后（如源模型 `my-qwen-ft` + 后缀 `int4` → `my-qwen-ft-int4`） | +| **量化模板** | 是 | 卡片式选择；模板名中 MU 编号越大，部署规格越小、成本越低，但潜在精度损失可能增加。**切换源模型会自动清空已选模板**（见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) “创建压缩任务”章节） | +| **校准数据** | 条件必填 | 仅当所选模板需校准输入时显示；最多选 5 个已发布数据集（不支持 OSS 挂载）；推荐选用与目标推理场景语义一致的数据（如客服场景用客服对话数据） | + +> **注意**：量化模板选择直接影响部署规格与精度平衡。文档中强调“模板名称中 MU 编号越大，部署规格越小、成本越低”，但未明确标注各模板对应的精度下降阈值或校准要求差异。开发者应结合业务测试集验证不同模板效果，而非仅依据 MU 编号决策。 ## 使用方式 -1. **前提条件**：工作空间中必须存在状态为 `SUCCEEDED` 的微调模型；若无可选模型，请确认已完成 [模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)，详见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 的“前提条件”章节。 -2. 控制台路径：**模型 > 模型训练 > 模型压缩** → 单击**创建压缩任务**。 -3. 配置参数后单击**开始压缩**（按钮仅在所有必填项完成时可用）。 -4. 任务创建后不可修改配置，务必在提交前确认量化模板与校准数据选择。 -5. 任务成功（`SUCCEEDED`）后，压缩后模型将出现在模型中心，可直接用于部署。 +1. **前提条件**：确保工作空间中已存在状态为“成功”的微调模型（参见 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) “前提条件”）。 +2. **入口路径**：控制台 → **模型** > **模型训练** > **模型压缩** → **创建压缩任务**。 +3. **配置并提交**：填写任务名称、选择源模型（仅展示可压缩模型）、指定后缀、选择量化模板、按需添加校准数据 → **开始压缩**。 +4. **监控与排查**： + - 任务列表页支持按状态（PENDING/QUEUING/RUNNING/SUCCEEDED/FAILED/CANCELING/CANCELED）、模板、时间等筛选； + - 详情页查看配置与错误信息，日志页支持下载全量日志、按级别着色（ERROR 红色）、自动刷新； + - 失败时优先检查详情页错误信息，再搜索日志中 `ERROR` 行，必要时提交工单并附任务 ID 与日志。 ## 限制和注意事项 -- **地域限制**：仅支持华北2（北京）地域。 -- **模型来源限制**：仅支持百炼平台内微调产出的自定义模型；基础模型、第三方模型、OSS 模型均不支持。 -- **不可逆性**：压缩后模型**不支持继续微调**，也**不支持二次压缩**；如需调整，必须回退至上游全精度微调模型重新提交任务。 -- **任务管理**： - - `PENDING` / `RUNNING` 状态可手动停止； - - `QUEUING` 状态不可删除； - - `SUCCEEDED` / `FAILED` / `CANCELED` 状态可删除（删除任务记录不影响已产出模型）。 -- **计费说明**：压缩任务本身限时免费（截止时间以控制台公告为准）；压缩后模型的部署费用按 MU 规格单独计费，与压缩免费期无关。 +- **模型来源限制**：仅支持百炼平台微调产出的自定义模型；基础模型、第三方模型、未完成微调的模型均不可压缩。 +- **操作不可逆性**：压缩后模型**不支持继续微调，也不支持二次压缩**；若需调整，必须回退至上游全精度微调模型重新发起压缩任务。 +- **任务不可修改**：创建后无法修改任何配置（包括量化模板、校准数据等），务必在点击“开始压缩”前确认。 +- **免费策略**：压缩任务本身限时免费（截止时间以控制台公告为准），但**压缩后模型的部署费用始终按 MU 规格计费**，不受免费期影响。 +- **地域与数据约束**：仅华北2（北京）可用；校准数据必须在百炼“数据管理”中创建并发布，不支持外部 OSS 数据集。 + +> **注意**：文档中多次强调“压缩不可逆”及“不支持二次压缩”，但未说明是否允许对同一源模型并发创建多个不同模板的压缩任务。实践中建议避免并发提交，以防资源冲突或状态混淆——此行为未在 [原文标题](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中明确定义，属隐含风险。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md index 46e3a89f..de8c799a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md @@ -1,84 +1,58 @@ # model context protocol -模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化接口协议，用于在大语言模型与外部工具（如地图、搜索、图表生成等服务）之间建立安全、可扩展的信息交互通道。它屏蔽了底层通信细节，使开发者无需为每个工具单独开发适配逻辑，即可在智能体或工作流中统一接入和编排多种能力。该协议基于开源 MCP 标准实现，支持云部署与自定义部署两种模式 [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md)。 +模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化接口协议，用于在大语言模型与外部工具（如地图、搜索、图表生成等服务）之间建立可互操作的信息通道。它屏蔽了底层通信细节，使开发者无需为每个工具单独开发适配逻辑，即可在智能体或工作流中声明式接入各类能力。该协议基于 Anthropic 提出的开源标准 [MCP 官网](https://modelcontextprotocol.io/) 实现，并已升级为 Streamable HTTP 协议以支持更稳定的外部集成。 ## 支持的模型/功能 -MCP 本身不绑定特定模型，而是作为能力接入层服务于百炼平台上的**智能体应用**和**工作流应用**。当前支持以下两类使用场景： +MCP 本身不绑定特定模型，而是作为**能力接入层**，供百炼平台内的以下两类应用调用： +- **智能体应用**：支持自动推理并动态调用最多 5 个已配置的 MCP 服务（如 `Amap Maps` 的路径规划、`Sequential Thinking` 的逻辑推理），无需显式指定工具； +- **工作流应用**：支持手动编排，每个 MCP 节点仅绑定一个具体工具（如 `maps_weather`），需通过前置大模型节点解析自然语言输入为结构化参数，再传递至 MCP 工具执行 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 -- **智能体应用**：大模型根据对话上下文自动判断是否调用、调用哪个 MCP 工具及传入参数，支持最多同时配置 5 个 MCP 服务 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 -- **工作流应用**：需显式添加 MCP 节点，并手动指定所用工具（如 `maps_weather`）、输入参数来源（如上游节点输出）和输出参数传递路径，适用于确定性、多步骤的工具链编排。 +当前官方已预置多种 MCP 服务，包括 Amap Maps（地理信息）、WebSearch（联网搜索）、Firecrawl（网页爬取）等，均支持一键开通使用；同时支持三类自定义部署方式： +- 使用脚本部署（npx/uvx 托管本地 MCP Server）； +- 从 AI 网关导入（将现有 RESTful API 封装为 MCP 工具）； +- 从阿里云 OpenAPI 导入（将 OSS/ECS 等云产品能力暴露为 MCP 工具） [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)。 -官方已预置并维护多种 MCP 服务，包括： -- Amap Maps（地理信息、路径规划、天气查询） -- WebSearch（联网搜索，含免费额度与计费规则） -- Firecrawl（网页爬取） -- Sequential Thinking（逻辑推理辅助） -- QuickChart（图表生成） - -此外，支持通过三种方式接入自定义 MCP 服务：脚本部署（npx/uvx）、AI 网关导入（封装 RESTful API）、阿里云 OpenAPI 导入（操作云资源） [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)。 - -> **注意**：文档 3 提到 MCP 协议已从旧版 SSE 升级为新版 Streamable HTTP 协议，而文档 2 和文档 4 的示例截图及部分配置项仍显示 SSE 相关字段（如 Cherry Studio 配置中类型标注为 `服务器发送事件 (sse)`）。实际部署时应以控制台最新 UI 和 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) 文档中明确的 `streamableHttp` 协议为准，避免因协议不匹配导致 `11200058` 或 `11200059` 错误。 +> **注意**：文档 4 明确指出 MCP 服务“**不能直接在调用千问 API 时接入**”，即 MCP 仅限百炼平台内智能体/工作流场景，不支持通过 DashScope SDK 直接调用千问模型时注入 MCP 工具 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 ## 关键参数 -MCP 服务配置与调用涉及以下核心参数： - -| 参数类别 | 参数名 | 说明 | 示例值 | -|----------|--------|------|--------| -| **服务元信息** | 服务名称、描述 | 仅用于控制台识别，不影响模型调用逻辑 | `"长期记忆"`, `"记录用户个性化信息"` | -| **连接配置** | `type` | 必填，指定通信协议类型，决定端点路径与请求方法 | `"stdio"`（本地）、`"sse"`（已逐步淘汰）、`"streamableHttp"`（推荐） | -| | `url` | 远程 MCP Server 地址（`type` 为 `streamableHttp` 时必填） | `"https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp"` | -| | `command` / `args` | `type` 为 `stdio` 时指定启动命令与参数 | `"npx"`, `["-y", "@modelcontextprotocol/server-memory"]` | -| **认证与安全** | `Authorization` header | 外部调用时必需，格式为 `Bearer ` | — | -| | KMS 凭据 | 涉及敏感密钥（如 `AMAP_MAPS_API_KEY`）时，必须通过 KMS 加密存储 | — | -| **工具级参数** | `tool.name` | 工具唯一标识符，模型调用时必须精确匹配 | `"maps_weather"`, `"web_search"` | -| | `tool.inputSchema` | JSON Schema 定义输入参数结构，影响模型参数生成准确性 | `{"type": "object", "properties": {"query": {"type": "string"}}}` | - -所有参数均需严格遵循 MCP 协议规范，否则将触发 `11200054`（协议解析错误）或 `11200060`（Bad Request）等错误码 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +| 参数类别 | 关键字段 | 说明 | +|----------|----------|------| +| **服务配置** | `type`（`stdio`/`sse`/`streamableHttp`） | 必须与接入端点严格匹配：`sse` 对应 `/sse`，`streamableHttp` 对应 `/mcp`；配置错误会导致 `11200058` 错误码 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) | +| **部署模式** | `基础模式` / `极速模式` | 基础模式按调用时长计费（0.000156 元/秒），无部署费；极速模式额外收取部署费（0.000036 元/秒），适合高频调用场景 [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md) | +| **安全凭证** | KMS 加密凭据 | 敏感参数（如 `AMAP_MAPS_API_KEY`）必须通过 KMS 凭据加密，不可明文填写 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) | +| **外部调用** | `DASHSCOPE_API_KEY` + `mcp_url` | 外部 SDK 集成时需提供百炼 API Key 和服务地址（如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`），且必须使用 `streamableHttp` 协议 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) | ## 使用方式 -### 1. 开通服务 -- 访问 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market)，选择目标服务（如 Amap Maps），点击“立即开通”。 -- 对于需密钥的服务（如商业化高德地图），在开通流程中通过 KMS 创建并关联加密凭据。 +### 平台内集成（智能体/工作流） +1. **开通服务**：前往 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market)，选择服务卡片点击「立即开通」； +2. **添加到应用**： + - 智能体：在应用编辑页「MCP 服务」区域添加，最多 5 个； + - 工作流：拖入「MCP 节点」，手动选择工具并绑定输入参数（如引用上游节点输出）； +3. **提示词优化**：明确指令工具名称与能力（例：“调用 Amap Maps MCP 服务规划杭州到上海的路线”），避免模糊表述导致调用失败 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 -### 2. 在智能体中集成 -- 创建智能体后，在「MCP 服务」配置页添加已开通的服务； -- 模型将依据提示词自动决策调用时机与参数，无需显式声明工具名（但提示词中明确工具能力可提升成功率）。 - -### 3. 在工作流中集成 -- 添加 MCP 节点，选择具体工具（如 `maps_weather`）； -- 手动配置输入参数（支持引用上游节点输出，如 `"引用：信息提取/result"`）； -- 输出结果需通过变量引用传递至后续节点（如大模型总结节点）。 - -### 4. 外部调用（第三方应用或 SDK） -- **集成至 Cherry Studio/Cursor**：在 MCP 服务详情页选择对应客户端，执行“一键配置”或手动导入 JSON 配置； -- **SDK 编程调用**：使用 `mcp.client.streamable_http` 客户端连接，配合 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)完成多轮工具调用循环（详见 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) 中的 Python 示例）。 +### 外部 SDK 集成 +1. 安装依赖：`pip install openai mcp`； +2. 初始化 `streamablehttp_client`，传入 `mcp_url` 和 `Authorization` 头； +3. 调用 `session.list_tools()` 获取工具列表，转换为 OpenAI `tools` 格式； +4. 在 `chat.completions.create` 中启用 `tools`，处理 `tool_calls` 并通过 `session.call_tool()` 执行 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 ## 限制和注意事项 -- **模型兼容性限制**：MCP 服务**仅支持在百炼平台的智能体或工作流应用中使用**，无法直接接入千问 API 的原始调用（如 `dashscope.ChatCompletion.create`）[MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -- **网络与权限限制**： - - 自定义 MCP 服务运行于函数计算 FC 环境，**无固定出口公网 IP**，访问云数据库等远程资源需配置 IP 白名单或 VPC 打通； - - **不支持访问用户本地资源**（如本地文件、硬件设备），此类服务应在本地部署。 -- **部署与更新限制**： - - 通过 `npx`/`uvx` 部署的服务，版本更新后**必须手动重新部署**，不会自动同步； - - 私有 npm/PyPI 仓库中的包暂不支持直接部署，需发布至公共仓库或改用 `streamableHttp` 连接远程服务。 -- **计费与限流**： - - 云部署服务（如 WebSearch）有明确 QPS 限制（如 15 QPS，主账号与 RAM 子账号共享）和调用费用（29 元/千次）； - - 自定义服务按“基础模式”（按调用时长计费）或“极速模式”（按部署+调用时长计费）计费，费率均为 0.000156 元/秒 [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md)。 -- **调试建议**： - - 遇到连接失败（如 `11200044`）或超时（如 `11200045`），优先使用 `curl` 测试服务地址连通性； - - 遇到协议错误（如 `11200054`），务必核对 `type` 与端点路径是否匹配（`streamableHttp` → `/mcp`，`sse` → `/sse`）； - - 模型调用失败时，首先检查提示词是否清晰表达工具意图，其次确认所选模型是否具备足够推理能力（推荐使用 Qwen-Max 或 Qwen3 系列）。 +- **网络限制**：自定义 MCP 服务托管于函数计算 FC，**无固定出口公网 IP**，访问云数据库等远程资源需配置 IP 白名单或 VPC 打通 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)； +- **本地资源不可达**：不支持访问用户本地文件、硬件或数据库，仅限云端可访问服务 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)； +- **版本同步**：通过 `npx/uvx` 部署的服务，上游包更新后**不会自动生效**，需手动重新部署 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)； +- **[Token](../concepts/token.md) 开销**：MCP 返回结果会作为上下文输入模型，**直接增加输入 [Token](../concepts/token.md) 数量**；丰富上下文也可能间接增加输出 [Token](../concepts/token.md) [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)； +- **协议兼容性**：旧版 SSE 服务需手动升级为 Streamable HTTP 协议，否则外部调用可能失败 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 ## 来源文档 - [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md) - [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) -- [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) +- [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md index 4cc49713..85be19a5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md @@ -1,38 +1,36 @@ # model data overview -百炼平台的模型数据体系围绕训练与评测两大核心场景构建，提供结构化、可管理的数据集支持。本文档汇总了当前支持的模型类型、关键数据格式参数、使用方式及限制条件，面向开发者提供可直接落地的技术参考。所有功能均需在华北2（北京）地域使用。 +百炼平台的模型数据体系为大模型训练与评测提供结构化、可管理的数据支撑，涵盖训练集（SFT/CPT/DPO/图生视频）、评测集及配套的数据处理能力。所有数据均需通过统一的数据管理界面上传与版本控制，地域限制为华北2（北京）。本文档聚焦数据格式、参数约束、使用路径及关键限制，面向开发者提供实操指引。 ## 支持的模型/功能 -- **训练集类型**：支持文本生成（SFT、DPO、CPT）、多模态理解（Qwen-VL 系列）、图生视频（首帧模式、首尾帧模式）三类训练任务。其中 SFT 支持 ChatML 格式多轮对话，DPO 支持偏好对标注，CPT 为纯文本预训练格式；图生视频训练集需严格按 ZIP 压缩包结构组织图像、视频及 `data.jsonl` 标注文件 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -- **评测集类型**：当前仅支持文本生成类单轮对话评测集（Excel 或 JSONL 格式），用于模型效果横向对比与迭代评估 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -- **数据处理能力**：提供数据清洗（如敏感信息打码、URL 移除）与数据增强（基于千问-Max 的 Few-Shot 生成）两类算子，**仅适用于 SFT-文本生成训练集（ChatML 格式）**，不支持 SFT-图片理解、DPO 或 CPT 数据集 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **训练集类型**：支持文本生成（SFT/CPT/DPO）、多模态理解（Qwen-VL 系列）、图生视频（首帧/首尾帧）三类训练任务。其中 SFT 支持 ChatML 格式多轮对话，CPT 为纯文本 JSONL，DPO 需包含 `chosen`/`rejected` 对比样本；图生视频训练集必须打包为 ZIP，含 `data.jsonl` 及对应图像/视频文件 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **评测集类型**：当前仅支持文本生成类单轮评测集（Excel 或 JSONL 格式），每条记录含 `Prompt` 和 `Completion` 字段，用于自动化或人工评分 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **数据处理能力**：提供数据清洗（如敏感信息打码、URL 移除）和数据增强（基于千问-Max 的 Few-Shot 生成）功能，**仅适用于 SFT-文本生成训练集（ChatML 格式）**，不支持 SFT-图片理解、DPO 或 CPT 数据 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 -> **注意**：文档 1 中称“支持图生视频（首帧）、（首尾帧）训练集”，而文档 2 明确指出“暂不支持[SFT-图片理解训练集]”，但未提及图生视频是否支持数据清洗/增强。结合上下文及控制台实际能力，图生视频类训练集**不支持任何数据清洗或增强操作**——该限制未在文档 1 中说明，属隐含约束。 +> **注意**：文档 1 中称“支持图生视频（首帧）”、“图生视频（首尾帧）”训练集，而文档 2 明确说明数据清洗/增强“暂不支持[SFT-图片理解训练集]和[DPO-文本生成训练集]”，且未提及图生视频数据处理能力。因此，**图生视频训练集不可进行任何数据清洗或增强操作**，该限制需在实际使用中严格遵守。 ## 关键参数 -- **`loss_weight`**：SFT（所有 assistant 行）和 DPO（`chosen` 字段）中支持，取值范围 `0.0 ~ 1.0`，用于调节单条样本训练权重；属邀测功能，需联系商务经理开通 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -- **视觉输入字段**：VL 模型要求 `system.content` 必须为数组格式 `[{"text": "..."}]`；图像/视频字段需显式声明 `resized_width`/`resized_height`；视频支持 `fps`（文件路径模式）或 `sample_fps`（帧列表模式）参数 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -- **坐标规范**：Qwen2.5-VL 使用绝对像素坐标，Qwen3-VL 使用 `[0, 999]` 归一化相对坐标，模型版本不匹配将导致物体定位失效。 -- **增强控制参数**：数据增强节点中 `指令生成依赖样本数`（few-shot 数量）、`生成样本数`（最大 2000 条/任务）、`过滤相似度阈值` 共同影响输出质量与多样性 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **`loss_weight`**：SFT（ChatML 和 Thinking 模式）及 DPO 的 `chosen` 字段中支持，取值范围 `0.0 ~ 1.0`，用于调节单条 assistant 输出或 chosen 样本的训练权重。该参数为邀测功能，需联系商务经理开通 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **视觉字段约束**：Qwen-VL 训练中，`system` 消息的 `content` 必须为数组格式 `[{"text":"..."}]`，不可用字符串；图像/视频字段需显式声明 `resized_width`/`resized_height`；Qwen3-VL 坐标为 `[0,999]` 相对坐标，Qwen2.5-VL 为像素绝对坐标。 +- **ZIP 包规范**：所有多模态/图生视频训练集必须为 ZIP 格式，最大 2 GB；`data.jsonl` 必须位于根目录；文件名仅支持 ASCII 字母、数字、下划线、连字符；图像单张 ≤ 1024px 宽高、≤ 10MB；图生视频图像/视频分辨率上限为 4096×4096 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 ## 使用方式 -- **数据集创建**：通过控制台 [数据管理](https://bailian.console.aliyun.com/#/efm/model_data) 统一上传 ZIP（VL/图生视频）或 JSONL/XLSX（文本）文件，训练集必须包含根目录 `data.jsonl`，图像/视频文件名全局唯一且不可嵌套路径。 -- **数据处理流程**：仅限 SFT 文本训练集，需先在控制台创建数据流（含清洗+增强节点），再基于该数据流启动任务；处理后自动生成新版本（如 V2），原数据集不受影响 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 -- **验证集构建**：图生视频验证集无需提供视频文件，仅需首帧/首尾帧图像 + `data.jsonl`，系统将在评估节点自动调用模型生成预览视频 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **创建与上传**：通过控制台 [数据管理](https://bailian.console.aliyun.com/#/efm/model_data) 页面上传训练集/评测集，系统自动校验格式与结构。文本类数据（SFT/CPT/DPO）支持 `.jsonl` 直传；多模态/图生视频需打包 ZIP 并确保目录结构合规。 +- **数据处理流程**：仅对 SFT-文本生成训练集有效。需先在“数据流”页签创建数据流（含开始→数据清洗→数据增强→结束节点），发布后在“任务列表”中选择目标训练集启动任务。处理结果将生成独立版本（如 V2），原数据集不受影响 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **评测执行**：上传文本生成评测集后，在 [模型评测](https://help.aliyun.com/zh/model-studio/model-evaluation-overview) 页面关联模型并启动评测任务，系统将基于每条 `Prompt` 进行推理，并比对 `Completion` 进行评分。 ## 限制和注意事项 -- **地域限制**：所有功能仅支持华北2（北京）地域，跨地域调用将失败。 -- **格式强约束**： - - VL 训练集 ZIP 包内文件名仅支持 ASCII 字符（a-z, A-Z, 0-9, `_`, `-`），大小上限 2 GB； - - 图生视频 ZIP 中 `data.jsonl` 必须位于根目录，图像/视频路径在 JSONL 中仅写文件名（如 `"image_1.jpg"`），**不可带子目录路径**； - - Excel 评测集仅支持单轮对话，多轮或复杂结构将解析失败。 -- **规模建议**：CPT 需 ≥10M Token；SFT 需 ≥1000 条优质样本；DPO 需 ≥100 条偏好对；低于阈值易导致调优效果不佳 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -- **API 缺失**：数据清洗与增强功能**暂无公开 API**，必须通过控制台操作 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 -- **模型兼容性**：图生视频训练集仅适配 Wan 系列模型；Qwen3.5-VL 及以后版本才支持视频文件路径模式；旧版 VL 模型不兼容新坐标规范。 +- **地域限制**：所有功能（数据上传、清洗、增强、训练、评测）**仅支持华北2（北京）地域**，跨地域调用将失败。 +- **格式与兼容性**： + - SFT 训练不支持 OpenAI 的 `name`、`weight` 参数； + - Excel 格式仅支持单轮 SFT 训练集（`.xls`/`.xlsx`），多轮必须用 `.jsonl`； + - 图生视频验证集无需提供视频文件，由系统自动调用模型生成预览。 +- **规模建议**：CPT 至少需 1000 万 [Token](../concepts/token.md) 预训练数据；SFT 微调建议 ≥1000 条高质量样本；DPO 偏好数据建议 ≥100 条 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **数据处理限制**：无 API 接口，仅支持控制台操作；数据增强每次最多生成 2000 条样本；增强过程依赖千问-Max 模型，不可更换 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md index cdc38f8d..9188652a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md @@ -1,54 +1,47 @@ # model deployment 1 -百炼平台提供三种模型部署方式：预置吞吐（PTU）、模型单元（MU）和按 Token 用量计费，分别面向高并发低延迟、资源隔离可定制、以及低成本验证等不同业务场景。所有部署均通过统一 API 接口或控制台完成，支持预置模型与 LoRA 微调模型，但全参微调模型暂不支持导入与部署。部署即计费，服务状态变更（如扩容、下线）需注意计费规则与权限约束。 +`model deployment 1` 是百炼平台面向生产环境的模型服务化核心能力，提供三种主流部署模式：预置吞吐（PTU）、模型单元（MU）和按 [Token](../concepts/token.md) 用量计费。其中 PTU 模式专为高并发、低延迟、流量可预估的场景设计，支持长输入与前缀缓存优化；MU 模式提供资源独占与性能自定义能力；[Token](../concepts/token.md) 用量模式适用于效果验证与轻量调用。所有模式均通过统一 API 接口调用，支持 OpenAI、Anthropic 和 DashScope 兼容协议。 ## 支持的模型/功能 -- **预置模型**：千问系列（Qwen3/2.5/Flash/Plus/Max/VL/Omni）、DeepSeek（v3/v3.2/v4-Pro/v4-Flash）、GLM（5.2/5.1/4.7）、MiniMax-M2.5、Kimi-K2.5、CosyVoice 等，详见 [模型部署简介](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) 中的计费表格。 -- **自定义模型**：仅支持 LoRA 微调模型导入与部署，需满足 rank ∈ {8,16,32,64}、词汇表与 chat_template 未修改、VL 模型 VIT 部分冻结等严格要求；全参微调模型明确不支持 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 -- **核心功能**： - - PTU 模式支持长输入（最高 256K token）与前缀缓存，自动应用阶梯系数与缓存折扣 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)； - - MU 模式支持 PD 分离计算（降低首 Token 延迟）、推理模式选择（Instruct/Thinking）、最长上下文与服务限流配置； - - Token 计费模式仅适用于经 SFT 训练后的 LoRA 模型，且仅限部分基础模型（如 qwen3-32b/qwen3-8b/qwen2.5-vl-7b 等）。 +- **PTU 部署**：支持 `glm-5.1`、`deepseek-v4-pro`、`qwen3.7-plus-2026-05-26` 等主流预置模型，最高支持 **256K 输入 token**（如 `glm-5.2` 达 1M），并启用前缀缓存优惠 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **模型单元（MU）部署**：支持全部预置模型及 LoRA 微调模型（含千问、GLM、DeepSeek、Kimi、MiniMax 等），支持 PD 分离计算模式以降低首 [Token](../concepts/token.md) 延迟 [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md)。 +- **Token 用量部署**：仅支持经 LoRA 微调后的部分基础模型（如 `qwen3-32b`、`qwen3-14b` 等），不支持全参微调模型或视觉语言模型（VL）的 LoRA 导入 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 +- **模型导入能力**：仅支持 LoRA 格式（`adapter_model.safetensors` + `adapter_config.json`），要求 rank ∈ {8,16,32,64}，且必须冻结 VIT（对 VL 模型）、禁用 vocab/chat_template 修改 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 -> **注意**：文档 1 中“支持模型”表格称“部分预置模型与所有调优后模型”支持模型单元计费，但文档 3 明确限定“仅支持导入 LoRA 模型”，且文档 4 的 API 示例中 `plan: "lora"` 实际对应 Token 计费（非 MU），三者存在术语混淆。实际支持情况以 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) 的 LoRA 限制为准：**只有符合规范的 LoRA 模型才能部署，且 MU/PTU/TOKEN 三种计费方式均仅对 LoRA 模型开放**。 +> **注意**：文档 2 中 `glm-5.1` 的输入上限标为 64K，但文档 1 明确其支持 200K；文档 2 表格中 `qwen3.7-plus-2026-05-26` 输入上限为 256K，与文档 1 一致。以文档 1 的实测能力为准，即 `glm-5.1` 实际支持 200K，控制台展示值可能滞后。 ## 关键参数 -| 参数 | 适用模式 | 说明 | 示例值 | -|------|----------|------|--------| -| `plan` | 全部 | 计费策略标识：`ptu` / `mu` / `lora`（注意：`lora` 此处指 Token 计费，非模型类型） | `"ptu"` | -| `ptu_capacity.input_tpm` / `output_tpm` | PTU | 预置吞吐额度（每分钟 Token 数），决定服务容量上限 | `{"input_tpm": 10000, "output_tpm": 1000}` | -| `deploy_spec` / `capacity` | MU | 模型单元规格（如 `"MU1"`）与副本数，直接关联算力与并发能力 | `"MU1"`, `4` | -| `enable_thinking` | MU | 是否启用思考模式（影响输出单价与性能） | `true` | -| `max_context_length` | MU | 最长上下文长度（部分模型支持，单位 token） | `10000` | -| `rpm_limit` / `tpm_limit` | MU | 服务级限流阈值（每分钟请求数 / 每分钟 Token 数） | `500`, `1000` | +| 参数 | PTU 模式 | MU 模式 | Token 用量模式 | +|------|----------|---------|----------------| +| **核心配置** | `input_tpm`, `output_tpm`（单位：token/分钟） | `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`, `max_context_length`, `rpm_limit`, `tpm_limit` | `plan: "lora"`，`capacity` 字段必须传但无效 | +| **缓存控制** | `provisioned_tokens`（含阶梯系数与缓存折扣）、`cached_tokens`（仅 OpenAI/DashScope 兼容格式返回） | 不支持前缀缓存 | 不支持前缀缓存 | +| **计费标识** | 响应头含 `x-dashscope-ptu-overflow:true`（溢出时），响应体含 `service_tier: "ptu-standard"` | 无专用额度字段，`service_tier` 不返回或为 `"default"` | 无专用额度字段，`service_tier` 不返回或为 `"default"` | -- PTU 模式不支持自定义 `max_context_length` 或限流，其吞吐与延迟由平台预置； -- Token 计费模式（`plan: "lora"`）的 `capacity` 参数无效，仅需填写占位值（如 `1`），扩缩容必须通过控制台申请 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **长输入阶梯系数**（仅 PTU）：`glm-5.1` 在 `[0,32K)` 区间系数为 1.0，`[32K,200K]` 区间输入系数升至 1.33、输出 1.17；`deepseek-v4-pro` 和 `qwen3.7-plus-2026-05-26` 无阶梯，全程系数为 1.0 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **缓存折扣率**（仅 PTU）：`glm-5.1` 和 `qwen3.7-plus-2026-05-26` 为 0.2（命中部分按 20% 折算），`deepseek-v4-pro` 为 0.08 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 ## 使用方式 -1. **控制台部署**：访问 [模型部署控制台](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_deploy/create)，选择模型、计费方式及对应参数（如 PTU 容量或 MU 规格），提交创建。 -2. **API 部署**（推荐自动化）： - - PTU：`POST /api/v1/deployments`，携带 `plan: "ptu"` 与 `ptu_capacity` 对象； - - MU：`POST /api/v1/deployments`，携带 `plan: "mu"`、`deploy_spec`、`capacity` 及可选 `enable_thinking` 等； - - Token 计费：`POST /api/v1/deployments`，携带 `plan: "lora"` 与占位 `capacity`。 -3. **状态查询与管理**：通过 `GET /api/v1/deployments/{deployed_model}` 获取状态（`RUNNING` 表示就绪），`DELETE /api/v1/deployments/{deployed_model}` 下线服务。 -4. **推理调用**：使用 `model` 参数指定部署服务 ID（即 `deployed_model` 字段值），而非基础模型名，例如 `model='qwen3-8b-ft-202511132025-0260'`。 +- **控制台部署**：登录 [百炼控制台 → 模型部署 → 创建部署](https://bailian.console.aliyun.com/#/efm/model_deploy/create)，选择模型、计费方式及对应参数（如 PTU 容量计算器、MU 规格、限流阈值等）。 +- **API 部署**（推荐自动化）： + - PTU：`POST /api/v1/deployments`，`"plan": "ptu"`，携带 `ptu_capacity` 对象； + - MU：`"plan": "mu"`，指定 `deploy_spec`、`capacity`、`enable_thinking` 等； + - Token 用量：`"plan": "lora"`，`capacity` 必填但忽略 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **推理调用**：使用 `deployed_model`（即部署后生成的专属服务 ID）作为 `model` 参数，通过 `/api/v1/services/{deployed_model}/completions` 或 SDK（如 `dashscope.Generation.call(model='xxx')`）发起请求，**无需修改 endpoint 或鉴权逻辑**。 ## 限制和注意事项 -- **权限约束**：API 部署需确保 API Key 所属业务空间已授权目标模型的部署权限，否则报错 `Workspace xxx does not have deployment privilege for model xxxx` [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 -- **计费刚性**：部署成功即开始计费，PTU/MU 无法中途切换计费方式，必须先下线再重建；PTU 预付费订单不可提前终止，首月退订按日单价 1.2 倍计费。 -- **额度溢出**：PTU 模式下，超出购买 TPM 或输入超模型上限（如 Qwen 128K）时，请求自动降级为按量计费，响应头含 `x-dashscope-ptu-overflow:true`，`service_tier` 字段不返回或为 `default` [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 -- **LoRA 导入限制**：OSS Bucket 必须添加 `bailian-datahub-access` 标签，且模型文件不得位于根目录；`adapter_model.safetensors` 中禁止出现 `visual` 相关权重参数 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 -- **地域限制**：API 部署当前仅支持华北2（北京）地域 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **PTU 溢出行为**：超出购买 TPM 或输入超过模型上限（如千问系列 128K、DeepSeek 系列 64K）时，请求**自动降级为按量计费**，API 响应中 `service_tier` 缺失或为 `"default"`，响应头含 `x-dashscope-ptu-overflow:true`，业务无感知但费用结构变化 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **模型单元约束**：MU 部署不支持思考模式与非思考模式动态切换（需在部署时固定），且 `max_context_length` 设置受基础模型原生上限约束（如 `qwen3-8b` 最高支持 128K，不可设为 256K）。 +- **LoRA 导入硬性限制**：不支持全参微调模型；若 `adapter_model.safetensors` 中存在 `visual.` 开头的权重键，则导入失败；`chat_template` 必须与开源基础模型完全一致，否则部署后效果异常 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 +- **地域与权限**：API 部署仅支持华北2（北京）地域；API Key 所属业务空间必须显式授权目标模型的部署权限，否则报错 `Workspace xxx does not have deployment privilege for model xxxx` [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 ## 来源文档 -- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) +- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) - [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md index 9826366b..cd56936b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md @@ -1,65 +1,64 @@ # model evaluation introduction -模型评测是百炼平台提供的核心能力评估功能，用于对文本生成类模型的推理结果进行结构化打分与对比分析。它通过可复用的评测维度定义评分规则，支持大模型自动评估、规则匹配和人工评审三种范式，帮助开发者量化模型表现、验证调优效果或支撑选型决策。所有评测均基于明确的输入（Prompt）、输出（Output）与参考答案（Completion）三元组展开。 +模型评测是百炼平台提供的模型能力量化评估功能，支持通过自定义或基线方式对文本生成类模型进行多维度打分与对比。它面向模型选型、调优验证、能力归因和持续监控等核心场景，提供 AI 自动评测、规则评估和人工评估三类评分机制，并可生成结构化报告与排行榜。该功能当前仅支持文本生成类模型，不支持多模态或语音类模型。 -## 支持的模型/功能 +## 支持的模型与功能 -百炼模型评测当前**仅支持文本生成类模型**，覆盖预置模型与调优后模型。评测功能分为两类： +- **支持模型类型**：仅限文本生成类模型（包括预置模型与调优后模型），详见[模型评测产品概览](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **评测方式**： + - **自定义评测**：用户上传评测数据集（EvaluationSet 类型）或推理结果集，自主创建评测维度并关联执行；支持全地域。 + - **基线评测**：使用平台预置公开数据集（如 C-Eval、GSM8K、BBH 等）快速评估基础能力；**仅北京地域可用**，且不支持下载结果、不显示综合得分列，详见[创建基线评测任务](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **评分器类型**：共五种，分为三大类： + - *大模型评估*（分类型/数值型）：依赖裁判模型（如千问-Max）进行语义级评判，产生 [Token](../concepts/token.md) 计费； + - *规则评估*（字符串匹配/文本相似度）：基于算法（ROUGE/BLEU/Cosine/Fuzzy Match 等）自动计算，零裁判模型费用； + - *人工评估-分类型*：由人工标注 Pass/Fail，无模型费用，但需人力投入。 + 各类型适用场景与参数差异详见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 -- **自定义评测**：使用用户上传的评测数据集（EvaluationSet 类型，含 Prompt 和 Completion 列）或已有的推理结果集，结合自定义创建的评测维度执行评分。支持三种评分方式： - - 大模型评估（AI 自动评测）：调用裁判模型（如千问-Max）进行语义级评判； - - 规则评估（自动化指标）：基于字符串匹配或 BLEU/ROUGE/余弦等算法计算分数； - - 人工评估（人工标注）：由评测人员按标签逐条标注 Pass/Fail。 - 详见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 - -- **基线评测**：仅在北京地域可用，使用平台预设的公开标准数据集（如 C-Eval、GSM8K、BBH），系统自动执行预置维度下的评分，不支持自定义维度或结果下载。其设计目标是快速获取基础能力基准分，与自定义评测形成互补 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 - -> **注意**：文档 1 中未提及基线评测的地域限制，而文档 2 明确指出“基线评测仅北京地域可用”，该信息以文档 2 为准。 +> **注意**：文档 1 称“当前仅支持文本生成类模型评测”，而文档 2 未明确限定模型类型，但其所有示例与参数说明均围绕文本生成展开，且未提及多模态输入/输出支持。因此以文档 1 的明确声明为准，确认功能边界。 ## 关键参数 -评测维度的核心参数依类型而异，需在创建时准确配置： - -- **通用参数**：维度名称（≤20 字符，必填）、描述（≤100 字符，选填）、类型（5 种之一，创建后不可更改）。 -- **大模型评估类型**（分类型/数值型）： - - 裁判模型（必填，推荐千问-Max）； - - 评分器 Prompt（必含 `${prompt}`、`${output}` 或 `${completion}` 至少一个变量）； - - 分类型：Pass/Fail 标签（互斥且不可重复）； - - 数值型：评分范围（整数区间，默认 0~5，最大值建议 ≤10）、通过阈值（小数，步长 0.1，默认 3.0）。 -- **规则评估类型**： - - 字符串匹配：比较操作符（相等/不相等/包含）、评测输入与模型输出（至少一侧含变量）； - - 文本相似度：评估指标（7 种算法可选，如 ROUGE-L 适用于摘要、BLEU 适用于翻译）、通过阈值（0~1，步长 0.01）。 -- **人工评估类型**：仅需配置 Pass/Fail 标签，无裁判模型调用。 - -所有维度模板均可在控制台[评测维度列表页](https://bailian.console.aliyun.com/#/efm/model_evaluate/dimension_template)统一管理，修改仅影响后续评测任务，已运行任务结果不变 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 +| 参数类别 | 参数名 | 说明 | 必填性 | 注意事项 | +|----------|--------|------|--------|----------| +| **维度通用** | 维度名称 | 最长 20 字符，建议采用“评估方面+评估方式”命名（如`回答准确性-LLM评分`） | 是 | 创建后可修改，不影响已关联任务 | +| | 描述 | 最长 100 字符，补充评判目标 | 否 | — | +| **大模型评估专用** | 裁判模型 | 如千问-Max，影响评分质量与费用 | 是（仅大模型评估） | 推荐千问-Max；费用按 [Token](../concepts/token.md) 计费 | +| | 评分器 Prompt | 含 `${prompt}`、`${output}`、`${completion}` 变量，至少引用一个 | 是（仅大模型评估） | 模糊 Prompt 易导致分数集中或区分度低，详见[配置评分器Prompt](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) | +| | 评分范围（数值型） | 整数区间，如 `0-5`，默认 `0-5` | 是（仅数值型） | 范围过大（如 `0-100`）会降低 LLM 评分一致性 | +| | 通过阈值 | 判定 Pass 的最低分（数值型）或相似度（规则型），步长 0.1（数值型）/0.01（相似度型） | 是（数值型/相似度型） | 与评分范围联动；3.0 是常用阈值，非强制 | +| | Pass/Fail 标签（分类型） | 标签互斥且穷尽，各标签 ≤20 字符 | 是（仅分类型） | 同一标签不可在 Pass 与 Fail 中重复出现 | ## 使用方式 -完整评测流程为四步闭环： -1. **准备数据集**：在数据管理模块上传 EvaluationSet 类型数据（含 Prompt 和 Completion 列），或准备已含 Output 的推理结果集； -2. **创建评测维度**：根据场景选择类型并配置参数（如大模型评估-数值型 + 千问-Max + 综合评测模板 + 0~5 分 + 阈值 3.0）； -3. **创建评测任务**：选择模型、指定数据来源（评测数据集或推理结果集）、关联维度、设置是否参与排行； -4. **查看结果**：在任务详情页的「指标统计」Tab 查看综合得分、通过率及分布图，在「数据明细」Tab 审查逐样本评分。 - -> **注意**：文档 1 提到“评分器类型创建后不可更改，选错只能删除重建”，而文档 2 补充说明“已关联该维度的评测任务不受影响”，该细节对运维安全至关重要，应严格遵循。 - -任务提交后状态流转为：待执行 → 进行中 → 评测完成/失败/终止。人工评估任务需全部标注完成后才变为“评测完成”状态 [原文标题](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +1. **准备数据**：在数据管理模块上传 `EvaluationSet` 类型数据集（含 `Prompt` 和 `Completion` 列），或准备已含 `Output` 的推理结果集。 +2. **创建维度**：在**模型评测 > 评测维度**页创建至少一个维度模板，选择类型并完成参数配置（如裁判模型、Prompt、评分范围等）。 +3. **创建任务**：在**模型评测 > 评测任务**页创建任务： + - 选择**自定义评测**（全地域）或**基线评测**（仅北京）； + - 选择被评测模型（预置或调优模型）； + - 配置数据来源（评测数据集 → 触发推理并计费；推理结果集 → 仅评分，不计推理费）； + - 关联已创建的维度； + - （可选）开启“参与排行”并绑定排行榜。 +4. **查看结果**：任务状态变为“评测完成”后，在详情页的**指标统计**Tab 查看综合得分、通过率及分布图；在**数据明细**Tab 查看逐条评分。人工评估任务需全部标注完成后才转为“评测完成”。 ## 限制和注意事项 -- **模型限制**：仅支持文本生成类模型，不支持多模态、语音或结构化输出模型。 -- **维度限制**：类型一旦创建不可修改；被排行榜绑定的维度删除后，将阻止新任务创建；已被评测任务引用的维度无法直接删除。 +- **地域限制**：基线评测功能仅在北京地域可用，其他地域控制台不显示该选项，属正常行为。 +- **模型限制**：仅支持文本生成类模型；不支持图像、音频等多模态模型评测。 +- **维度不可变**：评测维度的**类型**创建后不可修改，选错需删除重建；已关联任务不受影响，但排行榜若绑定该维度，删除后将阻止新任务创建。 - **费用说明**： - - 使用评测数据集时产生被评测模型推理费用（按 Token 计费）； - - 仅大模型评估维度产生裁判模型评分费用（按 Token 计费）； + - 使用评测数据集时，产生**被评测模型推理费用**（按输入/输出 [Token](../concepts/token.md) 计费）； + - 大模型评估维度产生**裁判模型评分费用**（按 Token 计费）； - 规则评估与人工评估无裁判模型费用； - - 推理结果集方式可规避被评测模型推理费用。 -- **成本优化建议**：优先用规则评估（零裁判模型费用）；先用 50–100 条数据小规模验证配置；保存并复用推理结果集避免重复推理。 -- **结果解读**：综合得分是各维度平均分，易掩盖维度间差异，应结合分数分布图与逐维度分析定位短板；1–3% 的分差通常属评测噪声，不宜作为决策依据。 + - 推理结果集方式不产生被评测模型推理费用。 +- **成本优化建议**： + - 先用 50–100 条数据小规模验证配置正确性； + - 优先选用规则评估（如 Function Calling 用字符串匹配、翻译用 BLEU）； + - 首次评测后下载推理结果集，后续复用以避免重复推理。 +- **API 支持**：当前模型评测功能**仅支持控制台操作，不提供公开 API/SDK**；如需自动化，可参考 PAI Judge Model API 替代方案。 ## 来源文档 -- [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - [模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) +- [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md index 2b78bd80..d8d8b607 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md @@ -1,58 +1,85 @@ # model experience -`model experience` 是百炼平台面向开发者提供的模型能力概览与使用指南，涵盖视觉理解、文本生成、多模态处理、语音/音频、3D生成及向量检索等核心AI能力。本文档聚焦于模型选型逻辑、关键参数约束、标准化调用方式及实际部署注意事项，所有信息均基于当前（2026年中）稳定可用的模型版本，不包含营销性描述或过时推荐。 +`model experience` 是百炼平台面向开发者提供的统一模型能力体验层，涵盖文本、图像、视频、语音、音乐、3D、多模态及向量/重排序等全栈AI模型服务。所有模型均通过标准化 API 接入，支持异步任务、流式响应、结构化输出与工具调用等核心能力，并按场景提供推荐选型路径。开发者可基于具体需求（如延迟敏感度、精度要求、成本约束）快速定位适配模型，无需关注底层基础设施。 -## 支持的模型/功能 +## 支持的模型与功能 -百炼平台提供覆盖全模态场景的模型体系，按能力域划分如下： +百炼平台当前提供覆盖多模态的模型矩阵，按能力域划分如下： -- **视觉理解**：支持图像OCR、视频理解、结构化输出及Function Calling。旗舰模型 `qwen3.7-plus` 支持1M上下文、2小时视频输入、2048张图片和64段视频；轻量模型 `qwen3.6-flash` 在保持相同上下文长度与功能集的前提下显著降低成本 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)。 -- **文本生成**：适用于AI编程、办公文档处理、长文本摘要等场景。`qwen3.7-plus` 和 `qwen3.6-flash` 均支持思考模式（`enable_thinking`）、Function Calling、内置工具（联网搜索、代码执行）及结构化JSON输出；超长文档处理推荐 `qwen-long`（10M上下文），但其不支持思考模式与内置工具 [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md)。 -- **图片与视频生成/编辑**：`wan2.7-image-pro` 支持4096×4096文生图与多图参考编辑；`happyhorse-1.1-t2v` 和 `wan2.7-t2v-2026-06-12` 均支持1080P有声视频生成，后者额外支持自定义音频文件注入 [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md)。 -- **语音与音乐**：S2S（语音转语音）模型如 `qwen3.5-omni-plus-realtime` 支持端到端音频理解与生成，兼具Function Calling与联网搜索能力；Fun-Music模型（`fun-music-v1`）支持[prompt](prompt.md)/lyrics双输入、性别选择及纯音乐生成，但仅限华北2（北京）地域 [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md)。 -- **3D与向量能力**：Tripo 3D模型（`Tripo/Tripo-P1.0`）需通过异步API调用，仅支持北京地域，且必须使用该地域API Key [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)；向量模型中，`text-embedding-v4` 为文本Embedding默认推荐，`qwen3-rerank` 支持最多500文档的纯文本重排序 [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。 +- **文本生成**：以 `qwen3.7-plus` 为旗舰，支持 100 万上下文、Function Calling、内置工具（联网搜索/代码解释器）、结构化 JSON 输出及逐步推理（`enable_thinking`）。轻量场景可选用 `qwen3.6-flash`，效果接近且成本更低 [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **图像生成与编辑**：`wan2.7-image-pro` 支持文生图（最高 4096×4096）、多图参考编辑、角色一致性生成；`qwen-image-2.0-pro` 支持负向提示词与单次最多 6 张变体；`z-image-turbo` 适用于低成本写实人像生成 [原文标题](../../raw/model-user-guide/model-experience/image-model.md)。 +- **视觉理解**：`qwen3.7-plus` 支持图像（最高 1600 万像素）、视频（最长 2 小时 / 2GB）、OCR 及结构化输出；专用 OCR 模型 `qwen3.5-ocr` 针对文档/手写内容优化 [原文标题](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **视频生成与编辑**：`happyhorse-1.1-t2v` 支持文生视频（1080P，3–15 秒），`wan2.7-i2v-2026-04-25` 支持首尾帧续写；`happyhorse-1.0-video-edit` 和 `wan2.7-videoedit` 分别覆盖基础编辑与特效/运镜复刻 [原文标题](../../raw/model-user-guide/model-experience/video-generate-edit-model.md)。 +- **语音合成（TTS）**：`qwen-audio-3.0-tts-plus` 支持指令控制（语速/情绪）；`cosyvoice-v3.5-plus` 同时支持声音复刻与声音设计；Qwen3-TTS 系列通过 `-realtime` 后缀区分 WebSocket 流式接入 [原文标题](../../raw/model-user-guide/model-experience/tts-model.md)。 +- **语音转语音（S2S）**：`qwen3.5-omni-plus-realtime` 提供端到端低延迟对话，支持音频语调感知；`qwen3.5-livetranslate-flash-realtime` 覆盖 60 种语言实时翻译（29 种输出语音） [原文标题](../../raw/model-user-guide/model-experience/s2s-model.md)。 +- **语音识别（ASR）**：`fun-asr`（非实时）支持说话人分离；`qwen3.5-omni-plus`（HTTP）支持 Prompt 注入领域上下文；`qwen3-asr-flash-realtime` 支持情感识别 [原文标题](../../raw/model-user-guide/model-experience/asr-model.md)。 +- **音乐生成**：`fun-music-v1` 支持 [prompt](prompt.md)/lyrics 输入、男声/女声选择及纯音乐模式（`is_instrumental=true`），仅限华北2（北京）地域 [原文标题](../../raw/model-user-guide/model-experience/fun-music.md)。 +- **3D 生成**：`Tripo/Tripo-P1.0`（快速预览）与 `Tripo/Tripo-H3.1`（影视级）支持文生3D、单图/多图生3D，需通过异步任务 API 调用 [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **向量与重排序**：`text-embedding-v4`（文本）与 `qwen3-vl-embedding`（多模态）支持跨模态检索；`qwen3-rerank` 用于 RAG 结果精排，支持 100+ 语言 [原文标题](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。 +- **全模态理解**：`qwen3.5-omni-plus` 统一处理文本/音频/图片/视频输入，支持 Function Calling、联网搜索及音视频分析；`qwen3-omni-flash` 为轻量替代方案，支持思考模式 [原文标题](../../raw/model-user-guide/model-experience/omni.md)。 -> **注意**：文档 2 中称 `qwen3.7-max` “不支持结构化输出”，但文档 1 明确列出 `qwen3.7-max-2026-06-08` 的结构化输出列为“不支持”，而 `qwen3.7-plus` 为“支持”。两者一致，无矛盾；但文档 2 表格中将 `qwen3.7-max` 的结构化输出标为“不支持”属正确表述，非错误。 +> **注意**：文档 3（视觉理解）与文档 11（全模态）均提及 `qwen3.5-omni-plus` 支持“联网搜索”，但文档 8（S2S）明确说明“Qwen3.5-Omni 实时（WebSocket）模式不支持此功能”，而文档 11 表格中 `qwen3.5-omni-plus-realtime` 的“联网搜索”列为“支持”。此处存在矛盾——实际能力以 API 文档为准：联网搜索仅在 HTTP 模式下可用，WebSocket 实时模式不可用。 ## 关键参数 -各模型共性关键参数如下（单位均为Token，除非特别注明）： - -| 参数 | 说明 | 典型值/范围 | 约束说明 | -|------|------|-------------|----------| -| `max_context` | 输入上下文长度上限 | `qwen3.7-plus`: 1M；`qwen-long`: 10M；`text-embedding-v4`: 8,192 | 超出将被截断，不报错 | -| `max_output_tokens` | 单次响应最大输出长度 | `qwen3.7-plus`: 64k；`qwen3-rerank`: 4,000/条 | 输出受模型能力与计费策略双重限制 | -| `max_image_count` / `max_video_count` | 单请求最大媒体数 | `qwen3.7-plus`: 2048图/64视频；`qwen3.5-omni-plus`: 256图/512视频 | 图像分辨率影响Token消耗：`h × w / (32 × 32) + 2` [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) | -| `texture_quality` / `geometry_quality` | Tripo 3D模型贴图与几何精度控制 | `standard` / `detailed`；`standard` / `ultra` | 仅 `Tripo/Tripo-H3.1` 支持 `geometry_quality` | -| `format` | 音频/视频输出格式 | `mp3` / `wav`；`720P` / `1080P` | `wav` 无损但体积大；视频输出帧率固定为24/30 fps | +各模型共性关键参数如下（具体值依模型而异）： + +- **`model`**：必需，指定模型 ID（如 `qwen3.7-plus`、`wan2.7-image-pro`）。 +- **`input`**：必需，结构因模型类型而异： + - 文本类：`{"messages": [...]}` 或 `{"prompt": "..."}`； + - 图像类：`{"prompt": "...", "image": "url"}` 或 `{"images": ["url1", "url2"]}`； + - 视频类：`{"prompt": "...", "video": "url"}`； + - 音频类：`{"audio": "url"}` 或 `{"prompt": "...", "audio": "url"}`； + - 3D 类：`{"prompt": "..."}`、`{"image": "url"}` 或 `{"images": [...]}`。 +- **`parameters`**：可选，控制生成行为： + - 文本：`temperature`（默认 0.8）、`top_p`（默认 0.8）、`max_tokens`； + - 图像：`texture_quality`（`standard`/`detailed`）、`size`（如 `"1024x1024"`）； + - 视频：`duration`（秒）、`fps`； + - TTS：`format`（`mp3`/`wav`）、`gender`（`male`/`female`）； + - 音乐：`is_instrumental`（`true`/`false`）、`format`； + - 3D：`texture_quality`、`geometry_quality`（仅 `Tripo-H3.1`）。 +- **`enable_thinking`**：布尔值，启用逐步推理（仅 Qwen3 及以上文本/全模态模型支持）。 +- **`X-DashScope-Async: enable`**：异步任务必需头（如 Tripo、Fun-Music），返回 `task_id` 后轮询结果。 ## 使用方式 -- **同步调用**：适用于文本生成、TTS、ASR、Embedding等低延迟场景。HTTP POST请求，`Content-Type: application/json`，模型ID置于`model`字段，输入数据置于`input`对象内（如`{"prompt": "..."}` 或 `{"audio_url": "..."}`）。 -- **异步调用**：适用于3D生成、长视频生成等耗时任务。首请求返回`task_id`，后续轮询 `GET /api/v1/tasks/{task_id}` 获取结果，状态流转为 `PENDING` → `RUNNING` → `SUCCEEDED`/`FAILED`，有效期24小时 [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 -- **流式调用**：WebSocket协议用于实时语音对话（`-realtime`后缀模型）、流式TTS/ASR。需维持长连接，服务端分块推送响应（如语音PCM片段或识别文本流）。 -- **多模态输入**：视觉/全模态模型接受混合输入。例如 `qwen3.5-omni-plus` 的`input`可同时含`text`、`audio_url`、`image_url`、`video_url`字段；Tripo模型则通过互斥字段`prompt`/`image`/`images`区分生成模式 [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **同步调用（HTTP）**：适用于低延迟要求场景（如聊天机器人），直接返回结果。示例： + ```bash + curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation' \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{"model":"qwen3.7-plus","input":{"messages":[{"role":"user","content":"你好"}]}}' + ``` +- **流式调用（WebSocket）**：适用于实时交互（如语音助手），需建立长连接并处理 `event: message` 流。模型名含 `-realtime` 后缀（如 `qwen3.5-omni-plus-realtime`）。 +- **异步调用（HTTP + 轮询）**：适用于耗时任务（如 3D 生成、长视频处理），先提交任务获 `task_id`，再 GET `/api/v1/tasks/{task_id}` 查询状态。Tripo 和 Fun-Music 必须使用此方式 [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **批量推理（HTTP）**：适用于高吞吐、低延迟容忍场景，通过 `/batch` 接口提交多请求，降低单位成本 [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 ## 限制和注意事项 -- **地域限制**：Tripo 3D模型、Fun-Music、部分S2S/ASR模型（如`qwen3.5-livetranslate-flash`）**仅支持华北2（北京）地域**，且必须使用该地域API Key与Endpoint [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 -- **功能互斥**：Qwen3.5-Omni系列在启用联网搜索时**不可同时启用Function Calling**；思考模式下**不支持生成语音输出**（仅文本） [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md)。 -- **旧版模型弃用**：Qwen2.5-VL、Qwen-Omni、Qwen-VL等旧系列模型已明确标注“不再作为首选推荐”，新项目应使用Qwen3.6或Qwen3.5系列 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)。 -- **音频规格硬约束**：Fun-ASR非实时模型支持最大12小时/2GB音频；Qwen3.5-Omni非实时模型限3小时/2GB；而Qwen3-omni-flash HTTP模式仅支持20分钟/100MB [语音识别](../../raw/model-user-guide/model-experience/asr-model.md)。 -- **语言覆盖差异**：Qwen3.5-Livetranslate支持60种语言（29种输出语音+文本），但Qwen3-Omni-Flash仅支持11种输出语言；方言支持因模型版本而异（如`fun-asr-realtime`支持数十种中文方言，而`paraformer-8k-v2`仅支持普通话） [全模态](../../raw/model-user-guide/model-experience/omni.md)。 +- **地域限制**：Tripo 3D 模型仅支持华北2（北京）；Fun-Music 仅限华北2（北京）且需邀测开通；部分模型（如 `wanx2.1-imageedit`）明确标注“仅支持北京地域” [原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)。 +- **输入约束**： + - 图像：单图最高 1600 万像素，[Token](../concepts/token.md) 消耗公式为 `h × w / (32 × 32) + 2`； + - 视频：`qwen3.7-plus` 最长 2 小时 / 2GB，`qwen3-vl-plus` 最长 1 小时 / 2GB； + - 音频：ASR 文件最大 12 小时 / 2GB，S2S 实时流无时长限制但单次输入建议 ≤2 小时。 +- **功能兼容性**： + - 思考模式（`enable_thinking`）与语音输出互斥：启用思考模式时，S2S/Qwen-Audio 模型不生成语音 [原文标题](../../raw/model-user-guide/model-experience/s2s-model.md)； + - Function Calling 与联网搜索不可同时开启（Qwen3.5-Omni HTTP 模式）； + - `qwen-long`（1000 万上下文）不支持 Function Calling、内置工具或思考模式。 +- **版本管理**：推荐使用快照版本（如 `qwen3.7-plus-2026-05-26`）保障稳定性；`-latest` 或无后缀版本可能随平台升级变更行为。 +- **旧版模型**：Qwen3、Qwen2.5 等系列已归为“旧版”，新项目应优先选用 Qwen3.6/Qwen3.7 或 Qwen3.5-Omni 系列 [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 ## 来源文档 -- [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) - [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md) - [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md) +- [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) - [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) -- [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [语音合成](../../raw/model-user-guide/model-experience/tts-model.md) +- [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) - [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) - [语音识别](../../raw/model-user-guide/model-experience/asr-model.md) -- [全模态](../../raw/model-user-guide/model-experience/omni.md) - [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) +- [全模态](../../raw/model-user-guide/model-experience/omni.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md index 252a68d3..9368b46e 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md @@ -1,41 +1,47 @@ # model high speed inference -百炼平台提供两种面向高吞吐与低延迟场景的推理加速能力：TPM 预留（保障专属容量）和快速模式（提升单请求输出速度）。二者定位不同，可独立使用或组合使用——TPM 预留解决“能不能稳定跑满”的问题，快速模式解决“单次响应够不够快”的问题。开发者应根据业务对容量确定性（如 SLA 要求）与响应时延（如 TPS/首 token 延迟）的优先级进行选型。 +百炼平台提供两类面向高吞吐、低延迟场景的推理加速能力：**快速模式（Fast mode）** 与 **TPM 预留（TPM Reservation）**。二者目标一致——提升服务稳定性与响应速度，但技术路径不同：前者通过模型级优化实现更高 TPS，后者通过资源独占保障确定性吞吐。开发者需根据业务对延迟敏感度、流量可预测性及成本结构选择合适方案。 ## 支持的模型/功能 -- **TPM 预留**：为指定模型锁定专属输入/输出吞吐量（单位：kTPM），确保高峰期不被公共资源限流影响。支持千问、GLM、DeepSeek、Kimi 等多个主流模型，具体列表见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档中的“支持的模型”表格。 -- **快速模式（Fast mode）**：Preview 阶段能力，通过优化推理调度与内存访问，提升单请求输出吞吐（TPS 达 80~100），适用于 AI 编程助手、Agent 多步推理等对首 token 和 token 流速敏感的场景。当前仅支持 `glm-5.2-fast-preview` 模型（[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 文档明确列出），其他模型暂未开放。 - -> **注意**：两篇文档对“模型支持范围”的描述存在明显差异——TPM 预留文档列出了十余个模型（如 `qwen3.7-max-2026-05-20`、`deepseek-v4-pro` 等），而快速模式文档仅声明 `glm-5.2-fast-preview` 可用。目前无证据表明其他模型已支持快速模式，因此以 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 的明确声明为准，不可自行尝试在非 listed 模型上添加 `-fast-preview` 后缀。 +- **快速模式**：当前仅支持 `glm-5.2-fast-preview` 模型（北京、新加坡地域），为预览阶段能力，模型 ID 即启用标识，无需额外参数 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)。 +- **TPM 预留**：支持多款主流模型，包括 `GLM-5.2`、`GLM-5.1`、`千问3.7-Max-2026-05-20`、`DeepSeek-v4-Pro`、`Kimi-K2.6` 等（具体以控制台实时列表为准），需创建后获取专属模型 code 才能调用 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 +- > **注意**：文档 1 中 `glm-5.2-fast-preview` 标注为 preview 阶段，而文档 2 中 `GLM-5.2`（无 `-fast-preview` 后缀）列为 TPM 预留支持模型。二者非同一模型变体：前者是专有高速推理版本，后者是标准模型的容量保障通道。不可混用 model ID。 ## 关键参数 | 能力类型 | 核心参数 | 说明 | |----------|----------|------| -| TPM 预留 | `input_tpm` / `output_tpm` | 单位为 kTPM（1 kTPM = 1,000 tokens/min），需按模型实际阶梯系数与缓存折扣估算，详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中的“容量计算器”与“长输入阶梯系数”表格。 | -| 快速模式 | 无显式参数 | 仅需将 `model` 设为 `glm-5.2-fast-preview`，并使用专属接入域名（如 `{workspace_id}.cn-beijing.maas.aliyuncs.com`），无需额外 query 或 header。 | +| 快速模式 | `model="glm-5.2-fast-preview"` | 唯一启用标识；接入域名固定为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（北京）或对应新加坡地域域名 | +| TPM 预留 | `model=""` | 创建后生成的唯一字符串；必须替换原 model ID；接入域名与标准 API 一致（如 `https://dashscope.aliyuncs.com/compatible-mode/v1`） | +| 共同参数 | `stream=true/false` | [流式输出](../concepts/streaming-output.md)时，快速模式返回 `delta.reasoning_content` 和 `delta.content` 字段；TPM 预留行为与标准模型一致 | ## 使用方式 -- **TPM 预留**：创建成功后，系统生成专属模型 code（如 `tpm-qwen37max-xxx`），**必须**在 API 请求中将 `model` 参数替换为此 code 才能生效。标准调用方式不变，但需注意预热期可能引入短暂延迟波动（见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) “创建 TPM 预留”章节示例代码注释）。 -- **快速模式**：直接使用 `model="glm-5.2-fast-preview"` 发起请求，并确保 base_url 指向对应地域的 MaaS 域名（如华北2为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。流式响应中需分别处理 `delta.reasoning_content` 和 `delta.content` 字段（见 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 的“使用示例”）。 +- **快速模式**：直接在请求中指定 `model: "glm-5.2-fast-preview"`，其余参数（如 `messages`, `stream`）与标准 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)一致。流式响应需分别处理 `reasoning_content` 与 `content` 字段 [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)。 +- **TPM 预留**：创建成功后，在控制台详情页复制专属模型 code，替换 API 请求中的 `model` 参数即可生效。首次调用存在短暂预热期，建议客户端实现重试或排队机制 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 +- **共用要求**：均需有效 `API_KEY` 与正确 `base_url`；快速模式强制使用 workspace 绑定域名，TPM 预留使用全局兼容域名。 ## 限制和注意事项 -- **TPM 预留**： - - 预留实例到期后 2 小时内仍可调用，2~14 小时内停止但可续费，14 小时后彻底删除且不可恢复； - - 缩容退订按 1.5 倍系数结算已用费用，公式见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) “计费与使用说明”； - - 超额请求自动降级至按量计费，不中断服务，但需监控“超额降级统计”避免成本失控。 +- **快速模式限制**： + - 仅限 `glm-5.2-fast-preview` 模型，不支持其他模型； + - 处于 preview 阶段，接口行为、计费策略或模型能力可能调整； + - 超出 TPM 额度时请求进入排队队列，而非立即拒绝。 + +- **TPM 预留限制**： + - 预留容量按 kTPM（千 tokens/分钟）购买，输入/输出 TPM 分开配置； + - 缩容退费按公式 `退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)` 计算； + - 实例到期后 2 小时内仍可调用，14 小时后彻底删除且不可恢复。 -- **快速模式**： - - 当前为 preview 阶段，接口行为、模型能力及计费规则可能调整，不建议用于生产环境 SLA 保障场景； - - 超出 TPM 额度时请求进入排队队列而非立即限流，可能导致端到端延迟升高，需评估业务容忍度； - - `glm-5.2-fast-preview` 返回结构含 `reasoning_content` 字段，与标准 `glm-5.2` 不兼容，客户端需适配解析逻辑。 +- **通用注意事项**： + - 快速模式与 TPM 预留**不可叠加使用**：`glm-5.2-fast-preview` 不支持 TPM 预留，TPM 预留仅作用于标准模型（如 `GLM-5.2`）； + - 缓存折扣仅影响输入容量计算（如 `glm-5.2` 缓存命中部分按 25% 折算），不影响输出； + - 超额处理逻辑不同：快速模式排队，TPM 预留自动降级至按量计费 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 ## 来源文档 -- [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - [快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) +- [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md index 17d81124..50b09f97 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md @@ -1,77 +1,61 @@ # model monitoring -模型监控是百炼平台提供的核心可观测性能力，用于实时跟踪模型调用行为、性能表现、成本消耗与异常事件。它覆盖从基础调用统计到细粒度 Token 追踪、从控制台可视化到 Prometheus 自建集成的全链路监控能力，适用于生产环境下的稳定性保障与成本精细化治理。所有监控数据默认按「模型 + 业务空间」维度聚合，主账号可跨空间查看，子账号仅限当前业务空间。 +模型监控是百炼平台提供的核心可观测性能力，用于实时跟踪模型调用行为、性能指标、成本消耗及异常事件。它面向生产环境提供细粒度的调用统计、多维指标监控、日志审计与主动告警能力，帮助开发者快速定位问题、优化成本并保障服务稳定性。该功能以业务空间为数据边界，默认按小时级聚合，高级监控支持分钟级洞察与 Prometheus 标准对接。 ## 支持的模型与功能 -- **监控覆盖范围**： - - **普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的全部模型（含基于其调优的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)）； - - **高级监控**（含分钟级指标、告警、Prometheus 接入）仅支持北京、新加坡、弗吉尼亚地域下的模型； - - **告警功能**仅支持北京、新加坡地域（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 +- **监控覆盖范围**：普通监控支持[所有公开模型及调优后的自定义模型](../../raw/model-user-guide/model-monitoring/model-telemetry.md)，包括大语言模型（如 `qwen-plus`、`qwen3-max`）、视觉模型、语音模型、全模态模型和向量模型；高级监控与告警功能当前仅限北京、新加坡、弗吉尼亚地域的模型（详见 [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 +- **核心功能**： + - **调用追踪**：记录请求/响应（限北京地域部分模型，见[支持请求和响应的模型](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）； + - **指标监控**：RPM、TPM、调用时长、首[Token](../concepts/token.md)延时、失败率、限流错误次数（429）、内容安全错误次数等； + - **[Token](../concepts/token.md) 消耗分析**：按业务空间维度汇总与单次调用级追踪（输入/输出/缓存/图像/音频/视频等细分用量类型）； + - **主动告警**：支持对成本突增、失败率飙升、延迟超阈值等场景配置多级通知（短信/邮件/钉钉/企业微信/Webhook）； + - **Grafana 与自建集成**：通过私有 Prometheus HTTP API 开放全部监控指标，支持标准 PromQL 查询（如 `model_usage{model="qwen-plus", workspace_id="..."}`）。 -- **核心功能模块**： - - **调用统计**：调用次数、失败次数、失败率、限流错误（429）、内容安全拦截次数； - - **性能指标**：RPM、TPM、调用时长、首Token延时、非首Token延时； - - **成本监控**：Token 消耗汇总与单次追踪（仅北京地域部分模型支持）； - - **日志审计**：输入/输出对话记录（仅北京地域且限于[指定模型列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）； - - **用量统计**：按业务空间维度的模型用量（含免费额度使用情况），延迟约 1 小时（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 - -> **注意**：文档 1 称“普通监控延迟通常为小时级”，而文档 2 明确用量统计延迟“约为 1 小时”；二者一致。但文档 1 中“新模型在首次数据同步完成后自动加入列表”未说明是否含自定义模型——文档 2 明确“调优后的模型”同样支持用量查看，故可推断其也纳入普通监控范围，无额外限制。 +> **注意**：文档1中“模型用量”页面的数据延迟为“约1小时”，而文档2明确区分普通监控（小时级）与高级监控（分钟级）。二者不矛盾，但需注意：**普通监控无法满足实时诊断需求，分钟级洞察必须启用[高级监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)**。 ## 关键参数与指标 -| 类别 | 指标名 | 说明 | 支持过滤 Label | -|--------|---------|------|----------------| -| 调用次数 | `model_call_count` | 总调用次数 | `user_id`, `apikey_id`, `workspace_id`, `model`, `protocol`, `sub_protocol`, `status_code`, `error_code` | -| 调用时长 | `model_call_duration`, `model_call_duration_p99` | 均值/P99 时长（秒） | 同上 | -| 首Token延时 | `model_first_token_duration` | 首包响应时间 | 同上 | -| 非首Token延时 | `model_generation_duration_per_token` | 每 Token 生成耗时 | 同上 | -| Token用量 | `model_usage` | 总 Token 数（支持 `usage_type` 过滤：`input_tokens`/`output_tokens`/`total_tokens` 等） | `usage_type`, `workspace_id`, `model`, `apikey_id` | +| 类别 | 指标名（Prometheus） | 说明 | 过滤标签（LabelKey）示例 | +|--------|----------------------|------|---------------------------| +| 调用统计 | `model_call_count` | 调用总次数 | `apikey_id`, `status_code`, `error_code` | +| 性能 | `model_call_duration_p99` | 调用时长P99 | `workspace_id`, `model`, `protocol` | +| 首[Token](../concepts/token.md)延时 | `model_first_token_duration` | 首包平均耗时 | `sub_protocol`（DEFAULT/ASYNC） | +| 用量 | `model_usage` | Token/图像张数/视频秒数等总和 | `usage_type`（`input_tokens`, `image_count`, `video_seconds` 等） | -所有指标均通过 Prometheus HTTP API 提供，需开启[高级监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)并配置 AccessKey 认证（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 +- `usage_type` 是关键过滤维度，取值包括 `total_tokens`、`input_tokens`、`output_tokens`、`cache_tokens`、`image_count`、`audio_count`、`video_count`、`duration`、`characters` 等，需结合模型类型选择（参见[模型用量统计单位说明](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 +- 所有指标均支持按 `workspace_id`、`model`、`apikey_id` 维度下钻，**不支持跨业务空间聚合或阿里云账号维度统计**（该限制在两篇文档中一致）。 ## 使用方式 -1. **控制台访问**： - - 普通监控入口：[模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)（北京）或对应地域控制台； - - 用量统计入口：[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)（仅业务空间维度）； - - 免费额度管理：[免费额度](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/free-quota)。 +1. **基础监控查看** + 进入控制台 [模型监控](https://bailian.console.aliyun.com/?tab=model#/model-telemetry) 页面 → 选择业务空间 → 查看「监控数据看板」及「模型监控」表格 → 点击目标模型操作列的 **监控** 或 **日志**。 -2. **日志与单次 Token 查看**（仅北京地域）： - - 需先在「模型监控配置」中开通**审计日志 + 推理日志**； - - 开通后，在模型列表点击「日志」页签，查看请求/响应及 `用量` 字段（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 +2. **启用高级能力（必选步骤）** + - 在模型监控页面右上角点击 **模型监控配置** → 开启 **性能和用量指标监控**（高级监控）； + - 如需日志审计（输入/输出详情），需额外开通 **审计日志** 和 **推理日志**（仅北京地域生效，且仅限[指定模型列表](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -3. **告警配置**： - - 仅北京、新加坡地域支持； - - 需先开启高级监控 → 进入[模型告警](https://bailian.console.aliyun.com/?tab=model#/model-alert) → 创建规则（支持短信/邮件/钉钉/Webhook 等通知方式）。 +3. **创建告警规则** + 进入 [模型告警](https://bailian.console.aliyun.com/?tab=model#/model-alert) 页面 → 点击 **创建告警规则** → 选择模型、模板、阈值与通知渠道（注意：告警功能仅在北京、新加坡地域可用）。 -4. **Grafana / 自建应用接入**： - - 获取 Prometheus HTTP API 地址（需开启高级监控）； - - 使用 `Basic` 认证（AccessKey:AccessKeySecret Base64 编码）调用 `/api/v1/query_range`； - - 示例：`GET {API}/api/v1/query_range?query=model_usage{workspace_id="xxx",model="qwen-plus"}&start=...&step=60s`（详见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 +4. **接入 Grafana / 自建系统** + - 获取 Prometheus HTTP API 地址（通过「模型监控配置」→「云监控Prometheus实例」→「查看详情」）； + - 使用 `Authorization: Basic base64Encode(AccessKey:AccessKeySecret)` 认证； + - 构造标准 PromQL 查询，例如： + `GET {API}/api/v1/query_range?query=model_usage{model="qwen-plus",usage_type="input_tokens"}&start=...&end=...&step=60s` ## 限制和注意事项 -- **地域限制严格**： - - 日志审计、单次 Token 追踪、告警、Prometheus 接入等功能**仅限北京、新加坡、弗吉尼亚地域**；其他地域仅提供小时级普通监控（如调用总量、失败率等基础卡片）。 - -- **模型兼容性差异**： - - 并非所有模型均支持全部监控能力。例如，历史对话（输入/输出日志）仅支持文档 1 列出的千问系列、开源及三方模型快照版本（如 `qwen3-max-2025-09-23`），旧版快照或未列型号不支持（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 - -- **数据延迟与范围**： - - 普通监控数据延迟约 **1 小时**；高级监控支持分钟级洞察； - - 控制台用量页面**最多查看最近 30 天数据**，更早数据需通过[费用与成本](https://billing-cost.console.aliyun.com/finance/expense-report/expense-detail-by-instance)查询（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 - -- **权限约束**： - - 主账号及具备足够权限的子账号可开通日志与高级监控； - - 子业务空间成员**无法切换查看其他业务空间数据**，仅限当前空间（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 - -- **用量单位差异**： - - 大语言模型按 **Token** 计费；视觉模型按 **张**（图像）、**秒**（视频）；语音模型按 **秒/字符/Token**（依模型而定）；全模态模型文本部分按 Token，其他模态按对应 Token 数（见 [模型用量 (raw/model-user-guide/model-monitoring/model-usage-statistics.md)](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)）。 +- **地域限制**：日志审计（请求/响应内容）、分钟级监控、告警功能仅支持 **华北2（北京）** 和 **新加坡** 地域；弗吉尼亚仅支持分钟级监控与 Prometheus 对接，**不支持告警与日志审计**。 +- **数据时效性**：普通监控数据延迟约1小时；高级监控数据延迟为分钟级（通常 ≤5 分钟），但日志从调用发生到可查存在分钟级延迟，需手动刷新。 +- **用量查询范围**：控制台内模型用量与监控页的 Token 消耗均**仅支持查询最近30天数据**；更早数据需通过[费用与成本](https://billing-cost.console.aliyun.com/finance/expense-report/expense-detail-by-instance)页面导出账单获取。 +- **权限隔离**：子业务空间成员**仅能查看本空间数据**，无法切换或跨空间筛选；主账号可查看全部空间。 +- **Token 计费口径差异**：不同模型类型计费单位不同（Token/张/秒/字符），务必参考[模型用量统计单位说明](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)匹配 `usage_type` 标签，避免误读指标。 +- **免费额度联动**：“免费额度用完即停”开关仅影响计费行为，**不影响监控数据采集**；即使额度耗尽，监控仍持续记录调用与失败（如返回 403 错误），可用于故障归因。 ## 来源文档 -- [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md) - [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md) +- [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md index 68323539..794f9fcf 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md @@ -1,58 +1,57 @@ # plug in -插件是百炼平台用于扩展大模型能力的核心机制，通过将外部工具（API）集成到模型推理链路中，解决大模型在实时信息获取、精确计算、代码执行、图像生成等场景下的固有局限。开发者可选用官方插件、三方插件或自定义插件，结合智能体应用、工作流应用或 Assistant API 进行调用。所有插件均需通过服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI` 授权方可使用。 +[插件](../concepts/plugin.md)是百炼平台用于扩展大模型能力的核心机制，通过将外部工具（API）集成到推理链路中，弥补大模型在实时信息获取、精确计算、代码执行、图像生成等方面的固有局限。[插件](../concepts/plugin.md)以“工具”为最小可调用单元，支持官方预置、三方市场及完全自定义三种来源，由大模型根据用户输入自主规划调用，或在工作流中显式编排执行。 ## 支持的模型/功能 -百炼当前支持以下模型调用插件能力： -- `qwen-turbo`、`qwen-plus`、`qwen-max`（文本模型） -- `qwen-vl-plus`、`qwen-vl-max`（多模态模型） +当前[插件](../concepts/plugin.md)能力仅对部分模型开放，**必须使用以下模型标识符之一**才能启用插件调用： -> **注意**：各模型对插件的兼容性存在差异，[插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md) 中列出的模型列表为截至文档发布时的兼容范围，**实际可用性请以控制台运行结果为准**；部分新模型（如 `qwen2.5` 系列）尚未明确列入该文档，需通过控制台实测验证。 +- `qwen-turbo`（通义千问-Turbo） +- `qwen-plus`（通义千问-Plus） +- `qwen-max`（通义千问-Max） +- `qwen-vl-max`（通义千问VL-Max） +- `qwen-vl-plus`（通义千问VL-Plus） -插件按来源分为三类： -- **官方插件**：预置于组件广场，开箱即用，无需配置参数。包括 `code_interpreter`（Python 执行）、`calculator`（数学计算）、`text_to_image`（文生图）、`quark_search`（实时搜索）、`generate_qrcode`（二维码生成）、`github_search`（GitHub 项目检索）等 [详见官方插件说明](../../raw/application-user-guide/plug-in/plugins.md)。 -- **三方插件**：来自阿里云云市场，覆盖商业服务、图像视频、教育等领域，开通后即可调用。 -- **自定义插件**：支持开发者接入自有 API，需定义插件 URL、工具路径、输入/输出参数及鉴权方式，完整流程见 [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md) 文档。 +> **注意**：文档 1 中称“各模型对插件的兼容性可能有差异”，但未明确列出不支持的模型；而文档 2 和 3 均未重复说明兼容性范围。实际开发中请以控制台运行结果为准，建议优先选用 `qwen-plus` 或 `qwen-max` 进行插件集成验证。该兼容性说明详见 [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 + +插件按来源分为三类，功能边界如下： + +- **官方插件**：开箱即用，无需配置参数。包括 `code_interpreter`（Python 执行）、`calculator`（数学计算）、`text_to_image`（文生图）、`quark_search`（实时搜索）、`generate_qrcode`（二维码生成）、`github_search`（GitHub 项目检索）等。详细功能与限制见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 +- **三方插件**：来自阿里云云市场，覆盖商业服务、教育、图像视频等领域，需开通后使用，同样免配置。 +- **自定义插件**：支持通过控制台创建或从云市场导入，可对接任意 HTTP API。需明确定义工具路径、鉴权方式、输入/输出参数结构，并完成调试与发布。完整流程参见 [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 ## 关键参数 -插件调用依赖以下核心参数，尤其在自定义插件和 API 集成中必须准确配置： +插件调用依赖以下核心参数，尤其在 API 集成场景中必须准确传递： -- **工具 ID（tool_id）**：唯一标识插件下的具体工具（如 `calculator`），用于 Assistant API 或工作流节点中指定调用目标。可通过插件详情页悬浮图标复制获取。 -- **插件 URL 与工具路径**：插件 URL 为域名根地址（如 `https://myapi.example.com`），工具路径为相对路径（如 `/query`），二者拼接构成完整调用地址。 -- **输入参数（input parameters）**： - - `传参方式` 必须明确设为 `大模型识别`（从用户输入提取）或 `业务透传`（由外部传入，通过 `biz_params` 或 `user_defined_params` 传递）； - - `参数名称` 和 `参数描述` 需语义清晰，直接影响大模型参数提取准确性； - - `类型` 支持 `String`、`Number`、`Object`（但 Object 子属性不可为空）。 -- **输出参数（output parameters）**：定义 API 返回数据中哪些字段被大模型用于生成最终回复，需精简且层级扁平。 -- **鉴权配置**：若 API 需鉴权，支持 `Header`（如 `Authorization: Bearer `）或 `Query`（如 `?api_key=xxx`）方式，`Type` 可选 `basic`/`bearer`/`appcode`。 +- **工具 ID（tool_id）**：唯一标识一个工具，如 `calculator`、`quark_search`。可在插件详情页的“插件工具”区域直接复制，[获取工具ID](../../raw/application-user-guide/plug-in/plugins.md) 有详细指引。 +- **输入参数（input parameters）**： + - `传参方式` 必须明确设为 `大模型识别`（由 LLM 从用户 query 中抽取）或 `业务透传`（由上游系统主动注入，通过 `biz_params` 传递）； + - 参数类型（String/Number/Object）及嵌套结构需严格匹配 API 接口契约，Object 类型子属性**不能为空**（见文档 3 错误码 130022）； + - 鉴权参数（如 `api_key`）若置于 Query，需在插件配置中指定 `参数名`；若置于 Header，则 `Type`（如 `bearer`）决定前缀格式。 +- **输出参数（output parameters）**：所有字段均为必填，描述需精简准确，便于大模型从 API 响应中提取关键字段并组织最终回复。 ## 使用方式 -插件可通过三种方式集成： - -1. **控制台可视化配置（推荐入门）**： - - 在 [插件市场](https://bailian.console.aliyun.com/#/plugin-market) 页面授权 `AliyunServiceRoleForSFMAccessCloudAPI` 角色（主账号直接授权；RAM 用户需先获 `ram:CreateServiceLinkedRole` 权限）； - - 官方/三方插件：单击“添加至智能体”，选择目标智能体应用（注意：官方插件仅支持与**同业务空间**的智能体关联）； - - 自定义插件：创建后需先发布为 MCP 服务，再在智能体编排页的 **MCP 区块** 中添加 [参考自定义插件文档](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 +插件可通过三种方式接入应用： -2. **工作流应用节点**：将插件作为独立节点拖入工作流画布，按需编排执行顺序，不依赖大模型自主决策。 +1. **智能体应用（Agent）**：在控制台应用编排页 → “MCP” 区块 → 添加已发布的插件（或其转换的 MCP 服务）。官方插件仅支持与**同业务空间**内的智能体关联；自定义插件需先发布为 MCP 服务。详见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) 的“调用插件”章节。 +2. **工作流应用（Workflow）**：将插件作为独立节点拖入画布，显式编排执行顺序，不依赖大模型自动决策。 +3. **Assistant API**：在请求 payload 的 `tools` 字段中声明工具列表（含 `type`、`function` 及 `function.name`），并在 `tool_choice` 中控制调用策略。具体格式参考 [Assistant API 文档](https://help.aliyun.com/zh/model-studio/quick-start-of-assistant-api) 中 `tools` 关键字说明。 -3. **API 调用**： - - Assistant API：在 `tools` 数组中声明工具 ID 及描述，模型自动规划调用； - - 智能体/工作流 API：通过 `biz_params` 传递业务透传参数或用户级鉴权 Token [详见 API 文档](https://help.aliyun.com/zh/model-studio/agent-and-workflow-application-api-reference)。 +> **注意**：首次使用插件前，主账号或 RAM 子账号**必须授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`**，否则无法访问插件市场或调用云市场 API。RAM 用户需额外授予 `ram:CreateServiceLinkedRole` 权限，操作细节见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) 和 [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 ## 限制和注意事项 -- **权限限制**：首次使用插件前，**必须完成 `AliyunServiceRoleForSFMAccessCloudAPI` 服务关联角色授权**，否则无法访问插件市场或调用任何插件 [详见官方和第三方插件文档](../../raw/application-user-guide/plug-in/plugins.md)。 -- **调用上限**：智能体应用最多支持添加 **10 个工具**；自定义插件中，`Object` 类型输入参数在 `GET` 请求下不被支持（仅 `POST` 允许）。 -- **功能边界**： - - `code_interpreter` 插件**不支持网络访问与本地文件上传**，可用依赖库已固化（如 `pandas`、`matplotlib`、`requests` 等）； - - `quark_search` 和 `github_search` 均**仅返回摘要、标题、链接，不支持访问网页或仓库详情页**； - - `text_to_image` 和 `quark_search` 为**限时免费，需单独申请开通**。 -- **调试要求**：自定义插件的工具必须经 **在线调试成功并发布为“已发布”状态** 后才能被应用调用；草稿或未启用状态的工具将导致调用失败。 -- **错误处理**：发布自定义工具时常见错误码 `130040`（参数描述缺失）、`130022`（Object 子属性为空或 GET 请求含 Object 参数）需严格按提示修正 [详见自定义插件错误码说明](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 +- **调用上限**：单次对话最多调用 10 个工具（含同一插件下的多个工具），且工具总调用次数受应用配额约束。 +- **安全限制**： + - `code_interpreter` 插件**禁止网络访问**（`requests` 等库不可用）及**本地文件上传**，仅支持内置依赖（如 `pandas`, `matplotlib`, `sympy`）； + - `quark_search` 和 `github_search` 仅返回网页标题、关键词/摘要、项目链接等元信息，**不支持抓取网页正文或项目源码**（见文档 1 和 2 的明确说明）。 +- **自定义插件部署要求**： + - 插件 URL 必须为 HTTPS 协议，且响应头需包含 `Access-Control-Allow-Origin: *` 或明确允许百炼域名； + - 工具路径必须以 `/` 开头，且拼接后构成合法 URL（如插件 URL `https://example.com` + 工具路径 `/query` → `https://example.com/query`）； + - 发布前必须通过在线调试验证连通性，未发布状态的工具无法被应用调用。 +- **权限与生命周期**：删除插件将**级联删除其下所有工具**，且已关联该插件的应用立即失效；编辑插件 URL 或鉴权配置后，必须重新测试并发布所有相关工具。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md index c5f97c2d..a669f10c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md @@ -1,57 +1,58 @@ # prompt -Prompt 是百炼平台中用于引导大语言模型生成预期输出的核心指令载体。通过结构化设计、模板化管理、样例增强与自动优化等能力，开发者可系统性提升模型输出的准确性、一致性与可控性。所有 Prompt 相关功能均需在华北2（北京）地域使用，且依赖业务空间（Workspace）上下文。 +Prompt 是百炼平台中驱动大语言模型行为的核心指令载体。通过结构化模板、自动优化、样例引导等多种机制，开发者可高效构建、复用和迭代高质量提示词，显著提升模型输出的准确性、一致性与可控性。所有 Prompt 相关能力均默认适用于华北2（北京）地域，跨地域使用需另行确认支持状态。 ## 支持的模型/功能 -百炼平台提供多种 Prompt 相关能力，覆盖从基础指令构造到高级场景适配的全链路： +百炼平台提供三类 Prompt 增强能力，面向不同开发阶段和精度要求： -- **Prompt 模板**：支持预置模板（如营销文案生成、摘要抽取）和自定义模板（文本生成、图片生成），后者可通过控制台或 API 创建，并支持 ICIO、CRISPE、RASCEF 等工程框架辅助构建 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 -- **Prompt 样例库**：通过少样本学习注入高质量问答对，引导模型输出风格与格式一致的结果；但该功能已停止维护，官方明确建议迁移到 RAG 表格库 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 -- **Prompt 自动优化**：基于大模型对原始 Prompt 进行结构重组、角色设定、指令增强与安全边界注入，不计费且数据不用于训练 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 -- **Prompt 反馈优化**：基于用户提供的输入-输出样例（query-answer pairs）进行多轮评估与迭代优化，推荐使用 `qwen-max` 作为推理模型，效果优于纯文本自动优化 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 +- **Prompt 模板**：支持预置模板（如营销文案生成、摘要抽取）和自定义模板（文本生成、图片生成），实现逻辑与内容分离。预置模板效果稳定、开箱即用；自定义模板支持基于 ICIO、CRISPE、RASCEF 等工程框架结构化构建，适用于金融风控、医疗咨询等强约束场景 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 +- **Prompt 自动优化**：基于大模型对原始 Prompt 进行结构重组、角色注入、指令增强和安全边界补充，不计费且数据不用于训练 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 +- **Prompt 反馈优化**：利用用户提供的输入-输出样例（建议 5–10 条）和评测数据集（建议 ≥20 条），在推理模型（推荐千问-max）上多轮评估、反思并生成带 few-shot 示例的优化 Prompt，适配真实业务场景 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 -> **注意**：文档 2 明确声明“Prompt样例库功能已不再维护”，而文档 1 和 3 中仍存在大量关于其创建、关联与调试的操作说明。实际开发中应以文档 2 的迁移指引为准，避免依赖已下线能力。 +> **注意**：Prompt 样例库功能已下线，官方明确要求迁移至 RAG 表格库，不再维护 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 ## 关键参数 | 参数 | 说明 | 来源/约束 | |------|------|-----------| -| `workspaceId` | 业务空间唯一标识，所有 Prompt 操作（模板获取、样例库关联、反馈优化）均需指定 | 必填，通过[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)获取 | -| `promptTemplateId` | 模板唯一 ID，用于 `GetPromptTemplate` 接口拉取内容 | 预置模板 ID 在控制台卡片中可见；自定义模板 ID 创建后生成 | -| `has_thoughts=true` | API 调用时启用样例检索过程日志（仅限已关联样例库的应用） | 仅影响响应中 `thoughts` 字段，非必需 | -| 召回片段数 | 单次请求注入上下文的样例数量，默认 5，最大 10 | 应用配置页可调，影响 Token 消耗与效果平衡 | -| 评测数据量 | Prompt 反馈优化中用于评估的 query-answer 对数量，建议 ≥20 条 | 数据越充分，优化效果越稳定 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) | +| `promptTemplateId` | 模板唯一标识符，用于 API 调用（如 `GetPromptTemplate`） | 控制台模板卡片或 API 响应中获取 | +| `workspaceId` | 业务空间 ID，调用所有 Prompt 相关 API 的必需参数 | 需通过 [获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id) 获取 | +| `variables` | 模板变量列表（如 `["platform", "topic"]`），用于运行时填充 | `GetPromptTemplate` 接口响应中返回 | +| `has_thoughts` | API 请求参数，设为 `true` 时返回样例检索详情（仅限历史样例库调试） | 已废弃，仅用于兼容旧调试流程 | +| 召回片段数 | 单次请求注入上下文的样例数量（默认 5，上限 10） | 仅适用于已停用的样例库功能 | ## 使用方式 ### 控制台操作 -- **模板创建与管理**：进入「组件管理 > 提示词」页面，支持自定义创建或基于 Prompt 工程框架（如 ICIO）生成；图片生成模板需分别填写正向/负向 Prompt [原文标题](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 -- **自动优化**：在「提示词 > 自动优化」页面粘贴原始 Prompt，点击「优化」后可复制结果或「保存为模板」。 -- **反馈优化**：在「提示词 > 反馈优化」页面配置初始 Prompt、样例集（5–10 条）、评测集（≥20 条），启动多轮优化任务。 +- **模板管理**：进入「应用开发 > 组件管理 > 提示词」，可创建、编辑、复制或删除自定义模板；预置模板在「提示词 > [插件](../concepts/plugin.md)市场」查看，支持一键复制、创建应用或调用 API 示例。 +- **自动优化**：在「提示词 > 自动优化」页面粘贴原始 Prompt，点击「优化」后可直接复制结果或「保存为模板」。 +- **反馈优化**：在「提示词 > 反馈优化」页面配置推理模型、初始 Prompt、样例数据（上传或选自样例库）及评测数据集，启动优化任务。 -### API 调用 -- 获取模板：调用 `GetPromptTemplate`，传入 `workspaceId` 和 `promptTemplateId`，返回含 `variables` 和 `content` 的 JSON 响应。 -- 创建模板：调用 `CreatePromptTemplate`，需指定 `name`、`type`（`text` 或 `image`）、`content`（文本模板）或 `positivePrompt`/`negativePrompt`（图片模板）。 -- 应用调用：若已关联样例库，可在请求体中设置 `has_thoughts: true` 查看检索详情。 +### API/SDK 调用 +- **获取模板**：调用 `GetPromptTemplate` 接口，传入 `workspaceId` 和 `promptTemplateId`，解析响应中的 `content` 与 `variables` 字段动态填充变量。 +- **创建应用**：将生成的 Prompt 作为 `system_prompt` 或 `user_prompt` 参数提交至智能体应用 API。 +- **调试验证**：在 OpenAPI 调试页选择 SDK V2.0（推荐），自动填充参数后运行示例代码 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 ## 限制和注意事项 -- **地域限制**：所有 Prompt 功能仅支持华北2（北京）地域，跨地域调用将失败。 +- **地域限制**：所有 Prompt 功能（模板、优化、样例库）当前仅支持华北2（北京）地域，跨地域调用将失败。 - **容量限制**： - - 单个 Prompt 模板内容最大 6144 字符（控制台编辑框右下角实时计数）； - - 单个样例库最多 300 条样例（已停用，仅作历史参考）； - - 批量导入样例文件 ≤20MB，单次 ≤100 条。 -- **Token 成本**：启用样例库或反馈优化会显著增加输入 Token（样例内容 + 用户 query），需纳入成本预估 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 -- **变量语法**：模板中使用 `${variable}` 占位符，渲染时需确保变量名与 `GetPromptTemplate` 返回的 `variables` 数组严格匹配。 -- **安全合规**：自动优化服务不存储用户 Prompt 数据，亦不用于模型训练，符合阿里云数据隐私政策 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 + - 自定义模板内容最大支持 6144 字符； + - 图片生成模板正向/负向 Prompt 各有独立长度限制（未明确定义，建议 ≤2048 字符）； + - 反馈优化评测数据集建议 ≥20 条，样例数据集建议 5–10 条且覆盖全部类别。 +- **安全与合规**： + - 自动优化过程不存储用户 Prompt，不用于模型训练； + - 所有模板变量需符合 `${variableName}` 格式，非法变量名将导致填充失败； + - 图片生成负向 Prompt 中禁止包含违法、违规或敏感词，否则触发内容审核拦截。 +- **成本影响**：启用反馈优化或历史样例库会显著增加输入 [Token](../concepts/token.md) 消耗（公式：`总输入 Token ≈ 用户查询 Token + 召回样例总 Token + 系统指令 Token`），需纳入费用预估。 ## 来源文档 - [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md) -- [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md) - [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt自动优化](../../raw/application-user-guide/prompt/optimize-prompt.md) +- [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md) - [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md index 33e0492e..e67a1227 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md @@ -1,46 +1,73 @@ # release notes -百炼平台的 Release Notes 汇总了模型上下架、功能迭代、API 变更及平台能力演进等关键动态，面向开发者提供可落地的技术更新概览。内容覆盖模型支持范围、核心参数变更、调用方式升级、已知限制与兼容性注意事项。所有信息均基于平台近期正式发布版本，建议开发者结合自身场景关注模型生命周期状态与接口兼容性。 +本页汇总百炼平台近期模型与功能更新，面向开发者提供关键变更、可用能力及使用约束的结构化参考。内容涵盖新上线模型、平台功能迭代、参数与接口变动，以及已知限制。所有信息均基于官方发布文档整理，建议结合具体 API 文档与 SDK 版本验证兼容性。 ## 支持的模型/功能 -- **新增模型**：2026年7月起，华北2（北京）地域陆续上线 `qwen-audio-3.0-realtime-plus`（实时多模态）、`vidu/viduq3-ad_reference2video`（参考生视频）、`happyhorse-1.1-t2v`（文生视频）、`fun-music-v1`（音乐生成）、`Tripo/Tripo-H3.1`（3D生成）等数十款模型，覆盖语音、图像、视频、3D、音乐及全模态场景。详见 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)。 -- **模型能力扩展**：Qwen3.7系列全面增强多模态交互混合智能体能力；Kimi K2.7 Code 系列新增高速档位（`kimi/kimi-k2.7-code-highspeed`），推理速度提升5~6倍；GLM-5.1 支持200K上下文与128K最大输出；DeepSeek-V4-Pro 支持 `cached_token` 单价调整为1元/百万token（见[文档 2](../../raw/model-user-guide/release-notes/newly-released-models.md)）。 -- **功能模块上线**：6月新增知识检索服务与知识问答服务（支持多知识库联合检索与混合排序）；6月上线智能体托管运行时 API；5月起模型调优支持强化学习（RL）、0代码安全合规强化、视频/图像/视觉理解模型类型；4月起多模态交互开发套件覆盖 Android/iOS Lite、Linux C++、RTOS C 等全端 SDK。 +- **新增模型**（2026年7月）： + - 实时多模态：`qwen-audio-3.0-realtime-plus`、`qwen-audio-3.0-realtime-flash`（端到端低延时语音交互）； + - 语音合成：`qwen-audio-3.0-tts-plus`（高品质）、`qwen-audio-3.0-tts-flash`（首包延时 ≤200ms）； + - Vidu 系列图像/视频生成：`vidu/vidu-image_reference2image`、`vidu/viduq3-ad_reference2video`、`vidu/viduq3-pro-fast_img2video` 等； + - Qwen 系列：`qwen3.7-max-2026-06-08`（新增视觉模态理解）、`qwen3.7-plus`（多模态交互混合智能体能力）； + - OCR：`qwen3.5-ocr`（128K上下文、多轮对话、卡证识别增强）。 + 完整列表详见 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)。 -> **注意**：文档 1 中提及“Qwen3-VL-8B-Instruct/Thinking 支持 SFT 调优”（2025年10月），但文档 2 中未列出该模型上架记录，且其命名与当前主流 Qwen3.5/Qwen3.6/Qwen3.7 系列不一致，建议以 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 中实际发布的模型列表为准，避免使用非公开快照模型。 +- **平台功能新增**（2026年6–7月）： + - 智能体托管运行时 API（[了解详情](https://help.aliyun.com/zh/model-studio/managed-agents-api-overview)）； + - 知识检索服务与知识问答服务（支持多知识库联合检索与混合排序）； + - Responses API 新增异步调用模式（`background=true`）； + - 模型导入功能国际站上线（支持从 OSS 导入 LoRA 微调模型）； + - Skill 能力包上线（支持添加官方或自定义技能）。 + 功能动态详情见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)。 + +> **注意**：文档 1 中 `qwen3.7-max-2026-06-08` 标注“具备多模态交互混合智能体能力”，但文档 2 未提及该模型的多模态输入支持；而文档 1 同期列出的 `qwen3.7-plus` 明确支持视觉参考生成代码等能力。建议以 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 中的模型规格说明为准，并在实际调用前验证输入模态兼容性。 ## 关键参数 -- **计费模式**：模型部署支持按模型单元（MU）时长计费（自2025年10月起），适用于 `qwen-flash`/`qwen-plus` 等预置模型；`deepseek-v4-pro` 的 `cached_token` 单价明确为 **1 元/百万 token**（标准 `input_token` 不变）。 -- **上下文与输出**：GLM-5.1 支持 200K 上下文与 128K 最大输出；Qwen3.5-OCR 上下文扩展至 128K；Qwen3.7-Max 系列仅支持纯文本输入，默认开启思考模式，支持显式缓存。 -- **性能指标**：`qwen-audio-3.0-tts-flash` 首包延时 ≤200ms；`qwen3.6-flash` 系列在代码智能体基准中大幅超越前代；`wan2.7-r2v` 支持单张多宫格故事板一键生成剧本化视频。 +- **模型输入约束**： + - `qwen3.7-max` 及 `qwen3.6-max-preview` 等 Max 系列模型**仅支持纯文本输入**，不接受图像或视频（见文档 1 中 2026-04-20 条目）； + - `kimi/kimi-k2.7-code` 仅支持思考模式； + - `qwen3.5-ocr` 上下文长度扩展至 128K，支持多轮对话； + - `qwen-audio-3.0-tts-flash` 首包延时控制在 200ms 以内。 + +- **部署与计费参数**： + - PTU 部署支持长输入与前缀缓存（2026-06-15 上线）； + - 模型部署支持按模型单元（MU）时长计费（2026-01-23 起）； + - `deepseek-v4-pro` 的 `cached_token` 单价调整为 1 元/百万 token（2026-04-29），标准 `input_token` 不变。 ## 使用方式 -- **API 调用**： - - Responses API 新增异步调用模式（`background=true`），适用于长耗时任务； - - 异步任务支持通过事件总线 EventBridge 主动推送完成事件（HTTP 回调或 RocketMQ），替代轮询； - - 新增临时 API Key 生成机制，适用于不可信环境，规避永久密钥泄露风险； - - 智能体托管运行时、知识检索/问答、Prompt 工程、数据连接等模块均已提供完整 API 文档（参见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)）。 -- **SDK 与集成**： - - 多模态交互开发套件提供 Android/iOS Lite、Android、iOS、Linux C++、RTOS C 等 SDK； - - Spring AI Alibaba 框架已支持调用百炼智能体与工作流应用； - - Codex 终端 AI 编程助手、Kilo CLI 工具均完成百炼接入适配。 +- **模型调用**： + - 所有模型通过统一推理 API 接入，支持 OpenAI Responses 与 Anthropic Messages 接口分类（2026-05-15 更新）； + - 新增 DashScope 智能体应用 API（2026-05-11），支持单轮/多轮、流式、文件问答与视觉理解； + - 异步任务支持事件总线 HTTP 回调与 RocketMQ 主动推送（2026-04-23），避免轮询。 + +- **开发集成**： + - 多模态交互开发套件提供 Java SDK（服务端）、Android/iOS Lite SDK、RTOS C SDK 及 Linux C++ SDK； + - Spring AI Alibaba 框架已支持调用百炼智能体与工作流应用（2026-06-01）； + - Codex 终端 AI 编程助手于 2026-06-24 接入百炼。 + +- **模型定制**： + - 模型调优支持图像生成（Wan/Wanx）、视觉理解（VL）、视频生成三类模型（2026-05-28 / 01-22 / 01-21）； + - 支持强化学习（RL）训练（邀约制，2026-05-31）及 0 代码安全合规强化（2026-05-04）。 ## 限制和注意事项 -- **模型下线**：2026年7月起分批下线部分老旧及长尾模型（如7月10日、7月9日通知），同时存在延期下线安排（7月6日通知）。具体清单与机制请严格参照 [模型下线机制说明](../../raw/model-user-guide/release-notes/model-release-notes.md)。 -- **地域与部署**：新模型（如 `qwen-audio-3.0-realtime-plus`、`vidu` 系列）当前仅部署于华北2（北京）地域，国际站用户需确认服务可用性；6月12日新增美国、德国、日本地域，但模型覆盖需单独验证。 -- **功能约束**： - - `qwen3.6-max-preview` 明确不支持图像与视频输入； - - `kimi/kimi-k2.7-code` 仅支持思考模式； - - `qwen3.5-omni-plus` 为全模态模型，但具体输入模态组合需查阅对应 API 文档； - - 免费额度用完即停功能启用后，将返回错误码 `AllocationQuota.FreeTierOnly`，需在客户端做好容错处理。 +- **地域与部署范围**：多数新模型（如 Vidu、Qwen-Audio 系列）当前仅限中国内地服务，美国、德国、日本等新地域于 2026-06-12 启用，需显式指定 region 参数（见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)）。 + +- **模型下线风险**： + - 2026-07-10 起已启动部分老旧模型下线通知（含长尾模型），具体清单与机制参见 [模型下线机制说明](https://help.aliyun.com/zh/model-studio/model-depreciation)； + - `qwen-turbo` 资源包已于 2026-06-28 启动退市，存量资源包到期后不可续购。 + +- **兼容性约束**： + - `qwen3.6-max-preview` 明确标注“> 不支持图像与视频输入”（文档 1，2026-04-20），与同系列 `qwen3.7-plus` 的多模态能力形成对比，调用前须核对模型文档； + - `kimi/kimi-k2.7-code-highspeed` 与 `kimi/kimi-k2.7-code` 功能一致但速度提升 5~6 倍，二者不可混用同一缓存策略。 + +- **免费额度与用量**：新人免费额度启用“用完即停”功能（2025-07-29 上线），耗尽后返回 `AllocationQuota.FreeTierOnly` 错误码，需主动升级付费计划。 ## 来源文档 -- [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) - [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) +- [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md index 03c8ebbc..5642e31d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md @@ -1,75 +1,58 @@ # security and compliance -阿里云百炼平台提供多层次的安全与合规能力，覆盖模型调用、数据传输、存储隔离及监管备案等关键环节。开发者可通过权限管理、AI安全护栏、端到端加密、私网访问及合规资质材料获取等机制，满足企业级安全要求与国内生成式AI监管规范（如《生成式人工智能服务管理暂行办法》）。所有能力均基于阿里云基础设施的合规底座（如SOC 2）构建，确保数据隐私、传输安全与责任可追溯。 +阿里云百炼平台提供多层次安全与合规能力，覆盖模型备案、数据传输加密、私网访问、内容安全防护、权限隔离及存储安全等关键维度。所有能力均面向企业级生产环境设计，开发者需根据自身业务场景（如C端上架、内部系统、高敏感数据处理）选择适配的组合方案，并独立承担《生成式人工智能服务管理暂行办法》等法规定义的服务提供者责任。 ## 支持的模型/功能 -- **AI安全护栏服务**：支持对文本和图片类模型的输入输出内容进行实时合规检测（涉黄、涉政、广告等），需显式启用 `X-DashScope-DataInspection` 请求头 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 -- **加密推理通道**：支持通过AES-RSA混合加密机制对请求体 `input` 字段加密，防止公网传输中敏感信息泄露；该功能仅适用于 DashScope Endpoint，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)不支持 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 -- **私网访问能力**： - - 普通业务空间：支持通过 PrivateLink 创建**接口终端节点**，实现 VPC 内资源直连百炼 API（限华北2北京、新加坡地域）[通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 - - 安全存储业务空间：需配置**反向终端节点** + MSE 网关 + 私有云资源（OSS/ADB/ES），构建完全隔离的数据存储与处理环境 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 -- **模型备案信息**：所有接入百炼的主流大模型（如千问、万相、DeepSeek、Moonshot 等）均已取得国家网信办算法备案号与大模型备案号，可在控制台或[模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)页面查询。 +- **已备案模型**：平台接入的千问、万相、智谱 AI、DeepSeek、Moonshot 等主流大模型均已取得国家网信办算法备案号及大模型备案号，完整清单见[模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)。其中千问系列备案主体为阿里巴巴达摩院(杭州)科技有限公司，万相视频生成算法备案主体为通义云启（杭州）信息技术有限公司。 +- **AI 安全护栏**：支持对文本和图片类模型的输入输出进行实时内容审核，自动识别涉黄、涉政、广告等违规内容，需在请求头中配置 `X-DashScope-DataInspection` 参数启用 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 +- **安全存储空间**：面向高合规要求客户，提供基于私网终端节点、MSE网关、OSS/ADB/ES三重资源隔离的安全存储业务空间，适用于金融、政务等敏感场景 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 +- **加密传输能力**：支持 AES+RSA 混合加密机制，对请求体中的 `input` 字段及响应结果全程加密，防止公网传输中敏感数据泄露 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 -> **注意**：文档 8 与文档 9 描述的私网访问路径存在适用范围差异——前者面向通用模型/API调用，后者专用于**安全存储业务空间**（需商务开通），二者网络架构、终端节点类型（接口 vs 反向）及依赖组件（无MSE vs 必须MSE）均不同，不可混用。 +> **注意**：文档 1 和文档 2 中关于万相的算法备案号存在不一致——文档 1 列出两个万相备案号（网信算备330110507206401230027号、网信算备330106003156001240091号），而文档 2 明确区分了“达摩院图像合成算法”（主体：达摩院）与“通义万相视频生成算法”（主体：通义云启），且后者发放日期为2024-12-20。建议以[千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)中按备案编号实时查询的结果为准。 ## 关键参数 -| 参数名 | 用途 | 来源/说明 | -|--------|------|-----------| -| `X-DashScope-DataInspection` | 启用AI安全护栏，值为 `{"input":"cip","output":"cip"}` | [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) | -| `X-DashScope-EncryptionKey` | 加密调用必需请求头，含 `public_key_id`、`encrypt_key`（RSA加密的AES密钥）、`iv` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | -| `enable_encryption=True` (Python) / `enableEncrypt(true)` (Java) | DashScope SDK 开箱即用加密开关 | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | -| `public_key_id`, `public_key` | 通过 `/api/v1/public-keys/latest` 接口获取，用于客户端RSA加密AES密钥 | [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) | +| 参数名 | 用途 | 示例值 | 来源 | +|--------|------|--------|------| +| `X-DashScope-DataInspection` | 启用AI安全护栏，控制输入/输出检查开关 | `{"input":"cip","output":"cip"}` | [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) | +| `X-DashScope-EncryptionKey` | 传输加密时携带RSA加密后的AES密钥、公钥ID及IV | `{"public_key_id":"1","encrypt_key":"...","iv":"..."}` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `enable_encryption=True` (Python) / `.enableEncrypt(true)` (Java) | DashScope SDK 启用自动加解密的开关 | `True` / `true` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `base_url` | 替换为终端节点服务域名实现私网调用 | `https://vpc-cn-beijing.dashscope.aliyuncs.com/compatible-mode/v1` | [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) | ## 使用方式 -1. **权限控制** - - 超级管理员通过全局管理菜单（如[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)）统一管控多业务空间模型授权、限流与API Key； - - 业务空间管理员在对应空间内配置模型调用/训练/部署权限，并管理用户控制台页面权限； - - API Key 继承归属业务空间的模型权限，不受用户控制台权限影响。 - -2. **启用AI安全护栏** - - 在[安全管理](https://bailian.console.aliyun.com/?globalset=1#/efm/global_set)页面完成服务授权； - - 所有调用请求必须携带 `X-DashScope-DataInspection` 头，否则不触发审核。 - -3. **加密推理调用** - - **SDK方式（推荐）**：Python/Java SDK 设置 `enable_encryption=True` 或 `.enableEncrypt(true)`，自动处理加解密； - - **HTTP方式**： - a) 调用 `/api/v1/public-keys/latest` 获取公钥； - b) 生成AES密钥并加密 `input` 字段； - c) 用RSA公钥加密AES密钥，构造 `X-DashScope-EncryptionKey` 头； - d) 解密响应体获取明文结果。 - -4. **私网访问配置** - - **普通场景**：在VPC中创建接口终端节点，关联 `com.aliyuncs.dashscope` 服务，替换API Base URL为终端节点域名； - - **安全存储场景**： - a) 创建反向终端节点并确认连接； - b) 配置MSE网关、可用区VIP及交换机网段； - c) 授权并绑定OSS/ADB/ES等私有云资源； - d) 激活业务空间。 +- **合规备案材料获取**：面向C端上架的应用，需准备所用模型的算法备案截图（通过[互联网信息服务算法备案系统](https://beian.cac.gov.cn/#/index)按备案编号查询）及阿里云合作协议；具有舆论属性的场景还需额外完成安全评估报告与自主算法备案 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 +- **启用内容安全**：开通AI安全护栏服务后，在调用请求头中添加 `X-DashScope-DataInspection`，值为 JSON 字符串 `{"input":"cip","output":"cip"}`；若仅需检查输入，可设为 `{"input":"cip"}`。 +- **启用传输加密**： + - 使用 DashScope SDK：设置 `enable_encryption=True`（Python）或 `.enableEncrypt(true)`（Java），SDK 自动处理密钥获取、加解密全流程； + - 使用 HTTP 调用：先调用 `/api/v1/public-keys/latest` 接口获取 RSA 公钥及 ID [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md)，再用该公钥加密 AES 密钥，最后将加密后 input 和密钥信息填入请求头。 +- **私网访问部署**：创建接口终端节点（类型：接口终端节点，服务：`com.aliyuncs.dashscope`），获取终端节点服务域名，替换原 API 的 `base_url` 域名即可；安全存储空间需额外配置反向终端节点 + MSE网关 + OSS/ADB/ES资源绑定 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 ## 限制和注意事项 -- **地域限制**：私网访问仅支持华北2（北京）和新加坡地域；美国（弗吉尼亚）地域暂不支持 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 -- **API Key 约束**：单个API Key仅归属一个地域内的一个业务空间和一个用户，不可转移；自2026年3月25日起，华北2（北京）新创建的API Key均归属主账号 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)。 -- **加密调用兼容性**：仅 DashScope Endpoint 支持加密，[OpenAI 兼容接口](../concepts/openai-compatible-api.md)（`/compatible-mode/v1`）不支持 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 -- **安全存储业务空间依赖**：OSS Bucket 若被释放，将导致安全存储空间**不可恢复**；ADB/ES 若停止计费或释放，相关模块（知识库、审计日志等）将不可用 [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md)。 -- **备案责任主体**：使用百炼模型的应用开发者是《生成式人工智能服务管理暂行办法》定义的“服务提供者”，须独立承担内容审核、用户保护、算法备案等全部法定义务，阿里云仅提供模型及备案信息支持 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 +- **模型备案责任归属**：阿里云百炼仅作为“服务技术支持者”完成算法备案，应用/小程序开发者是《生成式人工智能服务管理暂行办法》定义的“服务提供者”，须独立履行内容审核、用户标识、日志留存等全部法定义务 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 +- **地域与网络限制**： + - 私网终端节点仅支持华北2（北京）、新加坡地域，美国（弗吉尼亚）暂不支持 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)； + - 安全存储空间强制要求专有网络位于华北2（北京），且可用区需为G/H/L中的至少两个 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 +- **权限与配额约束**： + - 默认业务空间无法设置模型调用限流、训练或部署权限；精细化控制需新建业务空间并由超级管理员授权 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)； + - API Key 仅归属单个地域、单个业务空间、单个用户，不可跨空间转移；华北2（北京）地域新创建的 API Key 默认归属主账号 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)。 +- **加密与兼容性**：AES+RSA 加密机制**仅适用于 DashScope Endpoint**，OpenAI 兼容模式（`/compatible-mode/v1`）不支持该加密流程 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 ## 来源文档 -- [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md) - [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) -- [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) - [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md) -- [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) -- [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) -- [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) +- [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md) +- [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) - [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) - [配置可用区IP](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - [配置MSE云原生网关](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) +- [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) +- [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) +- [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md index bb115aed..aa84fddc 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md @@ -1,44 +1,41 @@ # skill -Skill 是百炼平台提供的可插拔能力包，用于扩展智能体在对话中自动处理特定任务的能力（如文件解析、数据清洗等），无需额外编码或工具集成。开发者可通过官方 Skill 快速启用通用能力，或通过自定义 ZIP 包构建业务专属 Skill。所有 Skill 均由智能体基于 `description` 语义匹配自动调用，调用准确性高度依赖元信息描述质量。 +Skill 是百炼平台提供的可插拔能力包，用于扩展智能体在对话中自动处理特定任务的能力（如文件解析、数据清洗等），无需额外编码或工具集成。开发者可通过官方 Skill 快速启用通用能力，或通过自定义 ZIP 包构建业务专属 Skill。其核心机制依赖 `SKILL.md` 中的语义描述驱动智能体自动识别与调用，[原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 详细说明了该机制的设计逻辑。 ## 支持的模型/功能 -- **官方 Skill**：平台预置、开箱即用的通用能力，覆盖 `.xlsx`/`.csv`/`.tsv` 等文件处理、PDF 文本提取、图像 OCR 等场景，由平台统一维护和更新，已添加的智能体会自动升级至最新版本。 -- **自定义 Skill**：通过上传符合规范的 ZIP 包实现，适用于官方 Skill 未覆盖的垂直场景（如行业专用格式解析、私有 API 封装等）。ZIP 包必须包含根目录下的 `SKILL.md` 文件，并满足 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中定义的结构与字段要求。 -- 所有 Skill 均不依赖特定大模型，其调用逻辑由百炼底层调度引擎根据用户输入语义与 `description` 匹配决定，与所选推理模型无关。 +- **官方 Skill**：由平台预置并维护，覆盖常见文件处理场景（如 `.xlsx`、`.csv` 解析与生成），开箱即用，无需配置。最新列表请参考控制台 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面，[原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 明确指出其持续更新特性。 +- **自定义 Skill**：通过上传符合规范的 ZIP 包实现，适用于行业特有格式（如医疗 DICOM 元数据提取）、私有协议解析等官方未覆盖场景。所有自定义 Skill 均需包含 `SKILL.md` 文件，且必须满足命名唯一性、10 MB 大小限制等要求，详见 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md)。 ## 关键参数 -关键参数全部定义在 ZIP 包根目录的 `SKILL.md` 文件中，采用 YAML 格式： - | 字段 | 必填 | 说明 | |------|------|------| -| `name` | 是 | Skill 唯一标识符，仅允许小写字母、数字和连字符（如 `invoice-parser`），同一账号下不可重复；该字段也作为版本管理的命名依据。 | -| `description` | 是 | **决定 Skill 是否被正确调用的核心字段**。需明确说明适用输入类型、支持操作、典型触发关键词及明确排除的不适用场景。描述质量直接影响匹配准确率，详见 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中的编写建议与完整示例。 | +| `name` | 是 | Skill 唯一标识符，仅支持小写字母、数字和连字符（如 `invoice-parser`）；同一账号下不可重复。 | +| `description` | 是 | 决定智能体是否调用该 Skill 的核心依据。必须明确说明：① 支持的输入类型（如 `.pdf`, JSON 数据流）；② 支持的操作（如“提取表格”“转为 Markdown”）；③ 触发关键词（如“帮我导出为 Excel”）；④ **不适用场景**（如“不处理扫描件 OCR”），否则易导致误调用。 | -> **注意**：`description` 中若未声明“不适用场景”，可能导致误触发；例如 xlsx Skill 明确排除产出 Word 或 HTML 的场景，此约束在 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 的示例中有严格体现，实际编写时必须遵循。 +> **注意**：`description` 的质量直接影响调用准确率，[原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 提供的 xlsx Skill 示例是当前唯一权威参考，其结构（含触发条件、输入输出约束、排除场景）应严格遵循。 ## 使用方式 1. **创建 Skill** - - 官方 Skill：直接在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面查看并添加，无需配置。 - - 自定义 Skill：按 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 要求准备 ZIP 包（含 `SKILL.md`，≤10 MB），在控制台 **组件 > Skill 管理 > 自定义 Skill** 中上传，系统约 2 分钟完成审查。 + - 官方 Skill：直接在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面选择添加。 + - 自定义 Skill：准备 ZIP 包（含 `SKILL.md` + 可执行代码/配置），在控制台 **组件 > Skill 管理 > 自定义 Skill** 中上传；审查约 2 分钟，通过后即可使用。 2. **添加到智能体** - 方式一：在 Skill 详情页点击 **添加到智能体**，选择目标应用。 - 方式二：进入智能体 **应用配置 > 技能** 区域，点击对应 Skill 右侧加号添加。 3. **测试与验证** - 在应用配置页右侧对话窗格中发送典型指令（如 `帮我清洗这份 CSV 数据，删除重复行并导出`），观察是否触发预期 Skill 并返回正确结果。 + 在应用配置页右侧对话窗格发送典型用户指令（如 `把附件里的销售数据按季度汇总成表格`），观察是否触发 Skill 并返回预期结果（如 `.xlsx` 文件下载链接）。 ## 限制和注意事项 -- ZIP 包大小上限为 **10 MB**，超限将导致上传失败。 -- `name` 字段全局唯一（同账号内），重名上传会拒绝，而非覆盖。 -- 自定义 Skill 版本更新需重新上传同名 ZIP 包，旧版本仍保留在历史记录中，但已添加该 Skill 的智能体会**自动切换至最新通过审查的版本**。 -- 官方 Skill 的 `description` 由平台维护，开发者不可修改；若发现官方 Skill 行为与文档描述不符，应以控制台实时展示的描述为准。 -- Skill 调用完全基于 `description` 的语义理解，**不支持正则匹配、硬编码关键词或条件分支逻辑**；复杂业务规则需在 Skill 内部代码中实现，而非依赖 `description` 控制流。 +- ZIP 包总大小 ≤ 10 MB，超限将被拒绝上传。 +- `name` 字段在账号维度全局唯一，重名上传会失败。 +- 自定义 Skill 更新需重新上传同名 ZIP 包，系统自动创建新版本；已添加该 Skill 的智能体会**自动切换至最新版本**（无需手动刷新配置）。 +- 官方 Skill 版本由平台统一升级，用户无法回滚或修改其 `description`。 +- 当前 Skill 仅支持同步执行（即阻塞式调用），不支持长时异步任务（如小时级数据训练）；此类需求需通过外部服务 + Callback 实现，不在 Skill 范畴内。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md index 1eafeb69..11be9bc3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md @@ -1,38 +1,49 @@ # start using -阿里云百炼平台提供低门槛、高灵活性的 AI 应用构建能力，支持零代码快速搭建私有知识问答应用，也支持高代码深度定制。开发者可通过控制台可视化配置或 API 编程方式接入模型、知识库、[长期记忆](../concepts/long-term-memory.md)等核心能力，适用于从原型验证到生产部署的全周期场景。本文档聚焦“开始使用”路径，梳理关键能力、参数与约束，帮助开发者高效启动。 +阿里云百炼平台提供低门槛、高灵活性的 AI 应用构建能力，支持零代码快速搭建私有知识问答应用，也兼容全代码集成场景。开发者可通过控制台可视化配置或 API 调用两种方式启动应用开发，核心路径包括模型选择、Prompt 设计、知识库接入与发布部署。本文档聚焦“开始使用”阶段的关键技术要素，面向实际开发需求提炼结构化指引。 ## 支持的模型/功能 -- **基础模型**：智能体应用和工作流应用均支持千问系列（如 `qwen-max`）、QwQ 系列（如 `qwq-plus`、`qwq-32b`）及 DeepSeek 系列模型；其中 QwQ 模型具备强推理能力，输出含显式思考链 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 -- **[多模态能力](../concepts/multi-modal.md)**：`qwen-vl-plus-latest`、`qwen-vl-plus-2025-01-25` 等视觉语言模型支持图文理解与生成；知识库支持导入图片、音视频文件，并启用“多模态回复增强”开关以解析图表内容 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **知识库类型**：分为**文档型**（PDF/DOCX/HTML/Excel）、**数据型**（RDS/DMS/自建 MySQL）、**图片型**三类；非结构化知识库支持离线 HTML、Excel 及自定义 metadata，结构化知识库支持图文检索与音视频解析 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **高级能力**：[长期记忆](../concepts/long-term-memory.md)（新）API 支持多应用共享、自动信息提取与语义检索；MCP 服务可作为插件集成至智能体或工作流；工作流应用支持异步运行模式与批量节点。 +- **基础模型**：智能体应用默认支持 `qwen-max`（文档推荐为“千问-Max”），同时已全面支持 `qwq-plus`、`qwq-32b`（工作流应用）、`qwen-vl-plus-latest` 及 `qwen-vl-plus-2025-01-25` 等多模态与推理增强模型 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 +- **知识库类型**：支持文档型（.docx/.pdf/.xlsx/.html 等）、音视频型（MP4/Audio/WAV）、图片型及结构化数据型（RDS/DMS/自建 MySQL）知识库 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **增强能力**： + - 多模态回复增强（需在智能体应用检索配置中开启）； + - [长期记忆](../concepts/long-term-memory.md)（新版 API 支持自动信息提取与用户画像管理）； + - MCP 工具集成（含预置服务与自定义 MCP）； + - 文件问答支持全文引用、切片检索、自定义处理三种模式。 -> **注意**：文档 1 中提及的“Assistant API（下线中）”已明确废弃，不应再用于新项目开发；当前推荐路径为智能体应用（Agent 2.0）或工作流应用 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 +> **注意**：文档 1 中提及“建议选择千问-Max”，但文档 2 明确指出智能体应用已支持 `qwq` 系列及 `qwen-vl-plus` 等新模型，且 `qwq` 模型具备更强的数学/代码推理能力。实际开发中应优先参考 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md) 的最新模型支持列表，而非文档 1 的示例性建议。 ## 关键参数 -- **知识库检索参数**：`初步向量检索TopK` 和 `初步关键词检索TopK` 可调低以减少送入排序模型的 Token 量，直接降低模型调用费用 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **知识库权重**：当智能体应用关联多个知识库时，可为每个知识库设置权重，系统优先召回高权重知识源 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **Prompt 配置**：System Prompt 定义角色与任务（如“你是一位阿里云百炼手机导购…”），直接影响模型行为边界；支持 FewShot Prompt 样例库提升回答准确性 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **[长期记忆](../concepts/long-term-memory.md)参数**：新版长期记忆支持自动提取对话关键信息、用户画像标签管理，无需手动构造记忆条目。 +- **知识库检索参数**： + - `初步向量检索TopK` 与 `初步关键词检索TopK`：可调低以减少送入排序模型的 [Token](../concepts/token.md) 量，直接降低模型调用费用 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)； + - 知识库权重：当应用关联多个知识库时，可按信息源重要性设置权重，系统优先召回高权重知识库内容； + - 切分策略：推荐使用“智能切分”，经评测对多数文档效果最优 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 +- **[长期记忆](../concepts/long-term-memory.md)参数**：新版[长期记忆](../concepts/long-term-memory.md) API 支持自动去重、语义检索及用户画像字段自定义，无需手动维护记忆条目。 +- **调用参数**： + - 同步调用（`Responses API`）适用于实时交互； + - 异步调用需设置 `background=true`，返回 Task ID 后通过任务中心查询结果。 ## 使用方式 -1. **零代码入门**：访问 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 创建智能体应用 → 选择模型（推荐 `qwen-max`）→ 设置 System Prompt 与欢迎语 → 添加知识库（支持直接上传文件，无需预创建连接器）→ 发布 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 -2. **API 调用**： - - 同步调用：使用 Responses API，兼容 OpenAI SDK，适用于实时交互场景； - - 异步调用：设置 `background=true`，返回 Task ID，通过 [任务中心](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/app-task-center) 查询结果； - - 知识库/长期记忆/工作流节点均提供独立 RESTful API（如 `CreateIndex`、`GetIndexMonitor`、`UpdateIndex`）。 -3. **调试与观测**：编辑智能体应用时可使用内置**知识库调试面板**实时验证检索效果；应用发布后可通过 [应用观测](https://bailian.console.aliyun.com/knowledge-base#/app-observe) 查看端到端处理链路与性能指标。 +1. **零代码快速启动**（适用于原型验证与业务试用）： + - 访问 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)，创建智能体应用； + - 配置 System Prompt（如“你是一位阿里云百炼手机导购…”）； + - 上传知识文档 → 创建知识库 → 在应用技能中绑定知识库； + - 发布前可使用右侧调试面板实时验证检索召回效果 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。 + +2. **API 集成开发**（适用于生产环境与定制化需求）： + - 调用 `CreateIndex` API 创建音视频/结构化知识库； + - 使用 `GetIndexMonitor` 和 `UpdateIndex` 管理知识库状态与配置； + - 通过 `Responses API`（同步/异步）调用已发布应用，兼容 OpenAI SDK 接口风格 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 ## 限制和注意事项 -- **计费变更**：知识库服务自 2026 年 1 月 4 日起正式商业化，费用 = 规格费 + 模型调用费；支持后付费与资源包两种模式，资源包需通过控制台单独开通 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **模型兼容性**：QwQ 系列模型在智能体应用中**不支持插件、流程编排与音视频交互能力**，仅适用于纯文本推理场景；DeepSeek 系列模型仅支持工作流与智能体应用，不支持旧版智能体编排应用 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **权限与分账**：知识库支持子账号开通与标签分账，但需提前配置服务关联角色（如 `AliyunServiceRoleForSFMTelemetry`）以启用应用观测功能 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 -- **文件限制**：单次上传文档大小上限为 100 MB；音视频文件需符合格式规范（MP4/MOV/AVI/WAV/MP3），且解析依赖 `qwen-vl` 系列模型。 +- **计费变更**：知识库服务自 2026 年 1 月 4 日起正式商业化，费用由规格费 + 模型调用费构成；支持后付费与资源包两种模式，资源包需通过控制台单独开通 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **模型兼容性**：`QwQ` 系列模型在智能体应用中**不支持[插件](../concepts/plugin.md)、流程、音视频交互能力**，仅限纯文本推理场景；而工作流应用则完整支持其多节点编排能力。 +- **知识库时效性**：非结构化知识库导入 Excel 文档时，若原始文件含复杂公式或宏，可能无法完全解析；结构化知识库从 RDS 同步数据时，需确保数据库账号具备 `SELECT` 权限。 +- **调试依赖**：知识库调试面板仅在编辑智能体应用时可用，工作流应用需通过任务中心查看节点执行日志。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/support.md b/skills/bailian-docs-llm-wiki/wiki/guides/support.md index 94574097..08aada60 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/support.md @@ -1,47 +1,52 @@ # support -阿里云百炼平台的 `support` 模块涵盖服务开通、计费、API/SDK 使用、模型能力边界及合规协议等核心支持事项。本文档面向开发者，系统梳理当前平台在功能支持、参数配置、调用方式、限制条件等方面的明确要求与实践指引，所有信息均基于最新公开文档与控制台行为验证。 +阿里云百炼平台的 `support` 模块涵盖服务开通、计费、API/SDK 使用、模型能力边界及合规协议等核心支持事项。它面向开发者提供可落地的技术指引与约束说明，而非泛泛的服务承诺。所有功能与限制均以控制台实际行为和最新 API 文档为准，历史文档中未同步更新的内容需谨慎参考。 ## 支持的模型/功能 -- **模型类型**：支持千问系列（Qwen-Turbo、Qwen-Max、Qwen3、Qwen-VL-Plus 等）及其他第三方模型，覆盖文本生成、多模态（图像训练）、RAG 增强等场景；其中 Qwen-VL-Plus 明确支持图片微调训练 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 -- **功能范围**： - - Completion API 与 Assistant API 均已上线，但 Assistant API **暂不支持 memory 配置**，且 **不支持单次调用中依次执行两个本地函数**（需拆分为两个独立 Assistant API 调用）。 - - RAG 功能可用，`doc_reference_type` 参数仅在旧版应用中生效；新版应用需通过控制台「展示回答来源」开关启用答案溯源能力 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 -- **数据对接**：当前**不支持直接对接 MySQL、Hive 等结构化数据库**，RDS 接入正在开发中。 - -> **注意**：文档 1 中“模型中心”第10条称“当前不支持”结构化数据对接，而文档 2 未涉及此内容；该限制仍有效，无更新说明。 +- **模型类型**：支持千问系列（Qwen-Turbo、Qwen-Max、Qwen3、Qwen-VL-Plus 等）、开源模型及第三方模型；其中 Qwen-VL-Plus 已支持图片训练 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **核心能力**： + - Completion API（基础文本生成） + - Assistant API（支持 function call，但**不支持连续调用多个本地函数**；当前**不支持 memory 配置**） + - RAG 增强（通过 `doc_reference_type` 参数或应用配置中的“展示回答来源”开关控制答案溯源，该参数仅在旧版应用中生效 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)） +- **数据对接**：当前**不支持直接对接 MySQL、Hive 等结构化数据源**，RDS 接入正在开发中。 ## 关键参数 -- **必需参数**：Completion API 调用必须包含 `AppId`、`Prompt`、`RequestId`；缺失或格式错误将返回错误码 `100004` [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 -- **幻觉抑制参数**：可通过降低 `temperature`、`top_k`、`top_p` 提升输出确定性；缩短 `max_tokens` 可防止冗余捏造；这些参数调整是降低模型幻觉的有效手段之一。 -- **RAG 相关参数**：`doc_reference_type` 仅对旧版应用生效，新版依赖控制台开关，参数设置无效。 +| 参数名 | 说明 | 注意事项 | +|--------|------|----------| +| `temperature` / `top_k` / `top_p` | 控制输出随机性与确定性 | 降低这些值可抑制幻觉，但可能削弱创造性；需结合任务人工评估效果 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) | +| `max_tokens` | 限制响应长度 | 过长易引发后半段捏造，适当截断可缓解幻觉 | +| `doc_reference_type` | 控制答案来源标注方式 | **仅在旧版本应用中生效**；新版本需在应用配置中开启“展示回答来源”开关，否则该参数无效 | + +> **注意**：文档 1 中提到“Assistant API 有 memory 相关的能力吗？当前暂不支持”，而部分早期 SDK 示例曾隐含 session state 语义，该能力**已明确废弃且无替代方案**，开发者须自行维护上下文。 ## 使用方式 -- **服务开通**：需以阿里云主账号在目标地域（如北京、新加坡）的[百炼控制台](https://bailian.console.aliyun.com/)开通，开通前须完成实名认证。 +- **服务开通**：需使用阿里云主账号，在目标地域（如北京、新加坡）的[百炼控制台](https://bailian.console.aliyun.com/)开通；未实名认证将被拦截 [常见问题 (raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 - **API 调用**： - - 支持 Python 和 Java SDK，安装方法详见官方指南； - - 请求需携带 `Authorization: Bearer `，Header 中 `Content-Type` 必须为 `application/json`； - - 错误码含义及处理方案请查阅 [错误码文档](https://help.aliyun.com/zh/model-studio/error-code)。 -- **计费与账单**： + - 必须携带 `Authorization: Bearer ` 及 `AppId`、`Prompt` 等必需字段； + - 错误码 `100004` 表示参数缺失或格式错误，需严格校验 JSON 结构与字段命名； + - SDK 仅官方支持 Python 和 Java，安装方式见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 +- **计费与资源管理**： - 后付费按分钟出账、按月结算； - - 扣款明细与发票申请均通过阿里云[费用与成本控制台](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)操作； - - 预付费支持节省计划与资源包，详情见 [节省计划与资源包](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 + - 部分模型支持预付费（节省计划/资源包），详情见[节省计划与资源包](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)； + - 万相会员**不支持百炼 API 调用**，二者计费体系完全独立。 ## 限制和注意事项 -- **服务关闭**：百炼服务开通后**不可主动关闭**；如需停用，仅能删除对应地域的 API-Key 以阻断调用。 -- **数据隐私与存储**： +- **数据与隐私**： - 所有传输数据经 AES-256 加密； - - 平台依法律法规存储调用日志，**不用于模型训练**； - - 控制台历史对话最多保留 100 条，未登录状态及推理报错对话不保存。 -- **模型能力边界**： - - 万相会员权益**不适用于百炼 API 调用**，二者计费体系完全独立； + - 阿里云**不会将用户数据用于模型训练**； + - 模型与应用调用日志依法律法规留存，具体条款见[《阿里云百炼服务协议》](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=a2ty02.30260209.aillm.1.d8bb74a10sknig) [相关协议 (raw/model-user-guide/support/related-agreements.md)](../../raw/model-user-guide/support/related-agreements.md)。 +- **功能限制**： + - 百炼控制台最多保留 **100 条历史对话记录**（未登录或推理报错对话不保存）； - 不支持为生成文本添加隐式标识； - - 无官方手机端 App，仅提供 Web 控制台访问。 -- **合规与协议**：使用前须阅读并接受 [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=a2c4g.2667824.0.0.6a2f6f83Ivpy5F) 及 [SLA 协议](https://terms.alicdn.com/legal-agreement/terms/b_end_product_protocol/20250923215800868/20250923215800868.html)，相关条款详见 [相关协议 (raw/model-user-guide/support/related-agreements.md)](../../raw/model-user-guide/support/related-agreements.md)。 + - 无官方手机端应用，仅支持 Web 访问； + - 自定义模型训练完成后**不支持导出**。 +- **SLA 与合规**： + - 服务可用性、响应延迟等承诺详见[阿里云百炼模型推理服务等级协议（SLA）](https://terms.alicdn.com/legal-agreement/terms/b_end_product_protocol/20250923215800868/20250923215800868.html) [相关协议 (raw/model-user-guide/support/related-agreements.md)](../../raw/model-user-guide/support/related-agreements.md)； + - 应用上架需完成[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)，合作协议需通过工单申请。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md index 4be0837d..33e45b57 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md @@ -1,39 +1,41 @@ # test 1 -`test 1` 是阿里云百炼平台面向开发者提供的核心计费与资源管理主题，涵盖模型调用、训练、部署的全链路成本控制机制。其核心围绕免费额度自动抵扣、多层级付费方案（按量、资源包、节省计划）及精细化账单溯源能力展开，旨在帮助开发者在保障业务连续性的同时实现成本可预测、可监控、可优化。所有计费行为均默认遵循“免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费”的严格抵扣顺序。 +`test 1` 是阿里云百炼平台面向开发者提供的核心计费与资源管理主题，涵盖模型调用、训练、部署的全链路成本控制机制。其核心围绕免费额度自动抵扣、多层级付费方案（按量、资源包、节省计划）及精细化账单溯源能力展开，适用于从新手试用到企业级规模化部署的各类场景。所有计费行为均严格遵循地域与服务部署范围约束，华北2（北京）为中国内地服务的默认且唯一支持免费额度的地域。 ## 支持的模型/功能 -- **支持免费额度的模型**：仅限华北2（北京）地域、服务部署范围为[中国内地](https://help.aliyun.com/zh/model-studio/regions/#080da663a75xh)的模型，例如 `qwen3.7-plus`、`qwen-max` 等主流文本生成模型；快照版本（如 `qwen3.7-plus-2026-05-26`）与基础版本视为独立模型，各自享有独立额度 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 -- **不支持免费额度的场景**：Batch调用、模型调优、模型部署、自定义模型（调优后或已部署模型）均不可使用免费额度抵扣 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 -- **支持的计费模型类型**：覆盖文本生成（千问、DeepSeek、GLM）、多模态（千问VL）、图像生成（万相）、视频生成（万相）、语音模型（千问语音）、向量/排序模型（text-embedding-v4、qwen3-rerank）等全品类，详见各模型价格表 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 -- **专属计费能力**：模型训练按训练Token计费（如千问VL、万相图生视频），模型部署支持两种模式——预置吞吐（按TPM时长）和模型单元（按算力规格小时）[模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 +- **实时推理**：支持千问（Qwen）、DeepSeek、GLM、Kimi、MiniMax 等主流文本生成模型，以及千问VL、万相（WanX）等多模态模型，覆盖非思考/思考模式、Batch调用（半价）、上下文缓存等特性 [原文标题](../../raw/model-user-guide/test-1/model-pricing.md)。 +- **模型训练**：支持文本生成（千问）、图像生成（万相）、视频生成（万相）三类模型微调，按训练[Token](../concepts/token.md)总量计费，计算逻辑依赖 `max_steps`、`Lstep` 或 `视频计费时长 × max_pixels × n_epochs` 等超参 [原文标题](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 +- **模型部署**：提供两种计费模式： + - *预置吞吐*：按输入/输出TPM（每分钟[Token](../concepts/token.md)数）与时长计费； + - *模型单元（MU）*：按算力规格（如 MU1 x 8）与使用时长（小时/月）计费，支持PD分离模式降低首[Token](../concepts/token.md)延迟 [原文标题](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 +- **成本优化工具**：支持AI通用型节省计划（跨模型、阶梯折扣）、其他模型节省计划（单模型、无折扣）、资源包（指定模型Token量）三类预购方案，抵扣顺序为：免费额度 > 资源包 > 其他模型节省计划 > AI通用型节省计划 > 按量付费 [原文标题](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 -> **注意**：文档 5 中 `qwen3.7-plus` 在华北2（北京）的“思考模式”输出单价标注为 `8元/百万Token`，而文档 2 中同模型在“模型部署计费”表格里输出单价为 `¥1.92/Per 1K TPM/小时`（即 `1920元/百万TPM/小时`），二者计量单位与场景不同（Token vs TPM），不构成矛盾；但需注意文档 2 明确说明部署计费不支持免费额度抵扣，而文档 1 强调免费额度仅适用于实时推理，此边界必须严格区分。 +> **注意**：文档 5 中 `qwen3.7-max` 在华北2（北京）标注“当前能力等同于 `qwen3.7-max-2026-05-20`”，但文档 2 的部署计费表中 `qwen3.7-max-2026-05-20` 被列为独立模型代码，且其部署单价（¥28.8/10K TPM/小时）与文档 5 中 `qwen3.7-max` 的输入单价（原价12元/百万Token）无直接换算关系。开发者需以控制台实际模型列表为准，避免因快照版本别名导致的配置错误。 ## 关键参数 -- **免费额度参数**：默认 100 万 Token/模型，有效期自开通或申请通过日起 90 天（2025年9月8日11点起新用户适用）；主账号与RAM子账号共享额度，不同模型间额度不互通 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 -- **阶梯计费参数**：部分模型（如 `qwen3-max`）按单次请求输入Token总量分档计价（如 0–32K、32K–128K），该次请求全部Token均按对应档位单价结算 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 -- **部署计费参数**： - - 预置吞吐：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`； - - 模型单元：`费用 = 使用时长（小时）× 模型单元数量 × 模型单元单价`，最小计费单位为分钟 [模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 -- **节省计划承诺参数**：AI 通用型节省计划以“动态月”为周期（非自然月），月承诺消费额从生效日起每满30天重置，当月未用完额度自动清零 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 +- **免费额度**：默认100万Token/模型，仅限华北2（北京）地域+中国内地部署范围，有效期90天（自2025年9月8日11点起新用户），不同模型（含带日期后缀的快照）额度完全独立 [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **Token单价**：按地域与模型ID差异化定价，例如 `qwen3.7-plus` 在华北2（北京）非思考模式下，0–256K输入Token单价为¥2/百万Token；同一模型在新加坡地域单价升至¥2.936/百万Token [原文标题](../../raw/model-user-guide/test-1/model-pricing.md)。 +- **TPM阈值**：部署时需指定输入/输出TPM上限，超出后自动降级至按量付费模式，并在响应Header中返回 `x-dashscope-ptu-overflow:true` [原文标题](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md)。 +- **节省计划承诺周期**：AI通用型节省计划以“动态月”为单位（非自然月），每月额度独立清零，不累积；例如3个月¥1000/月计划，每月仅可用¥1000，而非总计¥3000 [原文标题](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 ## 使用方式 -- **免费额度启用**：无需额外配置，开通百炼后系统自动发放，实时调用即自动优先抵扣；需确保使用通用 API Key（非 Token Plan/Coding Plan 专属 Key），否则不生效 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 -- **节省计划购买与抵扣**：通过 [AI 通用型节省计划购买页](https://common-buy.aliyun.com/?commodityCode=sfm_GenAI_spn_cn)下单，支持全预付/零预付；购买后立即生效，自动按抵扣顺序参与结算，无需绑定模型或API Key [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 -- **账单查询与归因**：通过[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页面，依据 `实例 ID（出账粒度）` 字段（格式：`ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`）精准定位费用来源 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 -- **成本防护配置**：在免费额度页面开启“免费额度用完即停”，可防止额度耗尽后意外扣费；同时建议设置[高额消费预警](https://usercenter2.aliyun.com/home/alarm-threshold)并绑定业务空间标签实现分账 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +1. **启用免费额度**：开通百炼后自动发放，无需额外操作；调用时系统自动优先抵扣，无需切换API Key [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +2. **配置成本控制**： + - 开启“免费额度用完即停”防止意外扣费（控制台 > 免费额度页面或模型详情页）； + - 购买AI通用型节省计划（推荐）或资源包，购买后立即生效，自动按抵扣顺序结算 [原文标题](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 +3. **调用与监控**： + - 使用通用API Key（非Token Plan/Coding Plan专属Key）确保免费额度生效； + - 通过[费用概览](https://bailian.console.aliyun.com/?tab=model#/costing-balance/overview)查看实时消费，通过[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)按 `ApiKeyID;业务空间ID;模型名称` 字段精准溯源费用 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 ## 限制和注意事项 -- **地域与部署范围强约束**：免费额度仅限华北2（北京）+中国内地部署范围；其他地域（如美国、新加坡）或全球/国际部署范围的同名模型无免费额度，且价格存在显著差异（如 `qwen3.7-max` 在新加坡单价为 18.736 元/百万Token） [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。 -- **额度耗尽后行为差异**：全新未认证用户额度用完将直接返回错误码 `AllocationQuota.FreeTierOnly` 并停止服务；已认证用户若未开启“免费额度用完即停”，则自动切换至按量付费，可能导致账户欠费 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。 -- **抵扣顺序刚性**：免费额度、资源包、节省计划的抵扣顺序不可更改；若某模型开启了“免费额度用完即停”，则即使存在未到期的节省计划，服务也会暂停，无法触发后续抵扣 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 -- **欠费影响全局**：账户整体欠费（可用额度 < 0）时，即使其他模型仍有免费额度或节省计划余额，所有服务均会暂停，必须结清欠费方可恢复 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 -- **账单延迟与溯源**：模型推理账单通常在调用结束后 2–10 分钟生成，批量/训练类任务为小时级出账；账单中“计费项”统一显示为“大模型文本消耗量”，须依赖 `实例 ID` 字段中的模型名称进行准确归因 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +- **地域强约束**：免费额度、部分模型部署及节省计划仅支持华北2（北京）；美国、新加坡等地域虽可调用模型，但无免费额度，且价格显著上浮（如 `qwen3.7-plus` 新加坡输入单价¥2.936 vs 北京¥2）。 +- **额度不互通**：同一账号下主账号与RAM子账号共享免费额度，但不同模型（如 `qwen-max` 与 `qwen-max-2026-05-17`）额度完全隔离，系统不会自动切换 [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **欠费影响全局**：账户欠费时，即使某模型仍有免费额度或节省计划余额，所有服务将暂停，必须结清欠费才能恢复 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +- **出账延迟**：模型推理账单分钟级生成（通常2–10分钟），批量推理、训练、知识库账单则为小时级，查询账单需预留缓冲时间 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md index 45d83bf4..694d9e94 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md @@ -1,64 +1,56 @@ # token plan guide -Token Plan 团队版是阿里云百炼面向企业团队提供的 AI 大模型订阅服务，以 Credits 为统一计量单位，支持文本生成与图像生成模型，兼容主流 AI 编程与智能体工具。服务基于多租户隔离架构，承诺不使用对话数据训练模型，并仅在华北2（北京）地域提供。开发者需严格按白名单模型 ID 调用，配套专属 API Key（`sk-sp-` 开头）与 Base URL 使用。 +[Token](../concepts/token.md) Plan 团队版是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持文本与图像生成模型，兼容主流 AI 编程及智能体工具。其核心面向团队协作场景，提供席位管理、用量分析与多模型灵活切换能力，所有调用均通过专属 API Key 和隔离 Base URL 进行鉴权与路由。 ## 支持的模型/功能 -Token Plan 团队版支持的模型为精确字符串白名单，**必须逐字符完全匹配**，版本号或子型号任何差异均视为不支持（如 `qwen3-coder-max` 不在列表中即不可用）[Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)。当前支持以下模型： +[Token](../concepts/token.md) Plan 团队版支持以下精确匹配的模型 ID（区分大小写，版本号必须完全一致）： +- **文本模型**：`qwen3.7-max`（限时活动）、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`deepseek-v4-pro`、`deepseek-v4-flash`、`deepseek-v3.2`、`kimi-k2.7-code`、`kimi-k2.6`、`kimi-k2.5`、`glm-5.2`、`glm-5.1`、`glm-5`、`MiniMax-M2.5`； +- **图像生成模型**：`qwen-image-2.0`、`qwen-image-2.0-pro`、`wan2.7-image`、`wan2.7-image-pro`。 -- **文本生成与视觉理解**：`qwen3.7-max`（限时活动）、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`kimi-k2.7-code`、`kimi-k2.6`、`kimi-k2.5`、`glm-5.2`、`glm-5.1`、`glm-5`、`MiniMax-M2.5`、`deepseek-v4-pro`、`deepseek-v4-flash`、`deepseek-v3.2` -- **图像生成**：`qwen-image-2.0`、`qwen-image-2.0-pro`、`wan2.7-image`、`wan2.7-image-pro` +> **注意**：文档 10（`raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md`）中列出的 `qwen3.5-plus`、`qwen3-coder-next` 等模型属于 Coding Plan 套餐，**不在 [Token](../concepts/token.md) Plan 团队版支持范围内**，该文档混淆了两个独立产品线。请严格以 [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) 中的白名单为准。 -> **注意**：文档 7（Coding Plan 概述）中列出的 `qwen3-coder-next`、`qwen3-coder-plus`、`qwen3-max-2026-01-23` 等模型**未出现在 Token Plan 团队版支持列表中**，不可用于 Token Plan 订阅。二者模型白名单独立，不可混用。 - -核心功能包括： -- **模型内置工具调用**：`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash` 支持通过 Responses API 直接调用联网搜索、代码解释器、网页抓取、以图搜图、文搜图五种工具，费用统一从套餐 Credits 抵扣 [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md)。 -- **图像生成能力**：需通过工具的扩展机制（如 Slash Command、Skill 或 Agent）接入 `multimodal-generation` API，**不可通过文本模型 Base URL 直接调用** [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)。 -- **视觉理解能力**：`qwen3.6-plus`、`qwen3.7-plus` 等原生支持图片输入；非视觉模型（如 `glm-5`）需通过 Skill/Agent 辅助实现，但该能力属于 Coding Plan 文档范畴，Token Plan 团队版默认不提供此类 Skill 配置说明。 +图像生成模型需通过工具扩展机制（如 Slash Command、Skill 或 Agent）接入，**不可直接通过文本模型 Base URL 调用**，详见 [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)。 +部分模型（`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`）原生支持联网搜索、代码解释器等 5 种内置工具，调用不额外收费，Credits 统一抵扣；其余模型需通过 MCP 服务接入工具能力，MCP 调用单独计费（如联网搜索 MCP 免费额度用尽后按 29 元/千次计费），详见 [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md)。 ## 关键参数 -| 参数 | 说明 | 取值/格式 | -|------|------|-----------| -| **API Key** | Token Plan 专属密钥 | 以 `sk-sp-` 开头，仅在创建或重置时完整显示一次，后续仅脱敏显示（如 `sk-sp-****`） | -| **Base URL** | 兼容 OpenAI/Anthropic 协议的端点 | OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | -| **Model ID** | 模型唯一标识符 | 必须严格匹配白名单（如 `qwen3.6-plus`），区分大小写，无空格 | -| **Credits** | 计费单位 | 按输入 tokens、缓存 tokens、输出 tokens 动态计算，优先抵扣坐席月度额度，再抵扣共享用量包 | - -> **注意**：文档 3（快速开始）明确指出 Token Plan、Coding Plan 和按量付费三者的 API Key 与 Base URL **完全隔离，不可混用**；误用通用 API Key（`sk-`）或 Coding Plan Base URL 将导致 401/403 错误或意外按量扣费 [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md)。 +| 参数 | 说明 | 来源 | +|------|------|------| +| **API Key** | Token Plan 专属密钥，格式为 `sk-sp-xxxxx`，仅在首次生成或重置时完整显示一次，后续仅脱敏显示（如 `sk-sp-****`）。丢失后需重置，原 Key 立即失效。 | [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) | +| **Base URL** | 必须配套使用：
- OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`
- Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) | +| **模型 ID** | 必须逐字符完全匹配白名单，禁止版本兼容推理（如 `qwen3.7-max` ≠ `qwen3.7-max-2026-01-23`）。 | [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) | +| **Credits 抵扣顺序** | 1. 坐席月度额度 → 2. 共享用量包（优先抵扣最近到期）→ 3. 全部用尽后服务暂停。 | [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) | ## 使用方式 -1. **订阅与分配**：在 [Token Plan 购买页面](https://common-buy.aliyun.com/token-plan/)完成坐席（标准/高级/尊享）订阅；管理员登录控制台，在「我的订阅」→「分配座席」为成员分配席位，系统自动生成专属 API Key。 -2. **配置工具**：将 API Key 和对应协议的 Base URL 配置至 AI 工具（如 Cursor、Qwen Code、Claude Code 等）。确认工具协议（OpenAI vs Anthropic）与 Base URL 后缀（`/compatible-mode/v1` vs `/apps/anthropic`）严格匹配。 -3. **调用模型**：直接使用支持的 Model ID 发起请求；如需工具调用，对 `qwen3.6-plus` 等模型启用 Responses API 即可自动触发；如需图像生成，按 [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) 文档配置 Slash Command 或 Skill。 -4. **管理用量**：通过控制台「用量分析」查看团队/成员/模型级 Credits 消耗趋势，或在「我的订阅」页面监控额度剩余百分比与重置时间。 +1. **订阅与分配**：主账号或已授权 RAM 用户登录 [Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)，完成套餐购买后，在「我的订阅」→「分配座席」中为成员分配席位，系统自动生成 API Key 和 Base URL。 +2. **配置工具**：将 API Key 和对应协议的 Base URL 配置至兼容工具（如 Cursor、Qwen Code、Claude Code 等），确保协议匹配（OpenAI 协议配 `/compatible-mode/v1`，Anthropic 协议配 `/apps/anthropic`）。 +3. **调用模型**： + - 文本模型：直接指定模型 ID（如 `qwen3.6-plus`）发起请求； + - 图像生成模型：需按工具规范配置扩展（如 Claude Code 的 Slash Command、OpenCode 的 Agent），调用 `multimodal-generation` API endpoint； + - 工具调用：对支持内置工具的模型，启用 Responses API 即可自动触发；对其他模型，需先开通 MCP 服务并配置至工具（如联网搜索 MCP 地址为 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`，鉴权使用百炼通用 API Key `sk-xxx`）。 ## 限制和注意事项 -- **地域限制**：仅支持华北2（北京）地域，海外调用需自行确保合规性 [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)。 -- **使用范围限制**：**仅限在兼容的 AI 编程与智能体工具中交互式使用**，禁止用于自动化脚本、应用后端或批量调用；违规可能导致订阅暂停或 API Key 封禁。 -- **API Key 规范**：每个席位绑定一个成员、一个 API Key，不可共享；丢失后需重置，原 Key 立即失效。 -- **额度规则**：坐席月度额度到期自动重置，不累积；共享用量包有效期 1 个月，到期清零；抵扣顺序为「坐席额度 → 共享用量包 → 服务暂停」。 -- **错误处理**： - - `404 model 'xxx' not found`：检查模型 ID 是否拼写正确且在白名单中； - - `401 InvalidApiKey`：确认使用 `sk-sp-` 开头 Key 及配套 Base URL； - - `429 Allocated quota exceeded`：可能因额度用尽或 TPS/TPM 限流触发，需加购用量包或实施请求平滑策略； - - `400 InvalidParameter: Range of input length`：输入超上下文长度，建议新建会话或切换更大上下文模型。 - -> **注意**：文档 4（常见问题）指出，Token Plan 团队版与 Coding Plan 是两个**完全独立的订阅计划，不支持相互转换**；退订重购会导致 API Key 和 Base URL 变更，需重新配置所有工具 [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md)。 +- **地域限制**：Token Plan 团队版目前仅支持 **华北2（北京）** 地域，跨地域调用将失败。 +- **使用范围限制**：仅限在兼容的 AI 编程与智能体工具中**交互式使用**，禁止用于自动化脚本、应用后端或批量调用。违规可能导致订阅暂停或 API Key 封禁。 +- **API Key 隔离**：Token Plan、Coding Plan 和按量付费三者的 API Key 与 Base URL 完全隔离，混用会导致 401/403 错误或意外按量扣费。 +- **席位绑定**：每个席位绑定唯一成员，不可共享；回收席位后原 API Key 失效，重新分配将生成新 Key。 +- **额度重置规则**：坐席月度 Credits 在订阅周期结束时重置，**不累积**；共享用量包有效期为 1 个月，到期自动清零，不随坐席周期重置。 +- **错误处理**：常见报错如 `404 model 'xxx' not found` 表示模型 ID 不在白名单或拼写错误；`401 InvalidApiKey` 通常因误用通用 API Key（`sk-xxx`）或 Base URL 不匹配导致，需核对 [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) 中的配置指引。 ## 来源文档 -- [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) - [团队管理](../../raw/model-user-guide/token-plan-guide/token-plan-team.md) +- [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) - [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md) +- [添加视觉理解能力](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) +- [联网搜索](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) +- [常见问题](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) - [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) - [Coding Plan概述](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) -- [联网搜索](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) -- [添加视觉理解能力](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) -- [常见问题](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md index b7f1bfd0..ba871901 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md @@ -1,69 +1,71 @@ # use cases -百炼平台的 use cases 覆盖从多模态内容生成、智能体与工作流构建，到深度研究、教育辅助及第三方模型集成等核心场景。这些用例均基于百炼统一 API 与模型服务层，支持开发者通过标准化接口快速落地生产级应用，无需关注底层基础设施运维。所有方案均提供开箱即用的部署路径与明确的成本预估。 +百炼平台提供覆盖文本、图像、视频、多模态及智能体工作流的全栈AI能力，支持从Prompt工程、模型调用、RAG构建到端到端应用部署的完整开发链路。开发者可基于预置模型快速验证场景，也可通过自定义训练、缓存优化与限流治理实现生产级落地。 ## 支持的模型/功能 -百炼提供两类核心能力：**阿里云自研模型**（如 Qwen 系列、Wan2.7、HappyHorse、Qwen3-VL）和**第三方直供模型**（如 DeepSeek、Kimi、GLM、MiniMax、MiMo、Stepfun、Vidu）。 -- **视觉生成**：万相（文生图 V1/V2、文生视频、图生视频）、HappyHorse（视频生成）、Vidu（视频生成）支持结构化提示词控制，覆盖主体、场景、运动、运镜、风格等维度 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)。 -- **多模态理解与生成**：Qwen3-VL 系列模型支撑解题与批改场景，具备 MathVista、MMMU 等权威评测 SOTA 能力 [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md)。 -- **深度推理与研究**：Qwen-Deep-Research 模型实现自动路径规划、多源交叉验证与结构化报告生成 [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md)。 -- **第三方模型集成**：DeepSeek（v3/v4-pro）、Kimi（k2.6/k2.7-code）、GLM（5.2）、MiniMax（M2.5/M2.7）、MiMo（v2.5-pro）、Stepfun（step-3.7-flash）均通过 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)或 DashScope SDK 接入，支持 `enable_thinking` 等非标参数控制推理模式。 +百炼支持多类原生与第三方大模型，并提供配套的视觉生成、深度研究、智能教学等垂直能力套件： -> **注意**：多个第三方模型文档（如 [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md)、[GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md)、[MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md)）均声明部分旧版本模型（如 deepseek-v3、glm-4.6、MiniMax-M2.1）将于 2026 年 7 月 9 日下架，且推荐迁移至 Qwen3 系列。但各文档未统一说明迁移后是否保留原模型特性（如上下文长度、联网搜索），实际选型需以控制台最新模型详情页为准。 +- **文本模型**：Qwen系列（如 `qwen3.7-max`、`qwen3-vl-plus`）、DeepSeek（`deepseek-v4-pro`）、Kimi（`kimi/kimi-k2.6`）、GLM（`ZHIPU/GLM-5.2`）、MiniMax（`MiniMax/MiniMax-M2.7`）、MiMo（`xiaomi/mimo-v2.5-pro`）、Step（`stepfun/step-3.7-flash`）等；所有第三方模型均需注意[下架时间](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md)，例如 `deepseek-v3` 系列将于 2026 年 7 月 9 日下架。 +- **视觉模型**：Wan2.7 图像/视频生成、HappyHorse 视频生成、Qwen3-VL 多模态理解（用于解题批改），详见 [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)。 +- **专用能力套件**： + - 深度研究：`Qwen-Deep-Research` 自动规划检索路径、多源交叉验证并生成结构化报告，适用于投资尽调与战略分析场景 [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md)； + - AI 教学：基于 `qwen3-vl-plus` 实现拍照解题、自动批改与题库生成，支持 33 种语言 [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md)； + - 智能体与工作流：支持 RAG（通过 LlamaIndex 集成知识库）、自主决策 Agent 及复杂对话流编排 [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md)。 + +> **注意**：文档中提及的 `qwen3-vl-plus` 在 [AI 解题 + 批改](../../raw/model-user-guide/use-cases/ai-homework-helper.md) 中明确为视觉模型，但部分第三方集成文档（如 Kimi、MiniMax）未说明其多模态能力支持情况，实际调用前请以控制台模型详情页为准。 ## 关键参数 -- **Prompt 控制**： - - 文生文：推荐使用 [Prompt 框架](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md)（背景/目的/风格/语气/受众/输出），避免模糊指令；平台提供一键优化工具，但会消耗 Token。 - - 文生图/视频：采用分层公式（基础：主体+场景+运动；进阶：主体描述+场景描述+运动描述+美学控制+风格化），支持 `prompt_extend`（V2 默认开启）、`negative_prompt`、`cache_control` 等参数。 -- **思考模式控制**：DeepSeek、Kimi、GLM、MiMo、Stepfun 等模型均支持 `enable_thinking` 参数（OpenAI SDK 需通过 `extra_body` 传入），开启后返回 `reasoning_content` 字段；部分模型（如 MiMo-v2.5-pro）默认开启，GLM-5.2 还支持 `reasoning_effort` 控制深度。 -- **缓存与限流**：显式缓存通过 `cache_control` 标记实现确定性命中；限流应对需结合 `X-DashScope-Wait-Timeout` 请求头（仅对 Traffic Burst 有效）与客户端流控策略。 +不同任务类型依赖特定参数组合，需严格遵循接口规范： + +- **Prompt 工程**： + - 文生图：使用 `prompt`（正向）与 `negative_prompt`（反向），V2 版本支持 `prompt_extend: true` 启用大模型智能扩写 [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md)； + - 文生视频/图生视频：采用结构化公式，如基础公式 `主体 + 场景 + 运动`，进阶公式需补充 `美学控制` 与 `风格化`；多镜头需显式指定 `镜头序号` 与 `时间戳` [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md)； + - Vidu 视频生成：强调句式简洁、避免主体分散，推荐按 `"主体/场景+场景描述+环境描述+艺术风格/媒介"` 结构组织提示词 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)。 + +- **模型推理**： + - 思考模式：DeepSeek、Kimi、GLM、MiMo、Step 等均支持 `enable_thinking` 参数（非 OpenAI 标准），需通过 `extra_body`（Python SDK）或顶层字段（Node.js）传入； + - 缓存控制：Anthropic 协议兼容工具（Claude Code、OpenCode、OpenClaw）默认注入 `cache_control`，支持对 system [prompt](prompt.md) 与最近 user message 进行显式缓存标记 [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)。 ## 使用方式 -1. **模型调用**： - - OpenAI 兼容模式：配置 `base_url`（如 `https://dashscope.aliyuncs.com/compatible-mode/v1`），使用标准 `chat.completions.create` 接口。 - - DashScope 原生模式：直接调用 `text-generation/generation` 或 `multimodal-generation/generation` 端点，需按模型类型选择 HTTP 地址与 SDK 配置。 -2. **工作流编排**： - - 可视化节点编排（如 HappyHorse 无限画布方案）支持拖拽连接文本、图像、视频生成节点 [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)。 - - RAG 应用通过 LlamaIndex 集成百炼知识库服务，使用 `DashScopeCloudIndex` 创建索引，`DashScopeCloudRetriever` 检索，`as_query_engine` 构建问答引擎。 -3. **部署与评测**： - - 自定义模型需完成调优→部署→评测三阶段闭环，部署为独占实例后方可调用；评测支持自动化指标计算与人工模板评估。 +- **快速启动**：所有方案均提供“15 分钟部署”指引，依赖函数计算（FC）或 ECS 构建 Web 服务，开箱即用（如深度研究方案、AI 教学方案）； +- **SDK 调用**： + - OpenAI 兼容模式：统一使用 `base_url="https://dashscope.aliyuncs.com/compatible-mode/v1"`，模型名按供应商前缀格式（如 `siliconflow/deepseek-v3.2`、`kimi/kimi-k2.6`）； + - DashScope 原生 SDK：无需配置 `base_url`（华北2默认），但跨地域需手动设置 `base_http_api_url`（如德国法兰克福需设为 `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1`）； +- **RAG 构建**：通过 LlamaIndex 集成百炼知识库，使用 `DashScopeCloudIndex` 创建索引，`DashScopeCloudRetriever` 获取检索器，`as_query_engine` 绑定 LLM [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md)； +- **文档转视频**：需本地安装 FFmpeg 与 Marp，依赖浏览器渲染生成演示文稿图片，再合成带语音与字幕的最终视频 [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md)。 ## 限制和注意事项 -- **地域与权限约束**：多数第三方模型（DeepSeek-硅基流动、Kimi、GLM-智谱、MiniMax、MiMo、Stepfun）仅支持华北2（北京）地域，且需对应地域的 API Key；部分模型（如 Kimi、GLM）在新加坡/东京等地域需配置 `WorkspaceId` 域名。 -- **输入输出限制**： - - DashScopeParse 文档解析支持单文件 ≤100MB、≤1000 页；函数计算部署方案有内存与超时限制（如深度研究方案为 15 分钟）。 - - Vidu 视频生成对提示词复杂度敏感，需避免主体物过多或句式模糊；万相图生视频需注意原始图片与运动描述的逻辑一致性（如火车方向需通过比例关系强化）。 -- **成本与计费**： - - 显式缓存首次写入产生 25% 额外开销，但后续命中可降本 90%；若未发生命中，总体成本高于不启用缓存。 - - 第三方模型调用按 Token 计费，部分模型（如 GLM 系列）提供 100 万免费 Token，但需注意免费额度是否跨模型共享。 -- **兼容性风险**：`enable_thinking` 等非 OpenAI 标准参数在不同 SDK 实现中存在差异（如 Python SDK 用 `extra_body`，Node.js SDK 作顶层参数），需严格参照各模型文档示例代码。 +- **限流策略**：百炼 API 按主账号维度、按模型独立限制 RPM/TPM、RPS/TPS 及 Traffic Burst。突发流量推荐优先启用 `X-DashScope-Wait-Timeout` 请求头实现服务端排队，而非简单重试 [限流应对最佳实践](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md)； +- **地域约束**：第三方模型（DeepSeek、Kimi、GLM、MiniMax、MiMo、Step）多数仅在华北2（北京）可用，开通服务与获取 API Key 必须匹配该地域；部分模型（如 Kimi、GLM）在新加坡、美国等地域亦支持，但需替换 `WorkspaceId` 并配置对应 `base_url`； +- **缓存成本**：显式缓存首次写入产生标准价格 25% 额外开销，后续命中节省 90% 成本；但 Claude Code 默认在 system [prompt](prompt.md) 中嵌入动态信息（如当前目录、日期），会降低跨会话命中率，建议启动时加 `--exclude-dynamic-system-prompt-sections` 参数 [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)； +- **模型生命周期**：多个第三方模型（DeepSeek、Kimi、GLM、MiniMax）已明确标注下架时间（2026年7月9日），文档中均给出迁移建议（统一推荐 `qwen3.7-plus` / `qwen3.7-max` / `qwen3.6-flash`），开发者应提前规划升级路径。 ## 来源文档 - [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md) - [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) - [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md) -- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md) -- [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md) +- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md) - [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](../../raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) -- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) - [限流应对最佳实践 ](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md) +- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) +- [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md) - [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) - [DeepSeek-硅基流动](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) -- [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) +- [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) -- [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [GLM-智谱](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) +- [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) - [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md index 758180af..4cdf2149 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md @@ -1,59 +1,64 @@ # use chat client or development tool -阿里云百炼支持通过多种主流 AI 开发工具和客户端接入模型服务，包括终端 CLI 工具（如 Hermes Agent、Qwen Code）、桌面 IDE（如 Cursor、Qoder CN）、开源平台（如 Dify）以及通用 HTTP 客户端（如 Postman）。所有工具均通过 OpenAI 或 Anthropic 兼容 API 协议对接，开发者可根据使用场景选择按量计费、Coding Plan 或 Token Plan 团队版三种计费方案。 +阿里云百炼平台支持通过多种主流 AI 开发工具和客户端接入模型服务，包括终端 CLI 工具（如 Hermes Agent、Qwen Code）、IDE [插件](../concepts/plugin.md)（如 Cline、Kilo CLI）、桌面应用（如 Cursor、Cherry Studio）以及开源平台（如 Dify、Qoder）。所有工具均通过 OpenAI 或 Anthropic 兼容协议对接，开发者可基于自身工作流选择合适工具，并按 [Token](../concepts/token.md) Plan 团队版、Coding Plan 或按量计费三种方案配置凭证。 ## 支持的模型/功能 -百炼支持的模型因计费方案而异，且需匹配对应协议（OpenAI 兼容或 Anthropic 兼容）与 Base URL。核心模型覆盖 Qwen 系列（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）、DeepSeek（如 `deepseek-v4-pro`）、Kimi（如 `kimi-k2.7-code`）、GLM（如 `glm-5.2`）及 MiniMax 等。部分模型（如 Qwen3 系列）支持思考模式（`enable_thinking: true`），需在请求体或配置中显式启用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)。 +百炼支持的模型因计费方案而异，且需匹配对应协议（OpenAI 兼容或 Anthropic 兼容）： -图像与视频生成类模型（如 `wan2.6-t2i`）**不适用**于常规聊天客户端，必须通过异步 API 调用，且仅支持直接 HTTP 请求（cURL/Postman）或 Dify 工作流等支持长轮询的平台 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md)。此外，Token Plan 团队版和 Coding Plan **明确禁止**用于工作流平台（Dify、n8n、Coze）、API 测试工具（Postman、Insomnia）或自定义后端应用——该限制在 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) 中有明确定义。 +- **[Token](../concepts/token.md) Plan 团队版**：支持 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus`、`qwen3.6-flash`、`deepseek-v4-pro`、`deepseek-v4-flash`、`kimi-k2.7-code`、`glm-5.2` 等文本生成模型；部分模型（如 Qwen3 系列）支持 `enable_thinking` 参数开启思考模式。[原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) 明确列出其 [Token](../concepts/token.md) Plan 配置中支持的全部模型 ID 及 thinking 启用方式。 +- **Coding Plan**：主要支持 `qwen3.7-plus`、`qwen3.6-plus` 等，不支持 Qwen3.7-max；仅提供 Chat/Completions API 接入，需使用旧版 Codex（如 0.80.0）[原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) 指出该限制。 +- **按量计费**：覆盖最广，除上述模型外，还支持万相（wanx）系列图像/视频生成模型（如 `wan2.6-t2i`），但需通过异步 API 调用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) 详细说明了该机制。 -> **注意**：文档 1 和文档 2 均列出 `qwen3.6-flash` 为 Token Plan 团队版支持模型，但文档 1 的 JSON 配置中其 `contextWindow` 为 `1000000`，而文档 4（OpenCode）中同模型未声明上下文长度；文档 7（Qwen Code）则明确要求 `qwen3.6-flash` 必须启用 `enable_thinking`。实际行为以控制台公布的[Token Plan 团队版支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)为准，建议以官方模型页描述为最终依据。 +> **注意**：文档间存在模型命名不一致问题。例如 Cursor 要求将 `kimi-k2.6` 写为 `kimi-k2-6`，而 OpenCode 和 Qwen Code 直接使用 `kimi-k2.6`；Dify [插件](../concepts/plugin.md)对 `qwen-turbo` 的权限校验也与官方模型列表存在偏差。实际配置时请以各工具文档的命名要求为准。 ## 关键参数 -所有工具共用三类核心参数： +所有工具共用以下核心参数，但字段名和协议适配方式不同： -- **API Key**：严格按计费方案隔离。Token Plan 团队版、Coding Plan 与按量计费的 API Key **互不通用**，混用将导致 401 错误。 -- **Base URL**：必须与 API Key 所属地域及计费方案完全匹配。例如： - - Token Plan 团队版（北京）：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） - - Coding Plan：`https://coding.dashscope.aliyuncs.com/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） - - 按量计费（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（OpenAI）或 `/apps/anthropic`（Anthropic） -- **Model ID**：部分工具（如 Cursor、Chatbox）要求对带点号的模型名做转换（如 `kimi-k2.6` → `kimi-k2-6`），详见各工具文档；Qwen3 系列模型在启用思考模式时，部分工具（如 Qwen Code）需在 `generationConfig.extra_body` 中设置 `"enable_thinking": true` [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)。 +- **API Key**：必须与所选计费方案严格匹配。Token Plan、Coding Plan 和按量计费的 API Key 互不通用，且按量计费 Key 必须与 Base URL 所在地域一致（如北京 Key 不可用于新加坡 endpoint）。 +- **Base URL**： + - OpenAI 兼容协议：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`（Token Plan）、`https://coding.dashscope.aliyuncs.com/v1`（Coding Plan）、`https://dashscope.aliyuncs.com/compatible-mode/v1`（按量，北京）。 + - Anthropic 兼容协议：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic`（Token Plan）、`https://coding.dashscope.aliyuncs.com/apps/anthropic`（Coding Plan）、`https://dashscope.aliyuncs.com/apps/anthropic`（按量，北京）。 +- **Model ID**：必须从对应方案的支持列表中选取，例如 Token Plan 不支持 `qwen-turbo`，Coding Plan 不支持 `qwen3.7-max`。 +- **高级参数**：`enable_thinking`（Qwen3 系列）、`max_tokens`（用于规避上下文超限）、`CLAUDE_CODE_MAX_CONTEXT_TOKENS`（Claude Code 扩展上下文至 1M）等，需在配置文件或 UI 中显式设置。 ## 使用方式 -1. **安装工具**：各工具提供标准化安装路径，如 `npm install -g`（Hermes Agent、Claude Code）、一键脚本（OpenClaw、QwenPaw）、GUI 下载（Cursor、Cherry Studio）或 VS Code 插件（Cline）。 -2. **配置凭证**：绝大多数工具通过编辑配置文件（如 `~/.hermes/config.yaml`、`~/.qwen/settings.json`）或图形化设置界面完成。环境变量（如 `OPENAI_API_KEY`）在 Codex 等工具中仍被广泛使用。 -3. **验证与调用**：配置后执行简单命令（如 `hermes chat -q "你好"`）或在 GUI 中发送测试消息。对于支持多模型的工具（如 Qoder、Cursor），需在对话界面手动切换模型，且免费版（如 Cursor Free）可能限制自定义模型调用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)。 -4. **高级能力**：部分工具（Qoder、Cline、Cursor）支持通过百炼 CLI 注册 Skills，实现自然语言驱动的代码生成、图像/视频生成等扩展能力，需提前全局安装 `bailian-cli` 并配置 API Key。 +配置流程高度统一，分为三步： + +1. **安装工具**：多数工具提供一键脚本（如 `curl -fsSL ... | bash`）、npm 全局安装（如 `npm install -g opencode-ai`）或 GUI 下载（如 Cursor、Cherry Studio）。 +2. **配置凭证**： + - CLI 工具（Hermes Agent、Qwen Code）通常提供交互式命令（如 `hermes config set` 或 `/auth`）或编辑 JSON/YAML 配置文件（路径见各文档）。 + - IDE [插件](../concepts/plugin.md)（Cline、Kilo CLI）通过图形化设置界面填写 Base URL、API Key 和 Model ID。 + - 桌面应用（Cursor、Chatbox）在 Settings > Models 中添加 OpenAI 兼容 Provider 并填入参数。 +3. **验证与调用**：执行简单命令（如 `hermes chat -q "你好"`）或发送测试消息，观察是否返回有效响应。部分工具（如 Qoder CN）要求先完成账号登录才能启用模型配置。 ## 限制和注意事项 -- **地域强绑定**：按量计费的 API Key 与 Base URL 必须属于同一地域（如北京 Key + 北京 URL），否则报错 401；Token Plan 团队版与 Coding Plan 的 Base URL 固定，无需选择地域。 -- **协议差异**：OpenAI 兼容端点（`/compatible-mode/v1`）接受标准 `/chat/completions` 请求；Anthropic 兼容端点（`/apps/anthropic`）需使用 `/messages` 接口及 `anthropic-messages` 协议，二者不可混用。 -- **免费额度限制**：按量计费新用户享免费额度，但**仅限华北2（北京）地域**的模型生效；使用新加坡或美国地域将立即产生费用 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)。 -- **模型兼容性**：Codex 对不同模型需区分 `wire_api`（`responses` vs `chat`），且仅新版支持 Qwen3 系列；旧版 Codex（v0.80.0）是 Coding Plan 的强制要求 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md)。 -- **违规风险**：将 Token Plan 团队版或 Coding Plan 的 API Key 用于 Dify、Postman 等非授权场景，可能触发订阅暂停或 Key 封禁 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)。 +- **套餐适用范围严格受限**：Token Plan 团队版和 Coding Plan **仅允许用于 AI 编程工具和 OpenClaw 类 Agent**，明确禁止用于 Dify、n8n、Postman 等工作流平台或 API 测试工具 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)。违规使用可能导致订阅暂停或 API Key 封禁。 +- **地域绑定强制**：按量计费的 API Key 与 Base URL 地域必须一致（如北京 Key + 北京 endpoint），否则返回 401 错误；免费额度也仅限华北2（北京）地域生效 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)。 +- **协议与版本兼容性**：Codex 对不同模型需切换 API 协议（Responses API vs Chat API）及版本（0.80.0），Claude Code 需跳过 Anthropic 官方登录验证，Cursor 免费版不支持自定义模型。这些细节均需严格遵循对应工具文档。 +- **模型能力差异**：Qwen3 系列支持思考模式（需 `enable_thinking: true`），而 DeepSeek、GLM 等模型默认不启用；图像/视频生成模型（如 wan2.6-t2i）必须使用异步调用流程，无法通过标准聊天接口直接使用。 ## 来源文档 - [OpenClaw](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) - [Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) -- [Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - [OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) -- [QwenPaw](../../raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) -- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) +- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) +- [QwenPaw](../../raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - [Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) - [Chatbox](../../raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md) -- [Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [Qoder](../../raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) +- [Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [Qoder CN（原 Lingma）](../../raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - [使用Postman或cURL调用图像/视频生成API](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) +- [Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/index.md b/skills/bailian-docs-llm-wiki/wiki/index.md index 08ed3957..584020ff 100644 --- a/skills/bailian-docs-llm-wiki/wiki/index.md +++ b/skills/bailian-docs-llm-wiki/wiki/index.md @@ -42,18 +42,18 @@ - [3d generation](api/3d-generation.md) — 1 篇源文档 - [application call](api/application-call.md) — 5 篇源文档 -- [application component api reference](api/application-component-api-reference.md) — 57 篇源文档 +- [application component api reference](api/application-component-api-reference.md) — 56 篇源文档 - [file management api](api/file-management-api.md) — 1 篇源文档 - [frameworks](api/frameworks.md) — 3 篇源文档 -- [image generation](api/image-generation.md) — 26 篇源文档 +- [image generation](api/image-generation.md) — 25 篇源文档 - [knowledge](api/knowledge.md) — 1 篇源文档 - [long term memory new](api/long-term-memory-new.md) — 1 篇源文档 - [managed agents api](api/managed-agents-api.md) — 7 篇源文档 - [model production](api/model-production.md) — 2 篇源文档 - [more](api/more.md) — 3 篇源文档 -- [more about models](api/more-about-models.md) — 6 篇源文档 +- [more about models](api/more-about-models.md) — 5 篇源文档 - [more models](api/more-models.md) — 6 篇源文档 -- [omni realtime api](api/omni-realtime-api.md) — 6 篇源文档 +- [omni realtime api](api/omni-realtime-api.md) — 5 篇源文档 - [preparations](api/preparations.md) — 4 篇源文档 - [qwen api reference](api/qwen-api-reference.md) — 1 篇源文档 - [toolkits and frameworks](api/toolkits-and-frameworks.md) — 10 篇源文档 @@ -62,20 +62,20 @@ ## 横切概念 -- [OpenAI 兼容接口](concepts/openai-compatible-api.md) — 关联 5 个主题 -- [Token 计量与管理](concepts/token.md) — 关联 5 个主题 +- [OpenAI 兼容接口](concepts/openai-compatible-interface.md) — 关联 5 个主题 +- [Token](concepts/token.md) — 关联 5 个主题 - [函数调用](concepts/function-calling.md) — 关联 4 个主题 -- [多模态能力](concepts/multi-modal.md) — 关联 5 个主题 -- [插件机制](concepts/plugin.md) — 关联 5 个主题 +- [插件](concepts/plugin.md) — 关联 5 个主题 - [检索增强生成](concepts/rag.md) — 关联 5 个主题 +- [模型上下文协议（MCP）](concepts/model-context-protocol.md) — 关联 5 个主题 - [流式输出](concepts/streaming-output.md) — 关联 5 个主题 -- [长期记忆](concepts/long-term-memory.md) — 关联 4 个主题 +- [长期记忆](concepts/long-term-memory.md) — 关联 5 个主题 ## 对比分析 -- [图像、视频与3D生成能力对比](comparisons/image-video-3d-generation.md) — 对比 3 个主题 -- [应用编排能力对比：托管智能体、应用组件与模型上下文协议](comparisons/application-orchestration.md) — 对比 3 个主题 -- [模型评估与监控体系对比](comparisons/model-evaluation-monitoring.md) — 对比 3 个主题 -- [模型部署方案对比：高并发推理、生产部署与压缩优化](comparisons/model-deployment-options.md) — 对比 3 个主题 +- [图像生成与视频生成对比](comparisons/image-vs-video-generation.md) — 对比 2 个主题 +- [应用开发框架对比：Managed Agents、Application Component 与 Toolkits](comparisons/application-frameworks.md) — 对比 3 个主题 +- [模型评估与监控能力对比](comparisons/model-evaluation-monitoring.md) — 对比 3 个主题 +- [模型部署方式对比：高并发推理、模型压缩与模型部署指南](comparisons/model-deployment-options.md) — 对比 3 个主题 - [长期记忆与知识库方案对比](comparisons/memory-solutions.md) — 对比 3 个主题 From f5a5ca1f5b4fdc009c468c1015324559345ddc15 Mon Sep 17 00:00:00 2001 From: bailian-bot Date: Thu, 23 Jul 2026 12:17:22 +0000 Subject: [PATCH 03/13] chore: update bailian-docs-llm-wiki (2026-07-23) --- skills/bailian-docs-llm-wiki/SKILL.md | 17 - skills/bailian-docs-llm-wiki/llms.txt | 985 +++++----- .../models/families.jsonl | 14 +- .../models/groups/Kimi-K2.json | 356 ++-- .../models/groups/MiniMax-M2.1.json | 190 +- .../groups/MiniMax-speech-market-place.json | 72 +- .../models/groups/aitryon-parsing-v1.json | 48 +- .../models/groups/aitryon-plus.json | 35 +- .../models/groups/aitryon-refiner.json | 115 +- .../models/groups/aitryon.json | 35 +- .../groups/animate-anyone-detect-gen2.json | 48 +- .../models/groups/animate-anyone-gen2.json | 48 +- .../groups/animate-anyone-template-gen2.json | 48 +- .../models/groups/cosyvoice.json | 298 +-- .../models/groups/deepseek.json | 382 +--- .../models/groups/embedding.json | 144 +- .../models/groups/emo-detect-v1.json | 48 +- .../models/groups/emo-v1.json | 54 +- .../models/groups/emoji-detect-v1.json | 48 +- .../models/groups/emoji-v1.json | 48 +- .../models/groups/facechain-facedetect.json | 34 +- .../models/groups/facechain-generation.json | 48 +- .../models/groups/farui-plus.json | 64 +- .../models/groups/fun-asr-flash.json | 42 +- .../models/groups/fun-asr-realtime.json | 71 +- .../models/groups/fun-asr.json | 34 +- .../models/groups/fun-music.json | 84 +- .../models/groups/glm-4.5.json | 851 ++++---- .../models/groups/glm-fast.json | 22 +- .../models/groups/gui-plus.json | 58 +- .../models/groups/gummy-chat-v1.json | 16 +- .../models/groups/gummy-realtime-v1.json | 18 +- .../models/groups/happyhorse-i2v.json | 118 +- .../models/groups/happyhorse-r2v.json | 112 +- .../models/groups/happyhorse-t2v.json | 112 +- .../models/groups/happyhorse-video-edit.json | 58 +- .../models/groups/image-erase-completion.json | 34 +- .../groups/image-instance-segmentation.json | 32 +- .../models/groups/image-out-painting.json | 40 +- .../groups/kimi-models-market-place.json | 256 ++- .../groups/kling-models-market-place.json | 196 +- .../models/groups/liveportrait-detect.json | 48 +- .../models/groups/liveportrait.json | 48 +- .../groups/minimax-models-market-place.json | 206 +- .../models/groups/paraformer-8k-v1.json | 22 +- .../models/groups/paraformer-8k-v2.json | 40 +- .../models/groups/paraformer-mtl-v1.json | 22 +- .../groups/paraformer-realtime-8k-v1.json | 22 +- .../groups/paraformer-realtime-8k-v2.json | 14 +- .../models/groups/paraformer-realtime-v1.json | 22 +- .../models/groups/paraformer-realtime-v2.json | 20 +- .../models/groups/paraformer-v1.json | 22 +- .../models/groups/paraformer-v2.json | 20 +- .../groups/pixverse-c1-market-place.json | 336 ++-- .../pixverse-capability-market-place.json | 81 - .../models/groups/pixverse-market-place.json | 328 ++-- .../groups/pixverse-v6-market-place.json | 346 ++-- .../models/groups/qvq-max.json | 60 +- .../models/groups/qvq-plus.json | 62 +- .../groups/qwen-audio-realtime-flash.json | 35 +- .../groups/qwen-audio-realtime-plus.json | 35 +- .../models/groups/qwen-audio-tts.json | 58 +- .../models/groups/qwen-coder-plus.json | 64 +- .../models/groups/qwen-coder-turbo.json | 64 +- .../models/groups/qwen-deep-research.json | 58 +- .../models/groups/qwen-doc-turbo.json | 76 +- .../models/groups/qwen-embedding.json | 318 ++- .../models/groups/qwen-flash-character.json | 64 +- .../models/groups/qwen-flash.json | 267 ++- .../models/groups/qwen-image-2.0-pro.json | 44 +- .../models/groups/qwen-image-2.0.json | 44 +- .../models/groups/qwen-image-3.0-pro.json | 53 + .../models/groups/qwen-image-edit-max.json | 32 +- .../models/groups/qwen-image-edit.json | 62 +- .../models/groups/qwen-image-max.json | 40 +- .../models/groups/qwen-image-plus.json | 80 +- .../models/groups/qwen-long.json | 146 +- .../models/groups/qwen-math-plus.json | 256 ++- .../models/groups/qwen-math-turbo.json | 62 +- .../models/groups/qwen-max.json | 87 +- .../models/groups/qwen-mt-flash.json | 39 +- .../models/groups/qwen-mt-image.json | 47 +- .../models/groups/qwen-mt-lite.json | 39 +- .../models/groups/qwen-mt-plus.json | 39 +- .../models/groups/qwen-mt-turbo.json | 45 +- .../groups/qwen-omni-turbo-realtime.json | 160 +- .../models/groups/qwen-omni-turbo.json | 218 ++- .../models/groups/qwen-plus-character.json | 64 +- .../models/groups/qwen-plus.json | 806 ++++++-- .../models/groups/qwen-rerank.json | 135 +- .../models/groups/qwen-tts-realtime.json | 100 +- .../models/groups/qwen-tts.json | 48 +- .../models/groups/qwen-turbo.json | 148 +- .../models/groups/qwen-vl-embedding.json | 126 +- .../models/groups/qwen-vl-max.json | 92 +- .../models/groups/qwen-vl-ocr.json | 224 ++- .../models/groups/qwen-vl-plus.json | 98 +- .../models/groups/qwen-voice-design.json | 30 +- .../models/groups/qwen-voice-enrollment.json | 24 +- .../models/groups/qwen2.5.json | 82 +- .../groups/qwen3-asr-flash-filetrans.json | 46 +- .../groups/qwen3-asr-flash-realtime.json | 40 +- .../models/groups/qwen3-asr-flash.json | 19 +- .../groups/qwen3-coder-30b-a3b-instruct.json | 127 +- .../qwen3-coder-480b-a35b-instruct.json | 126 +- .../models/groups/qwen3-coder-flash.json | 200 +- .../models/groups/qwen3-coder-plus.json | 204 +- .../qwen3-livetranslate-flash-realtime.json | 66 +- .../groups/qwen3-livetranslate-flash.json | 65 +- .../models/groups/qwen3-max.json | 519 +++-- .../groups/qwen3-omni-30b-a3b-captioner.json | 54 +- .../groups/qwen3-omni-flash-realtime.json | 88 +- .../models/groups/qwen3-omni-flash.json | 114 +- .../groups/qwen3-tts-flash-realtime.json | 20 +- .../models/groups/qwen3-tts-flash.json | 20 +- .../qwen3-tts-instruct-flash-realtime.json | 46 +- .../groups/qwen3-tts-instruct-flash.json | 55 +- .../models/groups/qwen3-tts-vc-realtime.json | 30 +- .../models/groups/qwen3-tts-vc.json | 28 +- .../models/groups/qwen3-tts-vd-realtime.json | 50 +- .../models/groups/qwen3-tts-vd.json | 30 +- .../models/groups/qwen3-vl-flash.json | 271 ++- .../models/groups/qwen3-vl-plus.json | 267 ++- .../models/groups/qwen3.5-flash.json | 328 +++- .../qwen3.5-livetranslate-flash-realtime.json | 60 +- .../models/groups/qwen3.5-ocr.json | 58 +- .../groups/qwen3.5-omni-flash-realtime.json | 78 +- .../models/groups/qwen3.5-omni-flash.json | 82 +- .../groups/qwen3.5-omni-plus-realtime.json | 78 +- .../models/groups/qwen3.5-omni-plus.json | 125 +- .../models/groups/qwen3.5-plus.json | 323 +++- .../models/groups/qwen3.5.json | 837 ++++---- .../models/groups/qwen3.6-flash.json | 262 ++- .../models/groups/qwen3.6-max.json | 144 +- .../models/groups/qwen3.6-plus.json | 266 ++- .../models/groups/qwen3.6.json | 256 ++- .../models/groups/qwen3.7-max.json | 301 +-- .../models/groups/qwen3.7-plus.json | 282 ++- .../models/groups/qwen3.json | 1641 ++++++---------- .../models/groups/qwq-plus.json | 74 +- .../models/groups/sambert.json | 1720 ++++------------- .../models/groups/shoemodel-v1.json | 34 +- .../models/groups/siliconflow-models.json | 210 +- .../models/groups/speech-biasing.json | 34 +- .../groups/stepfun-models-market-place.json | 59 +- .../groups/tongyi-intent-detect-v3.json | 58 +- .../groups/tongyi-xiaomi-analysis-flash.json | 58 +- .../groups/tongyi-xiaomi-analysis-pro.json | 58 +- .../groups/tripo-models-market-place.json | 172 -- .../groups/vanchin-models-market-place.json | 194 -- .../models/groups/video-style-transform.json | 48 +- .../models/groups/videoretalk.json | 42 +- .../vidu-image-models-market-place.json | 108 -- .../groups/vidu-models-market-place.json | 550 ------ .../models/groups/virtualmodel-v2.json | 34 +- .../models/groups/voice-enrollment.json | 8 +- .../models/groups/wan-image-edit.json | 231 +-- .../models/groups/wan-image-to-video.json | 811 +++----- .../models/groups/wan-reference-to-video.json | 99 +- .../models/groups/wan-text-to-image.json | 447 +---- .../models/groups/wan-text-to-video.json | 194 +- .../models/groups/wan-video-edit.json | 57 +- .../groups/wanx-background-generation-v2.json | 48 +- .../groups/wanx-poster-generation-v1.json | 34 +- .../groups/wanx-sketch-to-image-lite.json | 46 +- .../models/groups/wanx-style-repaint-v1.json | 42 +- .../models/groups/wanx-virtualmodel.json | 34 +- .../models/groups/wanx-x-painting.json | 38 +- .../models/groups/wanx2.1-vace-plus.json | 46 +- .../models/groups/wordart-semantic.json | 48 +- .../models/groups/wordart-texture.json | 48 +- .../groups/xiaomi-models-market-place.json | 54 +- .../models/groups/z-image-turbo.json | 42 +- .../groups/zhipu-models-market-place.json | 148 +- .../bailian-docs-llm-wiki/models/index.json | 72 +- skills/bailian-docs-llm-wiki/models/index.md | 15 +- .../bailian-docs-llm-wiki/models/models.jsonl | 581 +++--- .../api-bailian-2023-12-29-createindex.md | 4 +- .../api-bailian-2023-12-29-addchunk.md | 233 +++ .../api-bailian-2023-12-29-overview.md | 6 + .../api-bailian-2023-12-29-ram.md | 764 ++++++-- .../multimodal-products/product-billing.md | 2 +- ...pi-contactcenterai-2024-06-03-changeset.md | 0 ...centerai-2024-06-03-analyzeconversation.md | 0 ...contactcenterai-2024-06-03-analyzeimage.md | 0 ...i-contactcenterai-2024-06-03-createtask.md | 0 ...-contactcenterai-2024-06-03-createvocab.md | 0 ...-contactcenterai-2024-06-03-deletevocab.md | 0 ...api-contactcenterai-2024-06-03-getvocab.md | 0 ...pi-contactcenterai-2024-06-03-listvocab.md | 0 ...-contactcenterai-2024-06-03-updatevocab.md | 0 ...actcenterai-2024-06-03-analyzeaudiosync.md | 0 ...centerai-2024-06-03-generalanalyzeimage.md | 0 ...ontactcenterai-2024-06-03-gettaskresult.md | 0 ...ontactcenterai-2024-06-03-runcompletion.md | 0 ...enterai-2024-06-03-runcompletionmessage.md | 0 ...api-contactcenterai-2024-06-03-endpoint.md | 0 ...ces-for-automatic-work-order-generation.md | 2 +- ...rvice-quality-inspection-best-practices.md | 2 +- .../best-practices/summary-best-practices.md | 2 +- ...lysis-aio-application-through-task-type.md | 2 +- ...-ccai-dialogue-analysis-aio-application.md | 2 +- ...-style-request-body-signature-mechanism.md | 0 ...analysis-by-uploading-offline-task-data.md | 2 +- ...e-ram-users-for-ccai-dialogue-analysis.md} | 0 ...e-prompt-to-call-tongyi-xiaomi-ccai-aio.md | 2 +- ...plate-id-to-call-tongyi-xiaomi-ccai-aio.md | 2 +- .../product-activation.md | 2 +- .../technology-integration-scheme.md | 0 .../tongyi-xiaomi-ccai-update-announcement.md | 0 ...pi-bailianvoicebot-2025-01-01-changeset.md | 0 ...ailianvoicebot-2025-01-01-bridgewebcall.md | 0 ...anvoicebot-2025-01-01-createapplication.md | 0 ...bot-2025-01-01-createapplicationversion.md | 0 ...anvoicebot-2025-01-01-deleteapplication.md | 0 ...ilianvoicebot-2025-01-01-getapplication.md | 0 ...ianvoicebot-2025-01-01-listapplications.md | 0 ...oicebot-2025-01-01-listbackgroundmusics.md | 0 ...ailianvoicebot-2025-01-01-listnlumodels.md | 0 ...i-bailianvoicebot-2025-01-01-listvoices.md | 0 ...bailianvoicebot-2025-01-01-previewvoice.md | 0 ...ot-2025-01-01-publishapplicationversion.md | 0 ...anvoicebot-2025-01-01-updateapplication.md | 0 ...bot-2025-01-01-updateapplicationversion.md | 0 ...ianvoicebot-2025-01-01-createclonevoice.md | 0 ...ianvoicebot-2025-01-01-deleteclonevoice.md | 0 ...ilianvoicebot-2025-01-01-listclonevoice.md | 0 ...oicebot-2025-01-01-listclonevoicemodels.md | 0 ...ianvoicebot-2025-01-01-updateclonevoice.md | 0 ...ianvoicebot-2025-01-01-createvocabulary.md | 0 ...ianvoicebot-2025-01-01-deletevocabulary.md | 0 ...ianvoicebot-2025-01-01-exportvocabulary.md | 0 ...ailianvoicebot-2025-01-01-getvocabulary.md | 0 ...ianvoicebot-2025-01-01-importvocabulary.md | 0 ...ilianvoicebot-2025-01-01-listvocabulary.md | 0 ...ianvoicebot-2025-01-01-updatevocabulary.md | 0 ...voicebot-2025-01-01-disablesubscription.md | 0 ...lianvoicebot-2025-01-01-getsubscription.md | 0 ...nvoicebot-2025-01-01-updatesubscription.md | 0 ...bot-2025-01-01-createvoiceaccessprofile.md | 0 ...bot-2025-01-01-deletevoiceaccessprofile.md | 0 ...cebot-2025-01-01-listvoiceaccessprofile.md | 0 ...ianvoicebot-2025-01-01-listvoiceengines.md | 0 ...bot-2025-01-01-updatevoiceaccessprofile.md | 0 ...ilianvoicebot-2025-01-01-createvariable.md | 0 ...ilianvoicebot-2025-01-01-deletevariable.md | 0 ...bailianvoicebot-2025-01-01-listvariable.md | 0 ...ilianvoicebot-2025-01-01-updatevariable.md | 0 ...bot-2025-01-01-generatefileuploadparams.md | 0 ...bot-2025-01-01-getdatachannelcredential.md | 0 ...api-bailianvoicebot-2025-01-01-overview.md | 0 .../api-bailianvoicebot-2025-01-01-ram.md | 0 .../product-0verview.md | 0 .../api-anytrans-2025-07-07-changeset.md | 6 +- .../api-anytrans-2025-07-07-ram.md | 2 +- .../guidelines-for-use.md | 179 -- ...api-contactcenterai-2024-06-03-overview.md | 117 -- ...ces-for-automatic-work-order-generation.md | 141 -- ...rvice-quality-inspection-best-practices.md | 138 -- .../best-practices/summary-best-practices.md | 118 -- ...lysis-aio-application-through-task-type.md | 563 ------ ...-ccai-dialogue-analysis-aio-application.md | 225 --- ...analysis-by-uploading-offline-task-data.md | 197 -- ...e-prompt-to-call-tongyi-xiaomi-ccai-aio.md | 385 ---- ...plate-id-to-call-tongyi-xiaomi-ccai-aio.md | 404 ---- .../product-overview-1.md | 67 - .../application-management.md | 38 - ...an-application-based-on-a-custom-method.md | 60 - ...cation-based-on-dialogue-analysis-agent.md | 140 -- .../hot-phrase-management.md | 48 - .../using-the-knowledge-base.md | 74 - .../operation-guide.md | 85 - .../api-bailianchatbot-2024-11-05-ssechat.md | 0 .../api-bailianchatbot-2024-11-05-overview.md | 0 .../api-bailianchatbot-2024-11-05-ram.md | 0 ...orkflow-integration-video-understanding.md | 4 +- ...api-aimiaobi-2023-08-01-cancelaudittask.md | 0 .../api-aimiaobi-2023-08-01-queryaudittask.md | 0 ...api-aimiaobi-2023-08-01-submitaudittask.md | 0 .../api-aimiaobi-2023-08-01-overview.md | 36 +- .../application-after-sales-service-scope.md | 57 + .../application-related-agreements.md | 2 +- .../billing-for-knowledge-base.md | 92 +- .../knowledge-base/rag-knowledge-base.md | 42 +- .../application-introduction.md | 7 +- .../new-single-agent-application.md | 16 +- .../tripo-3d-generation-api-reference.md | 23 +- .../fun-asr-realtime-python-sdk.md | 10 +- ...recorded-speech-recognition-android-sdk.md | 12 +- ...sr-recorded-speech-recognition-http-api.md | 922 +-------- ...asr-recorded-speech-recognition-ios-sdk.md | 10 +- ...sr-recorded-speech-recognition-java-sdk.md | 28 +- ...-recorded-speech-recognition-python-sdk.md | 28 +- ...me-speech-recognition-for-fun-asr-flash.md | 561 ++++++ ...speech-recognition-for-fun-asr-realtime.md | 328 ++++ ...r-real-time-speech-recognition-java-sdk.md | 20 +- ...real-time-speech-recognition-python-sdk.md | 30 +- ...recorded-speech-recognition-android-sdk.md | 15 +- ...mer-recorded-speech-recognition-ios-sdk.md | 10 +- ...er-recorded-speech-recognition-java-sdk.md | 28 +- ...-recorded-speech-recognition-python-sdk.md | 28 +- ...recorded-speech-recognition-restful-api.md | 20 +- .../qwen-asr-api-reference.md | 86 +- .../qwen-asr-realtime-interaction-process.md | 8 +- .../cosyvoice-android-sdk.md | 15 +- .../cosyvoice-client-events.md | 49 +- .../cosyvoice-ios-sdk.md | 14 +- .../cosyvoice-java-sdk.md | 59 +- .../cosyvoice-python-sdk.md | 77 +- .../cosyvoice-websocket-api.md | 2 +- .../cosyvoice-tts-http-api.md | 20 +- .../cosyvoice-tts-java-sdk.md | 66 +- .../cosyvoice-tts-python-sdk.md | 20 +- .../voice-clone-design-http-api.md | 18 +- .../sound-reengraving/voice-clone-java-sdk.md | 109 +- .../voice-clone-python-sdk.md | 18 +- .../fun-audiochat-client-events.md | 2 +- ...> fun-audiochat-realtime-websocket-api.md} | 14 +- .../qwen-audio-realtime-server-events.md | 2 +- .../creative-poster-generation-api.md | 1 - ...anx-background-generation-api-reference.md | 12 +- ...ge-generation-and-editing-api-reference.md | 459 +++++ .../omni-realtime-api/client-events.md | 6 +- .../omni-realtime-java-sdk.md | 6 +- .../omni-realtime-python-sdk.md | 18 +- .../omni-realtime-api/server-events.md | 4 +- .../preparations/error-code.md | 20 +- .../preparations/get-api-key.md | 2 +- .../realtime-api-aoq-sdk-desc.md | 17 + .../aoq-audio-features.md | 1179 +++++++++++ .../aoq-connection-management.md | 171 ++ .../aoq-custom-audio-capture.md | 253 +++ .../aoq-custom-audio-playback.md | 235 +++ .../aoq-custom-video-input.md | 647 +++++++ .../aoq-media-stream-control.md | 115 ++ .../aoq-video-features.md | 1186 ++++++++++++ .../best-practice-aoq-omni-realtime.md | 520 +++++ .../best-practice-webrtc-multimodal-dialog.md | 512 +++++ .../best-practice-webrtc-omni-realtime.md | 388 ++++ .../realtime-api-overview.md | 180 ++ .../realtime-connect-model.md | 291 +++ .../realtime-sdk-download.md | 60 + .../realtime-token-authentication.md | 358 ++++ ...compatibility-with-openai-responses-api.md | 36 +- .../openai-compatible-conversations.md | 6 +- .../text-embedding-batch-api.md | 32 +- ...happyhorse-image-to-video-api-reference.md | 25 +- ...yhorse-reference-to-video-api-reference.md | 19 +- .../happyhorse-text-to-video-api-reference.md | 19 +- .../happyhorse-video-edit-api-reference.md | 19 +- .../kling-video-generation-api-reference.md | 23 +- .../pixverse-image-to-video-api-reference.md | 25 +- ...ixverse-keyframe-to-video-api-reference.md | 19 +- .../pixverse-lipsync-api-reference.md | 12 +- .../pixverse-motioncontrol-api-reference.md | 9 +- ...xverse-reference-to-video-api-reference.md | 14 +- .../pixverse-text-to-video-api-reference.md | 43 +- .../pixverse-upscale-api-reference.md | 9 +- .../vidu-image-to-video-api-reference.md | 12 +- .../vidu-keyframe-to-video-api-reference.md | 12 +- .../vidu-reference-to-video-api-reference.md | 24 +- .../vidu-text-to-video-api-reference.md | 32 +- .../image-to-video-general-api-reference.md | 92 +- .../legacy-image-to-video-api-reference.md | 95 +- ...o-by-first-and-last-frame-api-reference.md | 55 +- ...cy-wan-reference-to-video-api-reference.md | 65 +- .../legacy-wan-text-to-video-api-reference.md | 76 +- .../legacy-wanx-vace-api-reference.md | 41 +- .../text-to-video-api-reference.md | 2 +- .../wan-api-reference/wan-animate-mix-api.md | 10 +- .../wan-api-reference/wan-animate-move-api.md | 4 +- .../wan-video-editing-api-reference.md | 49 +- .../model-training-on-console.md | 20 +- .../model-training-overview.md | 2 +- .../wan-image-generation-finetune-guide.md | 138 ++ .../wan-video-generation-finetune-guide.md | 148 +- .../get-started-with-models/models.md | 254 +-- .../get-started-with-models/rate-limit.md | 112 +- .../get-started-with-models/regions.md | 12 +- .../model-data-overview/model-log-backflow.md | 330 ++++ .../model-deployment-introduction.md | 296 +-- .../ptu-long-input-and-cache.md | 30 +- .../model-experience/image-model.md | 22 +- .../model-experience/s2s-model.md | 116 +- .../model-experience/text-generation-model.md | 8 +- .../model-experience/tts-model.md | 10 +- .../tpm-reservation.md | 25 +- .../model-monitoring/model-telemetry.md | 54 +- .../release-notes/model-release-notes.md | 28 +- .../release-notes/newly-released-models.md | 46 +- .../support/after-sales-service-scope.md | 57 + .../faq-about-alibaba-cloud-model-studio.md | 4 + .../support/related-agreements.md | 2 +- .../test-1/bill-query-and-cost-management.md | 60 +- .../model-user-guide/test-1/model-pricing.md | 274 +-- .../model-training-and-deployment-billing.md | 282 ++- .../model-user-guide/test-1/new-free-quota.md | 9 +- .../savings-plan-and-resource-package.md | 22 +- .../coding-plan-guide/coding-plan.md | 34 +- .../add-vision-skill.md | 57 +- .../token-plan-harness-tool.md | 79 + .../token-plan-multimodal-gen.md | 78 +- .../token-plan-tool.md | 253 --- .../web-search-mcp.md} | 196 +- .../token-plan-guide/token-plan-faq.md | 292 --- .../token-plan-guide/token-plan-overview.md | 307 +-- .../token-plan-personal-faq.md | 172 ++ .../token-plan-personal-overview.md | 234 +++ .../token-plan-personal-quick-start.md | 81 + .../token-plan-team-faq.md | 205 ++ .../token-plan-team-management.md | 183 ++ .../token-plan-team-overview.md | 314 +++ .../token-plan-team-quickstart.md} | 29 +- .../token-plan-guide/token-plan-team.md | 272 --- .../deepseek-api.md | 8 +- .../kimi-api-by-moonshot-ai.md | 156 +- .../chatbox.md | 28 +- .../cherry-studio.md | 28 +- .../claude-code.md | 98 +- .../cline.md | 32 +- .../codex.md | 209 +- .../cursor.md | 22 +- .../hermes-agent.md | 59 +- .../kilo-cli.md | 114 +- .../lingma-agent.md | 16 +- .../more-tools.md | 26 +- .../openclaw.md | 185 +- .../opencode.md | 136 +- .../qoder-agent.md | 22 +- .../qwen-code.md | 128 +- .../qwenpaw.md | 24 +- .../wiki/api/3d-generation.md | 136 +- .../wiki/api/application-call.md | 198 +- .../application-component-api-reference.md | 246 +-- .../wiki/api/file-management-api.md | 50 +- .../wiki/api/frameworks.md | 169 +- .../wiki/api/image-generation.md | 114 +- .../wiki/api/knowledge.md | 68 +- .../wiki/api/long-term-memory-new.md | 255 +-- .../wiki/api/managed-agents-api.md | 187 +- .../wiki/api/model-production.md | 56 +- .../wiki/api/more-about-models.md | 140 +- .../wiki/api/more-models.md | 179 +- skills/bailian-docs-llm-wiki/wiki/api/more.md | 150 +- .../wiki/api/omni-realtime-api.md | 204 +- .../wiki/api/preparations.md | 96 +- .../wiki/api/qwen-api-reference.md | 47 +- .../wiki/api/realtime-api-user-guide.md | 74 + .../wiki/api/toolkits-and-frameworks.md | 163 +- .../wiki/api/vector-and-sort.md | 113 +- .../wiki/api/video-generation-api.md | 152 +- .../comparisons/api-invocation-comparison.md | 75 - .../comparisons/app-eval-vs-model-eval.md | 70 - .../app-evaluation-vs-monitoring.md | 65 - .../app-monitoring-vs-model-monitoring.md | 70 - .../comparisons/app-vs-model-evaluation.md | 64 - .../comparisons/app-vs-model-monitoring.md | 66 - .../application-call-vs-bailian-calling.md | 45 - .../application-calling-comparison.md | 57 - .../application-evaluation-vs-monitoring.md | 45 - .../application-frameworks-comparison.md | 69 + .../application-observation-vs-evaluation.md | 59 - .../application-vs-managed-agents.md | 56 - .../data-sources-for-rag-comparison.md | 61 - .../evaluation-app-vs-model-comparison.md | 52 - .../wiki/comparisons/evaluation-compare.md | 50 - .../wiki/comparisons/evaluation-comparison.md | 71 - .../extension-framework-comparison.md | 38 - ...ine-tuning-vs-compression-vs-deployment.md | 73 - .../comparisons/fine-tuning-vs-compression.md | 55 - .../fine-tuning-vs-model-compression.md | 55 - .../comparisons/generation-apis-comparison.md | 65 + .../generation-modalities-comparison.md | 54 - .../comparisons/getting-started-comparison.md | 60 - .../image-video-3d-generation-comparison.md | 78 - .../image-vs-video-vs-3d-generation.md | 52 - .../knowledge-base-vs-memory-library.md | 96 +- .../knowledge-data-management-comparison.md | 78 - .../comparisons/knowledge-memory-compare.md | 69 - .../knowledge-memory-comparison.md | 59 - .../wiki/comparisons/knowledge-vs-memory.md | 63 - .../managed-agents-guide-vs-api.md | 56 - .../managed-agents-vs-llm-application.md | 58 - .../comparisons/media-generation-compare.md | 48 - .../media-generation-comparison.md | 72 - .../memory-and-knowledge-solutions.md | 69 - .../memory-approaches-comparison.md | 59 - .../comparisons/memory-feature-comparison.md | 43 - .../model-call-vs-application-call.md | 43 - .../model-customization-methods-comparison.md | 64 - .../model-customization-options.md | 43 - .../comparisons/model-deployment-options.md | 57 + ...odel-deployment-vs-high-speed-inference.md | 45 - .../model-deployment-vs-inference.md | 51 - ...el-evaluation-vs-application-evaluation.md | 62 - .../model-evaluation-vs-monitoring.md | 60 - ...-experience-vs-production-vs-deployment.md | 55 - .../comparisons/model-lifecycle-comparison.md | 69 - ...el-monitoring-vs-application-monitoring.md | 54 - .../comparisons/model-optimization-compare.md | 42 - .../model-optimization-comparison.md | 85 - .../model-optimization-methods-comparison.md | 69 - .../monitoring-app-vs-model-comparison.md | 51 - .../wiki/comparisons/monitoring-compare.md | 60 - .../wiki/comparisons/monitoring-comparison.md | 57 - .../multimodal-generation-api-comparison.md | 76 - .../multimodal-generation-comparison.md | 71 - .../qwen-api-vs-app-call-vs-agents-api.md | 74 - ...-api-vs-omni-realtime-vs-managed-agents.md | 57 - .../comparisons/realtime-api-comparison.md | 70 + .../wiki/concepts/agent-application.md | 81 - .../wiki/concepts/agent-orchestration.md | 90 - .../wiki/concepts/api-key.md | 80 - .../wiki/concepts/async-invocation.md | 53 - .../wiki/concepts/billing.md | 85 - .../wiki/concepts/context-window.md | 32 - .../wiki/concepts/cross-session-memory.md | 112 -- .../wiki/concepts/dashscope-sdk.md | 99 - .../wiki/concepts/embedding.md | 54 - .../wiki/concepts/evaluation.md | 77 - .../concepts/fine-tuning-and-deployment.md | 92 - .../wiki/concepts/fine-tuning.md | 89 - .../wiki/concepts/function-calling.md | 53 +- .../wiki/concepts/knowledge-base.md | 87 - .../wiki/concepts/long-context.md | 45 - .../wiki/concepts/long-term-memory.md | 47 + .../wiki/concepts/mcp-and-tools.md | 94 - .../wiki/concepts/model-deployment.md | 144 +- .../wiki/concepts/model-selection.md | 83 - .../wiki/concepts/multi-modal.md | 71 + .../wiki/concepts/multimodal.md | 45 - .../concepts/openai-compatible-interface.md | 123 +- .../wiki/concepts/prompt-caching.md | 59 - .../wiki/concepts/prompt-engineering.md | 85 +- .../wiki/concepts/rag.md | 72 +- .../wiki/concepts/rate-limiting.md | 71 - .../wiki/concepts/region.md | 80 - .../wiki/concepts/rerank.md | 61 - .../wiki/concepts/sdk.md | 54 - .../wiki/concepts/security-and-compliance.md | 51 + .../wiki/concepts/streaming-output.md | 55 +- .../wiki/concepts/streaming.md | 47 - .../wiki/concepts/token-and-billing.md | 57 - .../wiki/concepts/token-billing.md | 69 - .../wiki/concepts/token.md | 73 +- .../wiki/concepts/vector-embedding.md | 57 - .../wiki/concepts/vpc-private-access.md | 55 - .../wiki/concepts/workflow.md | 75 - .../wiki/concepts/workspace.md | 61 - .../wiki/guides/application-evaluation.md | 177 +- .../wiki/guides/application-monitoring.md | 138 +- .../application-permission-management.md | 149 +- .../application-publishing-and-sharing.md | 125 +- .../wiki/guides/application-support.md | 98 +- .../wiki/guides/application-use-cases.md | 130 +- .../guides/bailian-application-calling.md | 238 +-- .../wiki/guides/data-connection-overview.md | 92 +- .../wiki/guides/fine-tuning.md | 97 +- .../wiki/guides/get-started-with-models.md | 115 +- .../wiki/guides/knowledge-base.md | 94 +- .../wiki/guides/llm-application.md | 138 +- .../wiki/guides/managed-agents.md | 84 +- .../wiki/guides/memory-library-overview.md | 184 +- .../wiki/guides/model-compression.md | 83 +- .../wiki/guides/model-context-protocol.md | 92 +- .../wiki/guides/model-data-overview.md | 191 +- .../wiki/guides/model-deployment-1.md | 110 +- .../guides/model-evaluation-introduction.md | 136 +- .../wiki/guides/model-experience.md | 114 +- .../wiki/guides/model-high-speed-inference.md | 112 +- .../wiki/guides/model-monitoring.md | 114 +- .../wiki/guides/plug-in.md | 162 +- .../wiki/guides/prompt.md | 128 +- .../wiki/guides/release-notes.md | 48 +- .../wiki/guides/security-and-compliance.md | 264 +-- .../wiki/guides/skill.md | 137 +- .../wiki/guides/start-using.md | 97 +- .../wiki/guides/support.md | 144 +- .../wiki/guides/test-1.md | 81 +- .../wiki/guides/token-plan-guide.md | 130 +- .../wiki/guides/use-cases.md | 83 +- .../use-chat-client-or-development-tool.md | 87 +- skills/bailian-docs-llm-wiki/wiki/index.md | 44 +- 583 files changed, 28550 insertions(+), 31513 deletions(-) create mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-changeset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/{lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md => use-and-authorize-ram-users-for-ccai-dialogue-analysis.md} (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/technology-integration-scheme.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio => official-application-lingque-ccai-dialogue-analysis-aio}/tongyi-xiaomi-ccai-update-announcement.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-voice-dialogue-robot => official-application-lingque-ccai-voice-dialogue-robot}/product-0verview.md (100%) delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-overview.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/{official-application-tongyi-xiaomi-ccai-chatbot-agent => official-application-voicepica-ccai-beebot-agent}/api-reference-5/api-bailianchatbot-2024-11-05-ram.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-cancelaudittask.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-queryaudittask.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser => api-aimiaobi-2023-08-01-dir-other}/api-aimiaobi-2023-08-01-submitaudittask.md (100%) create mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md rename skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/{qwen-audio-realtime-websocket-api.md => fun-audiochat-realtime-websocket-api.md} (67%) create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md rename skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/{coding-plan-guide => token-plan-best-practice}/add-vision-skill.md (77%) create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md rename skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/{coding-plan-guide/web-search-for-coding-plan.md => token-plan-best-practice/web-search-mcp.md} (51%) delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md rename skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/{token-plan-quickstart.md => token-plan-team-edition/token-plan-team-quickstart.md} (69%) delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/billing.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/region.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/security-and-compliance.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md diff --git a/skills/bailian-docs-llm-wiki/SKILL.md b/skills/bailian-docs-llm-wiki/SKILL.md index 7ec690da..836eff6a 100644 --- a/skills/bailian-docs-llm-wiki/SKILL.md +++ b/skills/bailian-docs-llm-wiki/SKILL.md @@ -103,7 +103,6 @@ description: >- | `Multimodal-Omni` | 全模态 | | `ME` | 多模态嵌入 | | `TR` | 翻译 | -| `3D-generation` | 3D 生成 | | `Realtime-Chatting` | Realtime-Chatting | 一个模型常常带多个 capability，`index.md` 中按 `capabilities[0]`（主能力）归类， @@ -148,22 +147,6 @@ description: >- | **按家族筛选**：按 primaryCapability / providers / itemCount / maxContextWindow 找家族 | `models/families.jsonl`（一行一家族，含 items[] 摘要） | | 模型家族总览 / 按能力分桶浏览 | `models/index.md` | | 主题页 / API 文档（按功能领域查找） | `wiki/index.md`（完整索引入口） | -| OpenAI 兼容接口 | `wiki/concepts/openai-compatible-interface.md` | -| API Key 鉴权 | `wiki/concepts/api-key.md` | -| 函数调用（Function Calling） | `wiki/concepts/function-calling.md` | -| 检索增强生成（RAG） | `wiki/concepts/rag.md` | -| 异步调用与任务轮询 | `wiki/concepts/async-invocation.md` | -| Token 与计费 | `wiki/concepts/token-and-billing.md` | -| 流式输出 | `wiki/concepts/streaming-output.md` | -| 业务空间（Workspace） | `wiki/concepts/workspace.md` | -| 模型调优与部署 | `wiki/concepts/fine-tuning-and-deployment.md` | -| MCP 与工具扩展 | `wiki/concepts/mcp-and-tools.md` | -| 模型微调、压缩与部署对比 | `wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md` | -| 模型评估与模型监控对比 | `wiki/comparisons/model-evaluation-vs-monitoring.md` | -| 图像、视频与 3D 生成对比 | `wiki/comparisons/image-vs-video-vs-3d-generation.md` | -| 应用评估与应用监控对比 | `wiki/comparisons/app-evaluation-vs-monitoring.md` | -| 知识库与记忆库对比 | `wiki/comparisons/knowledge-base-vs-memory-library.md` | -| 托管智能体：指南与 API 对比 | `wiki/comparisons/managed-agents-guide-vs-api.md` | > 实际文件名以 `wiki/index.md` 为准；上表若有出入应回到索引页查找。 diff --git a/skills/bailian-docs-llm-wiki/llms.txt b/skills/bailian-docs-llm-wiki/llms.txt index f711c7c8..762df236 100644 --- a/skills/bailian-docs-llm-wiki/llms.txt +++ b/skills/bailian-docs-llm-wiki/llms.txt @@ -4,208 +4,216 @@ ## 模型使用指南 -- **模型体验** - - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) - - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) - - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) - - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - - [语音合成](raw/model-user-guide/model-experience/tts-model.md) - - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - - [语音识别](raw/model-user-guide/model-experience/asr-model.md) - - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) - - [全模态](raw/model-user-guide/model-experience/omni.md) -- **开始使用** - - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - - [选择模型](raw/model-user-guide/get-started-with-models/models.md) - - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) - - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) - **产品计费** - [新人免费额度](raw/model-user-guide/test-1/new-free-quota.md) - [模型训练与部署计费](raw/model-user-guide/test-1/model-training-and-deployment-billing.md) - [节省计划与资源包](raw/model-user-guide/test-1/savings-plan-and-resource-package.md) - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) +- **开始使用** + - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) + - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) + - [选择模型](raw/model-user-guide/get-started-with-models/models.md) + - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) + - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) + - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) +- **Token Plan** + - **个人版** + - [概述](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md) + - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md) + - **团队版** + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md) + - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md) + - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md) + - [概述](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) + - **Coding Plan** + - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) + - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) + - **最佳实践** + - [接入 Harness 工具](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md) + - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) + - [联网搜索](raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md) + - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md) + - [Token Plan 概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - **接入客户端/开发工具** - - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) + - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) - - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) + - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) + - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) + - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) -- **Token Plan（团队版）** - - **最佳实践** - - [工具调用](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) - - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) - - **Coding Plan** - - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) - - [联网搜索](raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) - - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) - - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) - - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) - - [Token Plan（团队版）概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) - - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team.md) - - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-faq.md) + - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) + - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) +- **模型体验** + - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) + - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) + - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) + - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) + - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) + - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) + - [语音合成](raw/model-user-guide/model-experience/tts-model.md) + - [语音识别](raw/model-user-guide/model-experience/asr-model.md) + - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) + - [全模态](raw/model-user-guide/model-experience/omni.md) + - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) - **模型推理** - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) -- **模型部署** - - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) - - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) - - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型调优** - **千问模型调优** - [模型调优简介](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) - - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - [使用 API 或命令行进行模型调优](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) + - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - **语音合成模型调优** - [CosyVoice模型调优](raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) +- **模型部署** + - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) + - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) + - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) + - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型评测** - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) - [评测维度](raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - **模型压缩** - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) - **用量统计与性能监控** - - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) -- **模型数据** - - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) + - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) - **安全合规** + - **传输安全** + - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) + - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) + - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - **安全存储** - [配置终端节点并发起连接](raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) - [配置可用区IP](raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - - [配置MSE云原生网关](raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - [配置私有网络中的资源](raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - - **传输安全** - - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) - - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) + - [配置MSE云原生网关](raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - - [模型备案信息公示](raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) + - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) + - [模型备案信息公示](raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) - [合规资质与隐私说明](raw/model-user-guide/security-and-compliance/privacy-notice.md) +- **模型数据** + - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) + - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) + - [日志回流](raw/model-user-guide/model-data-overview/model-log-backflow.md) - **实践教程** - **三方模型调用教程** - - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek-阿里云](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) + - [DeepSeek-硅基流动](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) - [Kimi](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - [Kimi-月之暗面](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [GLM](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) + - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) + - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) - - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Stepfun-阶跃星辰](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) + - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - [高效搭建 AI 智能体与工作流应用](raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](raw/model-user-guide/use-cases/prompt-engineering-guide.md) - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [文生视频/图生视频Prompt指南](raw/model-user-guide/use-cases/text-to-video-prompt.md) + - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) +- **产品动态** + - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) + - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) - **服务支持** - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - [相关协议](raw/model-user-guide/support/related-agreements.md) -- **产品动态** - - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) - - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) + - [阿里云百炼平台售后服务范围说明](raw/model-user-guide/support/after-sales-service-scope.md) ## 应用使用指南 -- **开始使用** - - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **应用开发** - [应用类型介绍](raw/application-user-guide/llm-application/application-introduction.md) + - [新版智能体应用](raw/application-user-guide/llm-application/new-single-agent-application.md) - [智能体应用](raw/application-user-guide/llm-application/single-agent-application.md) - - [新版智能体应用（Agent 2.0）](raw/application-user-guide/llm-application/new-single-agent-application.md) + - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) - [高代码应用](raw/application-user-guide/llm-application/rich-code-application.md) - [文件问答](raw/application-user-guide/llm-application/file-q-a.md) - - [工作流应用](raw/application-user-guide/llm-application/workflow-application.md) - **Managed Agents** - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - - [构建 Agent](raw/application-user-guide/managed-agents/managed-agents-agent.md) - - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) + - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) + - [构建 Agent](raw/application-user-guide/managed-agents/managed-agents-agent.md) - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) +- **开始使用** + - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) + - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **Prompt** + - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) + - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) - - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - **记忆库** - - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) - - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) + - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) + - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) - **知识库（RAG）** - - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) + - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库日志与监控](raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) + - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) + - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - **数据连接** - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) +- **MCP** + - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) + - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) + - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) + - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) + - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) - **Skill** - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) - **插件** - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) -- **MCP** - - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) - - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) - - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) - **应用发布与分享** - - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) + - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) - [UI设计器](raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) - **应用调用** - - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) + - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) + - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) - **应用评测** - **新版应用评测** - [新版评测集](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) - [评测任务](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) - [标签管理](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评估器](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) + - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - [手动评测](raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - **应用广场** - **官方应用-通义拍照解题辅导** - **API参考** @@ -219,8 +227,8 @@ - **官方应用-通义音频播客生成** - **API参考** - **API目录** - - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) + - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) @@ -232,123 +240,79 @@ - [应用体验与发布](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-experience-and-publishing.md) - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - [指令列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/instruction-list.md) - - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) + - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [三方Agent接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-integration-a2a.md) - **SDK安装** - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) + - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) - - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) + - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) + - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) + - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) - **API参考** - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) - [调用官方Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/official-agent.md) - [调用三方语音模型](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/third-party-voice-integration.md) - - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) - - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) + - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) + - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) - **最佳实践** - **接入百炼及三方Agent** - [百炼及三方Agent直连调用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/agent-direct-call.md) - [接入百炼智能体应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-app.md) - [接入百炼工作流应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-workflow.md) - - **接入图像生成Agent** - - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) - - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) - **接入听悟智能纪要Agent** - [录音纪要Agent使用教程](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/recording-summary-agent-tutorial.md) - - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) + - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - **接入拍照问答Agent** - - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) + - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) + - **接入图像生成Agent** + - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) + - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) - [接入多模态备忘录Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/multimodal-memo-agent.md) - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) - [动作情绪控制实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/action-emotion-control-practice.md) - - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) + - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) - [基于RTOS SDK (License模式) 实现聊天能力](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/chat-capability-based-on-rtos-sdk.md) - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) - [多模态交互开发套件常见问题](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-faq.md) - - **官方应用-伶鹊CCAI-对话分析AIO** - - **使用指南** - - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) - - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) - - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) - - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) - - **API参考** - - **API目录** - - **热词管理** - - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) - - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) - - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) - - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) - - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) - - **不推荐或白名单开放** - - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) - - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) - - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) - - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) - - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) - - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) - - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) - - **最佳实践** - - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) - - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) - - **接口调用示例** - - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) - - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) - - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) - - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md) - - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) - - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) - - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - **官方应用-全妙轻应用系列** - **计费说明（全妙轻应用）** - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) - - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) + - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) + - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - **使用指南** - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) - - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) + - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) - - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) + - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - **开发文档** - **最佳实践** - [应用视频理解和一键成片的最佳实践](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-applying-video-understanding-and-one-click-film.md) @@ -358,73 +322,117 @@ - **数据结构** - [ModelUsage](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-struct-dir/api-quanmiaolightapp-2024-08-01-struct-modelusage.md) - **API目录** - - **电商零售推广文案写作** - - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) - **传媒/零售文章风格与格式学习** - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - **影视互娱剧本创作** - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) - - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) + - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) - **影视传媒视频理解** - - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) + - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) + - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) - [UpdateVideoAnalysisTasks - 视频理解-批量取消任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistasks.md) - - **影视传媒智能拆条** - - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) - - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) - - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) - - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) + - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) - **车机网络热点信息互动问答** - [RunHotTopicChat - 播报单（热榜）问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicchat.md) - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) - **泛企业VOC挖掘** - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - - **泛企业线索挖掘** - - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) - - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) - **网络内容安全审核** - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) + - **泛企业线索挖掘** + - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) + - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) - **作文批改** - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) - - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) + - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) - **其他** + - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) - [GenerateBroadcastNews - 播报单（热榜）热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-generatebroadcastnews.md) - - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) + - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) - - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) + - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) + - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC挖掘异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submitenterprisevocanalysistask.md) + - **影视传媒智能拆条** + - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) + - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) + - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) + - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) + - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) + - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) + - **电商零售推广文案写作** + - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) + - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-ram.md) + - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-changeset.md) - [全妙轻应用更新公告](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-update-announcement.md) - [常见问题](raw/application-user-guide/application-gallery/quanmiao-light-application-series/quanmiao-lightapp-faq.md) + - **官方应用-伶鹊CCAI-对话分析AIO** + - **使用指南** + - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) + - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) + - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) + - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) + - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) + - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) + - **API参考** + - **API目录** + - **热词管理** + - [CreateVocab - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md) + - [UpdateVocab - 修改热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md) + - [ListVocab - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md) + - [DeleteVocab - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md) + - [GetVocab - 获取热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md) + - **不推荐或白名单开放** + - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) + - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) + - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) + - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) + - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) + - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) + - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) + - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) + - **最佳实践** + - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) + - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) + - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) + - **接口调用示例** + - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) + - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) + - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) + - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) + - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md) + - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) + - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) + - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - **官方应用-伶鹊CCAI-客服对话Agent** - **API参考** - **API目录** - - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) + - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) @@ -432,150 +440,137 @@ - **API参考** - **API目录** - **MQ消息订阅配置** - - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) - - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) + - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) + - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) + - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) - **变量管理** - - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) - - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) - - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) - - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) + - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) + - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) + - [UpdateVariable - 更新变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md) + - [CreateVariable - 创建变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md) - **三方语音配置** - - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) - - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) - - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) - - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) - - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) - - **热词管理** - - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) - - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) - - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) - - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) - - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) - - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - [ListVoiceEngines - 获取三方语音引擎列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md) + - [UpdateVoiceAccessProfile - 更新三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md) + - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) + - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) + - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) - **克隆音管理** - - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) - - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) - - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) - - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) + - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) + - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) + - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) + - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) + - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) - **应用管理** - - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) - - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) - - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) - - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) - - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) - - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) - - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) - - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) - - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) - - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) - - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) - - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) - - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) - - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) - - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md) + - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) + - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) + - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) + - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) + - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) + - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) + - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) + - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) + - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) + - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) + - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) + - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) + - **热词管理** + - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) + - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) + - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) + - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) + - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) + - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) + - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) + - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) + - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) + - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) + - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) + - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md) - **通义点金** - **API参考** - **API目录** - **平台能力-文档库** - - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) - [UpdateDocumentChunk - 更新文档块内容](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocumentchunk.md) + - [GetAppConfig - 获取配置信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getappconfig.md) - [CreateLibrary - 创建文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createlibrary.md) - [GetLibraryList - 获取文档库列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrarylist.md) - [GetLibrary - 获取文档库详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrary.md) - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - [GetDocumentUrl - 获取文档的下载链接](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumenturl.md) + - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) - - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) - - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) + - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) - [RecallDocument - 文档召回](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-recalldocument.md) - - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) - [GetParseResult - 获取文档解析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getparseresult.md) + - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) - [UpdateLibrary - 更新文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatelibrary.md) - - [GetHistoryListByBizType - 根据业务类型获取对话历史记录](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-gethistorylistbybiztype.md) - - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - [RunLibraryChatGeneration - 文档库会话生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-runlibrarychatgeneration.md) + - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) + - [GetHistoryListByBizType - 根据业务类型获取对话历史记录](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-gethistorylistbybiztype.md) - **平台能力-应用** + - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [EndToEndRealTimeDialog - 语音实时对话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-endtoendrealtimedialog.md) - [RunDialogAnalysis - 会话分析结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rundialoganalysis.md) - - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) + - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) - - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [GetDialogAnalysisResult - 获取会话分析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoganalysisresult.md) - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) + - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) + - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) - - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) + - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) - - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) - [UpdateQaLibrary - 更新QA问答库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-updateqalibrary.md) - [SubmitChatQuestion - 提交问题列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-submitchatquestion.md) - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - [RunChatResultGeneration - 对话结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runchatresultgeneration.md) + - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) + - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - **其他** - [DashscopeAsyncTaskFinishEvent - Dashscope异步任务完成回调事件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-other/api-dianjin-2024-06-28-dashscopeasynctaskfinishevent.md) - - [服务接入点](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-ram.md) - [版本说明](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-changeset.md) - [产品简介](raw/application-user-guide/application-gallery/tongyi-dianjin/tongyi-dianjin-overview.md) - - **官方应用-通义数据挖掘** - - **API参考** - - **API目录** - - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) - - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) - - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) - - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) - - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) - - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) - - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - **官方应用-通义多模态翻译** - **API参考** - **API目录** - **文本翻译** - - [BatchTranslate - 批量文本翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-batchtranslate.md) - [TextTranslate - 文本翻译接口](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-texttranslate.md) + - [BatchTranslate - 批量文本翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-batchtranslate.md) - [SubmitLongTextTranslateTask - 提交长文本翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submitlongtexttranslatetask.md) + - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) - - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - **图片翻译** - - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) + - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) - **文档翻译** - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md) - [通义多模态翻译](raw/application-user-guide/application-gallery/official-application-tongyi-translate/official-application-tongyi-translate-overview.md) @@ -584,8 +579,8 @@ - **API参考** - **API目录** - [生成对话](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-chat-generate.md) - - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - [上传文件](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-file-upload.md) + - [对话文件管理](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-session-file-management.md) - [生成报告导出](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/deepsearch-report-export.md) - [对接自有知识库](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-list/docking-self-built-database.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-deepsearch/deepsearch-api-reference/deepsearch-api-overview.md) @@ -603,15 +598,28 @@ - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) - **通义 UI Agent** - [通义 UI Agent](raw/application-user-guide/application-gallery/ui-agent/ui-agent-api.md) + - **官方应用-通义数据挖掘** + - **API参考** + - **API目录** + - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) + - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) + - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) + - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) + - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) + - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) + - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) + - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - **官方应用-全妙解决方案类产品** - **妙笔、妙策和审校** - **使用指南** - **AI妙笔** - **功能界面** - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) + - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) - [妙笔首页概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/amb-homepage-overview.md) - [AI工具箱](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/ai-toolbox.md) @@ -622,14 +630,14 @@ - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) - [深度写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/deep-writing.md) - **文本写作指导** + - **政务公文写作指导** + - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) + - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) + - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - **传媒类文体写作指导** - [快速写一篇传媒稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/quick-media-writing-prompt.md) - [没有思路，要谋篇布局](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/use-amb-to-help-writing.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) - - **政务公文写作指导** - - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) - [常见FAQ](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/faq-for-using-quanmiao-series-products.md) - **更新公告** - **功能更新** @@ -637,10 +645,10 @@ - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2024年3月11更新-AI全妙系列 V2.2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-11-ai-quanmiao-v2-2.md) - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) - - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) + - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) + - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) - [计费说明（PPT生成）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/ppt-generation-billing.md) - [计费说明（妙策-自定义数据源）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-document-miaoce-custom-data-source.md) - [计费说明（视频混剪）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/billing-description-video-mixing.md) @@ -649,20 +657,12 @@ - [妙搜](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaodou-and-miaodu-guidelines-for-use/ai-miaosou.md) - [计费说明（妙搜和妙读）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaosou-and-miaodu/miaosou-miaodu-api-billing.md) - **开发文档** - - **最佳实践** - - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) - - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) - - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) - - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) - - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) - - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - **API参考** - **数据结构** - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) + - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) @@ -673,27 +673,33 @@ - [ListDialogues - 生成历史列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listdialogues.md) - [ListVersions - 获取版本信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listversions.md) - [GetProperties - 获取配置信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-getproperties.md) + - **通用接口-文件上传下载** + - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) + - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - **通用接口-异步任务管理** - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) - - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) - - **通用接口-文件上传下载** - - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) - - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) + - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) + - **通用接口-通用配置** + - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) + - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) + - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) + - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) + - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - **妙笔-创作文章** - - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) + - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) + - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) - - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) + - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunWriteToneGeneration - 文风改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritetonegeneration.md) - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - [RunSummaryGenerate - 摘要生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runsummarygenerate.md) + - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) @@ -701,22 +707,16 @@ - [ListBuildConfigs - 获取系统自定义预设](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-listbuildconfigs.md) - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) - [FeedbackDialogue - 反馈对话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-feedbackdialogue.md) - - **通用接口-通用配置** - - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) - - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) - - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) - - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) + - **妙笔-视频审校** + - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) + - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文体仿写** - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) - [DeleteStyleLearningResult - 删除自定义文体](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-deletestylelearningresult.md) - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) + - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) - - **妙笔-视频审校** - - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) - - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) - **妙笔-文章审校-规则库管理** - [SubmitAuditNote - 提交自定义规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-submitauditnote.md) - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) @@ -728,8 +728,8 @@ - **妙笔-文章审校-词库管理** - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) - - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) + - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) @@ -739,192 +739,200 @@ - [GetFactAuditUrl - 获取事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-getfactauditurl.md) - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - **妙笔-文章审校** - - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md) - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md) - - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md) - [GetSmartAuditResult - 查询智能审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-getsmartauditresult.md) - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) - [ExportAuditContentResult - 导出智能审校报告](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-exportauditcontentresult.md) + - **妙笔-文档管理** + - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) + - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) + - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) + - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) + - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) + - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) + - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) + - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - **妙笔-素材库** - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) - - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) + - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) - **妙笔-素材库-自定义文本** - [GetCustomText - 获取自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-getcustomtext.md) - [UpdateCustomText - 更新自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-updatecustomtext.md) - [ListCustomText - 获取自定义文本列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-listcustomtext.md) - [SaveCustomText - 保存自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-savecustomtext.md) - - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - - **妙笔-文档管理** - - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) - - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) - - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) - - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) - - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) - - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) - - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) - - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) + - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - **妙笔-视频混剪** - [GetClipsBuildInResource - 获取智能混剪内置资源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getclipsbuildinresource.md) + - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - [AsyncCreateClipsTimeLine - 创建剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstimeline.md) - [AsyncUploadVideo - 异步上传视频剪辑素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncuploadvideo.md) - - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) + - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) + - **妙策-自定义数据源** + - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) + - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) + - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) - **公文库检索** - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-选题热点** - - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) + - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) - - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) + - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) + - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) - [ListWebReviewPoints - 获取网友视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listwebreviewpoints.md) - - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) - - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) + - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - **妙策-自定义话题** - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) - - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) + - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) - - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) - - **妙策-自定义数据源** - - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) - - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) - - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) + - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) + - **妙策-新闻播报** + - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) + - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) + - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - **妙策-openapi** - [SubmitDocClusterTask - 提交内容聚合任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitdocclustertask.md) - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) + - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) - [SubmitCustomTopicSelectionPerspectiveAnalysisTask - 提交自定义热点选题视角分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitcustomtopicselectionperspectiveanalysistask.md) - - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - - **妙策-新闻播报** - - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) - - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) - - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) - **妙搜-智能搜索** - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) - - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) + - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) - - **妙策-企业VOC挖掘** - - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) - - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) - - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) - - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) - - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) - **妙搜-数据源** - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md) - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md) - - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) - - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md) + - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) - [AddDatasetDocument - 数据源-添加文档到数据集](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-adddatasetdocument.md) - [GetDatasetDocument - 数据源-获取文档详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdatasetdocument.md) - [UpdateDatasetDocument - 数据源-修改文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedatasetdocument.md) - [ListDatasetDocuments - 数据源-文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasetdocuments.md) - - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) + - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) + - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) + - **妙策-企业VOC挖掘** + - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) + - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) + - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) + - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) + - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) - **系统配置-干预配置** - [ListInterveneCnt - 获得所有干预项的数量](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenecnt.md) - [ListIntervenes - 列出干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenes.md) - - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [ImportInterveneFile - 同步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefile.md) - - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) + - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [ImportInterveneFileAsync - 异步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefileasync.md) + - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) - - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) + - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) + - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) + - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - [DeleteInterveneRule - 删除干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-deleteintervenerule.md) - [ExportIntervenes - 导出干预项内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-exportintervenes.md) - [GetInterveneImportTaskInfo - 获得干预项目导入任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneimporttaskinfo.md) - - **妙读-抽取类** - - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - **系统配置-信源管理** - - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) + - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - **妙读-基础操作类** - - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) + - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) - [UploadBook - 书籍上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploadbook.md) - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) - [DeleteDocs - 批量删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-deletedocs.md) - **妙读-生成类** - - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunMultiDocIntroduction - 多文档聚合摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runmultidocintroduction.md) - - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) + - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) + - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) - - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) - [RunCommentGeneration - 客户之声预测](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runcommentgeneration.md) - - **妙读-问答类** - - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) + - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) + - **妙读-抽取类** + - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - **妙读-其他** - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) + - **妙读-问答类** + - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) + - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - **深度写作** - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) - - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) - [GetDeepWriteTaskResult - 查询深度写作任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetaskresult.md) + - [CancelDeepWriteTask - 取消深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-canceldeepwritetask.md) - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) - **PPT生成** - [ListEnterprisePptTemplates - 查询企业专属PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listenterpriseppttemplates.md) + - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) - - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) - [ExportPptArtifact - 导出PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-exportpptartifact.md) - [GetPptArtifact - 查询PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifact.md) - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) - [ListPptArtifacts - 查询PPT作品列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listpptartifacts.md) - [InitiatePptCreation - 初始化用来创建PPT的会话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreation.md) + - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - [GetPptConfig - 获取PPT组件配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptconfig.md) - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) - - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - **标书生成** + - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) - - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) + - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - **其他** - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) - [SubmitSmartClipTask - 提交智能一键成片任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitsmartcliptask.md) + - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) - [SaveOrUpdateOssConfig - 配置-云存储-参数配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-saveorupdateossconfig.md) - - [CreateDataPermissions - 权限-批量添加](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdatapermissions.md) - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) + - [CreateDataPermissions - 权限-批量添加](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdatapermissions.md) + - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) + - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md) - [FetchParseDocumentLayoutTask - 获取排版任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-fetchparsedocumentlayouttask.md) + - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md) + - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md) - - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-ram.md) + - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) + - **最佳实践** + - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) + - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) + - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) + - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) + - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) + - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) + - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - **更多** - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) @@ -934,31 +942,28 @@ - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) - [官方应用-通义听悟Agent](raw/application-user-guide/application-gallery/official-application-tingwu-agent.md) - - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) - [官方应用-析言GBI](raw/application-user-guide/application-gallery/xiyan-gbi.md) + - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) - **应用观测** - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **权限管理** - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) - **实践教程** + - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [在网站上增加一个AI助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [基于本地知识库构建RAG应用](raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) - **服务支持** - [常见问题](raw/application-user-guide/application-support/application-faq.md) - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) + - [阿里云百炼平台售后服务范围说明](raw/application-user-guide/application-support/application-after-sales-service-scope.md) ## 模型 API 参考 -- **使用 API** - - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) - - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) - - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) - - [错误码](raw/model-api-reference/preparations/error-code.md) - **图像生成** - **千问** + - [千问-图像生成与编辑3.0 API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) @@ -966,55 +971,80 @@ - [万相-文生图V2版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) + - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-涂鸦作画API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - [万相-图像局部重绘API参考](raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) - - **Z-Image** - - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - **可灵** - [可灵-图像生成API参考](raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) - **Vidu** - [Vidu-图像生成API参考](raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) + - **Z-Image** + - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - **创意工具** - [人像风格重绘API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - - [图像画面扩展API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [虚拟模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) + - [图像画面扩展API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) + - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [图像背景生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) - - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) + - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) + - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [常见问题](raw/model-api-reference/image-generation/image-faq.md) +- **使用 API** + - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) + - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) + - [错误码](raw/model-api-reference/preparations/error-code.md) + - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) - **3D模型生成** - [Tripo-3D模型生成](raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) - **实时多模态** - - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) + - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) - [实时多模态交互流程](raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) + - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) + - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) - **更多模型** - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) + - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-OCR API参考](raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) + - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) + - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) +- **Realtime API** + - **快速开始** + - [SDK下载](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) + - [Token鉴权](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) + - [实现接通模型/应用](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) + - **最佳实践** + - [通过WebRTC使用多模态交互套件实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) + - [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) + - [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) + - **AOQ客户端API** + - **AOQ SDK功能介绍** + - [连接状态管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) + - [音频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) + - [媒体流发送管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) + - [自定义音频播放](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) + - [自定义音频采集](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) + - [视频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) + - [自定义视频输入](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) + - [AOQ SDK简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) + - [Realtime API简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) - **工具包/框架** - [OpenAI Chat接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) + - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) + - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) + - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) - **模型生产** @@ -1025,121 +1055,123 @@ - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [子业务空间的模型调用](raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - [上传本地文件获取临时URL](raw/model-api-reference/more-about-models/get-temporary-file-url.md) + - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - **视频生成** - **HappyHorse** - [HappyHorse-文生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) + - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - **万相** - **万相-早期视频模型（2.1-2.6）** - - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) + - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) - - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) + - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) + - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-视频换人API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) - - **爱诗** - - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - - [爱诗-视频对口型API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) - - [爱诗-视频动作模仿API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) + - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - **人像驱动** - [图生舞蹈视频-舞动人像AnimateAnyone](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [图生唱演视频-悦动人像EMO](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [图生播报视频-灵动人像LivePortrait](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) + - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) + - **爱诗** + - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) + - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) + - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) + - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) + - [爱诗-视频对口型API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) + - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) + - [爱诗-视频动作模仿API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) + - **可灵** + - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **Vidu** - [Vidu-文生视频API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - - **可灵** - - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) + - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) + - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - **音频** - **语音识别** - - **实时语音识别（Fun-ASR）** - - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) - - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) - - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) - - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) - **实时语音识别（Qwen-ASR-Realtime）** - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) + - **实时语音识别（Fun-ASR）** + - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) + - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) + - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) + - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) + - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) + - **非实时语音识别（Fun-ASR）** + - [Fun-ASR非实时语音识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) + - [Fun-ASR非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) + - [Fun-ASR非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) + - [Fun-ASR非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) + - [Fun-ASR非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) + - **非实时语音识别（Paraformer）** + - [Paraformer非实时语音识别HTTP API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) + - [Paraformer非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) + - [Paraformer非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) + - [Paraformer非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) + - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) + - [Paraformer非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) + - **定制热词** + - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) + - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) + - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - **实时语音识别（Paraformer）** - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) - - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) - - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) - - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) + - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) - - **录音文件识别（Fun-ASR）** - - [Fun-ASR录音文件识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - - [Fun-ASR录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) - - [Fun-ASR录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) - - [Fun-ASR录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - - [Fun-ASR录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) - - **录音文件识别（Paraformer）** - - [Paraformer录音文件识别RESTful API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md) - - [Paraformer录音文件识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - - [Paraformer录音文件识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) - - [Paraformer录音文件识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) - - [Paraformer录音文件识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) - - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) - - **定制热词** - - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) - - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) - - [录音文件识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) + - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) + - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) + - [非实时语音识别（Fun-ASR-Flash）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md) + - [非实时语音识别（Fun-ASR-Realtime）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md) + - [非实时语音识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - **语音合成** - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) - - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) + - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) - **实时语音合成（Qwen-TTS-Realtime）** - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) - - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) + - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) - **实时语音合成（Sambert）** - - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) + - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) + - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) - - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) - - **非实时语音合成（MiniMax）** - - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) + - **非实时语音合成（MiniMax）** + - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - **声音复刻** - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) @@ -1157,17 +1189,17 @@ - [音视频翻译-通义千问 API 参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/qwen3-livetranslate-flash-api.md) - **语音对话** - **实时语音对话** + - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话客户端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md) - - [Qwen-Audio 实时语音对话WebSocket API参考](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md) - [Qwen-Audio 实时语音对话服务端事件](raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md) - **向量与排序** - **通用文本向量** - [同步接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) - [批处理接口API详情](raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) - - **排序模型（Rerank）** - - [文本排序](raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - **多模态向量** - [Multimodal-Embedding API详情](raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) + - **排序模型（Rerank）** + - [文本排序](raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) - [文本生成模型API参考](raw/model-api-reference/qwen-api-reference.md) - [文件管理](raw/model-api-reference/file-management-api.md) @@ -1177,21 +1209,33 @@ - [API 总览与认证](raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) - - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) + - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - [File](raw/application-api-reference/managed-agents-api/files-api.md) - [Skill](raw/application-api-reference/managed-agents-api/skills-api.md) +- **应用调用** + - **Responses API** + - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) + - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) + - **DashScope API** + - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) + - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) + - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) +- **框架** + - **Spring AI Alibaba** + - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) + - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) + - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) - **应用组件** - **API目录** - **数据连接（原应用数据）** - - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) - - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [ListFile - 文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) + - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) @@ -1199,33 +1243,18 @@ - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) - [ChangeParseSetting - 修改类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddConnector - 新增连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) - - **知识库** - - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) - - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) - - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) - - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) - - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) + - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) + - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - **Prompt工程** - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) + - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) - **其他** - **长期记忆（旧）** - [CreateMemory - 创建长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) @@ -1233,36 +1262,40 @@ - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) + - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) + - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) - - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) + - [AddChunk - 新增切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md) + - **知识库** + - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) + - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) + - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) + - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) + - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) + - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) + - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) + - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) + - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) + - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) + - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) + - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) + - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) + - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) + - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - [授权信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) - [版本说明](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) -- **应用调用** - - **DashScope API** - - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - **Responses API** - - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) -- **框架** - - **Spring AI Alibaba** - - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) -- **长期记忆** - - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **更多** - - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) + - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](raw/application-api-reference/more/how-to-use-search-filters.md) +- **长期记忆** + - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - [知识检索与问答](raw/application-api-reference/knowledge.md) diff --git a/skills/bailian-docs-llm-wiki/models/families.jsonl b/skills/bailian-docs-llm-wiki/models/families.jsonl index 8bcd4a65..6fc4a92e 100644 --- a/skills/bailian-docs-llm-wiki/models/families.jsonl +++ b/skills/bailian-docs-llm-wiki/models/families.jsonl @@ -34,7 +34,7 @@ {"slug":"image-erase-completion","name":"图像擦除补全","description":"图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-erase-completion","name":"图像擦除补全","capabilities":["IG"]}],"detailPath":"groups/image-erase-completion.json"} {"slug":"image-instance-segmentation","name":"人物实例分割","description":"人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-instance-segmentation","name":"人物实例分割","capabilities":["IG"]}],"detailPath":"groups/image-instance-segmentation.json"} {"slug":"image-out-painting","name":"图像画面扩展","description":"图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"image-out-painting","name":"图像画面扩展","capabilities":["IG"]}],"detailPath":"groups/image-out-painting.json"} -{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["moonshot-ai"],"itemCount":4,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":262144} +{"slug":"kimi-models-market-place","name":"Kimi","description":"由月之暗面提供的Kimi系列模型的API服务。","primaryCapability":"TG","capabilities":["TG","VU","Reasoning"],"providers":["moonshot-ai"],"itemCount":5,"items":[{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","contextWindow":262144,"capabilities":["TG","VU","Reasoning"]},{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","contextWindow":262144,"capabilities":["TG","Reasoning","VU"]},{"model":"kimi/kimi-k3","name":"kimi/kimi-k3","contextWindow":1048576,"capabilities":["TG","VU","Reasoning"]}],"detailPath":"groups/kimi-models-market-place.json","maxContextWindow":1048576} {"slug":"kling-models-market-place","name":"可灵AI","description":"由可灵AI提供的高质量视频与图像生成及编辑模型。","primaryCapability":"VG","capabilities":["VG","IG"],"providers":["kling"],"itemCount":4,"items":[{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","capabilities":["IG"]},{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","capabilities":["IG"]},{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","capabilities":["VG"]},{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","capabilities":["VG"]}],"detailPath":"groups/kling-models-market-place.json"} {"slug":"liveportrait-detect","name":"灵动人像LivePortrait-detect","description":"LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","capabilities":["VG"]}],"detailPath":"groups/liveportrait-detect.json"} {"slug":"liveportrait","name":"灵动人像LivePortrait","description":"LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"liveportrait","name":"灵动人像LivePortrait","capabilities":["VG"]}],"detailPath":"groups/liveportrait.json"} @@ -54,18 +54,19 @@ {"slug":"pixverse-v6-market-place","name":"PixVerse V6","description":"由爱诗科技提供的PixVerse V系列视频大模型API服务。","primaryCapability":"VG","capabilities":["VG"],"providers":["pixverse"],"itemCount":4,"items":[{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","capabilities":["VG"]},{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","capabilities":["VG"]}],"detailPath":"groups/pixverse-v6-market-place.json"} {"slug":"qvq-max","name":"QVQ-Max","description":"千问QVQ视觉推理模型，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-max","name":"QVQ-Max","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-max.json","maxContextWindow":131072} {"slug":"qvq-plus","name":"Qwen-QVQ-Plus","description":"千问QVQ视觉推理模型增强版，支持视觉输入及思维链输出，在数学、编程、视觉分析、创作以及通用任务上都表现了更强的能力。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qvq-plus","name":"QVQ-Plus","contextWindow":131072,"capabilities":["Reasoning","VU"]}],"detailPath":"groups/qvq-plus.json","maxContextWindow":131072} -{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":8192} -{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":8192,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":8192} +{"slug":"qwen-audio-realtime-flash","name":"Qwen-Audio-Realtime-Flash","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Flash版更注重极致的响应速度","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","contextWindow":40960,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-flash.json","maxContextWindow":40960} +{"slug":"qwen-audio-realtime-plus","name":"Qwen-Audio-Realtime-Plus","description":"Qwen-Audio-Realtime 是一款登顶全球权威评测的下一代实时双工语音大模型，模型兼顾了模型智商与双工对话节奏,在保持流畅、自然的实时交互体验的同时,语音推理能力不打折扣;并通过并行推理和全向流式等工程优化,将端到端响应时延控制在低水平,实现\"又快又聪明\"的对话体验。Plus版本更注重高质量的回复结果。","primaryCapability":"Realtime-Chatting","capabilities":["Realtime-Chatting"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","contextWindow":40960,"capabilities":["Realtime-Chatting"]}],"detailPath":"groups/qwen-audio-realtime-plus.json","maxContextWindow":40960} {"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"} {"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072} {"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072} {"slug":"qwen-deep-research","name":"qwen-deep-research","description":"千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-deep-research","name":"qwen-deep-research","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-deep-research.json","maxContextWindow":1000000} {"slug":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","description":"快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen-doc-turbo.json","maxContextWindow":262144} -{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen-domain-model"],"itemCount":6,"items":[{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json"} +{"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen","qwen-domain-model"],"itemCount":7,"items":[{"model":"qwen3.7-text-embedding","name":"Qwen3.7-通用文本向量","contextWindow":131072,"capabilities":["TR"]},{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json","maxContextWindow":131072} {"slug":"qwen-flash-character","name":"Qwen-Flash-Character","description":"千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash-character","name":"Qwen-Flash-Character","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/qwen-flash-character.json","maxContextWindow":8192} {"slug":"qwen-flash","name":"Qwen-Flash","description":"Qwen3系列Flash模型，实现思考模式和非思考模式的有效融合，可在对话中切换模式。复杂推理类任务性能优秀，指令遵循、文本理解等能力显著提高。支持1M上下文长度，按照上下文长度进行阶梯计费。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash","name":"Qwen-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG"]}],"detailPath":"groups/qwen-flash.json","maxContextWindow":1000000} {"slug":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","description":"Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系列最强的文字渲染能力和真实质感。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0-pro.json"} {"slug":"qwen-image-2.0","name":"Qwen-Image-2.0","description":"Qwen-Image-2.0系列加速版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。加速版有效实现了模型效果和性能的最佳平衡。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","capabilities":["IG"]}],"detailPath":"groups/qwen-image-2.0.json"} +{"slug":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","description":"内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求**“好用”**——让图像生成真正成为可落地的生产力工具。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-3.0-pro.json"} {"slug":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","description":"千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit-max.json"} {"slug":"qwen-image-edit","name":"Qwen-Image-Edit-Plus","description":"千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image-edit","name":"Qwen-Image-Edit","capabilities":["IG"]},{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit.json"} {"slug":"qwen-image-max","name":"Qwen-Image-Max","description":"千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-max","name":"Qwen-Image-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-max.json"} @@ -117,7 +118,7 @@ {"slug":"qwen3-tts-vd","name":"Qwen3-TTS-VD","description":"Qwen3-TTS-VD模型是通义实验室最新推出的实时语音合成大模型，可对qwen3-voice-design服务设计的声音进行高保真实时语音合成，且同一音色支持11个语种的语音输出。该模型经过海量数据训练，合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","capabilities":["TTS"]}],"detailPath":"groups/qwen3-tts-vd.json"} {"slug":"qwen3-vl-flash","name":"Qwen3-VL-Flash","description":"Qwen3系列小尺寸视觉理解模型，实现思考模式和非思考模式的有效融合，效果优于开源版Qwen3-VL-30B-A3B，响应速度快。全面升级图像/视频理解，支持长视频长文档等超长上下文、空间感知与万物识别；具备视觉2D/3D定位能力，胜任复杂现实任务。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-flash.json","maxContextWindow":262144} {"slug":"qwen3-vl-plus","name":"Qwen3-VL-Plus","description":"Qwen3系列视觉理解模型，实现思考模式和非思考模式的有效融合，视觉智能体能力在OS World等公开测试集上达到世界顶尖水平。此版本在视觉coding、空间感知、多模态思考等方向全面升级；视觉感知与识别能力大幅提升，支持超长视频理解。","primaryCapability":"VU","capabilities":["VU","Reasoning"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","contextWindow":262144,"capabilities":["VU","Reasoning"]}],"detailPath":"groups/qwen3-vl-plus.json","maxContextWindow":262144} -{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","VU","TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","VU","TG"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} +{"slug":"qwen3.5-flash","name":"Qwen3.5-Flash","description":"Qwen3.5原生视觉语言系列Flash模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"Reasoning","capabilities":["Reasoning","TG","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","contextWindow":1000000,"capabilities":["Reasoning","TG","VU"]}],"detailPath":"groups/qwen3.5-flash.json","maxContextWindow":1000000} {"slug":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","description":"Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json","maxContextWindow":53248} {"slug":"qwen3.5-ocr","name":"Qwen3.5-OCR","description":"Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","contextWindow":65536,"capabilities":["VU"]}],"detailPath":"groups/qwen3.5-ocr.json","maxContextWindow":65536} {"slug":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash-realtime.json","maxContextWindow":262144} @@ -138,11 +139,10 @@ {"slug":"shoemodel-v1","name":"鞋靴模特","description":"鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"shoemodel-v1","name":"鞋靴模特","capabilities":["IG"]}],"detailPath":"groups/shoemodel-v1.json"} {"slug":"siliconflow-models","name":"SiliconFlow DeepSeek","description":"由硅基流动提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning"],"providers":["deepseek"],"itemCount":4,"items":[{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","contextWindow":163840,"capabilities":["TG"]},{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","contextWindow":163840,"capabilities":["TG","Reasoning"]},{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","contextWindow":163840,"capabilities":["TG","Reasoning"]}],"detailPath":"groups/siliconflow-models.json","maxContextWindow":163840} {"slug":"speech-biasing","name":"语音识别热词","description":"热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。","primaryCapability":"ASR","capabilities":["ASR"],"providers":["qwen"],"itemCount":1,"items":[{"model":"speech-biasing","name":"语音识别热词","capabilities":["ASR"]}],"detailPath":"groups/speech-biasing.json"} -{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} +{"slug":"stepfun-models-market-place","name":"StepFun推理模型","description":"由阶跃星辰StepFun提供的Step系列推理模型API服务","primaryCapability":"TG","capabilities":["TG","VU"],"providers":["stepfun"],"itemCount":1,"items":[{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","contextWindow":262144,"capabilities":["TG","VU"]}],"detailPath":"groups/stepfun-models-market-place.json","maxContextWindow":262144} {"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192} {"slug":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","description":"通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768} {"slug":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","description":"通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768} -{"slug":"tripo-models-market-place","name":"Tripo","description":"AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。","primaryCapability":"3D-generation","capabilities":["3D-generation"],"providers":["tripo"],"itemCount":2,"items":[{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","capabilities":["3D-generation"]},{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","capabilities":["3D-generation"]}],"detailPath":"groups/tripo-models-market-place.json"} {"slug":"vanchin-models-market-place","name":"Vanchin DeepSeek","description":"由快手万擎提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["deepseek"],"itemCount":6,"items":[{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","contextWindow":8192,"capabilities":["VU","TG"]},{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","contextWindow":131072,"capabilities":["TG"]},{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/vanchin-models-market-place.json","maxContextWindow":1048576} {"slug":"video-style-transform","name":"视频风格重绘","description":"视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"video-style-transform","name":"视频风格重绘","capabilities":["VG"]}],"detailPath":"groups/video-style-transform.json"} {"slug":"videoretalk","name":"声动人像VideoRetalk","description":"VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"videoretalk","name":"声动人像VideoRetalk","capabilities":["VG"]}],"detailPath":"groups/videoretalk.json"} diff --git a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json index 15fdc4d5..f2a97249 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json @@ -23,10 +23,39 @@ "prefix-completion" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -56,60 +85,23 @@ "contextWindow": 262144, "maxInputTokens": 229376, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "kimi-k2.7-code", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2.7-code\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.7-code\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2.7-code',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -s -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -s -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation\" \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H \"Content-Type: application/json\" \\\n -d '{\n \"model\": \"kimi-k2.7-code\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n }'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.comapi/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"你是谁\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.7-code',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2.7-code\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2948482.html" } } @@ -133,10 +125,39 @@ "model-experience" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.65", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -166,35 +187,22 @@ "latestOnlineAt": "2026-04-21T09:55:34.000+00:00", "contextWindow": 262144, "maxInputTokens": 229376, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2.6", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.6\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.6\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.6\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.6',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.6\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -217,11 +225,40 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2.5", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -250,35 +287,22 @@ "latestOnlineAt": "2026-01-30T01:49:03.000+00:00", "contextWindow": 262144, "maxInputTokens": 229376, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2.5", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi-k2.5\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi-k2.5\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi-k2.5\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"kimi-k2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'kimi-k2.5',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"kimi-k2.5\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -299,11 +323,28 @@ "function-calling" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi-k2-thinking", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -333,63 +374,26 @@ "maxInputTokens": 229376, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Kimi-K2-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"kimi-k2-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'kimi-k2-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"kimi-k2-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"kimi-k2-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"kimi-k2-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -410,11 +414,28 @@ "function-calling" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "Moonshot-Kimi-K2-Instruct", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -441,54 +462,27 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Moonshot-Kimi-K2-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2948482.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"Moonshot-Kimi-K2-Instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"Moonshot-Kimi-K2-Instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"Moonshot-Kimi-K2-Instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"Moonshot-Kimi-K2-Instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json index bd2b5e97..9531a790 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json @@ -11,94 +11,34 @@ "Text" ] }, - "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", + "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.1", - "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", - "capabilities": [ - "Reasoning", - "TG" - ], - "versionTag": "SNAPSHOT", - "maxOutputTokens": 32768, - "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", - "contextWindow": 204800, - "maxInputTokens": 172032, - "inferenceProvider": "bailian", - "name": "MiniMax-M2.1", - "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ + "model": "MiniMax-M2.5", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" - } - }, - "dashscope": { - "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" - } + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Text" - ], - "request_modality": [ - "Text" - ] - }, - "description": "MiniMax-M2.5是MiniMax推出的旗舰级开源大模型，经过数十万个真实复杂环境中的大规模强化学习训练，M2.5 在编程、工具调用和搜索、办公等生产力场景都达到或者刷新了行业的 SOTA。", - "features": [ - "model-experience", - "function-calling", - "cache" ], - "provider": "mini-max", - "limit": { - "message": "model not exist" - }, - "model": "MiniMax-M2.5", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -127,49 +67,95 @@ "latestOnlineAt": "2026-02-24T15:07:02.000+00:00", "contextWindow": 204800, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "MiniMax-M2.5", "docUrl": "https://help.aliyun.com/document_detail/3017140.html", - "predictConfig": [ + "samples": { + "openai": { + "default": { + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + } + }, + "dashscope": { + "default": { + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "MiniMax-M2.1是MiniMax推出的旗舰级开源大模型，聚焦真实世界复杂任务，以多语言编程与长链 Agent 能力为核心优势。", + "features": [ + "model-experience", + "function-calling", + "cache" + ], + "provider": "mini-max", + "model": "MiniMax-M2.1", + "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" } ], + "capabilities": [ + "Reasoning", + "TG" + ], + "versionTag": "SNAPSHOT", + "maxOutputTokens": 32768, + "latestOnlineAt": "2026-01-23T07:47:49.000+00:00", + "contextWindow": 204800, + "maxInputTokens": 172032, + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "MiniMax-M2.1", + "docUrl": "https://help.aliyun.com/document_detail/3017140.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.5\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.5',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"MiniMax-M2.1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'MiniMax-M2.1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.5\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.5\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"MiniMax-M2.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"MiniMax-M2.1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"MiniMax-M2.1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json index abeefc7d..f1d04dde 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json @@ -14,11 +14,22 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-2.8-turbo", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,7 +57,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-2.8-turbo", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -68,11 +78,22 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-2.8-hd", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -100,7 +121,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-2.8-hd", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -122,11 +142,22 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-02-turbo", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -154,7 +185,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-02-turbo", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -176,11 +206,22 @@ "description": "MiniMax 语音大模型能够根据上下文，智能预测文本的情绪、语调等信息，并生成超自然、高保真、个性化的语音。在社交、播客、有声书、新闻资讯、教育、数字人等多种场景中展现出强大的实力。", "features": [], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/speech-02-hd", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "9.9", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + }, + { + "priceUnit": "每万字符", + "price": "3.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -208,7 +249,6 @@ "maxInputTokens": 10000, "inferenceProvider": "mini-max", "name": "speech-02-hd", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json index 261b3171..4bacc726 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json @@ -14,49 +14,33 @@ "description": "图片分割模型是AI试衣OutfitAnyone的辅助模型，可对模特图、服饰图进行分割，用于试衣图片的前后处理。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-parsing-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-15T13:13:06.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣OutfitAnyone-图片分割", "docUrl": "https://help.aliyun.com/document_detail/2865249.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/vision/image-process/process' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"aitryon-parsing-v1\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250630/bakbqz/aitryon_parse_model.png\"\n },\n \"parameters\": {\n \"clothes_type\": [\"upper\"]\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json index 2516bb8d..9bec1d16 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json @@ -16,36 +16,33 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-04-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣-Plus版", "docUrl": "https://help.aliyun.com/document_detail/2881846.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-plus\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json index 5fa687a4..20c35639 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json @@ -14,36 +14,117 @@ "description": "图片精修是对AI试衣生成的效果图进行二次生成，输出还原度更高的精修试衣效果图。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon-refiner", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", - "inferenceProvider": "bailian", - "name": "AI试衣OutfitAnyone-图片精修", - "docUrl": "https://help.aliyun.com/document_detail/2796663.html", - "predictConfig": [ + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "图片生成数量<=25", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25 + }, + { + "rangeStart": 25, + "rangeName": "25<图片生成数量<=125", + "prices": [ + { + "priceUnit": "每张", + "price": "0.275", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 125 + }, { - "name": "sample_models" + "rangeStart": 125, + "rangeName": "125<图片生成数量<=250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 250 }, { - "name": "sample_suits" + "rangeStart": 250, + "rangeName": "250<图片生成数量<=1250", + "prices": [ + { + "priceUnit": "每张", + "price": "0.225", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 1250 }, { - "name": "sample_tops" + "rangeStart": 1250, + "rangeName": "1250<图片生成数量<=2500", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 2500 }, { - "name": "sample_bottoms" + "rangeStart": 2500, + "rangeName": "2500<图片生成数量<=25000", + "prices": [ + { + "priceUnit": "每张", + "price": "0.175", + "type": "image_number", + "priceName": "图片生成" + } + ], + "rangeEnd": 25000 + }, + { + "rangeStart": 25000, + "rangeName": "25000<图片生成数量", + "prices": [ + { + "priceUnit": "每张", + "price": "0.15", + "type": "image_number", + "priceName": "图片生成" + } + ] } ], + "capabilities": [ + "IG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2024-06-28T11:03:13.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", + "name": "AI试衣OutfitAnyone-图片精修", + "docUrl": "https://help.aliyun.com/document_detail/2796663.html", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon-refiner\",\n \"input\": {\n \"top_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-top.jpg\",\n \"bottom_garment_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-bottom.jpg\",\n \"person_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/sample-person.png\",\n \"coarse_image_url\": \"https://dashscope-swap.oss-cn-beijing.aliyuncs.com/aa-test/result.png\"\n },\n \"parameters\": {\n \"gender\": \"woman\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json index a1747bfc..de21e8c9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json @@ -16,36 +16,33 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "aitryon", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-24T10:29:05.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "AI试衣-基础版", "docUrl": "https://help.aliyun.com/document_detail/2796626.html", - "predictConfig": [ - { - "name": "sample_models" - }, - { - "name": "sample_suits" - }, - { - "name": "sample_tops" - }, - { - "name": "sample_bottoms" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"aitryon\",\n \"input\": {\n \"person_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/ubznva/model_person.png\",\n \"top_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/epousa/short_sleeve.jpeg\",\n \"bottom_garment_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250626/rchumi/pants.jpeg\" \n },\n \"parameters\": {\n \"resolution\": -1,\n \"restore_face\": true\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json index 5f83317b..bbdba13c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json @@ -12,49 +12,33 @@ "description": "AnimateAnyone-detect是辅助AnimateAnyone的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-detect-gen2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:28:33.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone-detect", "docUrl": "https://help.aliyun.com/document_detail/2786465.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"animate-anyone-detect-gen2\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {}\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json index 49465b13..0242409f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json @@ -14,49 +14,33 @@ "description": "AnimateAnyone是一款视频生成模型，可基于人物图片和动作模板生成人物全身动作视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:25:42.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone", "docUrl": "https://help.aliyun.com/document_detail/2786464.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-gen2\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/ythwqz/aa-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"template_id\": \"AACT.xxx.xxx-xxx.xxx\" \n },\n \"parameters\": {\n \"use_ref_img_bg\": false,\n \"video_ratio\": \"9:16\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json index a6232ced..5d5b1624 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json @@ -14,49 +14,33 @@ "description": "AnimateAnyone-Template是辅助AnimateAnyone的动作模板生成模型，可基于视频提取人物动作并制作模板。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "animate-anyone-template-gen2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:27:17.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "舞动人像AnimateAnyone-template", "docUrl": "https://help.aliyun.com/document_detail/2807955.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/aa-template-generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"animate-anyone-template-gen2\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241210/cwjmsz/1.mp4\"\n },\n \"parameters\": {}\n }'\n" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json index 62f0c0f7..33e35cc7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json +++ b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json @@ -15,10 +15,15 @@ "collectionTag": "", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3.5-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,36 +42,14 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-02-27T09:13:23.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3.5-flash大模型", "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-flash\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-flash\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -84,10 +67,15 @@ "description": "CosyVoice-v3.5-Plus是通义实验室CosyVoice系列的超高表现力语音合成大模型。对声音克隆和声音设计的语音合成效果进行全面升级，确保说话人高相似度的前提下，支持free-style指令控制，合成风格丰富多样。较之前版本大幅减少首包延迟，同时提高发音准确率，改善韵律和音质。支持跨多语种（中、英、德、法、俄、日、韩、葡、泰、印尼、越南）超自然听感实时语音合成。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3.5-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1.5", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -106,15 +94,14 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-02-27T09:13:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3.5-plus大模型", "docUrl": "https://help.aliyun.com/document_detail/2842586.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", - "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", + "python": "import os\nimport time\nimport dashscope\nfrom dashscope.audio.tts_v2 import VoiceEnrollmentService, SpeechSynthesizer\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ndashscope.api_key = os.getenv(\"\"DASHSCOPE_API_KEY\"\")\nif not dashscope.api_key:\n raise ValueError(\"\"DASHSCOPE_API_KEY environment variable not set.\"\")\n\nTARGET_MODEL = \"\"cosyvoice-v3.5-plus\"\" \nVOICE_PREFIX = \"\"myvoice\"\"\n# Public network accessible audio URL. Please replace it with your own.\nAUDIO_URL = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\"\n\nprint(\"\"--- Step 1: Creating voice enrollment ---\"\")\nservice = VoiceEnrollmentService()\ntry:\n voice_id = service.create_voice(\n target_model=TARGET_MODEL,\n prefix=VOICE_PREFIX,\n url=AUDIO_URL\n )\n print(f\"\"Voice enrollment submitted successfully. Request ID: {service.get_last_request_id()}\"\")\n print(f\"\"Generated Voice ID: {voice_id}\"\")\nexcept Exception as e:\n print(f\"\"Error during voice creation: {e}\"\")\n raise e\n\nprint(\"\"\\n--- Step 2: Polling for voice status ---\"\")\nmax_attempts = 30\npoll_interval = 10\nfor attempt in range(max_attempts):\n try:\n voice_info = service.query_voice(voice_id=voice_id)\n status = voice_info.get(\"\"status\"\")\n print(f\"\"Attempt {attempt + 1}/{max_attempts}: Voice status is '{status}'\"\")\n \n if status == \"\"OK\"\":\n print(\"\"Voice is ready for synthesis.\"\")\n break\n elif status == \"\"UNDEPLOYED\"\":\n print(f\"\"Voice processing failed with status: {status}. Please check audio quality or contact support.\"\")\n raise RuntimeError(f\"\"Voice processing failed with status: {status}\"\")\n time.sleep(poll_interval)\n except Exception as e:\n print(f\"\"Error during status polling: {e}\"\")\n time.sleep(poll_interval)\nelse:\n print(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n raise RuntimeError(\"\"Polling timed out. The voice is not ready after several attempts.\"\")\n\n\nprint(\"\"\\n--- Step 3: Synthesizing speech with the new voice ---\"\")\ntry:\n synthesizer = SpeechSynthesizer(model=TARGET_MODEL, voice=voice_id)\n text_to_synthesize = \"\"How is the weather today?\"\"\n \n audio_data = synthesizer.call(text_to_synthesize)\n print(f\"\"Speech synthesis successful. Request ID: {synthesizer.get_last_request_id()}\"\")\n\n output_file = \"\"my_custom_voice_output.mp3\"\"\n with open(output_file, \"\"wb\"\") as f:\n f.write(audio_data)\n print(f\"\"Audio saved to {output_file}\"\")\n\nexcept Exception as e:\n print(f\"\"Error during speech synthesis: {e}\"\")", + "java": "import com.alibaba.dashscope.audio.ttsv2.enrollment.Voice;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentParam;\nimport com.alibaba.dashscope.audio.ttsv2.enrollment.VoiceEnrollmentService;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Collections;\n\npublic class Main {\n public static void main(String[] args) {\n Constants.baseWebsocketApiUrl = \"\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference\"\";\n Constants.baseHttpApiUrl = \"\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\";\n\n String apiKey = System.getenv(\"\"DASHSCOPE_API_KEY\"\");\n String targetModel = \"\"cosyvoice-v3.5-plus\"\";\n String prefix = \"\"myvoice\"\";\n // Public network accessible audio URL. Please replace it with your own.\n String fileUrl = \"\"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/cosyvoice/cosyvoice-zeroshot-sample.wav\"\";\n String cloneModelName = \"\"voice-enrollment\"\";\n\n Voice myVoice = null;\n try {\n VoiceEnrollmentService service = new VoiceEnrollmentService(apiKey);\n myVoice = service.createVoice(\n targetModel,\n prefix,\n fileUrl,\n VoiceEnrollmentParam.builder()\n .model(cloneModelName)\n .languageHints(Collections.singletonList(\"\"zh\"\")).build());\n\n System.out.println(\"\"Voice creation submitted. Request ID: \"\" + service.getLastRequestId());\n System.out.println(\"\"Generated Voice ID: \"\" + myVoice.getVoiceId());\n } catch (Exception e) {\n System.out.println(\"\"Failed to create voice: \"\" + e.getMessage());\n }\n\n streamAudioDataToSpeaker(targetModel, myVoice.getVoiceId());\n System.exit(0);\n }\n\n public static void streamAudioDataToSpeaker(String targetModel, String voice) {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n .apiKey(System.getenv(\"\"DASHSCOPE_API_KEY\"\"))\n .model(targetModel)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"\"How is the weather today?\"\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"\"bye\"\");\n }\n if (audio != null) {\n File file = new File(\"\"output.mp3\"\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -132,10 +119,15 @@ "description": "合成能力：CosyVoice-v3-Flash是通义实验室CosyVoice系列最新版高性能的语音合成大模型，较之前版本在自然度、音质、韵律、情感表现力上有更好的表现。该模型支持文本至语音的实时流式合成。克隆能力：CosyVoice-v3-Flash也是通义实验室CosyVoice系列最新版的语音克隆大模型，较之前版本提升了发音准确性、音色相似度，并且增加了更多小语种支持（德、西、法、意、俄）。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -153,30 +145,8 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-17T06:35:36.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3-flash大模型", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { @@ -198,10 +168,15 @@ "description": "克隆能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音克隆大模型，具有更好的音质和复刻相似度，适用于更专业的场景。仅需提供5-20s的参考音频，即可迅速生成高度相似且听感自然的定制声音。合成能力：CosyVoice-v3-plus是通义实验室CosyVoice系列最新版的语音合成大模型，具有更好的音质和表现力，适用于更专业的场景。该模型支持文本至语音的实时流式合成。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-v3-plus", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -219,37 +194,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-03T01:45:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音生成CosyVoice-v3-plus大模型", "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "情感", - "key": "emotion", - "default": "neutral", - "tip": "合成音频说话的情感" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { @@ -271,45 +218,29 @@ "description": "声音复刻Cosyvoice大模型，依托先进的大模型技术进行特征提取，从而完成声音的复刻，且无需训练过程。仅需提供时长较短的音频，即可迅速生成高度相似且听感自然的定制声音。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "cosyvoice-clone-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", - "name": "声音复刻CosyVoice大模型", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "声音复刻CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2861519.html" }, { "inferenceMetadata": { @@ -320,50 +251,31 @@ "Text" ] }, - "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [ - "model-experience" - ], + "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v1", + "model": "cosyvoice-v2", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "capabilities": [ "TTS" ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", - "inferenceProvider": "bailian", - "name": "语音合成CosyVoice大模型", + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "语音生成cosyvoice-v2大模型", "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" } } } @@ -377,53 +289,39 @@ "Text" ] }, - "description": "cosyvoice-V2是通义实验室依托大规模预训练语言模型，在深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", - "features": [], + "description": "CosyVoice 是通义实验室依托大规模预训练语言模型，深度融合文本理解和语音生成的新一代生成式语音合成大模型，支持文本至语音的实时流式合成。", + "features": [ + "model-experience" + ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, - "model": "cosyvoice-v2", + "model": "cosyvoice-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "2", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "capabilities": [ "TTS" ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2025-05-27T10:10:42.000+00:00", - "inferenceProvider": "bailian", - "name": "语音生成cosyvoice-v2大模型", - "docUrl": "https://help.aliyun.com/document_detail/2817551.html", - "predictConfig": [ - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "字级别时间戳", - "key": "enableWordTimestamp", - "default": false + "versionTag": "MAJOR", + "latestOnlineAt": "2025-03-18T02:20:34.000+00:00", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ], + }, + "inferenceProvider": "aliyun-bailian", + "name": "语音合成CosyVoice大模型", + "docUrl": "https://help.aliyun.com/document_detail/2817551.html", "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v2\"\nvoice = \"longxiaochun_v2\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", - "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v2\";\n private static String voice = \"longxiaochun_v2\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# 若没有将API Key配置到环境变量中，需将your-api-key替换为自己的API Key\n# dashscope.api_key = \"your-api-key\"\n\nmodel = \"cosyvoice-v1\"\nvoice = \"longxiaochun\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"今天天气怎么样？\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "java": "import com.alibaba.dashscope.audio.ttsv2.SpeechSynthesisParam;\nimport com.alibaba.dashscope.audio.ttsv2.SpeechSynthesizer;\n\nimport java.io.File;\nimport java.io.FileOutputStream;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\n\npublic class Main {\n private static String model = \"cosyvoice-v1\";\n private static String voice = \"longxiaochun\";\n\n public static void streamAudioDataToSpeaker() {\n SpeechSynthesisParam param =\n SpeechSynthesisParam.builder()\n // 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将your-api-key替换为自己的API Key\n // .apiKey(\"your-api-key\")\n .model(model)\n .voice(voice)\n .build();\n\n SpeechSynthesizer synthesizer = new SpeechSynthesizer(param, null);\n ByteBuffer audio = null;\n try {\n audio = synthesizer.call(\"今天天气怎么样？\");\n } catch (Exception e) {\n throw new RuntimeException(e);\n } finally {\n synthesizer.getDuplexApi().close(1000, \"bye\");\n }\n if (audio != null) {\n File file = new File(\"output.mp3\");\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audio.array());\n } catch (IOException e) {\n throw new RuntimeException(e);\n }\n }\n }\n\n public static void main(String[] args) {\n streamAudioDataToSpeaker();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json index 00cd5780..602a9fff 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json +++ b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json @@ -71,59 +71,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V4-Pro", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -213,59 +160,6 @@ "name": "DeepSeek-V4-Flash", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 4000, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -364,18 +258,18 @@ ], "qpmInfo": { "model-default-actual": { - "count_limit_period": 15, + "count_limit_period": 60, "usage_limit": 1200000, "usage_limit_field": "total_tokens", - "count_limit": 3750, + "count_limit": 15000, "usage_limit_period": 60, "type": "model-default" }, "model-default": { - "count_limit_period": 15, + "count_limit_period": 60, "usage_limit": 1200000, "usage_limit_field": "total_tokens", - "count_limit": 3750, + "count_limit": 15000, "usage_limit_period": 60, "type": "model-default" } @@ -398,49 +292,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -527,49 +378,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Deepseek-V3.2-Exp", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -663,49 +471,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3.1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -811,43 +576,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -933,23 +661,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1052,23 +763,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1150,23 +844,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-7B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1248,23 +925,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-32B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1346,23 +1006,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-14B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { @@ -1424,23 +1067,6 @@ "inferenceProvider": "aliyun-bailian", "name": "DeepSeek-R1-Distill-Qwen-1.5B", "docUrl": "https://help.aliyun.com/document_detail/2868565.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/embedding.json b/skills/bailian-docs-llm-wiki/models/groups/embedding.json index 387fe2bf..fe9ab027 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/embedding.json @@ -14,10 +14,21 @@ "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。本模型（tongyi-embedding-vision-flash）是轻量化版本，在视觉向量化上具备极高性价比。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-embedding-vision-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -41,40 +52,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T09:09:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视觉向量-flash", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-flash\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-flash\",\n input=input\n)\n\nprint(resp)\n", "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-flash\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } @@ -93,10 +77,21 @@ "description": "Embedding-Vision是基于LLM底座的视觉多模态表征模型，具有以视觉为中心、领域性能优异（电商、 安防、相册/图库、自驾等）、高性价比的特点。兼容文本、图像、视频3种模态，可应用于以图搜图、以文搜图、以文搜视频，以视频搜视频等下游任务场景。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-embedding-vision-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -120,40 +115,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T09:09:31.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视觉向量-plus", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"tongyi-embedding-vision-plus\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"tongyi-embedding-vision-plus\",\n input=input\n)\n\nprint(resp)\n", "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"tongyi-embedding-vision-plus\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } @@ -172,10 +140,21 @@ "description": "通义实验室基于预训练多模态大模型构建的多模态向量模型。该模型根据用户的输入生成高维连续向量，这些输入可以是文本、图片或视频。多模态向量在可应用于图片搜索、文搜图、视频搜索、图片分类和视频内容审核等下游任务中。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "multimodal-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.9", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -202,40 +181,13 @@ "latestOnlineAt": "2024-12-23T11:55:31.000+00:00", "contextWindow": 0, "maxInputTokens": 512, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用多模态向量", "docUrl": "https://help.aliyun.com/document_detail/2712517.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"multimodal-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"multimodal-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"multimodal-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json index 114ba204..e4b79d5d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json @@ -12,49 +12,33 @@ "description": "EMO-Detect是辅助EMO的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emo-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:16:51.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "悦动人像EMO-detect", "docUrl": "https://help.aliyun.com/document_detail/2786463.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emo-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/aejgyj/input_audio.mp3\",\n \"face_bbox\":[302,286,610,593],\n \"ext_bbox\":[71,9,840,778]\n },\n \"parameters\": {\n \"style_level\": \"normal\"\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json index d2b6cb99..09b61cd4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json @@ -14,49 +14,39 @@ "description": "EMO是一款视频生成模型，可基于人物图片生成高质量的人物肖像动态视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emo-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_duration_1-1", + "priceName": "视频生成（1:1画幅视频）" + }, + { + "priceUnit": "每秒", + "price": "0.16", + "type": "video_duration_3-4", + "priceName": "视频生成（3:4画幅视频）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:16:42.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "悦动人像EMO", "docUrl": "https://help.aliyun.com/document_detail/2786461.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"emo-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/yhdvfg/emo-%E5%9B%BE%E7%89%87.png\"\n },\n \"parameters\": {\n \"ratio\": \"1:1\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json index ae9f1a1a..e9a48c75 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json @@ -12,49 +12,33 @@ "description": "表情包Emoji-Detect是辅助表情包Emoji生成的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emoji-detect-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-16T09:55:26.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "表情包Emoji-detect", "docUrl": "https://help.aliyun.com/document_detail/2865371.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-detect-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\"\n },\n \"parameters\": {\n \"ratio\":\"1:1\"\n }\n }'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json index 5c8c9829..dd438fae 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json @@ -14,49 +14,33 @@ "description": "表情包emoji是一款人脸动效视频生成模型，可基于人脸图片和预设的人脸动态模板，生成人脸动效视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "emoji-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-16T09:46:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "表情包Emoji", "docUrl": "https://help.aliyun.com/document_detail/2865374.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"emoji-v1\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250912/uopnly/emoji-%E5%9B%BE%E5%83%8F%E6%A3%80%E6%B5%8B.png\",\n \"driven_id\": \"mengwa_kaixin\",\n \"face_bbox\": [212,194,460,441],\n \"ext_bbox\": [63,30,609,575]\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json index b5948c96..b533698d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-facedetect.json @@ -12,45 +12,15 @@ "description": "对用户上传的人物图像进行检测，判断其中所包含的人脸是否符合facechain微调所需的标准，检测维度包括人脸数量、大小、角度、光照、清晰度等多维度，支持图像组输入，并返回每张图像对应的检测结果。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "facechain-facedetect", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:10:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "FaceChain人物图像检测", - "docUrl": "https://help.aliyun.com/document_detail/2712507.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] + "docUrl": "https://help.aliyun.com/document_detail/2712507.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json index d32c1391..7bd76785 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json @@ -14,45 +14,29 @@ "description": "基于人物形象训练已经得到的形象，可以继续通过人物生成写真模型完成该形象的写真生成，支持多种预设风格，包括证件照、商务写真等。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "facechain-generation", + "prices": [ + { + "priceUnit": "每张", + "price": "0.18", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:12:23.000+00:00", - "inferenceProvider": "bailian", - "name": "FaceChain人物写真生成", - "docUrl": "https://help.aliyun.com/document_detail/2712501.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "FaceChain人物写真生成", + "docUrl": "https://help.aliyun.com/document_detail/2712501.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json index ec9bd5b0..b49111d1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "farui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,50 +57,29 @@ "latestOnlineAt": "2024-05-14T13:32:40.000+00:00", "contextWindow": 12000, "maxInputTokens": 12000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "通义法睿-Plus-32K", "docUrl": "https://help.aliyun.com/document_detail/2778998.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"farui-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"farui-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"farui-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"farui-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"farui-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"farui-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json index 14412c98..7436007c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json @@ -14,10 +14,15 @@ "description": "百聆2026年6月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。支持context上下文能力，可转写5分钟以内的音频。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-flash-2026-06-15", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -35,40 +40,13 @@ ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-06-17T08:56:52.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-Flash-2026-06-15", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header \"Content-Type: application/json\" \\\n --header \"X-DashScope-SSE: enable\" \\\n --data '{\n \"model\": \"fun-asr-flash-2026-06-15\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"input_audio\",\n \"input_audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"format\": \"wav\",\n \"sample_rate\": \"16000\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/2869541.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json index 79fb2a3d..020604ef 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json @@ -14,10 +14,15 @@ "description": "通义实验室新一代端到端语音识别大模型的实时版，基于领先的自研语音技术，具备卓越的上下文感知和高精度语音转写能力。基于端到端架构，Fun-ASR 集成了创新的 RAG 技术，支持大规模热词自定义、敏感/语气词自动过滤、ITN 规范化、标点预测等多维功能，显著提升了整体识别准确率和语境贴合度。同时，Fun-ASR 支持中英文自由切换，多地区方言覆盖，具备更强的噪声鲁棒性，适应多样复杂环境。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -35,27 +40,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T11:05:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR实时语音识别", "docUrl": "https://help.aliyun.com/document_detail/2842554.html", - "predictConfig": [ - { - "name": "开启语义断句", - "key": "semantic_punctuation_enabled", - "default": false, - "tip": "开启语义断句后则将关闭VAD(语音活动检测)断句，具体见说明文档" - }, - { - "name": "VAD静音阈值", - "key": "max_sentence_silence", - "default": 1300, - "tip": "VAD(语音活动检测)断句的静音时长阈值(单位为ms)", - "range": [ - 200, - 6000 - ] - } - ], "samples": { "dashscope": { "default": { @@ -77,10 +64,15 @@ "description": "通义百聆推出的新一代轻量级实时语音识别模型，依托自研的先进语音技术架构，具备强大的上下文理解能力。专为中文电话客服场景设计：覆盖多地区方言口音，在低采样率、低信噪比环境下实现低延迟、高准确率的流式转写，满足高效部署需求。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-flash-8k-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -100,37 +92,10 @@ "versionTag": "MAJOR", "equivalentSnapshot": "fun-asr-flash-8k-realtime-2026-01-28", "latestOnlineAt": "2026-02-12T11:34:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-Flash-8k实时语音识别", "docUrl": "https://help.aliyun.com/document_detail/2842554.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json index eac84c5d..14f06761 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json @@ -14,10 +14,15 @@ "description": "百聆2026年4月更新的大模型ASR版本，全面支持汉语传统七大方言体系（官话/吴/湘/赣/客/闽/粤），并适配 20+ 地区口音官话。针对中文古诗词的韵律、节奏与文言表达特点进行专项优化，提升对古诗词内容的识别准确率，适用于文化传承、教育讲解、有声读物等场景。优化标点预测与文本归一化能力，使输出文本更符合书面表达习惯，数字、日期、金额等信息自动转换为标准格式，增强内容的可读性与专业性。同时语种扩展至英语、日语、韩语、越南语、泰语、印尼语、马来语、菲律宾语、印地语、阿拉伯语、法语、德语、西班牙语、葡萄牙语、俄语、意大利语、荷兰语、瑞典语、丹麦语、芬兰语、挪威语、希腊语、波兰语、捷克语、匈牙利语、罗马尼亚、保加利亚语、克罗地亚语、斯洛伐克语等，共计30个语种。此版本等同于2025年11月7日的快照版本。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -35,10 +40,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR语音识别", "docUrl": "https://help.aliyun.com/document_detail/2880903.html", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -59,10 +63,15 @@ "description": "百聆多语言语音识别大模型，支持超过31种语言，支持语种自由切换，出海用户首推，尤其东南亚出海。fun-asr为该模型的升级版本，建议切换使用fun-asr。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "fun-asr-mtl", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -80,10 +89,15 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-25T06:03:29.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Fun-ASR-MTL", "docUrl": "https://help.aliyun.com/document_detail/2978300.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json index 3751a275..e6046966 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json @@ -14,10 +14,15 @@ "description": "百聆音乐生成大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "fun-music-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.002", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -36,40 +41,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-05-06T12:15:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "音乐生成", "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "curl": "curl -X POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-v1\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030448.html" } } @@ -87,10 +65,15 @@ "description": "百聆音乐生成preview版大模型（Fun音乐大模型）支持输入开放性歌曲的创作要求或歌词，生成整首男/女声演唱的中文或英文歌曲。歌曲通俗易懂，情绪由浅入深，是人类灵感与大模型能力的完美结合。本次版本为预览快照版", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "fun-music-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.005", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -108,40 +91,13 @@ ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-06-01T07:20:31.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "音乐生成 Preview", "docUrl": "https://help.aliyun.com/document_detail/3030448.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", + "curl": "curl -X POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/music/generation' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"fun-music-preview\",\n \"input\": {\n \"prompt\": \"夏日清新民谣，木吉他与口琴伴奏，轻快节奏，适合旅行Vlog背景音乐\",\n \"gender\": \"female\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030448.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json index ad33f246..b8ca49e8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json @@ -19,10 +19,27 @@ "structured-outputs" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -51,71 +68,23 @@ "contextWindow": 1048576, "maxInputTokens": 1048576, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.2", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -138,9 +107,6 @@ "structured-outputs" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-5.1", "qpmInfo": { "model-default-actual": { @@ -160,6 +126,82 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" @@ -170,72 +212,24 @@ "latestOnlineAt": "2026-04-14T11:34:59.000+00:00", "contextWindow": 202745, "maxInputTokens": 202745, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.1", "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'glm-5.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -250,115 +244,111 @@ "Text" ] }, - "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", + "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-4.7", + "model": "glm-5", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, - "usage_limit": 1000000, + "usage_limit": 100000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 60, + "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 6, - "usage_limit": 1000000, + "usage_limit": 100000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 60, + "usage_limit_period": 6, "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" ], + "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 16384, - "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", + "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", "contextWindow": 202752, "maxInputTokens": 169984, - "offlineInfo": { - "inference": { - "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" - } - }, - "inferenceProvider": "bailian", - "name": "GLM-4.7", - "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], + "inferenceProvider": "aliyun-bailian", + "name": "GLM-5", + "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -373,110 +363,116 @@ "Text" ] }, - "description": "GLM-5是面向Coding与Agent场景的新一代大模型，在复杂系统工程与长程任务中达到开源 SOTA，真实编程体验逼近 Claude Opus 级别；基于 744B 新基座、异步强化学习与稀疏注意力，实现从“写代码”到“写工程”的全面升级。", + "description": "智谱最新旗舰，具备更强的编程能力与更稳定的多步骤推理/执行能力。总参数355B，支持长程任务规划、编码、工具协同，问答自然、写作沉浸、创意角色扮演能力强。", "features": [ "model-experience", "function-calling", "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, - "model": "glm-5", + "model": "glm-4.7", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, - "usage_limit": 100000, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 6, + "usage_limit_period": 60, "type": "model-default" }, "model-default": { "count_limit_period": 6, - "usage_limit": 100000, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", "count_limit": 50, - "usage_limit_period": 6, + "usage_limit_period": 60, "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "TG", "Reasoning" ], - "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 16384, - "latestOnlineAt": "2026-02-18T04:44:16.000+00:00", + "latestOnlineAt": "2025-12-25T06:06:48.000+00:00", "contextWindow": 202752, "maxInputTokens": 169984, - "inferenceProvider": "bailian", - "name": "GLM-5", - "docUrl": "https://www.alibabacloud.com/help/zh/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } - ], + }, + "inferenceProvider": "aliyun-bailian", + "name": "GLM-4.7", + "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.7\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.7\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.7\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.7\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -497,9 +493,6 @@ "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.6", "qpmInfo": { "model-default-actual": { @@ -519,6 +512,58 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=200k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 204800 + } + ], "capabilities": [ "Reasoning", "TG" @@ -530,75 +575,27 @@ "maxInputTokens": 169984, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.6", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.6\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.6\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.6\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.6\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -618,9 +615,6 @@ "model-experience" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.5", "qpmInfo": { "model-default-actual": { @@ -640,6 +634,46 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], "capabilities": [ "TG" ], @@ -648,71 +682,29 @@ "latestOnlineAt": "2025-08-06T11:38:01.000+00:00", "contextWindow": 131072, "maxInputTokens": 98304, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.5", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } @@ -732,9 +724,6 @@ "model-experience" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-4.5-air", "qpmInfo": { "model-default-actual": { @@ -754,6 +743,46 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + } + ], "capabilities": [ "TG" ], @@ -762,71 +791,29 @@ "latestOnlineAt": "2025-08-06T12:02:34.000+00:00", "contextWindow": 131072, "maxInputTokens": 98304, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "GLM-4.5-Air", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-4.5-air\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-4.5-air\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-4.5-air\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-4.5-air\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2974045.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json index 69f63105..81f893bb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json @@ -21,6 +21,26 @@ ], "provider": "zhipu-ai", "model": "glm-5.2-fast-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "56", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -51,7 +71,7 @@ ], "contextWindow": 1048576, "maxInputTokens": 1048576, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GLM-5.2-Fast-Preview", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "category": "Third-party", diff --git a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json index 3f5f3c4c..1741f402 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json @@ -15,10 +15,21 @@ "description": "GUI系列图形界面交互基础模型，针对手机端与电脑端图形界面理解与交互任务，性能优于开源版同类GUI模型。全面升级跨平台界面理解与多步任务规划，支持跨应用复杂任务；具备精细化动作执行与多角色多智能体协作能力，胜任真实复杂交互场景。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "gui-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -45,50 +56,23 @@ "latestOnlineAt": "2025-11-12T07:17:19.000+00:00", "contextWindow": 256000, "maxInputTokens": 254976, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "GUI-Plus", "docUrl": "https://help.aliyun.com/document_detail/2997010.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", - "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", + "node": "import OpenAI from \"openai\";\n\nconst systemPrompt = `# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by \\`action=key\\`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by \\`action=wait\\`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by \\`action=terminate\\`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.`;\n\nconst messages = [\n {\n role: \"system\",\n content: systemPrompt\n },\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n type: \"text\",\n text: \"帮我打开浏览器。\"\n }\n ]\n }\n];\n\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"gui-plus\",\n messages: messages,\n });\n console.log(response.choices[0].message.content);\n}\nmain();\n", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* \\`key\\`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* \\`type\\`: Type a string of text on the keyboard.\\\\n* \\`mouse_move\\`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click\\`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`left_click_drag\\`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* \\`right_click\\`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`middle_click\\`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`double_click\\`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* \\`triple_click\\`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* \\`scroll\\`: Performs a scroll of the mouse scroll wheel.\\\\n* \\`hscroll\\`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* \\`wait\\`: Wait specified seconds for the change to happen.\\\\n* \\`terminate\\`: Terminate the current task and report its completion status.\\\\n* \\`answer\\`: Answer a question.\\\\n* \\`interact\\`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by \\`action=key\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by \\`action=type\\`, \\`action=answer\\` and \\`action=interact\\`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by \\`action=mouse_move\\` and \\`action=left_click_drag\\`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by \\`action=scroll\\` and \\`action=hscroll\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by \\`action=wait\\`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by \\`action=terminate\\`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"帮我打开浏览器\"\n }\n ]\n }\n ]\n}\nEOF\n", + "python": "import os\nfrom openai import OpenAI\n\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": system_prompt,\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n },\n {\"type\": \"text\", \"text\": \"帮我打开浏览器。\"},\n ],\n },\n]\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(model=\"gui-plus\", messages=messages)\nprint(completion.choices[0].message.content)\n", "docUrl": "https://help.aliyun.com/document_detail/2997010.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d @- < XML tags:\\n\\n{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* key: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* type: Type a string of text on the keyboard.\\\\n* mouse_move: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* left_click: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* left_click_drag: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* right_click: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* middle_click: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* double_click: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* triple_click: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* scroll: Performs a scroll of the mouse scroll wheel.\\\\n* hscroll: Performs a horizontal scroll (mapped to regular scroll).\\\\n* wait: Wait specified seconds for the change to happen.\\\\n* terminate: Terminate the current task and report its completion status.\\\\n* answer: Answer a question.\\\\n* interact: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by action=key.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by action=type, action=answer and action=interact.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by action=mouse_move and action=left_click_drag.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by action=scroll and action=hscroll.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by action=wait.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by action=terminate.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\\n\\nFor each function call, return a json object with function name and arguments within XML tags:\\n\\n{\\\"name\\\": , \\\"arguments\\\": }\\n\\n\\n# Response format\\n\\nResponse format for every step:\\n1) Action: a short imperative describing what to do in the UI.\\n2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\nRules:\\n- Output exactly in the order: Action, .\\n- Be brief: one for Action.\\n- Do not output anything else outside those two parts.\\n- If finishing, use action=terminate in the tool call.\"\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"\n },\n {\n \"text\": \"帮我打开浏览器。\"\n }\n ]\n }\n ]\n }\n}\nEOF", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\nsystem_prompt = '''\n# Tools\n\nYou may call one or more functions to assist with the user query.\n\nYou are provided with function signatures within XML tags:\n\n{\"type\": \"function\", \"function\": {\"name\": \"computer_use\", \"description\": \"Use a mouse and keyboard to interact with a computer, and take screenshots.\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\n* The screen's resolution is 1000x1000.\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\", \"parameters\": {\"properties\": {\"action\": {\"description\": \"The action to perform. The available actions are:\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\n* `type`: Type a string of text on the keyboard.\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\n* `wait`: Wait specified seconds for the change to happen.\\n* `terminate`: Terminate the current task and report its completion status.\\n* `answer`: Answer a question.\\n* `interact`: Resolve the blocking window by interacting with the user.\", \"enum\": [\"key\", \"type\", \"mouse_move\", \"left_click\", \"left_click_drag\", \"right_click\", \"middle_click\", \"double_click\", \"triple_click\", \"scroll\", \"hscroll\", \"wait\", \"terminate\", \"answer\", \"interact\"], \"type\": \"string\"}, \"keys\": {\"description\": \"Required only by `action=key`.\", \"type\": \"array\"}, \"text\": {\"description\": \"Required only by `action=type`, `action=answer` and `action=interact`.\", \"type\": \"string\"}, \"coordinate\": {\"description\": \"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\", \"type\": \"array\"}, \"pixels\": {\"description\": \"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\", \"type\": \"number\"}, \"time\": {\"description\": \"The seconds to wait. Required only by `action=wait`.\", \"type\": \"number\"}, \"status\": {\"description\": \"The status of the task. Required only by `action=terminate`.\", \"type\": \"string\", \"enum\": [\"success\", \"failure\"]}}, \"required\": [\"action\"], \"type\": \"object\"}}}\n\n\nFor each function call, return a json object with function name and arguments within XML tags:\n\n{\"name\": , \"arguments\": }\n\n\n# Response format\n\nResponse format for every step:\n1) Action: a short imperative describing what to do in the UI.\n2) A single ... block containing only the JSON: {\"name\": , \"arguments\": }.\n\nRules:\n- Output exactly in the order: Action, .\n- Be brief: one for Action.\n- Do not output anything else outside those two parts.\n- If finishing, use action=terminate in the tool call.\n'''\n\nmessages = [{\n \"role\": \"system\",\n \"content\": system_prompt\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"},\n {\"text\": \"帮我打开浏览器。\"}]\n}]\n\nresponse = dashscope.MultiModalConversation.call(\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'gui-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n\n String systemPrompt = \"# Tools\\n\\n\" +\n \"You may call one or more functions to assist with the user query.\\n\\n\" +\n \"You are provided with function signatures within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"type\\\": \\\"function\\\", \\\"function\\\": {\\\"name\\\": \\\"computer_use\\\", \\\"description\\\": \\\"Use a mouse and keyboard to interact with a computer, and take screenshots.\\\\n* This is an interface to a desktop GUI. You do not have access to a terminal or applications menu. You must click on desktop icons to start applications.\\\\n* Some applications may take time to start or process actions, so you may need to wait and take successive screenshots to see the results of your actions. E.g. if you click on Firefox and a window doesn't open, try wait and taking another screenshot.\\\\n* The screen's resolution is 1000x1000.\\\\n* Make sure to click any buttons, links, icons, etc with the cursor tip in the center of the element. Don't click boxes on their edges unless asked.\\\", \\\"parameters\\\": {\\\"properties\\\": {\\\"action\\\": {\\\"description\\\": \\\"The action to perform. The available actions are:\\\\n* `key`: Performs key down presses on the arguments passed in order, then performs key releases in reverse order.\\\\n* `type`: Type a string of text on the keyboard.\\\\n* `mouse_move`: Move the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click`: Click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `left_click_drag`: Click and drag the cursor to a specified (x, y) pixel coordinate on the screen.\\\\n* `right_click`: Click the right mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `middle_click`: Click the middle mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `double_click`: Double-click the left mouse button at a specified (x, y) pixel coordinate on the screen.\\\\n* `triple_click`: Triple-click the left mouse button at a specified (x, y) pixel coordinate on the screen (simulated as double-click since it's the closest action).\\\\n* `scroll`: Performs a scroll of the mouse scroll wheel.\\\\n* `hscroll`: Performs a horizontal scroll (mapped to regular scroll).\\\\n* `wait`: Wait specified seconds for the change to happen.\\\\n* `terminate`: Terminate the current task and report its completion status.\\\\n* `answer`: Answer a question.\\\\n* `interact`: Resolve the blocking window by interacting with the user.\\\", \\\"enum\\\": [\\\"key\\\", \\\"type\\\", \\\"mouse_move\\\", \\\"left_click\\\", \\\"left_click_drag\\\", \\\"right_click\\\", \\\"middle_click\\\", \\\"double_click\\\", \\\"triple_click\\\", \\\"scroll\\\", \\\"hscroll\\\", \\\"wait\\\", \\\"terminate\\\", \\\"answer\\\", \\\"interact\\\"], \\\"type\\\": \\\"string\\\"}, \\\"keys\\\": {\\\"description\\\": \\\"Required only by `action=key`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"text\\\": {\\\"description\\\": \\\"Required only by `action=type`, `action=answer` and `action=interact`.\\\", \\\"type\\\": \\\"string\\\"}, \\\"coordinate\\\": {\\\"description\\\": \\\"(x, y): The x (pixels from the left edge) and y (pixels from the top edge) coordinates to move the mouse to. Required only by `action=mouse_move` and `action=left_click_drag`.\\\", \\\"type\\\": \\\"array\\\"}, \\\"pixels\\\": {\\\"description\\\": \\\"The amount of scrolling to perform. Positive values scroll up, negative values scroll down. Required only by `action=scroll` and `action=hscroll`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"time\\\": {\\\"description\\\": \\\"The seconds to wait. Required only by `action=wait`.\\\", \\\"type\\\": \\\"number\\\"}, \\\"status\\\": {\\\"description\\\": \\\"The status of the task. Required only by `action=terminate`.\\\", \\\"type\\\": \\\"string\\\", \\\"enum\\\": [\\\"success\\\", \\\"failure\\\"]}}, \\\"required\\\": [\\\"action\\\"], \\\"type\\\": \\\"object\\\"}}}\\n\" +\n \"\\n\\n\" +\n \"For each function call, return a json object with function name and arguments within XML tags:\\n\" +\n \"\\n\" +\n \"{\\\"name\\\": , \\\"arguments\\\": }\\n\" +\n \"\\n\\n\" +\n \"# Response format\\n\\n\" +\n \"Response format for every step:\\n\" +\n \"1) Action: a short imperative describing what to do in the UI.\\n\" +\n \"2) A single ... block containing only the JSON: {\\\"name\\\": , \\\"arguments\\\": }.\\n\\n\" +\n \"Rules:\\n\" +\n \"- Output exactly in the order: Action, .\\n\" +\n \"- Be brief: one for Action.\\n\" +\n \"- Do not output anything else outside those two parts.\\n\" +\n \"- If finishing, use action=terminate in the tool call.\";\n\n MultiModalMessage systemMsg = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", systemPrompt))).build();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i2/O1CN016iJ8ob1C3xP1s2M6z_!!6000000000026-2-tps-3008-1758.png\"),\n Collections.singletonMap(\"text\", \"帮我打开浏览器。\"))).build();\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"gui-plus\")\n .messages(Arrays.asList(systemMsg, userMessage))\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}\n", "docUrl": "https://help.aliyun.com/document_detail/2997010.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json index 69390930..4b7651be 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "gummy-chat-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -40,13 +45,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "一句话识别及翻译V1.0", "docUrl": "https://help.aliyun.com/document_detail/2866122.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json index eefec95c..284c4893 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "gummy-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00015", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,14 +44,13 @@ "latestOnlineAt": "2025-03-04T04:07:10.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "实时语音识别及翻译V1.0", "docUrl": "https://help.aliyun.com/document_detail/2865393.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json index 2e70e68e..16dc8b05 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json @@ -12,16 +12,43 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "HappyHorse-1.1-I2V支持图生视频，进一步提升画面质感、动态表现与跨片段一致性。模型能够更精准地理解输入图像并延续创作意图，在人物皮肤质感、ID跨片段保持、动作流畅度、文字渲染稳定性以及音画同步上带来显著改善，输出更真实自然、细节丰富且一致性更高的高质量视频。", "features": [ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -45,40 +72,14 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T03:16:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } @@ -99,15 +100,28 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -115,7 +129,7 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -127,39 +141,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T17:07:07.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-I2V", "docUrl": "https://help.aliyun.com/document_detail/3029821.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-i2v\",\n \"input\": {\n \"prompt\": \"一只猫在草地上奔跑\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029821.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json index 8b69d7f3..bb531658 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json @@ -17,10 +17,23 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,45 +57,14 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T03:16:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-R2V", "docUrl": "https://help.aliyun.com/document_detail/3030778.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", "docUrl": "https://help.aliyun.com/document_detail/3030778.html" } } @@ -103,15 +85,28 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -119,7 +114,7 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -131,44 +126,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-26T12:42:22.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-R2V", "docUrl": "https://help.aliyun.com/document_detail/3030778.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-r2v\",\n \"input\": {\n \"prompt\": \"[Image 1]中身着红色旗袍的女性，镜头先以侧面中景勾勒旗袍修身剪裁与S型曲线，随即切换至低角度仰拍，捕捉她轻抬玉手展开[Image 2]中的折扇的同时，[Image 3]中的流苏耳坠随头部转动轻盈摆动的细节，最后推近至面部特写，定格在她指尖轻点扇骨、眼波流转间的含蓄风情，多视角全方位展现东方韵味。\",\n \"media\": [\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/mvzfud/hh-v2v-girl.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/fvuihk/hh-v2v2-folding-fan.jpg\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260424/imerii/hh-v2v-earrings.jpg\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'\n\n", "docUrl": "https://help.aliyun.com/document_detail/3030778.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json index edb5bce7..88af0269 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json @@ -16,10 +16,23 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.1-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.6, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "discount": 0.6, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -43,45 +56,14 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-06-16T06:14:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.1-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", "category": "Visual", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.1-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } @@ -101,22 +83,35 @@ "model-experience" ], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-t2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "type": "model-default", "async_user_concurrency_limit": 5 }, "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "type": "model-default", "async_user_concurrency_limit": 5 } @@ -127,44 +122,13 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-21T09:03:16.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-T2V", "docUrl": "https://help.aliyun.com/document_detail/3029820.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-t2v\",\n \"input\": {\n \"prompt\": \"一座由硬纸板和瓶盖搭建的微型城市，在夜晚焕发出生机。一列硬纸板火车缓缓驶过，小灯点缀其间，照亮前路。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"ratio\": \"16:9\",\n \"duration\": 5\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3029820.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json index 01a4bf23..0e68aaf6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json @@ -8,6 +8,7 @@ "Video" ], "request_modality": [ + "Text", "Image", "Video" ] @@ -15,15 +16,28 @@ "description": "HappyHorse-1.0-Video-Edit支持视频编辑，自然语言指令编辑视频，可参考最多5张图片局部或全局编辑视频元素，能够精准复刻视频动态过程，实现更强表现能力。", "features": [], "provider": "happyhorse", - "limit": { - "message": "model not exist" - }, "model": "happyhorse-1.0-video-edit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "discount": 0.8, + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.6", + "discount": 0.8, + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -31,7 +45,7 @@ "model-default": { "count_limit_period": 1, "async_user_queue_limit": 500, - "count_limit": 10, + "count_limit": 5, "async_task_timeout": 180, "type": "model-default", "async_user_concurrency_limit": 5 @@ -43,42 +57,14 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-26T07:51:50.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "HappyHorse-1.0-Video-Edit", "docUrl": "https://help.aliyun.com/document_detail/3030779.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], + "category": "Visual", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"happyhorse-1.0-video-edit\",\n \"input\": {\n \"prompt\": \"让视频中的马头人身角色穿上图片中的条纹毛衣\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260409/dozxak/Wan_Video_Edit_33_1.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260415/hynnff/wan-video-edit-clothes.webp\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3030779.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json index 05868c52..bd5ffed3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-erase-completion.json @@ -14,49 +14,19 @@ "description": "图像擦除补全通过指定图像mask中要删除的人体、宠物、物品、文字、水印等图像区域，在保留背景的同时移除图像中的一个或多个人物、物体、文字等元素，此功能不支持输入prompt的消除。擦除补全技术结合了计算机视觉、AIGC inpainting等先进技术，可以在多种场景下应用，从而满足用户对隐私保护、内容创作和图像编辑等方面需求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-erase-completion", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-08-19T01:17:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "图像擦除补全", "docUrl": "https://help.aliyun.com/document_detail/2840907.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'X-DashScope-DataInspection: enable' \\\n--data-raw '{\n \"model\": \"image-erase-completion\",\n \"input\": {\n \"image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E5%8E%9F%E5%9B%BE.png\",\n \"mask_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E6%93%A6%E9%99%A4.png\",\n \"foreground_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E5%9B%BE%E7%89%87%E6%93%A6%E9%99%A42-%E4%BF%9D%E7%95%99.png\"\n },\n \"parameters\":{\n \"dilate_flag\":true\n }\n}' \n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json index e90a0e3b..6b98da94 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-instance-segmentation.json @@ -14,45 +14,15 @@ "description": "人物实例分割运用了检测和分割技术，不仅能够在图像中识别出不同的对象，而且还能准确地画出每一个对象边界的像素级掩码（mask）。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-instance-segmentation", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-08-19T01:17:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "人物实例分割", "docUrl": "https://help.aliyun.com/document_detail/2840906.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json index e6182d40..82349a0d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json +++ b/skills/bailian-docs-llm-wiki/models/groups/image-out-painting.json @@ -14,49 +14,25 @@ "description": "图像画面大模型，对输入图像进行画面自由扩展，支持旋转画面，支持按照扩展系数和扩展像素数两种方式进行扩图。用户可以通过指定宽度、高度画面扩展比例或者左、右、上、下的扩展的像素值来控制画面扩展，可用于创意娱乐、辅助作图、画面设计、影视后期制作等场景。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "image-out-painting", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-24T10:29:57.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "图像画面扩展", "docUrl": "https://help.aliyun.com/document_detail/2796845.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/out-painting' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"image-out-painting\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\"\n },\n \"parameters\":{\n \"x_scale\":2,\n \"y_scale\":2,\n \"best_quality\":false,\n \"limit_image_size\":true\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json index 1646ab45..9a55992a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json @@ -2,6 +2,90 @@ "name": "Kimi", "description": "由月之暗面提供的Kimi系列模型的API服务。", "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text", + "Image", + "Video" + ] + }, + "description": "Kimi K3 是 Kimi 迄今能力最强的旗舰模型，拥有 2.8 万亿参数，基于 KDA 混合线性注意力机制（Kimi Delta Attention）和注意力残差（Attention Residuals）技术构建，原生支持视觉理解，并拥有 100 万 token 上下文窗口。它是全球首个开源的 3 万亿级别模型，面向长程编程、知识工作和推理等前沿智能场景而设计。", + "features": [ + "function-calling", + "structured-outputs", + "cache", + "prefix-completion" + ], + "provider": "moonshot-ai", + "model": "kimi/kimi-k3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "usage_limit": 3000000, + "usage_limit_field": "total_tokens", + "count_limit": 500, + "usage_limit_period": 60, + "type": "model-default" + } + }, + "capabilities": [ + "TG", + "VU", + "Reasoning" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "maxOutputTokens": 1048576, + "latestOnlineAt": "2026-07-17T08:23:12.000+00:00", + "contextWindow": 1048576, + "maxInputTokens": 1048576, + "inferenceProvider": "moonshot-ai", + "name": "kimi/kimi-k3", + "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", + "category": "Third-party", + "samples": { + "openai": { + "completionsAPI": { + "curl": "curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kimi/kimi-k3\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://dashscope.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"kimi/kimi-k3\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://dashscope.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"kimi/kimi-k3\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()", + "docUrl": "https://help.aliyun.com/document_detail/3021620.html" + } + } + } + }, { "inferenceMetadata": { "response_modality": [ @@ -21,10 +105,27 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.7-code-highspeed", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "13", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -57,33 +158,6 @@ "inferenceProvider": "moonshot-ai", "name": "kimi/kimi-k2.7-code-highspeed", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -114,10 +188,27 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.7-code", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -150,33 +241,6 @@ "inferenceProvider": "moonshot-ai", "name": "kimi/kimi-k2.7-code", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -207,10 +271,27 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.6", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.1", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -244,19 +325,6 @@ "inferenceProvider": "moonshot-ai", "name": "Kimi/Kimi K2.6", "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { @@ -287,11 +355,28 @@ "cache" ], "provider": "moonshot-ai", - "limit": { - "message": "model not exist" - }, "model": "kimi/kimi-k2.5", "iconUrl": "https://img.alicdn.com/imgextra/i4/O1CN01KzHLBW1LISVEUaotl_!!6000000001276-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -324,19 +409,6 @@ "inferenceProvider": "moonshot-ai", "name": "Kimi/Kimi K2.5", "docUrl": "https://help.aliyun.com/document_detail/3021620.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json index 4a04f994..d2ced42d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json @@ -15,11 +15,34 @@ "description": "智能分镜可读懂剧本场景流转，自动调度机位和景别。原生多模态框架支持音画一致性。打破时长限制，多镜头故事创作更自由。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-video-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,31 +69,10 @@ "latestOnlineAt": "2026-03-26T13:45:28.000+00:00", "inferenceProvider": "kling", "name": "Kling Video 3.0", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026701.html" } } @@ -90,11 +92,46 @@ "description": "新增“全能参考”，支持3-8秒视频或多图锚定角色元素。可匹配原声及口型驱动，实现角色本色呈现。视频一致性更强，表现更灵动。支持音画同步、智能分镜。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-omni-video-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.9", + "type": "720P_no_reference_video", + "priceName": "视频生成（720P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "1080P_no_reference_video", + "priceName": "视频生成（1080P 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "720P_no_audio_no_reference_video", + "priceName": "视频生成（720P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.8", + "type": "1080P_no_audio_no_reference_video", + "priceName": "视频生成（1080P 无声 无参考视频）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "type": "720P_no_audio_reference_video", + "priceName": "视频生成（720P 无声 有参考视频）" + }, + { + "priceUnit": "每秒", + "price": "1.2", + "type": "1080P_no_audio_reference_video", + "priceName": "视频生成（1080P 无声 有参考视频）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -121,31 +158,10 @@ "latestOnlineAt": "2026-03-26T13:45:46.000+00:00", "inferenceProvider": "kling", "name": "Kling Video 3.0 Omni", - "predictConfig": [ - { - "name": "mode", - "key": "mode", - "default": "pro" - }, - { - "name": "audio", - "key": "audio", - "default": false - }, - { - "name": "duration", - "key": "duration", - "default": 5, - "range": [ - 3, - 15 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"kling/kling-v3-omni-video-generation\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"mode\": \"std\",\n \"aspect_ratio\": \"16:9\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026701.html" } } @@ -164,11 +180,22 @@ "description": "支持最多10张参考图，可锁定主体、元素和色调，保证风格一致。融合风格转绘、人像/角色参考、多图融合及局部重绘，操作灵活。人像细节真实，整体画面细腻丰富，色彩氛围兼具影视感。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-image-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -195,28 +222,10 @@ "latestOnlineAt": "2026-03-26T13:45:34.000+00:00", "inferenceProvider": "kling", "name": "Kling Image 3.0", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026706.html" } } @@ -235,11 +244,28 @@ "description": "解锁影视级叙事画面，新增系列组图及2K/4K直出。深度解析提示词视听元素，精确响应创作指令。支持自由多参考图及全面效果升级，适合分镜、剧情概念图及场景设定。", "features": [], "provider": "kling", - "limit": { - "message": "model not exist" - }, "model": "kling/kling-v3-omni-image-generation", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_1k", + "priceName": "图片生成（1K）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_type_2k", + "priceName": "图片生成（2K）" + }, + { + "priceUnit": "每秒", + "price": "0.4", + "type": "image_type_4k", + "priceName": "图片生成（4K）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -266,28 +292,10 @@ "latestOnlineAt": "2026-03-26T13:45:39.000+00:00", "inferenceProvider": "kling", "name": "Kling Image 3.0 Omni", - "predictConfig": [ - { - "name": "aspect_ratio", - "key": "aspect_ratio", - "default": "16:9" - }, - { - "name": "resolution", - "key": "resolution", - "default": "1k" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"kling/kling-v3-omni-image-generation\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"aspect_ratio\": \"1:1\",\n \"resolution\": \"1k\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3026706.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json index a4387422..97cf97fd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json @@ -12,49 +12,33 @@ "description": "LivePortrait-detect是辅助LivePortrait的图像检测模型，用于检测图片中的人物形象是否符合视频生成要求。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "liveportrait-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:17:01.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "灵动人像LivePortrait-detect", "docUrl": "https://help.aliyun.com/document_detail/2856727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"liveportrait-detect\",\n \"input\": {\n \"image_url\":\"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json index 62fa8b89..9c6fa97b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json @@ -14,49 +14,33 @@ "description": "LivePortrait是一款视频生成模型，可基于人物图片生成轻量化的人物肖像动态视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "liveportrait", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.02", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-11-07T14:17:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "灵动人像LivePortrait", "docUrl": "https://help.aliyun.com/document_detail/2856730.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"liveportrait\",\n \"input\": {\n \"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/ynhjrg/p874909.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250911/mbeygv/%E7%B4%A0%E6%8F%8F%E7%94%B7%E5%AD%A9.mp3\"\n },\n \"parameters\": {\n \"template_id\": \"normal\",\n \"eye_move_freq\": 0.5,\n \"video_fps\":30,\n \"mouth_move_strength\":1,\n \"paste_back\": true,\n \"head_move_strength\":0.7\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json index 43ee7b5c..d750913e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json @@ -19,10 +19,27 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.84", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -54,33 +71,6 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M3", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -107,10 +97,27 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.7", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.42", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -143,37 +150,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.7", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.7\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } @@ -194,11 +174,28 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.5", "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -230,37 +227,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.5", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } @@ -281,11 +251,28 @@ "cache" ], "provider": "mini-max", - "limit": { - "message": "model not exist" - }, "model": "MiniMax/MiniMax-M2.1", "iconUrl": "https://img.alicdn.com/imgextra/i1/O1CN01EFGi131NqT95FPDqc_!!6000000001621-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.21", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -317,37 +304,10 @@ "inferenceProvider": "mini-max", "name": "MiniMax/MiniMax-M2.1", "docUrl": "https://help.aliyun.com/document_detail/3021647.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"MiniMax/MiniMax-M2.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3021647.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json index e3116a64..95fb8e84 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "ASR" ], @@ -28,19 +33,12 @@ "offlineInfo": { "inference": { "announceUrl": "mhttps://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-8k-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json index 13d9ccb2..c499ffb6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json @@ -14,45 +14,23 @@ "description": "Paraformer最新中文语音识别模型，模型结构升级，具有更好的识别效果,支持8kHz电话语音识别，仅支持中文热词。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-19T11:29:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-8k-v2", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json index cd175230..66df088d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-mtl-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "ASR" ], @@ -28,19 +33,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-mtl-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json index b99160b0..ddfe5da2 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json @@ -14,10 +14,15 @@ "description": "Paraformer中文实时语音识别模型，支持8kHz电话客服等场景下的实时语音识别。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-8k-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "Realtime-ASR" ], @@ -26,19 +31,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-8k-v1", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json index a3ab4c5e..dd3a1678 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json @@ -16,19 +16,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-8k-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-31T08:01:43.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-8k-v2", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json index 1d4292c4..7c751e45 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "Realtime-ASR" ], @@ -28,19 +33,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-v1", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json index 4b95da69..25bf199e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json @@ -16,25 +16,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-realtime-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00024", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "Realtime-ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer实时语音识别-v2", "docUrl": "https://help.aliyun.com/document_detail/2712536.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json index ea6b8ea1..911d0ca5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-v1", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "ASR" ], @@ -28,19 +33,12 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118331", - "offlineTime": "2026-09-07 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-v1", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json index d0e2040e..c7643257 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json @@ -16,25 +16,23 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "paraformer-v2", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00008", + "type": "content_duration", + "priceName": "音频时长" + } + ], "capabilities": [ "ASR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-05T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Paraformer语音识别-v2", "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "过滤语气词", - "key": "disfluency_removal_enabled", - "tip": "过滤语气词，默认为关闭false。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json index f484629a..6e68510e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json @@ -15,11 +15,58 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，r2v（多主体参考生成视频）输入2-7张图像，智能融合不同主体，同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力和想象力、更接近影视专业水准的打斗动作和术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合多主体群像、多人对话、多人交互等复杂剧情，适合中景、全景镜头。\n如果输入了1张多宫格分镜图片（最高支持九宫格），则可以一键生成连续分镜长视频。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-r2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -47,37 +94,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-r2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -95,11 +115,58 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -127,41 +194,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-t2v", "docUrl": "https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -180,11 +216,58 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -212,37 +295,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-kf2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -261,11 +317,58 @@ "description": "C1是PixVerse在26年3月底推出的影视行业大模型，it2v（图片生成视频）模型除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征。相比V6可增强提示词，拥有更强的想象力、更接近影视专业水准的打斗动作、术法特效。支持15秒长视频、音乐和视频直出、支持多种语言文字。适合单人特写、单人独白、定格/慢动作、空镜转场等短时长镜头。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-c1-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.39", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.71", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.18", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.24", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.3", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.56", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -291,41 +394,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-C1-it2v", "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-c1-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json index 9c8f461b..8594769d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json @@ -48,33 +48,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-upscale", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -143,33 +116,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-motioncontrol", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -228,33 +174,6 @@ "inferenceProvider": "pixverse", "name": "pixverse/pixverse-lipsync", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json index 02f27899..424d6161 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json @@ -14,11 +14,58 @@ "description": "输入文字描述，秒级生成与语义精准匹配的高质量视频，支持多种风格。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -43,37 +90,10 @@ "latestOnlineAt": "2026-03-19T06:58:48.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-t2v", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -92,11 +112,58 @@ "description": "输入2–7张图像，智能融合不同主体，保持风格统一与动作协调，轻松构建丰富叙事场景，提升内容可控性与创意自由度。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-r2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -121,37 +188,10 @@ "latestOnlineAt": "2026-03-19T06:43:04.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-r2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -170,11 +210,58 @@ "description": "在任意两张图片之间实现无缝转换，实现更流畅自然的场景过渡，打造视觉冲击力强的画面效果。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -199,37 +286,10 @@ "latestOnlineAt": "2026-03-19T06:56:40.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-kf2v", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -248,11 +308,58 @@ "description": "上传任意图片，自由定制剧情、节奏与风格，生成生动连贯的视频。PixVerse V5.6 是爱诗科技自研的视频生成大模型，在文生视频与图生视频能力上实现全面升级。模型在画面清晰度、复杂运动稳定性与音画协同方面显著提升，多角色对话场景下嘴型与台词同步更准确，情绪表达更自然。同时优化构图、光影与质感一致性，整体生成质量进一步提升。PixVerse V5.6 在 Artificial Analysis 文生视频与图生视频榜单中位列全球第一梯队。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v5.6-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.47", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.44", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -277,37 +384,10 @@ "latestOnlineAt": "2026-03-19T06:59:33.000+00:00", "inferenceProvider": "pixverse", "name": "PixVerse-V5.6-it2v", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v5.6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json index 0740d773..b44d8299 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json @@ -14,11 +14,58 @@ "description": "V6是PixVerse在26年3月底推出的新模型，t2v（文字生成视频）模型可通过提示词精准控制视频画面，精确还原各类镜头语言，推、拉、摇、移、跟随等运镜方式流畅自然，视角切换精准可控。支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-t2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,46 +93,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-t2v", "docUrl": "○ https://help.aliyun.com/document_detail/3025608.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-t2v\",\n \"input\": {\n \"prompt\": \"一只小猫在月光下奔跑\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025608.html" } } @@ -104,10 +115,57 @@ "description": "V6是PixVerse在26年3月底推出的新模型，r2v（多主体参考生成视频）模型全球排名第二，输入2-7张图像，智能融合不同主体，适合复杂的中景、远景视频镜头。同时拥有t2v（文字生成视频）的提示词控制能力，和it2v（图片生成视频）的一致性保持能力、更强的情绪表现力、和更流畅的高速运动画面。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-r2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -134,37 +192,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-r2v", "docUrl": "https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-r2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260320/knsple/wan-r2v-role-frame.jpg\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/qpzxps/wan-r2v-object4.png\"\n },\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260129/wfjikw/wan-r2v-backgroud5.png\"\n }\n ],\n \"prompt\": \"男人坐在靠窗的椅子上，手持吉他，在咖啡厅旁演奏一首舒缓的美国乡村民谣\"\n },\n \"parameters\": {\n \"size\": \"1280*720\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025612.html" } } @@ -183,11 +214,58 @@ "description": "V6是PixVerse在26年3月底推出的新模型，kf2v（首尾帧生成视频）模型可将任意两张图片衔接，视频转场更加流畅自然，支持15秒长视频、音乐和视频直出、支持多种语言文字。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-kf2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -215,37 +293,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-kf2v", "docUrl": "https://help.aliyun.com/document_detail/3025612.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-kf2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\n },\n {\n \"type\": \"last_frame\",\n \"url\": \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n }\n ],\n \"prompt\": \"一只小猫从窗台向下跳跃，轻盈地落在沙发上，然后好奇地环顾四周。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"watermark\": false\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025611.html" } } @@ -264,11 +315,58 @@ "description": "V6是PixVerse在26年3月底推出的新模型，it2v（图片生成视频）模型全球排名第二，it2v除了拥有t2v（文字生成视频）的提示词控制能力外，还能高度还原参考图片的色彩、饱和度、场景和人物特征，拥有更强的人物情绪、高速运动表现力。支持15秒长视频、音乐和视频直出、支持多种语言文字。在电商产品特写、广告宣传片、模拟c4d建模展示产品结构等场景下可一键直出。", "features": [], "provider": "pixverse", - "limit": { - "message": "model not exist" - }, "model": "pixverse/pixverse-v6-it2v", "iconUrl": "", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.21", + "type": "video_ratio_360p", + "priceName": "视频生成（360P）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.36", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.68", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "360P_no_audio", + "priceName": "视频生成（360P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.21", + "type": "540P_no_audio", + "priceName": "视频生成（540P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.27", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.53", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -296,46 +394,10 @@ "inferenceProvider": "pixverse", "name": "PixVerse-V6-it2v", "docUrl": "https://help.aliyun.com/document_detail/3025609.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "540P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - }, - { - "name": "shot_type", - "key": "shot_type", - "default": "single" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"pixverse/pixverse-v6-it2v\",\n \"input\": {\n \"media\": [\n {\n \"type\": \"image_url\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260121/zlpocv/wan-i2v-haigui.webp\"\n }\n ],\n \"prompt\": \"镜头从海龟下方缓缓上移，海龟悠然游动，腹部细节清晰可见。\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 5,\n \"audio\": false,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025609.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json index 121259d4..796918f8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json @@ -19,10 +19,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qvq-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -52,54 +63,27 @@ "maxInputTokens": 106496, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QVQ-Max", "docUrl": "https://help.aliyun.com/document_detail/2877996.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-max\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-max',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-max\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-max\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json index 1273bf44..84557be9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json @@ -19,10 +19,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qvq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -52,56 +63,27 @@ "maxInputTokens": 106496, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QVQ-Plus", "docUrl": "https://help.aliyun.com/document_detail/2877996.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qvq-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true}\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n # 如果没有配置环境变量，请用百炼API Key替换：api_key=\"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\n\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qvq-plus\", # 此处以 qvq-max 为例，可按需更换模型名称\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qvq-plus',\n messages: messages,\n stream: true\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qvq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qvq-plus\", # 此处以qvq-max为例，可按需更换模型名称。\n messages=messages,\n stream=True,\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.19.0\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(MultiModalConversationResult message) {\n String re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\n List> content = message.getOutput().getChoices().get(0).getMessage().getContent();\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (Objects.nonNull(content) && !content.isEmpty()) {\n Object text = content.get(0).get(\"text\");\n finalContent.append(content.get(0).get(\"text\"));\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(text);\n }\n }\n public static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\n return MultiModalConversationParam.builder()\n // 如果使用华北2（北京）地域的模型，需要使用华北2（北京）地域的 API Key，获取链接：https://bailian.console.alibabacloud.com/?tab=model#/api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n // 此处以 qvq-max 为例，可按需更换模型名称\n .model(\"qvq-plus\")\n .messages(Arrays.asList(Msg))\n .incrementalOutput(true)\n .build();\n }\n\n public static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\n throws NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\n MultiModalConversationParam param = buildMultiModalConversationParam(Msg);\n Flowable result = conv.streamCall(param);\n result.blockingForEach(message -> {\n handleGenerationResult(message);\n });\n }\n public static void main(String[] args) {\n try {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMsg = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\n Collections.singletonMap(\"text\", \"请解答这道题\")))\n .build();\n streamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n } catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json index 52c4ecf2..389bf217 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json @@ -68,41 +68,14 @@ ], "modelAlias": "", "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:47.013+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, + "maxOutputTokens": 8192, + "latestOnlineAt": "2026-07-14T06:59:47.000+00:00", + "contextWindow": 40960, + "maxInputTokens": 16384, "inferenceProvider": "aliyun-bailian", "name": "千问实时语音对话大模型3.0（极速版）", "docUrl": "https://help.aliyun.com/document_detail/3041584.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json index e02c2e41..3f7b4f04 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json @@ -68,41 +68,14 @@ ], "modelAlias": "", "versionTag": "MAJOR", - "maxOutputTokens": 4096, - "latestOnlineAt": "2026-07-14T06:59:43.526+00:00", - "contextWindow": 8192, - "maxInputTokens": 4096, + "maxOutputTokens": 8192, + "latestOnlineAt": "2026-07-14T06:59:43.000+00:00", + "contextWindow": 40960, + "maxInputTokens": 16384, "inferenceProvider": "aliyun-bailian", "name": "千问实时语音大模型 （标准版）", "docUrl": "https://help.aliyun.com/document_detail/3041584.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json index f354c773..c7029617 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json @@ -45,37 +45,10 @@ "name": "qwen-audio-3.0-tts-plus", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-plus\"\n\n#Please enter the correct voice below.\nvoice = \"\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", "docUrl": "https://help.aliyun.com/document_detail/2938790.html" } } @@ -124,37 +97,10 @@ "name": "qwen-audio-3.0-tts-flash", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\nvoice = \"longanfengyue\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", + "python": "# coding=utf-8\n\nimport dashscope\nfrom dashscope.audio.tts_v2 import *\n\n# If the API Key is not configured in the environment variable, your-api-key needs to be replaced with your own API Key\n# dashscope.api_key = \"your-api-key\"\n\ndashscope.base_websocket_api_url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/inference'\n\nmodel = \"qwen-audio-3.0-tts-flash\"\n\n#Please enter the correct voice below.\nvoice = \"\"\n\nsynthesizer = SpeechSynthesizer(model=model, voice=voice)\naudio = synthesizer.call(\"How is the weather today?\")\n\nwith open('output.mp3', 'wb') as f:\n f.write(audio)", "docUrl": "https://help.aliyun.com/document_detail/2938790.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json index 11d030a0..ed3a26b6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json @@ -17,10 +17,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-coder-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -49,57 +60,28 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Coder-Plus", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json index 9d331b94..1f7b4cb5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json @@ -17,10 +17,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-coder-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -49,57 +60,28 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Coder-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-coder-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-coder-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-coder-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-coder-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-coder-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-coder-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json index 419b305c..18a61bd0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json @@ -14,10 +14,21 @@ "description": "千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-deep-research", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "54", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "163", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -44,50 +55,23 @@ "latestOnlineAt": "2025-08-22T14:05:23.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "qwen-deep-research", "docUrl": "https://help.aliyun.com/document_detail/2975991.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-deep-research\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-deep-research\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-deep-research\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-deep-research\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-deep-research\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-deep-research\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-deep-research\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-deep-research\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-deep-research\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-deep-research\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-deep-research\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-deep-research\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json index a5b2499e..4f4a37c3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json @@ -16,10 +16,39 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-doc-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -46,50 +75,23 @@ "latestOnlineAt": "2025-07-23T13:21:02.000+00:00", "contextWindow": 262144, "maxInputTokens": 253952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Doc-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2948885.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-doc-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-doc-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-doc-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-doc-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-doc-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-doc-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json index f4d5508c..a3b4c73d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json @@ -2,6 +2,70 @@ "name": "Qwen-Embedding", "description": "基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。", "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Text" + ], + "request_modality": [ + "Text" + ] + }, + "description": "是通义实验室基于Qwen3.7训练的多语言文本统一向量模型，相较text-embedding-v4版本在文本检索、聚类、分类性能大幅提升；在MTEB多语言、中英、Code检索等评测任务上效果提升20%；支持256~2560维用户自定义向量维度。", + "collectionTag": "qwen3.7", + "features": [ + "model-experience" + ], + "provider": "qwen", + "model": "qwen3.7-text-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2400, + "usage_limit_period": 6, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 6, + "usage_limit": 100000, + "usage_limit_field": "total_tokens", + "count_limit": 2400, + "usage_limit_period": 6, + "type": "model-default" + } + }, + "capabilities": [ + "TR" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-15T02:24:33.000+00:00", + "contextWindow": 131072, + "maxInputTokens": 131072, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.7-通用文本向量", + "docUrl": "https://help.aliyun.com/document_detail/2842587.html", + "category": "Embeddings", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"qwen3.7-text-embedding\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"qwen3.7-text-embedding\",\ninput=input_texts\n)\nprint(resp)", + "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"qwen3.7-text-embedding\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" + } + } + } + }, { "inferenceMetadata": { "response_modality": [ @@ -16,10 +80,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v4", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,19 +119,14 @@ "versionTag": "MAJOR", "latestOnlineAt": "2025-06-05T03:07:20.000+00:00", "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v4", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", "category": "Embeddings", - "predictConfig": [ - { - "name": "topK" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v4\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v4\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v4\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -73,10 +143,21 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -100,40 +181,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-07-12T09:44:51.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v3", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v3\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v3\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v3\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -150,10 +204,21 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -177,40 +242,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:03:19.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v2", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v2\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -227,10 +265,21 @@ "description": "通用文本向量，是通义实验室基于LLM底座的多语言文本统一向量模型，面向全球多个主流语种，提供高水准的向量服务，帮助开发者将文本数据快速转换为高质量的向量数据。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.35", + "type": "embedding_token_batch", + "priceName": "向量输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -254,40 +303,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:02:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-v1", "docUrl": "https://help.aliyun.com/document_detail/2712515.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -304,49 +326,27 @@ "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-async-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:05:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-async-v2", "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v2\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v2\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v2\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } @@ -363,49 +363,27 @@ "description": "通用文本向量的批处理接口，通过这个接口客户可以以文本方式一次性的提交大批量的向量计算请求，在系统完成所有的计算之后，大模型服务平台会将结果信息存储在结果文件中供客户下载解析。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "text-embedding-async-v1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "capabilities": [ "TR" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T09:04:41.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通用文本向量-async-v1", "docUrl": "https://help.aliyun.com/document_detail/2712516.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"text-embedding-async-v1\",\n\"input\": {\n\"texts\":[\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"] \n}'", "python": "import dashscope\nfrom http import HTTPStatus\ninput_texts = \"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\"\n\nresp = dashscope.TextEmbedding.call(\nmodel=\"text-embedding-async-v1\",\ninput=input_texts\n)\nprint(resp)", "java": "import java.util.Arrays;\nimport java.util.concurrent.Semaphore;\nimport com.alibaba.dashscope.common.ResultCallback;\nimport com.alibaba.dashscope.embeddings.TextEmbedding;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingParam;\nimport com.alibaba.dashscope.embeddings.TextEmbeddingResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic final class Main {\npublic static void main(String[] args) {\ntry {\nTextEmbeddingParam param = TextEmbeddingParam\n.builder()\n.model(\"text-embedding-async-v1\") \n.texts(Arrays.asList(\"衣服的质量杠杠的，很漂亮，不枉我等了这么久啊，喜欢，以后还来这里买\")) \n.build();\nTextEmbedding textEmbedding = new TextEmbedding();\nTextEmbeddingResult result = textEmbedding.call(param);\n\nSystem.out.println(result);\n\n} catch (ApiException | NoApiKeyException e) {\nSystem.out.println(e.getMessage());\n}\n}\n}" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json index 6a62a6d2..9f1cee04 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json @@ -18,10 +18,27 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-flash-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.05", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -48,50 +65,23 @@ "latestOnlineAt": "2026-01-13T04:04:02.000+00:00", "contextWindow": 8192, "maxInputTokens": 8000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Flash-Character", "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-flash-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-flash-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-flash-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json index b9ac0286..554f773f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json @@ -23,9 +23,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-flash", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,197 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.188", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -54,82 +242,23 @@ "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json index 1ee17628..2a1085ab 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json @@ -17,11 +17,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-2.0-pro", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -41,40 +46,15 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-2.0-pro-2026-04-22", "latestOnlineAt": "2026-04-22T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-2.0-Pro", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0-pro\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0-pro\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json index 2b28d8f3..eb692b65 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json @@ -17,11 +17,16 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-2.0", "iconUrl": "", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -41,40 +46,15 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-2.0-2026-03-03", "latestOnlineAt": "2026-03-03T11:31:36.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-2.0", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "2048*2048", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-2.0\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-2.0\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-2.0\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json new file mode 100644 index 00000000..f90bf284 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-3.0-pro.json @@ -0,0 +1,53 @@ +{ + "name": "Qwen-Image-3.0-Pro", + "description": "内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求**“好用”**——让图像生成真正成为可落地的生产力工具。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Image", + "Text" + ] + }, + "description": "内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求“好用”——让图像生成真正成为可落地的生产力工具。", + "features": [ + "model-experience" + ], + "provider": "qwen", + "model": "qwen-image-3.0-pro", + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "count_limit": 1, + "type": "model-default" + }, + "model-default": { + "count_limit_period": 60, + "count_limit": 1, + "type": "model-default" + } + }, + "capabilities": [ + "IG" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-07-20T14:02:23.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen-Image-3.0-Pro", + "docUrl": "https://help.aliyun.com/document_detail/3047054.html", + "category": "Visual", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-3.0-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"画面是一张竖幅户外人像摄影，整体从上到下呈现温暖的午后街景氛围。顶部左侧到上方大面积被深绿色藤蔓和橙色小花覆盖，花叶从建筑檐口自然垂落，受阳光照射的叶片呈黄绿色高光，阴影处则偏深绿，形成浓密而柔和的背景层次。左上至中上区域是一块深蓝色横向招牌，招牌表面较暗、略带磨砂质感，上面以白色哥特体大字写着 Il Messaggero，文字位于画面左侧偏上，部分被前景花叶轻微遮挡，字体高对比、带装饰性尖角和粗细变化。招牌下方是报刊亭或书报摊的玻璃展示窗，黑色金属框架将橱窗分隔成多个矩形区域，内部陈列着许多报纸、杂志和书刊封面，但大多因景深虚化和光线反射而难以辨读，形成浅色纸张与深色边框交错的背景纹理。画面右上方是强烈的逆光区域，阳光从街道尽头照入，背景建筑被虚化成米灰色块面，边缘柔和，呈现明显的浅景深效果。画面中部偏右是一名年轻成年女性的半身至膝上人像，她回头面向镜头微笑，身体略向右转，肩背朝向观者，姿态自然放松。她有长而浓密的黑色波浪卷发，发丝被逆光勾勒出金色轮廓光，发梢在右侧向外散开，显得轻盈蓬松。她肤色白皙，脸型柔和偏鹅蛋形，眉形细致，眼睛明亮，眼妆清透，睫毛明显，面部带有自然高光，唇部为柔和珊瑚红色，笑容露齿，表情亲切明朗。她佩戴小巧耳饰，身穿黑色细肩带露背连衣裙，面料颜色深黑、轮廓简洁，细肩带从肩部向背部延伸，背部线条清晰。画面下部偏左到中部，她双手抱着一束玫瑰花，花束体积较大，主要由橙色、杏色、粉色和浅桃色玫瑰组成，花瓣层层卷曲，边缘被阳光照亮，绿色叶片和长花茎从花束下方垂出，花束与黑色裙装形成鲜明色彩对比。右侧背景是一条被阳光照亮的城市街道，地面呈暖灰与金黄色调，远处建筑、街边设施和一个模糊的红色圆形交通标志位于右下远景，均因焦外虚化而只保留色块和轮廓。整张照片采用暖色胶片感处理，带有细腻颗粒、柔和对比和明显逆光边缘光，人物位于视觉焦点，背景报刊亭、花藤、街道和阳光共同营造出浪漫、明亮、都市漫步式的氛围。\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"prompt_extend\": true\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3047054.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json index dfdfec60..827da9ab 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json @@ -17,10 +17,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit-max", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -38,22 +43,21 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-01-15T12:28:13.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit-Max", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-max\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-max\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-max\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json index ec102b64..fdb26226 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,22 +42,21 @@ ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-10-30T09:10:49.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit-Plus", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit-plus\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 2,\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n n=2,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n parameters.put(\"n\", 2);\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit-plus\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -72,10 +76,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-edit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.3", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -93,22 +102,21 @@ ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-09-21T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Edit", "docUrl": "https://help.aliyun.com/document_detail/2976416.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - } - ], "samples": { "dashscope": { "default": { "curl": "curl --location 'https://{Domain}/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"qwen-image-edit\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"\n },\n {\n \"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"\n },\n {\n \"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"negative_prompt\": \"\",\n \"watermark\": false\n }\n}'", - "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import json\nimport os\nfrom dashscope import MultiModalConversation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"},\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"},\n {\"text\": \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-edit\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=True,\n negative_prompt=\"\"\n)\n\nprint(json.dumps(response, ensure_ascii=False))", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.io.IOException;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\npublic class QwenImageEdit {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey=\"sk-xxx\"\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void call() throws ApiException, NoApiKeyException, UploadFileException, IOException {\n\n MultiModalConversation conv = new MultiModalConversation();\n\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/thtclx/input1.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/iclsnx/input2.png\"),\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/gborgw/input3.png\"),\n Collections.singletonMap(\"text\", \"图1中的女生穿着图2中的黑色裙子按图3的姿势坐下\")\n )).build();\n\n Map parameters = new HashMap<>();\n parameters.put(\"watermark\", true);\n parameters.put(\"negative_prompt\", \"\");\n\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(apiKey)\n .model(\"qwen-image-edit\")\n .messages(Collections.singletonList(userMessage))\n .parameters(parameters)\n .build();\n\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException | IOException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json index 5f460769..00f7fbb7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-max", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -44,32 +49,19 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen-image-max-2025-12-30", "latestOnlineAt": "2025-12-30T07:06:12.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Max", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-max\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-max\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json index 55add199..ce2053da 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json @@ -16,10 +16,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -47,32 +52,19 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T10:49:31.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image-Plus", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image-plus\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } @@ -91,10 +83,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.25", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -122,32 +119,19 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-08-13T12:58:52.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Image", "docUrl": "https://help.aliyun.com/document_detail/2975126.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, - { - "name": "size", - "key": "size", - "default": "1328*1328", - "tip": "输出分辨率" - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" + "python": "import json\nimport os\nimport dashscope\nfrom dashscope import MultiModalConversation\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"一副典雅庄重的对联悬挂于厅堂之中，房间是个安静古典的中式布置，桌子上放着一些青花瓷，对联上左书“义本生知人机同道善思新”，右书“通云赋智乾坤启数高志远”， 横批“智启通义”，字体飘逸，中间挂在一着一副中国风的画作，内容是岳阳楼。\"}\n ]\n }\n]\n\n# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\"\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nresponse = MultiModalConversation.call(\n api_key=api_key,\n model=\"qwen-image\",\n messages=messages,\n result_format='message',\n stream=False,\n watermark=False,\n prompt_extend=True,\n negative_prompt='',\n size='1328*1328'\n)\n\nif response.status_code == 200:\n print(json.dumps(response, ensure_ascii=False))\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")\n print(\"请参考文档：https://www.alibabacloud.com/help/zh/model-studio/error-code\")" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json index 27b295ec..2313f981 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json @@ -16,10 +16,33 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-long-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,50 +69,29 @@ "latestOnlineAt": "2025-03-19T02:45:13.000+00:00", "contextWindow": 10000000, "maxInputTokens": 10000000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Long-Latest", "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -109,10 +111,33 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-long", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -139,50 +164,23 @@ "latestOnlineAt": "2024-05-20T14:57:26.000+00:00", "contextWindow": 10000000, "maxInputTokens": 10000000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Long", "docUrl": "https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-long\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-long\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-long\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-long\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-long\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-long\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json index 89c3ed99..f1b5fafe 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -46,50 +57,29 @@ "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -109,10 +99,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-0919", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -140,50 +141,29 @@ "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-2024-09-19", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0919\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0919\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0919\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0919\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0919\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0919\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -203,10 +183,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -233,50 +224,29 @@ "latestOnlineAt": "2024-09-18T16:00:00.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-Latest", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-latest\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-latest\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-latest\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-latest\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-latest\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -296,10 +266,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-plus-0816", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -327,50 +308,29 @@ "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", "contextWindow": 4096, "maxInputTokens": 3072, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Plus-2024-08-16", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-plus-0816\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-plus-0816\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-plus-0816\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-plus-0816\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-plus-0816\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-plus-0816\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json index fc73582a..95a2dccd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json @@ -17,10 +17,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-math-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -49,57 +60,28 @@ "maxInputTokens": 3072, "offlineInfo": { "inference": { - "announceUrl": { - "cn_hangzhou": "https://www.aliyun.com/notice/118177" - }, + "announceUrl": "https://www.aliyun.com/notice/118177", "offlineTime": "2026-07-13 23:59:59" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Math-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2849934.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-math-turbo\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-math-turbo\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-math-turbo\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-math-turbo\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-math-turbo\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-math-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json index b8bd3cb9..946f1650 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json @@ -22,10 +22,39 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9.6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 15, @@ -52,62 +81,24 @@ "latestOnlineAt": "2024-10-15T05:39:20.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-max\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-max\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-max\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-max\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-max\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json index 6c36cbab..754c5177 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -47,30 +58,22 @@ "latestOnlineAt": "2025-11-06T08:07:15.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Flash", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-flash\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-flash\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-flash\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-flash\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-flash\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json index d6a8ceda..7c7f1c3e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json @@ -14,10 +14,15 @@ "description": "专注做图片翻译的模型服务，能将中、英、日等11个语言的图片翻译到指定的语言，精准还原图片排版和内容信息，支持术语定义、敏感词过滤、商品主体检测等自定义功能，提供灵活、准确、高效的图像本地化服务。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.003", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,40 +44,18 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-08-22T09:53:46.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Image", "docUrl": "https://help.aliyun.com/document_detail/2977163.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-mt-image\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i2/O1CN01XsvEqj1fNlMqLNBHR_!!6000000003995-0-tps-5933-2930.jpg\",\n \"source_lang\": \"en\",\n \"target_lang\": \"ja\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json index edd7ca5f..b1e3862a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-lite", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,30 +57,22 @@ "latestOnlineAt": "2025-11-19T11:49:54.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Lite", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-lite\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-lite\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-lite\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-lite\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-lite\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json index bc356ae0..627c99cd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5.4", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,30 +57,22 @@ "latestOnlineAt": "2025-07-22T06:16:44.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Plus", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-plus\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-plus\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-plus\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-plus\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-plus\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json index a9872aef..1c77270d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json @@ -16,10 +16,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-mt-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.95", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,30 +57,28 @@ "latestOnlineAt": "2025-07-22T06:16:54.000+00:00", "contextWindow": 16384, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-MT-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2860790.html", - "predictConfig": [ - { - "name": "translation_options", - "key": "translation_options", - "default": "{\"source_lang\": \"Chinese\", \"target_lang\": \"English\"}", - "tip": "支持的语言详见https://bailian.console.aliyun.com/#/model-market/detail/qwen-mt-turbo?tabKey=sdk”。" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": \"看完这个视频我没有笑\"\n }\n ],\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n}\ncompletion = client.chat.completions.create(\n model=\"qwen-mt-turbo\",\n messages=messages,\n extra_body={\n \"translation_options\": translation_options\n }\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-mt-turbo\", \n messages: [\n { role: \"user\", content: \"我看到这个视频后没有笑\" }\n ],\n translation_options: {\n source_lang: \"auto\",\n target_lang: \"English\"\n }\n});\nconsole.log(completion.choices[0].message.content)" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", - "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation \\\n-H \"Authorization: $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-mt-turbo\",\n \"input\": {\n \"messages\": [\n {\n \"content\": \"我看到这个视频后没有笑\",\n \"role\": \"user\"\n }\n ]\n },\n \"parameters\": {\n \"translation_options\": {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\"\n }\n }\n}'", + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n {\n \"role\": \"user\",\n \"content\": \"我看到这个视频后没有笑\"\n }\n]\ntranslation_options = {\n \"source_lang\": \"auto\",\n \"target_lang\": \"English\",\n}\nresponse = dashscope.Generation.call(\n # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen-mt-turbo\", # 此处以qwen-mt-turbo为例，可按需更换模型名称\n messages=messages,\n result_format='message',\n translation_options=translation_options\n)\nprint(response.output.choices[0].message.content)", + "java": "// DashScope SDK 版本需要不低于 2.20.6\nimport java.lang.System;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.aigc.generation.TranslationOptions;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"我看到这个视频后没有笑\")\n .build();\n TranslationOptions options = TranslationOptions.builder()\n .sourceLang(\"auto\")\n .targetLang(\"English\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-mt-turbo\")\n .messages(Collections.singletonList(userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .translationOptions(options)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n e.printStackTrace();\n } finally {\n System.exit(0);\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json index ddd80c4a..4b9adfaf 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json @@ -18,10 +18,45 @@ "description": "千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -48,41 +83,20 @@ "latestOnlineAt": "2025-05-08T11:50:42.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } @@ -104,10 +118,45 @@ "description": "千问全新多模态理解生成大模型实时版，此版本为动态更新版本。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -134,41 +183,20 @@ "latestOnlineAt": "2025-05-08T12:29:07.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Realtime-Latest", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen-omni-turbo-realtime-latest'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen-omni-turbo-realtime-latest\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json index d8bba37b..e3181535 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json @@ -21,10 +21,99 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "vision_input_token_cache", + "priceName": "输入：图片/视频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "text_input_token_cache", + "priceName": "输入：文本（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "audio_input_token_cache", + "priceName": "输入：音频（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "text_input_token_batch", + "priceName": "输入：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "type": "audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "vision_input_token_batch", + "priceName": "输入：图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "multi_output_token_batch", + "priceName": "输出：文本+音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.25", + "type": "multiin_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "purein_text_output_token_batch", + "priceName": "输出：文本（Batch File，输入仅包含文本时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -51,42 +140,21 @@ "latestOnlineAt": "2025-02-14T14:56:44.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } @@ -110,10 +178,45 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-omni-turbo-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.5", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "50", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -140,42 +243,21 @@ "latestOnlineAt": "2025-02-14T14:55:08.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Omni-Turbo-Latest", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-omni-turbo-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-omni-turbo-latest\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-omni-turbo-latest\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json index fff3c32a..39d23237 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json @@ -19,10 +19,27 @@ "structured-outputs" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-character", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -49,50 +66,23 @@ "latestOnlineAt": "2025-03-20T09:16:40.000+00:00", "contextWindow": 32768, "maxInputTokens": 32768, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-Character", "docUrl": "https://help.aliyun.com/document_detail/2874763.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-character\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-character\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-character\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-character\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-character\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-character\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json index ffe268ea..bbb4e97e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json @@ -23,9 +23,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,365 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.24", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.96", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "thinking_input_token_cache_creation_5m", + "priceName": "显式缓存创建（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "thinking_input_token_cache_read", + "priceName": "显式缓存命中（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "thinking_input_token_batch_chat", + "priceName": "输入（思考模式 Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "discount": 0.5, + "type": "thinking_output_token_batch_chat", + "priceName": "输出（思考模式 Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -54,82 +410,23 @@ "latestOnlineAt": "2025-06-23T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -155,9 +452,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-latest", "qpmInfo": { "model-default-actual": { @@ -177,6 +471,173 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "32", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "TG" @@ -186,82 +647,23 @@ "latestOnlineAt": "2025-07-30T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 995904, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-Latest", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-latest\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-plus-latest',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-latest\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-latest\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -281,10 +683,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-1220", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -312,50 +725,29 @@ "latestOnlineAt": "2024-12-26T12:24:51.000+00:00", "contextWindow": 131072, "maxInputTokens": 129024, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-2024-12-20", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-1220\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-1220\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-1220\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-1220\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-1220\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-1220\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -375,10 +767,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-plus-0112", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -406,50 +809,29 @@ "latestOnlineAt": "2025-01-15T11:28:32.000+00:00", "contextWindow": 131072, "maxInputTokens": 129024, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Plus-2025-01-12", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-plus-0112\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen-plus-0112\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen-plus-0112\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen-plus-0112\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-plus-0112\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-plus-0112\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json index 6fa115a4..20ba1fca 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json @@ -16,10 +16,21 @@ "description": "Qwen3-VL-Rerank重排模型，它能够深入理解文本、图片、视频的丰富多模态信息。在初步检索获得结果后，Qwen3-VL-Rerank 能够运用其先进的跨模态关联能力，对候选项目进行智能化的二次排序，将最相关的结果置于显要位置。通用用于提升跨模态搜索的准确率、优化图搜和视频检索的精准度、辅助图像聚类的分组质量、以及实现复杂多模态信息的高效检索和精确打标。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,40 +55,14 @@ "versionTag": "MAJOR", "latestOnlineAt": "2026-01-29T10:23:42.000+00:00", "maxInputTokens": 120000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Rerank", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-rerank\",\n \"input\": {\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } @@ -94,10 +79,15 @@ "description": "基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐开源Qwen3-Rerank系列模型", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-rerank", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 10, @@ -124,40 +114,14 @@ "latestOnlineAt": "2025-10-21T08:21:26.000+00:00", "contextWindow": 30000, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "千问3-Rerank", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-rerank\",\n \"input\": {\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } @@ -174,10 +138,15 @@ "description": "gte-rerank-v2是通义实验室研发的多语言文本统一排序模型，面向全球多个主流语种，提供高水平的文本排序服务。通常用于语义检索、RAG等场景，可以简单、有效地提升文本检索的效果。给定查询 (Query) 和一系列候选文本 (documents)，模型会根据与查询的语义相关性从高到低对候选文本进行排序。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "gte-rerank-v2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -204,40 +173,14 @@ "latestOnlineAt": "2025-03-20T08:32:00.000+00:00", "contextWindow": 30000, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "深度文本重排序", "docUrl": "https://help.aliyun.com/document_detail/2780056.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"input\":{\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ]\n },\n \"parameters\": {\n \"return_documents\": true,\n \"top_n\": 5\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"gte-rerank-v2\",\n \"query\": \"什么是文本排序模型\",\n \"documents\": [\n \"文本排序模型广泛用于搜索引擎和推荐系统中，它们根据文本相关性对候选文本进行排序\",\n \"量子计算是计算科学的一个前沿领域\",\n \"预训练语言模型的发展给文本排序模型带来了新的进展\"\n ],\n \"top_n\": 5,\n \"return_documents\": true\n}'", + "docUrl": "https://help.aliyun.com/document_detail/2780056.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json index aa990dde..26cb0c69 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json @@ -14,10 +14,21 @@ "description": "Qwen-TTS实时模型是通义实验室“qwen系列”模型中的语音合成模型。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,40 +57,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -101,10 +85,21 @@ "description": "Qwen-TTS实时模型是通义实验室千问模型中语音合成利器，始终与最新快照版能力相同。具备双向上下文感知能力，可以低延迟高保真完成多音色、方言及长文本的双向流式生成。本模型是动态更新版本，模型更新不会提前通知。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-realtime-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -133,40 +128,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Realtime-Latest", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json index a0f026ff..386ac268 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json @@ -14,10 +14,21 @@ "description": "千问系列首个语音合成模型，支持中文、英文、中英混合输入。自适应根据输入文本调整输出语气，音色真实自然，支持流式输出。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,14 +57,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS", "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], "samples": { "dashscope": { "default": { @@ -75,10 +85,21 @@ "description": "模型是动态更新版本，等同于最新版本快照模型，模型更新时不会提前通知。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-tts-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "qwen_tts_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -107,14 +128,13 @@ "maxInputTokens": 512, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1934", - "offlineTime": "2026-07-06 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118332", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-TTS-Latest", "docUrl": "https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc", - "predictConfig": [], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json index af0c0b42..e7f53f6d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json @@ -20,10 +20,75 @@ "structured-outputs" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-turbo", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "thinking_input_token_cache", + "priceName": "输入（思考模式缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "thinking_output_token_batch", + "priceName": "思考模式输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "thinking_input_token_batch", + "priceName": "输入（思考模式 Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 15, @@ -53,87 +118,28 @@ "maxInputTokens": 98304, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen-turbo\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen-turbo',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-turbo\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen-turbo\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen-turbo\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json index 21d91205..4a818fa6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json @@ -10,27 +10,40 @@ "Image" ] }, - "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [], + "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [ + "model-experience" + ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen2.5-vl-embedding", + "model": "qwen3-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 60000, + "usage_limit": 120000, "usage_limit_field": "total_usage", - "count_limit": 20, + "count_limit": 40, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 60000, + "usage_limit": 120000, "usage_limit_field": "total_usage", - "count_limit": 20, + "count_limit": 40, "usage_limit_period": 6, "type": "model-default" } @@ -39,43 +52,18 @@ "ME" ], "versionTag": "MAJOR", - "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", - "inferenceProvider": "bailian", - "name": "Qwen2.5-VL-Embedding", + "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", + "maxInputTokens": 32000, + "offlineInfo": {}, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-VL-Embedding", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } } } @@ -88,29 +76,38 @@ "Image" ] }, - "description": "基于Qwen3-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", - "features": [ - "model-experience" - ], + "description": "基于Qwen2.5-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景", + "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3-vl-embedding", + "model": "qwen2.5-vl-embedding", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "embedding_image_token", + "priceName": "图片输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "embedding_token", + "priceName": "文本输入" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 120000, + "usage_limit": 60000, "usage_limit_field": "total_usage", - "count_limit": 40, + "count_limit": 20, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 120000, + "usage_limit": 60000, "usage_limit_field": "total_usage", - "count_limit": 40, + "count_limit": 20, "usage_limit_period": 6, "type": "model-default" } @@ -119,23 +116,16 @@ "ME" ], "versionTag": "MAJOR", - "latestOnlineAt": "2026-01-21T03:39:38.000+00:00", - "maxInputTokens": 32000, - "offlineInfo": {}, - "inferenceProvider": "bailian", - "name": "Qwen3-VL-Embedding", + "latestOnlineAt": "2025-10-21T08:21:32.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen2.5-VL-Embedding", "docUrl": "https://help.aliyun.com/document_detail/2842587.html", - "predictConfig": [ - { - "name": "topK" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --silent --location --request POST 'ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", - "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen3-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", - "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen3-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" + "curl": "curl --silent --location --request POST '[workspace-id].cn-beijing.maas.aliyuncs.com' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen2.5-vl-embedding\",\n \"input\": {\n \"contents\": [ \n {\"text\": \"多模态向量模型\"}\n ]\n }\n ]\n }\n}'", + "python": "import dashscope\n\ntext = \"通用多模态表征模型示例\"\ninput = [{'text': text}]\nresp = dashscope.MultiModalEmbedding.call(\n model=\"qwen2.5-vl-embedding\",\n input=input\n)\n\nprint(resp)\n", + "java": "import com.alibaba.dashscope.embeddings.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\n\nimport java.util.Arrays;\nimport java.util.List;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n String text = \"通用多模态表征模型示例\";\n\n MultiModalEmbeddingItemText textContent = new MultiModalEmbeddingItemText(text);\n List contents = Arrays.asList(textContent);\n MultiModalEmbeddingParam param = MultiModalEmbeddingParam.builder()\n .model(\"qwen2.5-vl-embedding\")\n .contents(contents)\n .build();\n MultiModalEmbedding multiModalEmbedding = new MultiModalEmbedding();\n MultiModalEmbeddingResult result = multiModalEmbedding.call(param);\n\n System.out.println(result);\n } catch (UploadFileException | NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n }\n}\n" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json index e57df8de..c07de66c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json @@ -23,10 +23,45 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.32", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "150", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 5, @@ -55,60 +90,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-VL-Max", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-max\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-max\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-max\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-max',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-max\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json index 14c496eb..1239f40f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json @@ -18,24 +18,47 @@ "batch" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr-latest", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, - "usage_limit": 600000, + "usage_limit": 3000000, "usage_limit_field": "total_tokens", - "count_limit": 20, + "count_limit": 100, "usage_limit_period": 6, "type": "model-default" }, "model-default": { "count_limit_period": 1, - "usage_limit": 600000, + "usage_limit": 3000000, "usage_limit_field": "total_tokens", - "count_limit": 20, + "count_limit": 100, "usage_limit_period": 6, "type": "model-default" } @@ -48,49 +71,28 @@ "latestOnlineAt": "2025-09-22T16:00:00.000+00:00", "contextWindow": 38192, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR-Latest", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-latest\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-latest\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-latest\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-latest',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-latest\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -110,10 +112,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr-1028", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -141,49 +154,28 @@ "latestOnlineAt": "2024-11-14T13:51:21.000+00:00", "contextWindow": 34096, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR-2024-10-28", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr-1028\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr-1028\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr-1028\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr-1028',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr-1028\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -204,10 +196,33 @@ "batch" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -235,49 +250,28 @@ "latestOnlineAt": "2025-11-19T16:00:00.000+00:00", "contextWindow": 38192, "maxInputTokens": 30000, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-OCR", "docUrl": "https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json index 9e9b6ac1..fcb69a54 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json @@ -24,10 +24,45 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-vl-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.16", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -56,66 +91,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QwenVL-Plus", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "vl_high_resolution_images", - "key": "vl_high_resolution_images", - "default": false, - "tip": "是否提高输入图片的默认Token上限" - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-plus\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen-vl-plus\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-plus\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen-vl-plus',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen-vl-plus\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json index 526e2ad2..40311a6c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json @@ -14,11 +14,16 @@ "description": "千问voice-design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-voice-design", "iconUrl": "", + "prices": [ + { + "priceUnit": "每次", + "price": "0.2", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,26 +44,19 @@ "latestOnlineAt": "2025-12-12T07:41:45.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Voice-Design", "docUrl": "https://help.aliyun.com/document_detail/3000986.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", - "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", - "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n}'", + "python": "import requests\nimport base64\nimport os\n\ndef create_voice_and_play():\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n \n if not api_key:\n print(\"错误: 未找到DASHSCOPE_API_KEY环境变量，请先设置API Key\")\n return None, None, None\n \n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n \n data = {\n \"model\": \"qwen-voice-design\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vd-realtime-2025-12-16\",\n \"voice_prompt\": \"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\",\n \"preview_text\": \"各位听众朋友，大家好，欢迎收听晚间新闻。\",\n \"preferred_name\": \"announcer\",\n \"language\": \"zh\"\n },\n \"parameters\": {\n \"sample_rate\": 24000,\n \"response_format\": \"wav\"\n }\n }\n \n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n \n try:\n response = requests.post(\n url,\n headers=headers,\n json=data,\n timeout=60\n )\n \n if response.status_code == 200:\n result = response.json()\n \n voice_name = result[\"output\"][\"voice\"]\n print(f\"音色名称: {voice_name}\")\n \n base64_audio = result[\"output\"][\"preview_audio\"][\"data\"]\n \n audio_bytes = base64.b64decode(base64_audio)\n \n filename = f\"{voice_name}_preview.wav\"\n \n with open(filename, 'wb') as f:\n f.write(audio_bytes)\n \n print(f\"音频已保存到本地文件: {filename}\")\n print(f\"文件路径: {os.path.abspath(filename)}\")\n \n return voice_name, audio_bytes, filename\n else:\n print(f\"请求失败，状态码: {response.status_code}\")\n print(f\"响应内容: {response.text}\")\n return None, None, None\n \n except requests.exceptions.RequestException as e:\n print(f\"网络请求发生错误: {e}\")\n return None, None, None\n except KeyError as e:\n print(f\"响应数据格式错误，缺少必要的字段: {e}\")\n print(f\"响应内容: {response.text if 'response' in locals() else 'No response'}\")\n return None, None, None\n except Exception as e:\n print(f\"发生未知错误: {e}\")\n return None, None, None\n\nif __name__ == \"__main__\":\n voice_name, audio_data, saved_filename = create_voice_and_play()\n \n if voice_name:\n print(f\"\\n成功创建音色 '{voice_name}'\")\n print(f\"音频文件已保存: '{saved_filename}'\")\n print(f\"文件大小: {os.path.getsize(saved_filename)} 字节\")\n else:\n print(\"\\n音色创建失败\")", + "java": "import com.google.gson.JsonObject;\nimport com.google.gson.JsonParser;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.util.Base64;\n\npublic class Main {\n public static void main(String[] args) {\n Main example = new Main();\n example.createVoice();\n }\n\n public void createVoice() {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonBody = \"{\\n\" +\n \" \\\"model\\\": \\\"qwen-voice-design\\\",\\n\" +\n \" \\\"input\\\": {\\n\" +\n \" \\\"action\\\": \\\"create\\\",\\n\" +\n \" \\\"target_model\\\": \\\"qwen3-tts-vd-realtime-2025-12-16\\\",\\n\" +\n \" \\\"voice_prompt\\\": \\\"沉稳的中年男性播音员，音色低沉浑厚，富有磁性，语速平稳，吐字清晰，适合用于新闻播报或纪录片解说。\\\",\\n\" +\n \" \\\"preview_text\\\": \\\"各位听众朋友，大家好，欢迎收听晚间新闻。\\\",\\n\" +\n \" \\\"preferred_name\\\": \\\"announcer\\\",\\n\" +\n \" \\\"language\\\": \\\"zh\\\"\\n\" +\n \" },\\n\" +\n \" \\\"parameters\\\": {\\n\" +\n \" \\\"sample_rate\\\": 24000,\\n\" +\n \" \\\"response_format\\\": \\\"wav\\\"\\n\" +\n \" }\\n\" +\n \"}\";\n\n HttpURLConnection connection = null;\n try {\n URL url = new URL(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\");\n connection = (HttpURLConnection) url.openConnection();\n\n connection.setRequestMethod(\"POST\");\n connection.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n connection.setRequestProperty(\"Content-Type\", \"application/json\");\n connection.setDoOutput(true);\n connection.setDoInput(true);\n\n \n try (OutputStream os = connection.getOutputStream()) {\n byte[] input = jsonBody.getBytes(\"UTF-8\");\n os.write(input, 0, input.length);\n os.flush();\n }\n\n \n int responseCode = connection.getResponseCode();\n if (responseCode == HttpURLConnection.HTTP_OK) {\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getInputStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n response.append(responseLine.trim());\n }\n }\n\n \n JsonObject jsonResponse = JsonParser.parseString(response.toString()).getAsJsonObject();\n JsonObject outputObj = jsonResponse.getAsJsonObject(\"output\");\n JsonObject previewAudioObj = outputObj.getAsJsonObject(\"preview_audio\");\n\n \n String voiceName = outputObj.get(\"voice\").getAsString();\n System.out.println(\"音色名称: \" + voiceName);\n\n \n String base64Audio = previewAudioObj.get(\"data\").getAsString();\n\n \n byte[] audioBytes = Base64.getDecoder().decode(base64Audio);\n\n \n String filename = voiceName + \"_preview.wav\";\n saveAudioToFile(audioBytes, filename);\n\n System.out.println(\"音频已保存到本地文件: \" + filename);\n\n } else {\n StringBuilder errorResponse = new StringBuilder();\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(connection.getErrorStream(), \"UTF-8\"))) {\n String responseLine;\n while ((responseLine = br.readLine()) != null) {\n errorResponse.append(responseLine.trim());\n }\n }\n\n System.out.println(\"请求失败，状态码: \" + responseCode);\n System.out.println(\"错误响应: \" + errorResponse.toString());\n }\n\n } catch (Exception e) {\n System.err.println(\"请求发生错误: \" + e.getMessage());\n e.printStackTrace();\n } finally {\n if (connection != null) {\n connection.disconnect();\n }\n }\n }\n\n private void saveAudioToFile(byte[] audioBytes, String filename) {\n try {\n File file = new File(filename);\n try (FileOutputStream fos = new FileOutputStream(file)) {\n fos.write(audioBytes);\n }\n System.out.println(\"音频已保存到: \" + file.getAbsolutePath());\n } catch (IOException e) {\n System.err.println(\"保存音频文件时发生错误: \" + e.getMessage());\n e.printStackTrace();\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json index b1ef359c..8e1ec12b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json @@ -14,10 +14,15 @@ "description": "千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出10个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-voice-enrollment", + "prices": [ + { + "priceUnit": "次", + "price": "0.01", + "type": "tts_vc_model", + "priceName": "声音复刻及声音设计" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,20 +43,19 @@ "latestOnlineAt": "2025-11-27T05:44:15.000+00:00", "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1938", - "offlineTime": "2026-09-07 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118331", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen-Voice-Enrollment", "docUrl": "https://help.aliyun.com/document_detail/2975034.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", - "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", - "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": \"qwen3-tts-vc-realtime-2025-11-27\",\n \"preferred_name\": \"guanyu\",\n \"audio\": {\n \"data\": \"https://xxx.wav\"\n }\n }\n}'", + "python": "import os\nimport requests\nimport base64, pathlib\n\ntarget_model = \"qwen3-tts-vc-realtime-2025-11-27\"\npreferred_name = \"guanyu\"\naudio_mime_type = \"audio/mpeg\"\n\nfile_path = pathlib.Path(\"input.mp3\")\nbase64_str = base64.b64encode(file_path.read_bytes()).decode()\ndata_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\nurl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n\npayload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\n \"data\": data_uri\n }\n }\n}\n\nheaders = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n}\n\nresp = requests.post(url, json=payload, headers=headers)\n\nif resp.status_code == 200:\n data = resp.json()\n voice = data[\"output\"][\"voice\"]\n print(f\"voice name is: {voice}\")\nelse:\n print(\"Failed: \", resp.status_code, resp.text)", + "java": "import com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.util.Base64;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2025-11-27\";\n private static final String PREFERRED_NAME = \"guanyu\";\n private static final String AUDIO_FILE = \"input.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n\n public static String toDataUrl(String filePath) throws Exception {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static void main(String[] args) {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n String apiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\";\n\n try {\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(apiUrl).openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(\"UTF-8\"));\n }\n\n int status = con.getResponseCode();\n InputStream is = (status >= 200 && status < 300)\n ? con.getInputStream()\n : con.getErrorStream();\n\n StringBuilder response = new StringBuilder();\n try (BufferedReader br = new BufferedReader(new InputStreamReader(is, \"UTF-8\"))) {\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n }\n\n System.out.println(\"HTTP status: \" + status);\n System.out.println(\"Response is: \" + response.toString());\n\n if (status == 200) {\n Gson gson = new Gson();\n JsonObject jsonObj = gson.fromJson(response.toString(), JsonObject.class);\n String voice = jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n System.out.println(\"voice name is: \" + voice);\n }\n\n } catch (Exception e) {\n e.printStackTrace();\n }\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json index cc1c7767..56fc3814 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json @@ -20,10 +20,45 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen2.5-omni-7b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "38", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "76", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -50,42 +85,21 @@ "latestOnlineAt": "2025-03-26T12:01:58.000+00:00", "contextWindow": 32768, "maxInputTokens": 30720, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen2.5-Omni-7B", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen2.5-omni-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen2.5-omni-7b\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen2.5-omni-7b\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json index d6052ad9..018c8b38 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash-filetrans", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -36,42 +41,15 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-11-17T13:12:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash-Filetrans", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", - "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", - "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\":[\n 0\n ],\n \"language\": \"zh\", \n \"enable_itn\": false, \n \"corpus\": {\n \"text\": \"张三，李四，王五\"\n }\n }\n}'\n\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header 'Authorization: Bearer $DASHSCOPE_API_KEY' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json'", + "python": "import os\nimport time\nimport requests\nimport json\n\n\nAPI_URL_SUBMIT = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\"\nAPI_URL_QUERY_BASE = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/\"\n\n\ndef main():\n # If no environment variable is configured, please replace the downlink with the Bailian API Key: api_key = \"sk-xxx\"\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\",\n \"X-DashScope-Async\": \"enable\"\n }\n\n\n payload = {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n # \"language\": \"zh\",\n \"enable_itn\": False\n # \"corpus\": {\n # \"text\": \"\"\n # }\n }\n }\n\n\n try:\n submit_resp = requests.post(API_URL_SUBMIT, headers=headers, data=json.dumps(payload))\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if submit_resp.status_code != 200:\n print(f\"Failed! HTTP code: {submit_resp.status_code}\")\n print(submit_resp.text)\n return\n\n resp_data = submit_resp.json()\n output = resp_data.get(\"output\")\n if not output or \"task_id\" not in output:\n print(\"resp_data:\", resp_data)\n return\n\n task_id = output[\"task_id\"]\n print(f\"任务已提交，task_id: {task_id}\")\n\n\n finished = False\n while not finished:\n time.sleep(2)\n\n query_url = API_URL_QUERY_BASE + task_id\n try:\n query_resp = requests.get(query_url, headers=headers)\n except requests.RequestException as e:\n print(f\"Failed: {e}\")\n return\n\n if query_resp.status_code != 200:\n print(f\"Failed! HTTP code: {query_resp.status_code}\")\n print(query_resp.text)\n return\n\n query_data = query_resp.json()\n output = query_data.get(\"output\")\n if output and \"task_status\" in output:\n status = output[\"task_status\"]\n print(f\"status: {status}\")\n\n if status.upper() in (\"SUCCEEDED\", \"FAILED\", \"UNKNOWN\"):\n finished = True\n print(\"task finished：\")\n print(json.dumps(query_data, indent=2, ensure_ascii=False))\n else:\n print(\"query data:\", query_data)\n\n\nif __name__ == \"__main__\":\n main()", + "java": "import com.google.gson.Gson;\nimport com.google.gson.annotations.SerializedName;\nimport okhttp3.*;\n\nimport java.io.IOException;\nimport java.util.concurrent.TimeUnit;\n\npublic class Main {\n private static final String API_URL_SUBMIT = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/asr/transcription\";\n private static final String API_URL_QUERY = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/\";\n private static final Gson gson = new Gson();\n\n public static void main(String[] args) {\n // If no environment variable is configured, please replace the downlink with the Bailian API Key: String apiKey = \"sk-xxx\"\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n OkHttpClient client = new OkHttpClient();\n\n String payloadJson = \"\"\"\n {\n \"model\": \"qwen3-asr-flash-filetrans\",\n \"input\": {\n \"file_url\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"\n },\n \"parameters\": {\n \"channel_id\": [0],\n \"enable_itn\": false\n }\n }\n \"\"\";\n\n RequestBody body = RequestBody.create(payloadJson, MediaType.get(\"application/json; charset=utf-8\"));\n Request submitRequest = new Request.Builder()\n .url(API_URL_SUBMIT)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"Content-Type\", \"application/json\")\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .post(body)\n .build();\n\n String taskId = null;\n\n try (Response response = client.newCall(submitRequest).execute()) {\n if (response.isSuccessful() && response.body() != null) {\n String respBody = response.body().string();\n ApiResponse apiResp = gson.fromJson(respBody, ApiResponse.class);\n if (apiResp.output != null) {\n taskId = apiResp.output.taskId;\n System.out.println(\"task_id: \" + taskId);\n } else {\n System.out.println(\"respBody: \" + respBody);\n return;\n }\n } else {\n System.out.println(\"Failed! HTTP code: \" + response.code());\n if (response.body() != null) {\n System.out.println(response.body().string());\n }\n return;\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n\n boolean finished = false;\n while (!finished) {\n try {\n TimeUnit.SECONDS.sleep(2);\n } catch (InterruptedException e) {\n Thread.currentThread().interrupt();\n return;\n }\n\n String queryUrl = API_URL_QUERY + taskId;\n Request queryRequest = new Request.Builder()\n .url(queryUrl)\n .addHeader(\"Authorization\", \"Bearer \" + apiKey)\n .addHeader(\"X-DashScope-Async\", \"enable\")\n .addHeader(\"Content-Type\", \"application/json\")\n .get()\n .build();\n\n try (Response response = client.newCall(queryRequest).execute()) {\n if (response.body() != null) {\n String queryResponse = response.body().string();\n ApiResponse apiResp = gson.fromJson(queryResponse, ApiResponse.class);\n\n if (apiResp.output != null && apiResp.output.taskStatus != null) {\n String status = apiResp.output.taskStatus;\n System.out.println(\"task status: \" + status);\n if (\"SUCCEEDED\".equalsIgnoreCase(status)\n || \"FAILED\".equalsIgnoreCase(status)\n || \"UNKNOWN\".equalsIgnoreCase(status)) {\n finished = true;\n System.out.println(\"task finished: \");\n System.out.println(queryResponse);\n }\n } else {\n System.out.println(\"query response: \" + queryResponse);\n }\n }\n } catch (IOException e) {\n e.printStackTrace();\n return;\n }\n }\n }\n\n static class ApiResponse {\n @SerializedName(\"request_id\")\n String requestId;\n Output output;\n }\n\n static class Output {\n @SerializedName(\"task_id\")\n String taskId;\n @SerializedName(\"task_status\")\n String taskStatus;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json index 9afac99e..55ce754a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash-realtime", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00033", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -36,36 +41,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-10-27T10:00:46.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2989727.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json index 0d5c8ec0..7df8bfc9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-asr-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.00022", + "type": "content_duration", + "priceName": "音频时长" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -37,15 +42,15 @@ "modelAlias": "qwen3-asr-flash", "versionTag": "MAJOR", "latestOnlineAt": "2025-09-08T05:39:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-ASR-Flash", "docUrl": "https://help.aliyun.com/document_detail/2979031.html", - "predictConfig": [], "samples": { "dashscope": { "default": { - "python": "import os\nimport dashscope\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\npublic class Main {\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"system\",\n \"content\": [\n # 此处用于配置定制化识别的Context\n {\"text\": \"\"},\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"audio\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\"},\n ]\n }\n]\nresponse = dashscope.MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-asr-flash\",\n messages=messages,\n result_format=\"message\",\n asr_options={\n # \"language\": \"zh\", # 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n \"enable_lid\":True,\n \"enable_itn\":False\n }\n)\nprint(response)", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport java.util.HashMap;\nimport java.util.Map;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder()\n .role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"audio\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3\")))\n .build();\n\n MultiModalMessage sysMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n // 此处用于配置定制化识别的Context\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"\")))\n .build();\n\n Map asrOptions = new HashMap<>();\n asrOptions.put(\"enable_lid\", true);\n asrOptions.put(\"enable_itn\", false);\n // asrOptions.put(\"language\", \"zh\"); // 可选，若已知音频的语种，可通过该参数指定待识别语种，以提升识别准确率\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-asr-flash\")\n .message(userMessage)\n .message(sysMessage)\n .parameter(\"asr_options\", asrOptions)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/2986952.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json index 53b49766..01975fab 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json @@ -17,9 +17,6 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-30b-a3b-instruct", "qpmInfo": { "model-default-actual": { @@ -39,6 +36,84 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.25", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "37.5", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -49,54 +124,28 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-30B-A3B-Instruct", "docUrl": "https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-30b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-30b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-30b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json index d9010d47..e3c85793 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json @@ -17,9 +17,6 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-480b-a35b-instruct", "qpmInfo": { "model-default-actual": { @@ -39,6 +36,84 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -49,54 +124,27 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-480B-A35B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-480b-a35b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-480b-a35b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-480b-a35b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-480b-a35b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json index 71e5e762..b61014ad 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json @@ -19,9 +19,6 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-flash", "qpmInfo": { "model-default-actual": { @@ -41,6 +38,156 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -49,56 +196,23 @@ "latestOnlineAt": "2025-08-04T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 997952, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-Flash", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-flash\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-flash\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-flash\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-flash\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-flash\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json index 1d369815..b8057431 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json @@ -19,9 +19,6 @@ "cache" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-plus", "qpmInfo": { "model-default-actual": { @@ -41,6 +38,156 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "200", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG" ], @@ -51,60 +198,27 @@ "maxInputTokens": 997952, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Coder-Plus", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-plus\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-plus\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-plus\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-plus\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-plus\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-plus\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json index 3b4b1748..9359f9c8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json @@ -17,10 +17,33 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "64", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "240", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -47,40 +70,19 @@ "latestOnlineAt": "2025-09-23T11:11:30.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-LiveTranslate-Flash-Realtime", "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", "docUrl": "https://help.aliyun.com/document_detail/2983281.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json index 73d436d0..6386eb81 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json @@ -16,10 +16,33 @@ "description": "Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，Qwen3-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂19种语言，会说10种语言以及8种中文方言。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3-livetranslate-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -46,36 +69,14 @@ "latestOnlineAt": "2025-12-04T12:21:57.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", - "name": "Qwen3-LiveTranslate-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2999748.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-LiveTranslate-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2999748.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json index 95e6b638..bc82e62a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json @@ -11,7 +11,74 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Completions API", + "name": "search_strategy:agent_max", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent_max", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY", + "tag": "限时优惠" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], "description": "千问3系列Max模型，相较preview版本在智能体编程与工具调用方向进行了专项升级。本次发布的正式版模型达到领域SOTA水平，适配场景更加复杂的智能体需求。", "collectionTag": "qwen3", "features": [ @@ -24,9 +91,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-max", "qpmInfo": { "model-default-actual": { @@ -46,6 +110,197 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.125", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "8.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.7", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG", "Reasoning" @@ -58,93 +313,34 @@ "maxInputTokens": 258048, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3-max\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3-max\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -159,7 +355,21 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3系列Max模型Preview版本，实现思考模式和非思考模式的有效融合。思考模式下在智能体编程能力、常识知识推理能力、数学/科学/通用类推理等能力上均有显著增强。", "collectionTag": "qwen3", "features": [ @@ -167,9 +377,6 @@ "cache" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-max-preview", "qpmInfo": { "model-default-actual": { @@ -189,6 +396,83 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG", "Reasoning" @@ -201,86 +485,27 @@ "maxInputTokens": 258048, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-max-preview\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-max-preview\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-max-preview\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-max-preview\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-max-preview\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json index 98eec78f..271a47c9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json @@ -15,10 +15,21 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-30b-a3b-captioner", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -45,36 +56,15 @@ "latestOnlineAt": "2025-09-16T16:38:17.000+00:00", "contextWindow": 65536, "maxInputTokens": 32768, - "inferenceProvider": "bailian", - "name": "Qwen3-Omni-30b-a3b-Captioner", - "docUrl": "https://help.aliyun.com/document_detail/2980468.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" } - ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3-Omni-30b-a3b-Captioner", + "docUrl": "https://help.aliyun.com/document_detail/2980468.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json index 50dfdaf6..263adc45 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json @@ -19,10 +19,45 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.2", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "18.9", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.9", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "8.3", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.2", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "75.1", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -50,49 +85,20 @@ "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", "contextWindow": 65536, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Omni-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json index 494aeafe..5e54cd14 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json @@ -22,10 +22,75 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-omni-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "text_input_token", + "priceName": "输入：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "vision_input_token", + "priceName": "输入：图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "type": "purein_text_output_token", + "priceName": "输出：文本（输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "multiin_text_output_token", + "priceName": "输出：文本（输入包含图片/音频/视频时）" + }, + { + "priceUnit": "每百万tokens", + "price": "62.6", + "type": "multi_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "thinking_text_input_token", + "priceName": "输入：文本（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "15.8", + "type": "thinking_audio_input_token", + "priceName": "输入：音频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "thinking_vision_input_token", + "priceName": "输入：图片/视频（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6.9", + "type": "thinking_purein_text_output_token", + "priceName": "输出：文本（思考模式下，输入仅包含文本时）" + }, + { + "priceUnit": "每百万tokens", + "price": "12.7", + "type": "thinking_multiin_text_output_token", + "priceName": "输出：文本（思考模式下，输入包含图片/音频/视频时）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -54,50 +119,15 @@ "latestOnlineAt": "2025-12-03T16:00:00.000+00:00", "contextWindow": 65536, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Omni-Flash", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "开启深度思考，开启后将不支持音频输出" - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json index fd92af23..e4dc9381 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,16 +43,9 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen3-tts-flash-realtime-2025-11-27", "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json index e1cf65b5..9195d43d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json @@ -17,10 +17,15 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-flash", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -41,16 +46,9 @@ "shortDescription": "韵律拟人，低延迟，支持十种语言和国内多种方言输出", "equivalentSnapshot": "qwen3-tts-flash-2025-11-27", "latestOnlineAt": "2025-11-26T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Flash", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json index dd146c2e..587cc5ae 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-instruct-flash-realtime", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,36 +42,15 @@ "modelAlias": "qwen3-tts-instruct-flash-realtime", "versionTag": "MAJOR", "latestOnlineAt": "2026-01-21T07:33:55.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-instruct-flash-realtime", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json index 875e188b..8c3f0c4c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json @@ -15,11 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-instruct-flash", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -39,46 +44,20 @@ "versionTag": "MAJOR", "equivalentSnapshot": "qwen3-tts-instruct-flash-2026-01-26", "latestOnlineAt": "2026-02-10T02:56:41.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-Instruct-Flash", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageHint", - "default": "" - }, - { - "name": "音量", - "key": "volume", - "default": 50, - "tip": "数值越大，合成音频声音越大", - "range": [ - 0, - 100 - ] - }, - { - "name": "语速", - "key": "speechRate", - "default": 1, - "tip": "数值越大，合成音频语速越快", - "range": [ - 0.5, - 2 - ] - }, - { - "name": "指令控制", - "key": "instructions", - "tip": "仅支持中英文，通过自然语言合成语音的语气、语速、情感及人物性格，需要具体客观的描述文字，如：\n· 请用非常激昂且高亢的语气说话，表现出获得重大成功后的狂喜与激动。\n· 语速请保持中等偏慢，语气要显得优雅、知性，给人以从容不迫的安心感。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", - "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "python": "import os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\ntext = \"Dear listeners, hello everyone. Welcome to the evening news.\"\n\nresponse = dashscope.MultiModalConversation.call(\n model=\"qwen3-tts-instruct-flash\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=\"Cherry\",\n instructions='The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.',\n optimize_instructions=True,\n stream=False\n)\nprint(response)", + "java": "import com.alibaba.dashscope.aigc.multimodalconversation.AudioParameters;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.io.FileOutputStream;\nimport java.io.InputStream;\nimport java.net.URL;\n\npublic class Main {\n private static final String MODEL = \"qwen3-tts-instruct-flash\";\n public static void call() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(MODEL)\n .text(\"Today is a wonderful day to build something people love!\")\n .voice(AudioParameters.Voice.CHERRY)\n .parameter(\"instructions\",\"The speaking speed is fast and there is a distinct upward inflection, which is suitable for introducing fashionable products.\")\n .parameter(\"optimize_instructions\",true)\n .build();\n MultiModalConversationResult result = conv.call(param);\n String audioUrl = result.getOutput().getAudio().getUrl();\n System.out.print(audioUrl);\n\n // 下载音频文件到本地\n try (InputStream in = new URL(audioUrl).openStream();\n FileOutputStream out = new FileOutputStream(\"downloaded_audio.wav\")) {\n byte[] buffer = new byte[1024];\n int bytesRead;\n while ((bytesRead = in.read(buffer)) != -1) {\n out.write(buffer, 0, bytesRead);\n }\n } catch (Exception e) {\n System.out.println(\"\\nError message: \" + e.getMessage());\n }\n }\n public static void main(String[] args) {\n try {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n call();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json index 0c9eee16..35e45139 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vc-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,22 +42,21 @@ "modelAlias": "qwen3-tts-vc-realtime-0115", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-01-14T11:23:59.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-vc-realtime-2026-01-15", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", - "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "# DashScope SDK Version>=1.23.9，Python Version >=3.10\n# coding=utf-8\n# Installation instructions for pyaudio:\n# APPLE Mac OS X\n# brew install portaudio\n# pip install pyaudio\n# Debian/Ubuntu\n# sudo apt-get install python-pyaudio python3-pyaudio\n# or\n# pip install pyaudio\n# CentOS\n# sudo yum install -y portaudio portaudio-devel && pip install pyaudio\n# Microsoft Windows\n# python -m pip install pyaudio\n\nimport pyaudio\nimport os\nimport requests\nimport base64\nimport pathlib\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import QwenTtsRealtime, QwenTtsRealtimeCallback, AudioFormat\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\nTEXT_TO_SYNTHESIZE = [\n 'Today is a wonderful day to build something people love!'\n]\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"The audio file does not exist {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n\n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"Failed to create voice: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"The voice response failed to be resolved: {e}\")\n\ndef init_dashscope_api_key():\n dashscope.api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self._player = pyaudio.PyAudio()\n self._stream = self._player.open(\n format=pyaudio.paInt16, channels=1, rate=24000, output=True\n )\n\n def on_open(self) -> None:\n print('[TTS] has been established')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self._stream.stop_stream()\n self._stream.close()\n self._player.terminate()\n print(f'[TTS] close, code={close_status_code}, msg={close_msg}')\n\n def on_event(self, response: dict) -> None:\n try:\n event_type = response.get('type', '')\n if event_type == 'session.created':\n print(f'[TTS] session begin: {response[\"session\"][\"id\"]}')\n elif event_type == 'response.audio.delta':\n audio_data = base64.b64decode(response['delta'])\n self._stream.write(audio_data)\n elif event_type == 'response.done':\n print(f'[TTS] response complete, Response ID: {qwen_tts_realtime.get_last_response_id()}')\n elif event_type == 'session.finished':\n print('[TTS] session end')\n self.complete_event.set()\n except Exception as e:\n print(f'[Error] callback error: {e}')\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n print('Qwen TTS Realtime ...')\n\n callback = MyCallback()\n qwen_tts_realtime = QwenTtsRealtime(\n model=DEFAULT_TARGET_MODEL,\n callback=callback,\n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n qwen_tts_realtime.connect()\n \n qwen_tts_realtime.update_session(\n voice=create_voice(VOICE_FILE_PATH),\n response_format=AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode='server_commit'\n )\n\n for text_chunk in TEXT_TO_SYNTHESIZE:\n print(f'[send text]: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n\n print(f'[Metric] session_id={qwen_tts_realtime.get_session_id()}, '\n f'first_audio_delay={qwen_tts_realtime.get_first_audio_delay()}s')", + "java": "// Java DashScope SDK Version >= 2.20.9\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.Gson;\nimport com.google.gson.JsonObject;\n\nimport javax.sound.sampled.*;\nimport java.io.*;\nimport java.net.HttpURLConnection;\nimport java.net.URL;\nimport java.nio.file.*;\nimport java.nio.charset.StandardCharsets;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n private static final String TARGET_MODEL = \"qwen3-tts-vc-realtime-2026-01-15\";\n private static final String PREFERRED_NAME = \"guanyu\";\n \n private static final String AUDIO_FILE = \"voice.mp3\";\n private static final String AUDIO_MIME_TYPE = \"audio/mpeg\";\n private static String[] textToSynthesize = {\n \"Today is a wonderful day to build something people love!\"\n };\n\n public static String toDataUrl(String filePath) throws IOException {\n byte[] bytes = Files.readAllBytes(Paths.get(filePath));\n String encoded = Base64.getEncoder().encodeToString(bytes);\n return \"data:\" + AUDIO_MIME_TYPE + \";base64,\" + encoded;\n }\n\n public static String createVoice() throws Exception {\n String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n String jsonPayload =\n \"{\"\n + \"\\\"model\\\": \\\"qwen-voice-enrollment\\\",\"\n + \"\\\"input\\\": {\"\n + \"\\\"action\\\": \\\"create\\\",\"\n + \"\\\"target_model\\\": \\\"\" + TARGET_MODEL + \"\\\",\"\n + \"\\\"preferred_name\\\": \\\"\" + PREFERRED_NAME + \"\\\",\"\n + \"\\\"audio\\\": {\"\n + \"\\\"data\\\": \\\"\" + toDataUrl(AUDIO_FILE) + \"\\\"\"\n + \"}\"\n + \"}\"\n + \"}\";\n\n HttpURLConnection con = (HttpURLConnection) new URL(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\").openConnection();\n con.setRequestMethod(\"POST\");\n con.setRequestProperty(\"Authorization\", \"Bearer \" + apiKey);\n con.setRequestProperty(\"Content-Type\", \"application/json\");\n con.setDoOutput(true);\n\n try (OutputStream os = con.getOutputStream()) {\n os.write(jsonPayload.getBytes(StandardCharsets.UTF_8));\n }\n\n int status = con.getResponseCode();\n System.out.println(\"HTTP status: \" + status);\n\n try (BufferedReader br = new BufferedReader(\n new InputStreamReader(status >= 200 && status < 300 ? con.getInputStream() : con.getErrorStream(),\n StandardCharsets.UTF_8))) {\n StringBuilder response = new StringBuilder();\n String line;\n while ((line = br.readLine()) != null) {\n response.append(line);\n }\n System.out.println(\"response: \" + response);\n\n if (status == 200) {\n JsonObject jsonObj = new Gson().fromJson(response.toString(), JsonObject.class);\n return jsonObj.getAsJsonObject(\"output\").get(\"voice\").getAsString();\n }\n throw new IOException(\"failed: \" + status + \" - \" + response);\n }\n }\n\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws Exception {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(TARGET_MODEL)\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n\n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // Processing when the connection is established\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // Processing at the time of session creation\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n \n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n break;\n case \"session.finished\":\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // Handling when the connection is closed\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n .voice(createVoice())\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n\n\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json index 0d15a100..c30acbf4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json @@ -15,11 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vc-2026-01-22", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,20 +43,19 @@ "modelAlias": "qwen3-tts-vc-0122", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-02-10T03:00:44.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-VC-2026-01-22", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" + "python": "import os\nimport requests\nimport base64\nimport pathlib\nimport dashscope\n\n\nDEFAULT_TARGET_MODEL = \"qwen3-tts-vc-2026-01-22\"\nDEFAULT_PREFERRED_NAME = \"guanyu\"\nDEFAULT_AUDIO_MIME_TYPE = \"audio/mpeg\"\nVOICE_FILE_PATH = \"voice.mp3\"\n\n\ndef create_voice(file_path: str,\n target_model: str = DEFAULT_TARGET_MODEL,\n preferred_name: str = DEFAULT_PREFERRED_NAME,\n audio_mime_type: str = DEFAULT_AUDIO_MIME_TYPE) -> str:\n api_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n file_path_obj = pathlib.Path(file_path)\n if not file_path_obj.exists():\n raise FileNotFoundError(f\"音频文件不存在: {file_path}\")\n\n base64_str = base64.b64encode(file_path_obj.read_bytes()).decode()\n data_uri = f\"data:{audio_mime_type};base64,{base64_str}\"\n\n url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/audio/tts/customization\"\n payload = {\n \"model\": \"qwen-voice-enrollment\",\n \"input\": {\n \"action\": \"create\",\n \"target_model\": target_model,\n \"preferred_name\": preferred_name,\n \"audio\": {\"data\": data_uri}\n }\n }\n headers = {\n \"Authorization\": f\"Bearer {api_key}\",\n \"Content-Type\": \"application/json\"\n }\n\n resp = requests.post(url, json=payload, headers=headers)\n if resp.status_code != 200:\n raise RuntimeError(f\"create voice failed: {resp.status_code}, {resp.text}\")\n\n try:\n return resp.json()[\"output\"][\"voice\"]\n except (KeyError, ValueError) as e:\n raise RuntimeError(f\"failed: {e}\")\n\n\nif __name__ == '__main__':\n dashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n text = \"今天天气怎么样？\"\n \n response = dashscope.MultiModalConversation.call(\n model=DEFAULT_TARGET_MODEL,\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n text=text,\n voice=create_voice(VOICE_FILE_PATH),\n stream=False\n )\n print(response)" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json index 518a47eb..2ab0d68f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json @@ -15,10 +15,15 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vd-realtime-2026-01-15", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -37,42 +42,21 @@ "modelAlias": "qwen3-tts-vd-realtime-0115", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-01-14T11:14:10.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "qwen3-tts-vd-realtime-2026-01-15", "docUrl": "https://help.aliyun.com/document_detail/2938790.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-realtime-2026-01-15',\n callback=callback, \n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-realtime-2026-01-15\")\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json index f95a31a7..a44db4e3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json @@ -15,11 +15,16 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-tts-vd-2026-01-26", "iconUrl": "", + "prices": [ + { + "priceUnit": "每万字符", + "price": "0.8", + "type": "cosy_tts_number", + "priceName": "语音合成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,21 +43,20 @@ "modelAlias": "qwen3-tts-vd-0126", "versionTag": "SNAPSHOT", "latestOnlineAt": "2026-02-10T02:59:43.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-TTS-VD-2026-01-26", "docUrl": "https://help.aliyun.com/document_detail/2879134.html", - "predictConfig": [ - { - "name": "语言", - "key": "languageType", - "default": "Chinese" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", - "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" + "python": "import os\nimport base64\nimport threading\nimport time\nimport dashscope\nfrom dashscope.audio.qwen_tts_realtime import *\n\nqwen_tts_realtime: QwenTtsRealtime = None\ntext_to_synthesize = [\n '今天天气怎么样？'\n]\n\nDO_VIDEO_TEST = False\n\ndef init_dashscope_api_key():\n \"\"\"\n Set your DashScope API-key. More information:\n https://github.com/aliyun/alibabacloud-bailian-speech-demo/blob/master/PREREQUISITES.md\n \"\"\"\n if 'DASHSCOPE_API_KEY' in os.environ:\n dashscope.api_key = os.environ[\n 'DASHSCOPE_API_KEY'] # load API-key from environment variable DASHSCOPE_API_KEY\n else:\n dashscope.api_key = 'your-dashscope-api-key' # set API-key manually\n\n\n\nclass MyCallback(QwenTtsRealtimeCallback):\n def __init__(self):\n self.complete_event = threading.Event()\n self.file = open('result_24k.pcm', 'wb')\n\n def on_open(self) -> None:\n print('connection opened, init player')\n\n def on_close(self, close_status_code, close_msg) -> None:\n self.file.close()\n print('connection closed with code: {}, msg: {}, destroy player'.format(close_status_code, close_msg))\n\n def on_event(self, response: str) -> None:\n try:\n global qwen_tts_realtime\n type = response['type']\n if 'session.created' == type:\n print('start session: {}'.format(response['session']['id']))\n if 'response.audio.delta' == type:\n recv_audio_b64 = response['delta']\n self.file.write(base64.b64decode(recv_audio_b64))\n if 'response.done' == type:\n print(f'response {qwen_tts_realtime.get_last_response_id()} done')\n if 'session.finished' == type:\n print('session finished')\n self.complete_event.set()\n except Exception as e:\n print('[Error] {}'.format(e))\n return\n\n def wait_for_finished(self):\n self.complete_event.wait()\n\n\nif __name__ == '__main__':\n init_dashscope_api_key()\n\n print('Initializing ...')\n\n callback = MyCallback()\n\n qwen_tts_realtime = QwenTtsRealtime(\n model='qwen3-tts-vd-2026-01-26',\n callback=callback, \n url='wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n )\n\n qwen_tts_realtime.connect()\n qwen_tts_realtime.update_session(\n # 将下面的音色替换为实际的设计音色\n voice = 'your_design_voice',\n response_format = AudioFormat.PCM_24000HZ_MONO_16BIT,\n mode = 'server_commit' \n )\n for text_chunk in text_to_synthesize:\n print(f'send texd: {text_chunk}')\n qwen_tts_realtime.append_text(text_chunk)\n time.sleep(0.1)\n qwen_tts_realtime.finish()\n callback.wait_for_finished()\n print('[Metric] session: {}, first audio delay: {}'.format(\n qwen_tts_realtime.get_session_id(), \n qwen_tts_realtime.get_first_audio_delay(),\n ))", + "java": "// Dashscope SDK 版本不低于2.21.16\nimport com.alibaba.dashscope.audio.qwen_tts_realtime.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.LineUnavailableException;\nimport javax.sound.sampled.SourceDataLine;\nimport javax.sound.sampled.AudioFormat;\nimport javax.sound.sampled.DataLine;\nimport javax.sound.sampled.AudioSystem;\nimport java.io.FileNotFoundException;\nimport java.io.IOException;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.CountDownLatch;\nimport java.util.concurrent.atomic.AtomicReference;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\n\npublic class Main {\n static String[] textToSynthesize = {\n \"今天天气怎么样？\"\n };\n\n \n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n\n \n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n\n \n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n\n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n \n Thread.sleep(audioLength - 10);\n }\n\n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n\n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n\n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n\n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n }\n\n public static void main(String[] args) throws InterruptedException, LineUnavailableException, FileNotFoundException {\n QwenTtsRealtimeParam param = QwenTtsRealtimeParam.builder()\n .model(\"qwen3-tts-vd-2026-01-26\")\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .build();\n AtomicReference completeLatch = new AtomicReference<>(new CountDownLatch(1));\n final AtomicReference qwenTtsRef = new AtomicReference<>(null);\n \n \n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n \n QwenTtsRealtime qwenTtsRealtime = new QwenTtsRealtime(param, new QwenTtsRealtimeCallback() {\n @Override\n public void onOpen() {\n // 连接建立时的处理\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n // 会话创建时的处理\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n // 实时播放音频\n audioPlayer.write(recvAudioB64);\n break;\n case \"response.done\":\n // 响应完成时的处理\n break;\n case \"session.finished\":\n // 会话结束时的处理\n completeLatch.get().countDown();\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n // 连接关闭时的处理\n }\n });\n qwenTtsRef.set(qwenTtsRealtime);\n try {\n qwenTtsRealtime.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n QwenTtsRealtimeConfig config = QwenTtsRealtimeConfig.builder()\n // 替换下面的音色\n .voice(\"your_desigin_voice\")\n .responseFormat(QwenTtsRealtimeAudioFormat.PCM_24000HZ_MONO_16BIT)\n .mode(\"server_commit\")\n .build();\n qwenTtsRealtime.updateSession(config);\n for (String text:textToSynthesize) {\n qwenTtsRealtime.appendText(text);\n Thread.sleep(100);\n }\n qwenTtsRealtime.finish();\n completeLatch.get().await();\n qwenTtsRealtime.close();\n \n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json index 8f009242..aa275109 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json @@ -23,9 +23,6 @@ "cache" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-flash", "qpmInfo": { "model-default-actual": { @@ -45,6 +42,197 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.075", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.015", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.375", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.03", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.06", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "VU", "Reasoning" @@ -58,85 +246,26 @@ "maxInputTokens": 260096, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Flash", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-flash\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-flash\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-flash',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-flash\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-flash\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json index f662f875..a28289a1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json @@ -24,9 +24,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-plus", "qpmInfo": { "model-default-actual": { @@ -46,6 +43,197 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.875", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.15", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.3", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "VU", "Reasoning" @@ -57,81 +245,22 @@ "latestOnlineAt": "2026-01-25T16:00:00.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-Plus", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-plus\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-plus\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-plus',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-plus\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-plus\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json index 5d0abd08..42799f67 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Flash模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步；响应速度快，兼具推理速度和性能。", "collectionTag": "qwen3.5", "features": [ @@ -26,9 +92,6 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-flash", "iconUrl": "", "qpmInfo": { @@ -49,10 +112,183 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "0.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.02", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", - "VU", - "TG" + "TG", + "VU" ], "modelAlias": "", "versionTag": "MAJOR", @@ -60,88 +296,30 @@ "latestOnlineAt": "2026-02-23T03:28:01.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json index 63a65f52..56cf0c2d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json @@ -17,10 +17,33 @@ "collectionTag": "qwen3.5", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-livetranslate-flash-realtime", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "translate_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "translate_vision_input_token", + "priceName": "输入：图片" + }, + { + "priceUnit": "每百万tokens", + "price": "100", + "type": "translate_multi_text_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "160", + "type": "translate_multi_output_token", + "priceName": "输出：音频" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -48,41 +71,14 @@ "latestOnlineAt": "2026-05-19T08:25:22.000+00:00", "contextWindow": 53248, "maxInputTokens": 49152, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-LiveTranslate-Flash-Realtime", "docUrl": "https://www.alibabacloud.com/help/en/document_detail/2983281.html", "category": "Audio", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", + "python": "# pip install websocket-client\nimport json\nimport websocket\nimport os\n\nAPI_KEY=os.getenv(\"DASHSCOPE_API_KEY\")\nAPI_URL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-livetranslate-flash-realtime\"\n\nheaders = [\n\"Authorization: Bearer \" + API_KEY\n]\n\ndef on_open(ws):\nprint(f\"Connected to server: {API_URL}\")\ndef on_message(ws, message):\ndata = json.loads(message)\nprint(\"Received event:\", json.dumps(data, indent=2))\ndef on_error(ws, error):\nprint(\"Error:\", error)\n\nws = websocket.WebSocketApp(\nAPI_URL,\nheader=headers,\non_open=on_open,\non_message=on_message,\non_error=on_error\n)\n\nws.run_forever()", "docUrl": "https://help.aliyun.com/document_detail/2983281.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json index 91c827b8..f6d0ac91 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json @@ -17,10 +17,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-ocr", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -48,49 +59,22 @@ "contextWindow": 65536, "maxInputTokens": 49152, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-OCR", "docUrl": "https://help.aliyun.com/document_detail/2860683.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3.5-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3.5-ocr\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-ocr\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3.5-ocr',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3.5-ocr\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json index 9954766f..f7b2ad78 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json @@ -15,7 +15,21 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", "collectionTag": "Qwen3.5", "features": [ @@ -23,11 +37,34 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-flash-realtime", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "27", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "107", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.3", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -56,42 +93,15 @@ "latestOnlineAt": "2026-03-30T03:54:16.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Flash-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2880812.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-flash-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-flash-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json index 7a872831..438c6296 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json @@ -15,18 +15,55 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", "collectionTag": "Qwen3.5", "features": [ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-flash", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "72", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.2", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "13.3", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -55,45 +92,16 @@ "latestOnlineAt": "2026-03-30T03:58:51.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Flash", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json index e20db8f9..2d288870 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json @@ -15,7 +15,21 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。", "collectionTag": "Qwen3.5", "features": [ @@ -23,11 +37,34 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-plus-realtime", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "80", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "300", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "60", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -56,42 +93,15 @@ "latestOnlineAt": "2026-03-30T03:54:11.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Plus-Realtime", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", - "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", + "python": "# 依赖：dashscope >= 1.23.9，pyaudio\nimport os\nimport base64\nimport time\n\nimport pyaudio\nfrom dashscope.audio.qwen_omni import MultiModality, AudioFormat,OmniRealtimeCallback,OmniRealtimeConversation\nimport dashscope\n\n\nurl = f'wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime'\n# 配置 API Key，若没有设置环境变量，请用 API Key 将下行替换为 dashscope.api_key = \"sk-xxx\"\ndashscope.api_key = os.getenv('DASHSCOPE_API_KEY')\n# 指定音色\nvoice = 'Ethan'\n# 指定模型\nmodel = 'qwen3.5-omni-plus-realtime'\n# 指定模型角色\ninstructions = \"你是个人助理小云，请用幽默风趣的方式回答用户的问题\"\nclass SimpleCallback(OmniRealtimeCallback):\n def __init__(self, pya):\n self.pya = pya\n self.out = None\n def on_open(self):\n # 初始化音频输出流\n self.out = self.pya.open(\n format=pyaudio.paInt16,\n channels=1,\n rate=24000,\n output=True\n )\n def on_event(self, response):\n if response['type'] == 'response.audio.delta':\n # 播放音频\n self.out.write(base64.b64decode(response['delta']))\n elif response['type'] == 'conversation.item.input_audio_transcription.completed':\n # 打印转录文本\n print(f\"[User] {response['transcript']}\")\n elif response['type'] == 'response.audio_transcript.done':\n # 打印助手回复文本\n print(f\"[LLM] {response['transcript']}\")\n\n# 1. 初始化音频设备\npya = pyaudio.PyAudio()\n# 2. 创建回调函数和会话\ncallback = SimpleCallback(pya)\nconv = OmniRealtimeConversation(model=model, callback=callback, url=url)\n# 3. 建立连接并配置会话\nconv.connect()\nconv.update_session(output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], voice=voice, instructions=instructions)\n# 4. 初始化音频输入流\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\n# 5. 主循环处理音频输入\nprint(\"对话已开始，对着麦克风说话 (Ctrl+C 退出)...\")\ntry:\n while True:\n audio_data = mic.read(3200, exception_on_overflow=False)\n conv.append_audio(base64.b64encode(audio_data).decode())\n time.sleep(0.01)\nexcept KeyboardInterrupt:\n # 清理资源\n conv.close()\n mic.close()\n callback.out.close()\n pya.terminate()\n print(\"\\n对话结束\")", + "java": "// DashScope Java SDK 版本不低于2.20.9\n\nimport com.alibaba.dashscope.audio.omni.*;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.google.gson.JsonObject;\nimport javax.sound.sampled.*;\nimport java.io.IOException;\nimport java.nio.ByteBuffer;\nimport java.util.Arrays;\nimport java.util.Base64;\nimport java.util.Queue;\nimport java.util.concurrent.ConcurrentLinkedQueue;\nimport java.util.concurrent.atomic.AtomicBoolean;\nimport java.util.concurrent.atomic.AtomicReference;\n\npublic class OmniServerVad {\n // RealtimePcmPlayer 类定义开始\n public static class RealtimePcmPlayer {\n private int sampleRate;\n private SourceDataLine line;\n private AudioFormat audioFormat;\n private Thread decoderThread;\n private Thread playerThread;\n private AtomicBoolean stopped = new AtomicBoolean(false);\n private Queue b64AudioBuffer = new ConcurrentLinkedQueue<>();\n private Queue RawAudioBuffer = new ConcurrentLinkedQueue<>();\n \n // 构造函数初始化音频格式和音频线路\n public RealtimePcmPlayer(int sampleRate) throws LineUnavailableException {\n this.sampleRate = sampleRate;\n this.audioFormat = new AudioFormat(this.sampleRate, 16, 1, true, false);\n DataLine.Info info = new DataLine.Info(SourceDataLine.class, audioFormat);\n line = (SourceDataLine) AudioSystem.getLine(info);\n line.open(audioFormat);\n line.start();\n decoderThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n String b64Audio = b64AudioBuffer.poll();\n if (b64Audio != null) {\n byte[] rawAudio = Base64.getDecoder().decode(b64Audio);\n RawAudioBuffer.add(rawAudio);\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n playerThread = new Thread(new Runnable() {\n @Override\n public void run() {\n while (!stopped.get()) {\n byte[] rawAudio = RawAudioBuffer.poll();\n if (rawAudio != null) {\n try {\n playChunk(rawAudio);\n } catch (IOException e) {\n throw new RuntimeException(e);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n } else {\n try {\n Thread.sleep(100);\n } catch (InterruptedException e) {\n throw new RuntimeException(e);\n }\n }\n }\n }\n });\n decoderThread.start();\n playerThread.start();\n }\n \n // 播放一个音频块并阻塞直到播放完成\n private void playChunk(byte[] chunk) throws IOException, InterruptedException {\n if (chunk == null || chunk.length == 0) return;\n \n int bytesWritten = 0;\n while (bytesWritten < chunk.length) {\n bytesWritten += line.write(chunk, bytesWritten, chunk.length - bytesWritten);\n }\n int audioLength = chunk.length / (this.sampleRate*2/1000);\n // 等待缓冲区中的音频播放完成\n Thread.sleep(audioLength - 10);\n }\n \n public void write(String b64Audio) {\n b64AudioBuffer.add(b64Audio);\n }\n \n public void cancel() {\n b64AudioBuffer.clear();\n RawAudioBuffer.clear();\n }\n \n public void waitForComplete() throws InterruptedException {\n while (!b64AudioBuffer.isEmpty() || !RawAudioBuffer.isEmpty()) {\n Thread.sleep(100);\n }\n line.drain();\n }\n \n public void shutdown() throws InterruptedException {\n stopped.set(true);\n decoderThread.join();\n playerThread.join();\n if (line != null && line.isRunning()) {\n line.drain();\n line.close();\n }\n }\n } // RealtimePcmPlayer 类定义结束\n \n public static void main(String[] args) throws InterruptedException, LineUnavailableException {\n String imageB64 = null;\n\n OmniRealtimeParam param = OmniRealtimeParam.builder()\n .model(\"qwen3.5-omni-plus-realtime\")\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 如果没有配置环境变量，请用您的 API Key 将下行修改为.apikey(\"sk-xxx\")\n .apikey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .url(\"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime\")\n .build();\n\n RealtimePcmPlayer audioPlayer = new RealtimePcmPlayer(24000);\n OmniRealtimeConversation conversation = null;\n final AtomicReference conversationRef = new AtomicReference<>(null);\n conversation = new OmniRealtimeConversation(param, new OmniRealtimeCallback() {\n @Override\n public void onOpen() {\n System.out.println(\"connection opened\");\n }\n @Override\n public void onEvent(JsonObject message) {\n String type = message.get(\"type\").getAsString();\n switch(type) {\n case \"session.created\":\n System.out.println(\"start session: \" + message.get(\"session\").getAsJsonObject().get(\"id\").getAsString());\n break;\n case \"conversation.item.input_audio_transcription.completed\":\n System.out.println(\"question: \" + message.get(\"transcript\").getAsString());\n break;\n case \"response.audio_transcript.delta\":\n System.out.println(\"got llm response delta: \" + message.get(\"delta\").getAsString());\n break;\n case \"response.audio.delta\":\n String recvAudioB64 = message.get(\"delta\").getAsString();\n audioPlayer.write(recvAudioB64);\n break;\n case \"input_audio_buffer.speech_started\":\n System.out.println(\"======VAD Speech Start======\");\n audioPlayer.cancel();\n break;\n case \"response.done\":\n System.out.println(\"======RESPONSE DONE======\");\n if (conversationRef.get() != null) {\n System.out.println(\"[Metric] response: \" + conversationRef.get().getResponseId() +\n \", first text delay: \" + conversationRef.get().getFirstTextDelay() +\n \" ms, first audio delay: \" + conversationRef.get().getFirstAudioDelay() + \" ms\");\n }\n break;\n default:\n break;\n }\n }\n @Override\n public void onClose(int code, String reason) {\n System.out.println(\"connection closed code: \" + code + \", reason: \" + reason);\n }\n });\n conversationRef.set(conversation);\n try {\n conversation.connect();\n } catch (NoApiKeyException e) {\n throw new RuntimeException(e);\n }\n OmniRealtimeConfig config = OmniRealtimeConfig.builder()\n .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT))\n .voice(\"Ethan\")\n .enableTurnDetection(true)\n .enableInputAudioTranscription(true)\n .InputAudioTranscription(\"gummy-realtime-v1\")\n .build();\n conversation.updateSession(config);\n long last_photo_time = System.currentTimeMillis();\n try {\n // 创建音频格式\n AudioFormat audioFormat = new AudioFormat(16000, 16, 1, true, false);\n // 根据格式匹配默认录音设备\n TargetDataLine targetDataLine =\n AudioSystem.getTargetDataLine(audioFormat);\n targetDataLine.open(audioFormat);\n // 开始录音\n targetDataLine.start();\n ByteBuffer buffer = ByteBuffer.allocate(1024);\n long start = System.currentTimeMillis();\n // 录音50s并进行实时转写\n while (System.currentTimeMillis() - start < 50000) {\n int read = targetDataLine.read(buffer.array(), 0, buffer.capacity());\n if (read > 0) {\n buffer.limit(read);\n String audioB64 = Base64.getEncoder().encodeToString(buffer.array());\n // 将录音音频数据发送给流式识别服务\n conversation.appendAudio(audioB64);\n buffer = ByteBuffer.allocate(1024);\n // 录音速率有限，防止cpu占用过高，休眠一小会儿\n Thread.sleep(20);\n }\n }\n } catch (Exception e) {\n e.printStackTrace();\n }\n conversation.commit();\n conversation.createResponse(null, null);\n conversation.close(1000, \"bye\");\n audioPlayer.waitForComplete();\n audioPlayer.shutdown();\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2880812.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json index 6adf2613..72ace6d3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json @@ -15,18 +15,96 @@ "Audio" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Completions API", + "name": "search_strategy:agent", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "search_strategy:agent", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", "collectionTag": "Qwen3.5", "features": [ - "web-search" + "web-search", + "function-calling", + "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-omni-plus", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "53", + "type": "omni_audio_input_token", + "priceName": "输入：音频" + }, + { + "priceUnit": "每百万tokens", + "price": "213", + "type": "omni_audio_output_token", + "priceName": "输出：文本+音频（输出的文本不计费）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "omni_no_audio_input_token", + "priceName": "输入：文本/图片/视频" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "omni_no_audio_output_token", + "priceName": "输出：文本" + }, + { + "priceUnit": "每百万tokens", + "price": "26.5", + "type": "omni_audio_input_token_batch", + "priceName": "输入：音频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.5", + "type": "omni_no_audio_input_token_batch", + "priceName": "输入：文本/图片/视频（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "omni_no_audio_output_token_batch", + "priceName": "输出：文本（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "53", + "discount": 0.5, + "type": "omni_audio_input_token_batch_chat", + "priceName": "输入：音频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "7", + "discount": 0.5, + "type": "omni_no_audio_input_token_batch_chat", + "priceName": "输入：文本/图片/视频（Batch Chat）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "discount": 0.5, + "type": "omni_no_audio_output_token_batch_chat", + "priceName": "输出：文本（Batch Chat）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -55,45 +133,16 @@ "latestOnlineAt": "2026-03-30T03:58:25.000+00:00", "contextWindow": 262144, "maxInputTokens": 196608, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Omni-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867839.html", "category": "Multimodal", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "音色选择", - "key": "voice" - }, - { - "name": "内容输出", - "key": "modalities", - "default": [ - "text" - ], - "tip": "设置模型返回的模态" - }, - { - "name": "top_p", - "key": "top_p", - "default": 1, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-plus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-plus\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/2867839.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json index fc319d5d..2dae9b2d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5原生视觉语言系列Plus模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在多项任务评测中，3.5系列均展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。", "collectionTag": "qwen3.5", "features": [ @@ -26,9 +92,6 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-plus", "iconUrl": "", "qpmInfo": { @@ -49,6 +112,179 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.08", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,88 +297,29 @@ "latestOnlineAt": "2026-02-15T09:15:31.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json index b83e2fe2..ce14d6e2 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json @@ -13,40 +13,143 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", "collectionTag": "qwen3.5", "features": [ "model-experience", "function-calling", "structured-outputs", "web-search", - "prefix-completion", - "fine-tuning" + "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-27b", + "model": "qwen3.5-397b-a17b", + "iconUrl": "", "qpmInfo": { "model-default-actual": { - "count_limit_period": 6, - "usage_limit": 1000000, + "count_limit_period": 1, + "usage_limit": 500000, "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, + "count_limit": 10, + "usage_limit_period": 30, "type": "model-default" }, "model-default": { - "count_limit_period": 6, - "usage_limit": 1000000, + "count_limit_period": 1, + "usage_limit": 500000, "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, + "count_limit": 10, + "usage_limit_period": 30, "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -55,97 +158,32 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", + "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "trainingTypes": { - "sft": [ - "lora", - "full" - ] - }, - "inferenceProvider": "bailian", - "name": "Qwen3.5-27B", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-397B-A17B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -162,8 +200,74 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列397B-A17B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。在语言理解、逻辑推理、代码生成、智能体任务、图像理解、视频理解、图形用户界面（GUI）等多种任务中，均展现出与当前顶尖前沿模型相媲美的卓越性能。具备强大的代码生成与智能体能力，对于各类智能体场景具有良好的泛化性。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", "collectionTag": "qwen3.5", "features": [ "model-experience", @@ -173,29 +277,65 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-397b-a17b", - "iconUrl": "", + "model": "qwen3.5-35b-a3b", "qpmInfo": { "model-default-actual": { - "count_limit_period": 1, - "usage_limit": 500000, + "count_limit_period": 6, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, + "count_limit": 60, + "usage_limit_period": 60, "type": "model-default" }, "model-default": { - "count_limit_period": 1, - "usage_limit": 500000, + "count_limit_period": 6, + "usage_limit": 1000000, "usage_limit_field": "total_tokens", - "count_limit": 10, - "usage_limit_period": 30, + "count_limit": 60, + "usage_limit_period": 60, "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3.2", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12.8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -204,91 +344,32 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-15T09:18:22.000+00:00", + "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-397B-A17B", + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-35B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-397b-a17b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-397b-a17b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-397b-a17b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-397b-a17b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-397b-a17b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-397b-a17b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-397b-a17b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -305,21 +386,85 @@ "Video" ] }, - "builtInToolMultiPrices": [], - "description": "Qwen3.5系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现接近于Qwen3.5-27B。", + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], + "description": "Qwen3.5系列27B原生视觉语言Dense模型，融合了线性注意力机制；响应速度快，兼具推理速度和性能。该模型的综合能力接近于Qwen3.5-122B-A10B。", "collectionTag": "qwen3.5", "features": [ "model-experience", "function-calling", "structured-outputs", "web-search", - "prefix-completion" + "prefix-completion", + "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, - "model": "qwen3.5-35b-a3b", + "model": "qwen3.5-27b", "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -338,6 +483,46 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -346,91 +531,39 @@ "modelAlias": "", "versionTag": "SNAPSHOT", "maxOutputTokens": 65536, - "latestOnlineAt": "2026-02-23T03:27:40.000+00:00", + "latestOnlineAt": "2026-02-23T03:42:27.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", - "name": "Qwen3.5-35B-A3B", + "trainingTypes": { + "sft": [ + "lora", + "full" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Qwen3.5-27B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], + "category": "Cost-optimized", "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-27b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-27b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } @@ -447,7 +580,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.5系列122B-A10B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。该模型的综合表现仅次于Qwen3.5-397B-A17B，文本能力显著优于Qwen3-235B-2507，视觉能力优于Qwen3-VL-235B。", "collectionTag": "qwen3.5", "features": [ @@ -458,9 +657,6 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.5-122b-a10b", "qpmInfo": { "model-default-actual": { @@ -480,6 +676,46 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6.4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "VU", @@ -491,88 +727,29 @@ "latestOnlineAt": "2026-02-23T03:28:11.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.5-122B-A10B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.5-122b-a10b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-122b-a10b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.5-122b-a10b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.5-122b-a10b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.5-122b-a10b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.5-122b-a10b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.5-122b-a10b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json index 33ca24e5..b42c7d6c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Flash模型，模型效果相较3.5-Flash显著提升。本模型重点提升agentic coding能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；视觉方面在空间智能能力上显著增强，物体定位与目标检测提升尤为突出。", "collectionTag": "qwen3.6", "features": [ @@ -26,9 +92,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-flash", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,118 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "3.6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.12", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "7.2", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "14.4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.48", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "4.8", + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "28.8", + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -59,88 +234,29 @@ "latestOnlineAt": "2026-04-16T14:02:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Flash", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-flash\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-flash\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-flash',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-flash\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json index 79bc5b4f..cfbc14e0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json @@ -22,9 +22,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-max-preview", "qpmInfo": { "model-default-actual": { @@ -44,6 +41,70 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "9", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "54", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "11.25", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.9", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "15", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "90", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "18.75", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "Reasoning", "TG" @@ -56,87 +117,28 @@ "maxInputTokens": 245760, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1950", - "offlineTime": "2026-09-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118344", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识，当前暂时免费提供服务，即将于2025年8月25日0时起转计费模式，按量后付费" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-max-preview\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-max-preview',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-max-preview\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-max-preview\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json index 102483c2..5b652f21 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果相较3.5系列显著提升。模型在Agentic coding、前端编程、Vibe coding等代码能力、多模态万物识别、OCR、物体定位等能力上显著增强。", "collectionTag": "qwen3.6", "features": [ @@ -26,9 +92,6 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-plus", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,122 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "48", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "Reasoning", "VU", @@ -60,89 +239,30 @@ "latestOnlineAt": "2026-04-01T11:55:34.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json index 20e285d4..5f91ea86 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.6系列35B-A3B原生视觉语言模型，基于混合架构设计，融合了线性注意力机制与稀疏混合专家模型，实现了更高的推理效率。模型效果相较3.5-35B-A3B显著提升了agentic coding能力、数学推理和代码推理能力、空间智能能力、物体定位与目标检测能力。", "collectionTag": "qwen3.6", "features": [ @@ -24,10 +90,21 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-35b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10.8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -57,89 +134,30 @@ "latestOnlineAt": "2026-04-16T14:01:43.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-35B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-35b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-35b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.6-35b-a3b\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.6-35b-a3b\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.6-35b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/3016809.html" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-35b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-35b-a3b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-35b-a3b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } } @@ -165,10 +183,21 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.6-27b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -198,82 +227,23 @@ "latestOnlineAt": "2026-04-22T12:50:52.000+00:00", "contextWindow": 262144, "maxInputTokens": 260096, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.6-27B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - }, - { - "name": "result_format", - "key": "result_format", - "default": "message", - "tip": "返回结果格式" - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.6-27b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.6-27b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.6-27b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.6-27b',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.6-27b\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json index e56dd891..49527a64 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json @@ -11,7 +11,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型，当前开放纯文本模型能力供体验。Qwen3.7是面向智能体时代的新一代旗舰模型，核心优势在于智能体能力的广度与深度：在编程、办公与生产力、长周期自主执行方面均能出色胜任各项任务。", "collectionTag": "qwen3.7", "features": [ @@ -23,10 +63,70 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-max", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "discount": 0.5, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2.4", + "discount": 0.5, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "18", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "15", + "discount": 0.5, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.5, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -56,77 +156,24 @@ "latestOnlineAt": "2026-05-21T06:37:11.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -141,7 +188,47 @@ "Text" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中规模最大、综合能力最强的Max模型预览版，仅支持思考模式，开放纯文本模型能力供体验。主要优化面向用户的通用对话场景，例如知识问答、指令跟随、创意写作等。", "collectionTag": "qwen3.7", "features": [ @@ -151,10 +238,21 @@ "web-search" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-max-preview", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "36", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -183,76 +281,29 @@ "latestOnlineAt": "2026-05-19T16:00:00.000+00:00", "contextWindow": 1000000, "maxInputTokens": 991808, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Max-Preview", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-max-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-max-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" }, "responsesAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\": \"9.9和9.11哪个大？\"\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-max-preview\",\n input=\"9.9和9.11哪个大？\"\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-max-preview\",\n input: \"9.9和9.11哪个大？\"\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-max-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3.7-max-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-max-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json index c1fce0bc..e7f11c96 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json @@ -13,7 +13,73 @@ "Video" ] }, - "builtInToolMultiPrices": [], + "builtInToolMultiPrices": [ + { + "supportedApi": "Responses API", + "name": "code_interpreter", + "docUrl": "https://help.aliyun.com/document_detail/2990719.html", + "type": "code_interpreter", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "i2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022096.html", + "type": "i2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "48", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "t2i_search", + "docUrl": "https://help.aliyun.com/document_detail/3022091.html", + "type": "t2i_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "24", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_extractor", + "docUrl": "https://help.aliyun.com/document_detail/3016816.html", + "type": "web_extractor", + "prices": [ + { + "priceUnit": "", + "price": "0", + "currency": "CNY" + } + ] + }, + { + "supportedApi": "Responses API", + "name": "web_search", + "docUrl": "https://help.aliyun.com/document_detail/2867560.html", + "type": "web_search", + "prices": [ + { + "priceUnit": "千次调用", + "price": "4", + "currency": "CNY" + } + ] + } + ], "description": "Qwen3.7系列中高性价比Plus模型，在强大文本能力的基础上全面升级了视觉-语言能力，同时保持了在编码、工具使用和生产力工作流方面的完整智能体能力。其核心特色为多模态交互混合智能体能力，能够感知真实世界场景、读取屏幕并操作 GUI、基于视觉参考生成代码、端到端导航移动应用。", "collectionTag": "qwen3.7", "features": [ @@ -26,9 +92,6 @@ "structured-outputs" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3.7-plus", "qpmInfo": { "model-default-actual": { @@ -48,6 +111,144 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2.5", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.2", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 262144 + }, + { + "rangeStart": 262144, + "rangeName": "256k<输入<=1m", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.8, + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.8, + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.2", + "discount": 0.8, + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "discount": 0.8, + "type": "input_token_cache_creation_5m", + "priceName": "显式缓存创建" + }, + { + "priceUnit": "每百万tokens", + "price": "0.6", + "discount": 0.8, + "type": "input_token_cache_read", + "priceName": "显式缓存命中" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "discount": 0.5, + "type": "input_token_batch_chat", + "priceName": "输入（Batch Chat)" + }, + { + "priceUnit": "每百万tokens", + "price": "24", + "discount": 0.5, + "type": "output_token_batch_chat", + "priceName": "输出（Batch Chat)" + } + ], + "rangeEnd": 1000000 + } + ], "capabilities": [ "TG", "Reasoning", @@ -61,83 +262,30 @@ "contextWindow": 1000000, "maxInputTokens": 991808, "offlineInfo": {}, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3.7-Plus", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", "category": "Flagship", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "node": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3.7-plus',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices || chunk.choices.length === 0) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3.7-plus\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" }, "responsesAPI": { - "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", + "node": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.responses.create({\n model: \"qwen3.7-plus\",\n input: \"9.9和9.11哪个大？\",\n enable_thinking: true // 启用思考模式\n });\n\n // 遍历输出项\n for (const item of response.output) {\n if (item.type === \"reasoning\") {\n console.log(\"【推理过程】\");\n for (const summary of item.summary) {\n console.log(summary.text.substring(0, 500));\n }\n console.log();\n } else if (item.type === \"message\") {\n console.log(\"【最终答案】\");\n console.log(item.content[0].text);\n }\n }\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1/responses \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\": \"9.9和9.11哪个大？\",\n \"enable_thinking\": true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v2/apps/protocols/compatible-mode/v1\",\n)\n\nresponse = client.responses.create(\n model=\"qwen3.7-plus\",\n input=\"9.9和9.11哪个大？\",\n extra_body={\n \"enable_thinking\": True # 启用思考模式\n }\n)\n\n# 遍历输出项\nfor item in response.output:\n if item.type == \"reasoning\":\n # 打印推理过程摘要\n print(\"【推理过程】\")\n for summary in item.summary:\n print(summary.text[:500]) # 截取前500字符\n print()\n elif item.type == \"message\":\n # 打印最终答案\n print(\"【最终答案】\")\n print(item.content[0].text)", "docUrl": "https://help.aliyun.com/document_detail/3016808.html" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", - "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3.7-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象?\"}]\n }]\nresponse = dashscope.MultiModalConversation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model='qwen3.7-plus',\n messages=messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])\n", + "java": "import java.util.Arrays;\nimport java.util.Collections;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n \n static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\n public static void simpleMultiModalConversationCall()\n throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(\n Collections.singletonMap(\"image\", \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241022/emyrja/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象?\"))).build();\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3.7-plus\")\n .messages(Arrays.asList(userMessage))\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\"));\n }\n public static void main(String[] args) {\n try {\n simpleMultiModalConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json index b3a42bb4..9b09f528 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json @@ -21,10 +21,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-32b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -52,73 +63,26 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-32B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-32b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-32b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-32b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-32b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -142,10 +106,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-32b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -174,58 +149,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-32B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-32b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-32b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-32b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-32b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-32b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -250,10 +193,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-30b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -283,73 +237,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-30B-A3B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-30b-a3b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-30b-a3b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-30b-a3b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -374,10 +281,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-30b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -406,58 +324,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-30B-A3B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-30b-a3b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-30b-a3b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-30b-a3b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-30b-a3b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-30b-a3b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -483,10 +369,27 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-8b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -523,72 +426,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-8B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-8b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-8b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-8b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-8b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -614,10 +470,27 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-8b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -653,57 +526,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-8B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-8b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-8b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-8b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-8b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-8b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -728,10 +569,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-235b-a22b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -761,73 +613,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-235B-A22B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", - "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"请解答这道题\"\n }\n ]\n }\n ],\n \"stream\":true,\n \"stream_options\":{\"include_usage\":true},\n \"enable_thinking\": true,\n \"thinking_budget\": 81920\n}'", + "python": "from openai import OpenAI\nimport os\n\n# 初始化OpenAI客户端\nclient = OpenAI(\n api_key = os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n)\n\nreasoning_content = \"\" # 定义完整思考过程\nanswer_content = \"\" # 定义完整回复\nis_answering = False # 判断是否结束思考过程并开始回复\nenable_thinking = False\n# 创建聊天完成请求\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"这道题怎么解答？\"},\n ],\n },\n ],\n stream=True,\n # enable_thinking 参数开启思考过程，thinking_budget 参数设置最大推理过程 Token 数\n extra_body={\n 'enable_thinking': True,\n \"thinking_budget\": 81920},\n\n # 解除以下注释会在最后一个chunk返回Token使用量\n # stream_options={\n # \"include_usage\": True\n # }\n)\n\nif enable_thinking:\n print(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20 + \"\\n\")\n\nfor chunk in completion:\n # 如果chunk.choices为空，则打印usage\n if not chunk.choices:\n print(\"\\nUsage:\")\n print(chunk.usage)\n else:\n delta = chunk.choices[0].delta\n # 打印思考过程\n if hasattr(delta, 'reasoning_content') and delta.reasoning_content != None:\n print(delta.reasoning_content, end='', flush=True)\n reasoning_content += delta.reasoning_content\n else:\n # 开始回复\n if delta.content != \"\" and is_answering is False:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n is_answering = True\n # 打印回复过程\n print(delta.content, end='', flush=True)\n answer_content += delta.content\n\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(reasoning_content)\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(answer_content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = '';\nlet answerContent = '';\nlet isAnswering = false;\nlet enable_thinking = true;\n\nlet messages = [\n {\n role: \"user\",\n content: [\n { type: \"image_url\", image_url: { \"url\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\" } },\n { type: \"text\", text: \"解答这道题\" },\n ]\n}]\n\nasync function main() {\n try {\n const stream = await openai.chat.completions.create({\n model: 'qwen3-vl-235b-a22b-thinking',\n messages: messages,\n stream: true,\n // 注意：在 Node.js SDK，enable_thinking 这样的非标准参数作为顶层属性传递的，无需放在 extra_body 中\n enable_thinking: enable_thinking,\n thinking_budget: 81920\n\n });\n\n if (enable_thinking){console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');}\n\n for await (const chunk of stream) {\n if (!chunk.choices?.length) {\n console.log('\\nUsage:');\n console.log(chunk.usage);\n continue;\n }\n\n const delta = chunk.choices[0].delta;\n\n // 处理思考过程\n if (delta.reasoning_content) {\n process.stdout.write(delta.reasoning_content);\n reasoningContent += delta.reasoning_content;\n }\n // 处理正式回复\n else if (delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", - "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", - "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-H 'X-DashScope-SSE: enable' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"请解答这道题\"}\n ]\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"incremental_output\": true,\n \"thinking_budget\": 50\n }\n}'", + "python": "import os\nimport dashscope\nfrom dashscope import MultiModalConversation\n# dashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nmessages = [\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"},\n {\"text\": \"解答这道题？\"}\n ]\n }\n]\n\nresponse = MultiModalConversation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"qwen3-vl-235b-a22b-thinking\",\n messages=messages,\n stream=True,\n # enable_thinking 参数开启思考过程\n enable_thinking=True,\n # thinking_budget 参数设置最大推理过程 Token 数\n thinking_budget=81920,\n\n)\n\n# 定义完整思考过程\nreasoning_content = \"\"\n# 定义完整回复\nanswer_content = \"\"\n# 判断是否结束思考过程并开始回复\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in response:\n # 如果思考过程与回复皆为空，则忽略\n message = chunk.output.choices[0].message\n reasoning_content_chunk = message.get(\"reasoning_content\", None)\n if (chunk.output.choices[0].message.content == [] and\n reasoning_content_chunk == \"\"):\n pass\n else:\n # 如果当前为思考过程\n if reasoning_content_chunk != None and chunk.output.choices[0].message.content == []:\n print(chunk.output.choices[0].message.reasoning_content, end=\"\")\n reasoning_content += chunk.output.choices[0].message.reasoning_content\n # 如果当前为回复\n elif chunk.output.choices[0].message.content != []:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content[0][\"text\"], end=\"\")\n answer_content += chunk.output.choices[0].message.content[0][\"text\"]\n\n# 如果您需要打印完整思考过程与完整回复，请将以下代码解除注释后运行\n# print(\"=\" * 20 + \"完整思考过程\" + \"=\" * 20 + \"\\n\")\n# print(f\"{reasoning_content}\")\n# print(\"=\" * 20 + \"完整回复\" + \"=\" * 20 + \"\\n\")\n# print(f\"{answer_content}\")", + "java": "// dashscope SDK的版本 >= 2.21.10\nimport java.util.*;\n\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\n\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\n\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n\n// 若使用新加坡地域的模型，请取消下列注释\n// static {Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";}\n\nprivate static final Logger logger = LoggerFactory.getLogger(Main.class);\nprivate static StringBuilder reasoningContent = new StringBuilder();\nprivate static StringBuilder finalContent = new StringBuilder();\nprivate static boolean isFirstPrint = true;\n\nprivate static void handleGenerationResult(MultiModalConversationResult message) {\nString re = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\nString reasoning = Objects.isNull(re)?\"\":re; // 默认值\n\nList> content = message.getOutput().getChoices().get(0).getMessage().getContent();\nif (!reasoning.isEmpty()) {\nreasoningContent.append(reasoning);\nif (isFirstPrint) {\nSystem.out.println(\"====================思考过程====================\");\nisFirstPrint = false;\n}\nSystem.out.print(reasoning);\n}\n\nif (Objects.nonNull(content) && !content.isEmpty()) {\nObject text = content.get(0).get(\"text\");\nfinalContent.append(content.get(0).get(\"text\"));\nif (!isFirstPrint) {\nSystem.out.println(\"\\n====================完整回复====================\");\nisFirstPrint = true;\n}\nSystem.out.print(text);\n}\n}\npublic static MultiModalConversationParam buildMultiModalConversationParam(MultiModalMessage Msg) {\nreturn MultiModalConversationParam.builder()\n// 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n.apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n.model(\"qwen3-vl-235b-a22b-thinking\")\n.messages(Arrays.asList(Msg))\n.enableThinking(true)\n.thinkingBudget(500)\n.incrementalOutput(true)\n.build();\n}\n\npublic static void streamCallWithMessage(MultiModalConversation conv, MultiModalMessage Msg)\nthrows NoApiKeyException, ApiException, InputRequiredException, UploadFileException {\nMultiModalConversationParam param = buildMultiModalConversationParam(Msg);\nFlowable result = conv.streamCall(param);\nresult.blockingForEach(message -> {\nhandleGenerationResult(message);\n});\n}\npublic static void main(String[] args) {\ntry {\nMultiModalConversation conv = new MultiModalConversation();\nMultiModalMessage userMsg = MultiModalMessage.builder()\n.role(Role.USER.getValue())\n.content(Arrays.asList(Collections.singletonMap(\"image\", \"https://img.alicdn.com/imgextra/i1/O1CN01gDEY8M1W114Hi3XcN_!!6000000002727-0-tps-1024-406.jpg\"),\nCollections.singletonMap(\"text\", \"请解答这道题\")))\n.build();\nstreamCallWithMessage(conv, userMsg);\n// 打印最终结果\n// if (reasoningContent.length() > 0) {\n// System.out.println(\"\\n====================完整回复====================\");\n// System.out.println(finalContent.toString());\n// }\n} catch (ApiException | NoApiKeyException | UploadFileException | InputRequiredException e) {\nlogger.error(\"An exception occurred: {}\", e.getMessage());\n}\nSystem.exit(0);\n}\n}" } } } @@ -852,10 +657,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-vl-235b-a22b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -884,58 +700,26 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-VL-235B-A22B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2845871.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"type\": \"image_url\", \"image_url\": {\"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"}},\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"}\n ]\n }]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3-vl-235b-a22b-instruct\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n },\n },\n {\"type\": \"text\", \"text\": \"请仅输出图像中的文本内容。\"},\n ],\n },\n ],\n)\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI({\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n});\n\nasync function main() {\n const response = await openai.chat.completions.create({\n model: \"qwen3-vl-235b-a22b-instruct\",\n messages: [\n {\n role: \"user\",\n content: [{\n type: \"image_url\",\n image_url: {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"\n }\n },\n {\n type: \"text\",\n text: \"请仅输出图像中的文本内容。\"\n }\n ]\n }\n ]\n });\n console.log(response.choices[0].message.content);\n}\nmain()" } }, "dashscope": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", - "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", - "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n \"model\": \"qwen3-vl-235b-a22b-instruct\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"},\n {\"text\": \"图中描绘的是什么景象？\"}\n ]\n },\n {\n \"role\": \"assistant\",\n \"content\": [\n {\"text\": \"图中是一名女子和一只拉布拉多犬在沙滩上玩耍。\"}\n ]\n },\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"写一首七言绝句描述这个场景\"}\n ]\n }\n ]\n }\n}'", + "python": "import os\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nmessages = [\n{\n \"role\": \"user\",\n \"content\": [\n {\"image\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\"},\n {\"text\": \"请仅输出图像中的文本内容。\"}]\n}]\nresponse = dashscope.MultiModalConversation.call(\n #若没有配置环境变量， 请用百炼API Key将下行替换为： api_key =\"sk-xxx\"\n api_key = os.getenv('DASHSCOPE_API_KEY'),\n model = 'qwen3-vl-235b-a22b-instruct',\n messages = messages\n)\nprint(response.output.choices[0].message.content[0][\"text\"])", + "java": "import java.util.ArrayList;\nimport java.util.Arrays;\nimport java.util.Collections;\nimport java.util.List;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam;\nimport com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult;\nimport com.alibaba.dashscope.common.MultiModalMessage;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final String modelName = \"qwen3-vl-235b-a22b-instruct\"; \n public static void MultiRoundConversationCall() throws ApiException, NoApiKeyException, UploadFileException {\n MultiModalConversation conv = new MultiModalConversation();\n MultiModalMessage systemMessage = MultiModalMessage.builder().role(Role.SYSTEM.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"You are a helpful assistant.\"))).build();\n MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"image\", \"https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg\"),\n Collections.singletonMap(\"text\", \"图中描绘的是什么景象？\"))).build();\n List messages = new ArrayList<>();\n messages.add(systemMessage);\n messages.add(userMessage);\n MultiModalConversationParam param = MultiModalConversationParam.builder()\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\")) \n .model(modelName)\n .messages(messages)\n .build();\n MultiModalConversationResult result = conv.call(param);\n System.out.println(\"第一轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); // add the result to conversation\n messages.add(result.getOutput().getChoices().get(0).getMessage());\n MultiModalMessage msg = MultiModalMessage.builder().role(Role.USER.getValue())\n .content(Arrays.asList(Collections.singletonMap(\"text\", \"做一首诗描述这个场景\"))).build();\n messages.add(msg);\n param.setMessages((List)messages);\n result = conv.call(param);\n System.out.println(\"第二轮输出： \"+result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get(\"text\")); }\n\n public static void main(String[] args) {\n try {\n MultiRoundConversationCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -953,10 +737,21 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-next-80b-a3b-instruct", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -986,58 +781,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Next-80B-A3B-Instruct", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-next-80b-a3b-instruct\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-next-80b-a3b-instruct\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-instruct\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-instruct\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1056,10 +819,21 @@ "collectionTag": "qwen3", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-next-80b-a3b-thinking", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -1090,72 +864,25 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-Next-80B-A3B-Thinking", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-next-80b-a3b-thinking',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-next-80b-a3b-thinking\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-next-80b-a3b-thinking\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-next-80b-a3b-thinking\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1179,10 +906,21 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1221,58 +959,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B-Instruct-2507", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-30b-a3b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-30b-a3b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1293,10 +999,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1326,73 +1043,26 @@ "maxInputTokens": 126976, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B-Thinking-2507", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1412,10 +1082,21 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b-thinking-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1443,73 +1124,26 @@ "contextWindow": 131072, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B-Thinking-2507", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_search", - "key": "enable_search", - "tip": "联网搜索补充互联网知识" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b-thinking-2507',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-thinking-2507\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-thinking-2507\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-thinking-2507\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1532,10 +1166,21 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b-instruct-2507", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -1564,59 +1209,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B-Instruct-2507", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-235b-a22b-instruct-2507\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-235b-a22b-instruct-2507\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b-instruct-2507\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b-instruct-2507\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1640,10 +1253,33 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-235b-a22b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1673,75 +1309,27 @@ "maxInputTokens": 129024, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-235B-A22B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-235b-a22b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-235b-a22b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-235b-a22b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-235b-a22b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-235b-a22b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1765,10 +1353,33 @@ "prefix-completion" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-30b-a3b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.75", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "7.5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 30, @@ -1799,74 +1410,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-30B-A3B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-30b-a3b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-30b-a3b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-30b-a3b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-30b-a3b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-30b-a3b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -1891,10 +1454,39 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-32b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "20", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "40", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1937,75 +1529,27 @@ }, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-32B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-32b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-32b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-32b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-32b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2030,10 +1574,39 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-14b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "30", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -2073,75 +1646,27 @@ }, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-14B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-14b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-14b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-14b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-14b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2166,10 +1691,39 @@ "fine-tuning" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-8b", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.5", + "type": "thinking_input_token", + "priceName": "输入（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "5", + "type": "thinking_output_token", + "priceName": "输出（思考）" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "ft", + "priceName": "调优" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -2210,74 +1764,26 @@ "offlineInfo": { "inference": { "announceUrl": "https://www.aliyun.com/notice/118345", - "offlineTime": "2026-07-08 00:00:00" + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Qwen3-8B", "docUrl": "https://help.aliyun.com/document_detail/2712576.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "stop", - "key": "stop", - "tip": "用于控制生成时遇到某些内容则停止。您可传入多个字符串。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwen3-8b\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'qwen3-8b',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-8b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-8b\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-8b\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -2298,9 +1804,6 @@ "model-experience" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen3-coder-next", "qpmInfo": { "model-default-actual": { @@ -2320,6 +1823,65 @@ "type": "model-default" } }, + "multiPrices": [ + { + "rangeStart": 0, + "rangeName": "输入<=32k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 32768 + }, + { + "rangeStart": 32768, + "rangeName": "32k<输入<=128k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 131072 + }, + { + "rangeStart": 131072, + "rangeName": "128k<输入<=256k", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2.5", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "10", + "type": "output_token", + "priceName": "输出" + } + ], + "rangeEnd": 262144 + } + ], "capabilities": [ "TG" ], @@ -2331,54 +1893,27 @@ "maxInputTokens": 204800, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/zh/notice/detail?id=1949", - "offlineTime": "2026-07-08 00:00:00" + "announceUrl": "https://www.aliyun.com/notice/118345", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义千问3-Coder-Next", "docUrl": "https://help.aliyun.com/document_detail/2850166.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3-coder-next\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"qwen3-coder-next\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3-coder-next\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"qwen3-coder-next\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwen3-coder-next\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwen3-coder-next\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json index 302d2793..eb494d3b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json @@ -20,10 +20,33 @@ "batch" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwq-plus", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.8", + "type": "input_token_batch", + "priceName": "输入（Batch File）" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "output_token_batch", + "priceName": "输出（Batch File）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -52,54 +75,27 @@ "maxInputTokens": 98304, "offlineInfo": { "inference": { - "announceUrl": "https://www.alibabacloud.com/notice/detail?id=1841", - "offlineTime": "2026-07-13 23:59:59" + "announceUrl": "https://www.aliyun.com/notice/118177", + "offlineTime": "2026-10-10 00:00:00" } }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "QwQ-Plus", "docUrl": "https://help.aliyun.com/document_detail/2870973.html", "category": "Older", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"qwq-plus\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'qwq-plus',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwq-plus\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"qwq-plus\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"qwq-plus\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/sambert.json b/skills/bailian-docs-llm-wiki/models/groups/sambert.json index 86b2dd73..b9233e1d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/sambert.json +++ b/skills/bailian-docs-llm-wiki/models/groups/sambert.json @@ -14,45 +14,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiyue-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:01.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知悦", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -74,45 +52,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiyuan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-20T09:14:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知媛", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -134,45 +90,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiying-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:43.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知颖", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -194,45 +128,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiye-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:07.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知晔", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -254,45 +166,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiya-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:17:55.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知雅", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -314,45 +204,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhixiao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知笑", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -374,45 +242,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhixiang-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:13.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知祥", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -434,45 +280,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiwei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:04.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知薇", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -494,45 +318,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiting-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:16.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知婷", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -554,45 +356,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhistella-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:19:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知莎", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -614,45 +394,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhishuo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:18.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知硕", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -674,45 +432,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhishu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:05.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知树", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -734,45 +470,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiru-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:22.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知茹", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -794,45 +508,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiqian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:08.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知倩", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -854,45 +546,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiqi-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:25.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知琪", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -914,45 +584,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhinan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T08:10:47.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知楠", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -974,45 +622,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhina-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:46.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知娜", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1034,45 +660,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知墨", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1094,45 +698,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhiming-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知茗", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1154,45 +736,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimiao-emo-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:38.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知妙（多情感）", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1214,45 +774,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhimao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知猫", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1274,45 +812,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhilun-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知伦", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1334,45 +850,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhijing-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:41.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知婧", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1394,45 +888,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhijia-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知佳", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1454,45 +926,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhihao-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:45.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知浩", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1514,45 +964,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhigui-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知柜", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1574,45 +1002,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhifei-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:49.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知飞", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1634,45 +1040,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhide-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:17.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知德", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1694,45 +1078,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhida-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:52.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知达", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1754,45 +1116,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-zhichu-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:12.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-知厨", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1814,45 +1154,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-waan-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:56.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Waan", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1874,45 +1192,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-perla-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Perla", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1934,45 +1230,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-indah-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:18:59.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Indah", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1994,45 +1268,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-hanna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:09.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Hanna", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2054,45 +1306,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-eva-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Eva", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2114,45 +1344,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-donna-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:23.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Donna", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2174,45 +1382,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-clara-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:21:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Clara", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2234,45 +1420,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-cindy-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:26.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Cindy", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2294,45 +1458,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-camila-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:27.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Camila", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2354,45 +1496,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-cally-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:58.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Cally", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2414,45 +1534,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-brian-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Brian", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2474,45 +1572,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-betty-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:16:48.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Betty", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -2534,45 +1610,23 @@ "description": "提供SAMBERT+NSFGAN深度神经网络算法与传统领域知识深度结合的文字转语音服务，兼具读音准确，韵律自然，声音还原度高，表现力强的特点。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "sambert-beth-v1", + "prices": [ + { + "priceUnit": "每万字符", + "price": "1", + "type": "tts_text_number", + "priceName": "Sambert 语音合成" + } + ], "capabilities": [ "TTS" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-18T02:20:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Sambert语音合成-Beth", "docUrl": "https://help.aliyun.com/document_detail/2712458.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json index e4546f6c..ccc4abba 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/shoemodel-v1.json @@ -14,49 +14,19 @@ "description": "鞋靴模特支持输入多视角鞋靴系列图片，同时对输入模特模板图的鞋子区域进行鞋靴AI试穿，实现模特鞋靴布局重绘生成，最终生成图片的效果, 布局自然、细节丰富、画面细腻、试穿结果逼真。可用于模特商品图设计、新鞋AI试穿、模特穿戴布局重绘等场景。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "shoemodel-v1", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-21T02:47:24.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "鞋靴模特", "docUrl": "https://help.aliyun.com/document_detail/2804662.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"shoemodel-v1\",\n \"input\": {\n \"template_image_url\": \"https://img.alicdn.com/imgextra/i1/O1CN01EyPuz31d79mKv75CI_!!6000000003688-49-tps-1120-1680.webp\",\n \"shoe_image_url\": [\"https://img.alicdn.com/imgextra/i2/O1CN01zTIls120gdcrI7dX2_!!6000000006879-49-tps-1120-1493.webp\"]\n },\n \"parameters\": \n {\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https:/[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json index 848f0974..13c1be04 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json +++ b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json @@ -17,11 +17,22 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3.2", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "3", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -52,53 +63,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3.2", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -119,11 +87,22 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3.1-terminus", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "12", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -154,47 +133,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3.1-Terminus", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -215,11 +157,22 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-v3-0324", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -249,47 +202,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-V3-0324", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.6, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "presence_penalty", - "key": "presence_penalty", - "default": 0.95, - "tip": "用户控制模型生成时整个序列中的重复度。提高该参数时可以降低模型生成的重复度。", - "range": [ - -2, - 2 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-v3-0324\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } @@ -310,11 +226,22 @@ "prefix-completion" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "siliconflow/deepseek-r1-0528", "iconUrl": "https://img.alicdn.com/imgextra/i2/O1CN01zDsrfb1WbCTjLP89d_!!6000000002806-2-tps-56-56.png", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "16", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -346,27 +273,10 @@ "inferenceProvider": "siliconflow", "name": "SiliconFlow DeepSeek-R1-0528", "docUrl": "https://help.aliyun.com/document_detail/3014912.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "max_tokens", - "key": "max_tokens", - "default": 4000, - "tip": "最终回答的最大长度（不含思维链输出）", - "range": [ - 1, - 8192 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"siliconflow/deepseek-r1-0528\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3014912.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json index 3896d924..290d0f82 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json +++ b/skills/bailian-docs-llm-wiki/models/groups/speech-biasing.json @@ -14,9 +14,6 @@ "description": "热词是指用户可以预先定义的一组特定词汇或短语，这些词汇或短语在识别、翻译过程中会被赋予更高的优先级。针对您的特定业务领域，如果有部分词汇的语音识别、翻译效果不够好，可以将这些关键词或短语添加为热词进行优先识别或翻译，从而提升识别、翻译效果。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "speech-biasing", "qpmInfo": { "model-default-actual": { @@ -35,36 +32,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "语音识别热词", - "docUrl": "https://help.aliyun.com/document_detail/2712535.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ] + "docUrl": "https://help.aliyun.com/document_detail/2712535.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json index 58eca4da..0abb94be 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json @@ -21,10 +21,27 @@ "batch" ], "provider": "stepfun", - "limit": { - "message": "model not exist" - }, "model": "stepfun/step-3.7-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1.35", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "8.1", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "0.27", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 6, @@ -44,7 +61,8 @@ } }, "capabilities": [ - "TG" + "TG", + "VU" ], "versionTag": "MAJOR", "maxOutputTokens": 262144, @@ -55,39 +73,6 @@ "inferenceProvider": "stepfun", "name": "stepfun/step-3.7-flash", "docUrl": "https://help.aliyun.com/document_detail/3036697.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 1, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json index 115a1609..c2e7b9af 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json @@ -14,10 +14,21 @@ "description": "意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "tongyi-intent-detect-v3", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,50 +55,23 @@ "latestOnlineAt": "2024-12-12T11:33:25.000+00:00", "contextWindow": 8192, "maxInputTokens": 8192, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "意图分类模型", "docUrl": "https://help.aliyun.com/document_detail/2861138.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"tongyi-intent-detect-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"tongyi-intent-detect-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-intent-detect-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"tongyi-intent-detect-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-intent-detect-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json index 1510e8ce..29c2688e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json @@ -14,10 +14,21 @@ "description": "通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-xiaomi-analysis-flash", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "0.2", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "0.4", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -44,49 +55,22 @@ "latestOnlineAt": "2026-01-09T03:23:03.000+00:00", "contextWindow": 32768, "maxInputTokens": 28672, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义晓蜜-对话分析-flash", "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-flash\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } }, "dashscope": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-flash\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-flash\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json index 706099a7..b8f1c727 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json @@ -14,10 +14,21 @@ "description": "通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。", "features": [], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "tongyi-xiaomi-analysis-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "1", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "2.7", + "type": "output_token", + "priceName": "输出" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -44,49 +55,22 @@ "latestOnlineAt": "2026-01-09T03:31:49.000+00:00", "contextWindow": 32768, "maxInputTokens": 28672, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "通义晓蜜-对话分析-pro", "docUrl": "https://help.aliyun.com/document_detail/3015075.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", - "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", - "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n}'", + "python": "from openai import OpenAI\nimport os\n\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0,\n extra_body={\n \"top_k\": 1\n }\n)\n\nprint(completion.choices[0].message.content)", + "java": "import com.openai.client.OpenAIClient;\nimport com.openai.client.okhttp.OpenAIOkHttpClient;\nimport com.openai.core.JsonValue;\nimport com.openai.models.chat.completions.ChatCompletion;\nimport com.openai.models.chat.completions.ChatCompletionCreateParams;\n\npublic class Main {\n public static void main(String[] args) {\n try {\n OpenAIClient client = OpenAIOkHttpClient.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .baseUrl(\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\")\n .build();\n \n ChatCompletionCreateParams params = ChatCompletionCreateParams.builder()\n .model(\"tongyi-xiaomi-analysis-pro\")\n .addUserMessage(getUserMessage())\n .temperature(0.0)\n .putAdditionalBodyProperty(\"top_k\", JsonValue.from(1))\n .build();\n \n ChatCompletion chatCompletion = client.chat().completions().create(params);\n String content = chatCompletion.choices().get(0).message().content().orElse(\"未返回有效内容\");\n System.out.println(content);\n\n } catch (Exception e) {\n System.err.println(\"Error Message: \" + e.getMessage());\n }\n }\n\n private static String getUserMessage() {\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } }, "dashscope": { "default": { - "curl": "curl --location \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"tongyi-xiaomi-analysis-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\"\n }\n ],\n \"temperature\": 0.0,\n \"top_k\": 1\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "from http import HTTPStatus\nimport dashscope\nimport os\n\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\"\n\nanalysis_prompt = f\"\"\"\n请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\n\"\"\"\n\nresponse = dashscope.Generation.call(\n api_key=os.getenv('DASHSCOPE_API_KEY'),\n model=\"tongyi-xiaomi-analysis-pro\",\n messages=[\n {\n 'role': 'user',\n 'content': analysis_prompt\n }\n ],\n temperature=0.0,\n top_k=1,\n result_format=\"message\"\n)\n\nif response.status_code == HTTPStatus.OK:\n print(response.output.choices[0].message.content)\nelse:\n print('Request id: %s, Status code: %s, error code: %s, error message: %s' % (\n response.request_id, response.status_code,\n response.code, response.message\n ))", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(getContent())\n .build();\n\n GenerationParam param = GenerationParam.builder()\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"tongyi-xiaomi-analysis-pro\")\n .messages(Arrays.asList(userMsg))\n .temperature(0.0f)\n .topK(1)\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"Error message: \"+e.getMessage());\n }\n }\n\n private static String getContent() {\n\n String analysisPrompt = \"请帮我提取下面对话中的关键信息：[1] 客服：您好，欢迎致电AB电商平台，请问有什么可以帮您？[2] 客户：你好，我上周在你们店买了一台料理机，现在运行时有点异响。\";\n\n return analysisPrompt;\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json deleted file mode 100644 index a467552c..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json +++ /dev/null @@ -1,172 +0,0 @@ -{ - "name": "Tripo", - "description": "AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "3D-Generation" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。", - "features": [ - "function-calling", - "structured-outputs", - "batch" - ], - "provider": "tripo", - "limit": { - "message": "model not exist" - }, - "model": "Tripo/Tripo-P1.0", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "3D-generation" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T04:07:42.000+00:00", - "inferenceProvider": "tripo", - "name": "Tripo-P1.0", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "3D-Generation" - ], - "request_modality": [ - "Text" - ] - }, - "description": "Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。", - "features": [ - "function-calling", - "batch", - "structured-outputs" - ], - "provider": "tripo", - "limit": { - "message": "model not exist" - }, - "model": "Tripo/Tripo-H3.1", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - }, - "model-default": { - "count_limit_period": 60, - "async_user_queue_limit": 500, - "count_limit": 5, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 10 - } - }, - "capabilities": [ - "3D-generation" - ], - "modelAlias": "", - "versionTag": "MAJOR", - "latestOnlineAt": "2026-04-27T04:08:07.000+00:00", - "inferenceProvider": "tripo", - "name": "Tripo-H3.1", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], - "samples": { - "dashscope": { - "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", - "docUrl": "https://help.aliyun.com/document_detail/3030679.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json index a7fcb0ab..9db909f9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json @@ -69,33 +69,6 @@ "inferenceProvider": "vanchin", "name": "vanchin/deepseek-v4-pro", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -175,49 +148,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3.2-think", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -295,49 +225,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3.1-Terminus", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.95, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.6, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": false, - "tip": "推理模式" - }, - { - "name": "thinking_budget", - "key": "thinking_budget", - "default": 4000, - "tip": "思维链的最大输出tokens数量", - "range": [ - 1, - 32768 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -416,33 +303,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-V3", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -521,33 +381,6 @@ "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-R1", "docUrl": "https://help.aliyun.com/document_detail/3027089.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { @@ -618,33 +451,6 @@ "maxInputTokens": 8192, "inferenceProvider": "vanchin", "name": "Vanchin/DeepSeek-OCR", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json index d9aa0e65..def2ebc4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json +++ b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json @@ -14,49 +14,33 @@ "description": "视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "video-style-transform", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_540p", + "priceName": "视频生成（540P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:06.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "视频风格重绘", "docUrl": "https://help.aliyun.com/document_detail/2846319.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"video-style-transform\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250704/viwndw/%E5%8E%9F%E8%A7%86%E9%A2%91.mp4\"\n },\n \"parameters\": {\n \"style\": 0,\n \"video_fps\": 15,\n \"min_len\": 540\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json index b12ee860..df5f005a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json +++ b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json @@ -15,49 +15,27 @@ "description": "VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "videoretalk", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.08", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-12-10T06:23:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "声动人像VideoRetalk", "docUrl": "https://help.aliyun.com/document_detail/2860466.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"videoretalk\",\n \"input\": {\n \"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/pvegot/input_video_01.mp4\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250717/aumwir/stella2-%E6%9C%89%E5%A3%B0%E4%B9%A67.wav\",\n \"ref_image_url\": \"\"\n },\n \"parameters\": {\n \"video_extension\": false\n }\n }'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/ \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" " } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json index deec3f50..f89dda08 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json @@ -62,33 +62,6 @@ "name": "Vidu-image_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -158,33 +131,6 @@ "name": "ViduQ3-fast_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -254,33 +200,6 @@ "name": "ViduQ2-Pro_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -338,33 +257,6 @@ "name": "ViduQ2-fast_reference2image", "docUrl": "https://help.aliyun.com/document_detail/3045893.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json index c01401c5..fddb6d41 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json @@ -56,33 +56,6 @@ "name": "ViduQ3-Ad_reference2video", "docUrl": "https://help.aliyun.com/document_detail/3045891.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -140,33 +113,6 @@ "name": "ViduQ3-Drama_reference2video", "docUrl": "https://help.aliyun.com/document_detail/3045892.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -230,33 +176,6 @@ "name": "ViduQ3-Pro-fast_img2video", "docUrl": "https://help.aliyun.com/document_detail/3026198.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -327,37 +246,6 @@ "latestOnlineAt": "2026-03-26T13:47:00.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -427,37 +315,6 @@ "latestOnlineAt": "2026-03-26T13:45:57.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -530,33 +387,6 @@ "latestOnlineAt": "2026-03-26T13:46:33.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -627,37 +457,6 @@ "latestOnlineAt": "2026-03-26T13:46:46.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -727,37 +526,6 @@ "latestOnlineAt": "2026-03-26T13:46:51.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 16 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -830,33 +598,6 @@ "latestOnlineAt": "2026-03-26T13:46:28.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -927,32 +668,6 @@ "latestOnlineAt": "2026-03-26T13:46:18.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1023,33 +738,6 @@ "latestOnlineAt": "2026-03-26T13:46:02.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1121,32 +809,6 @@ "latestOnlineAt": "2026-03-26T13:46:23.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1217,32 +879,6 @@ "latestOnlineAt": "2026-03-26T13:46:08.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Turbo_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1312,32 +948,6 @@ "latestOnlineAt": "2026-03-26T13:45:21.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2_text2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1410,33 +1020,6 @@ "latestOnlineAt": "2026-03-26T13:46:14.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Turbo_start-end2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1507,32 +1090,6 @@ "latestOnlineAt": "2026-03-26T13:45:51.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2_reference2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1602,33 +1159,6 @@ "latestOnlineAt": "2026-04-27T08:35:14.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1698,33 +1228,6 @@ "latestOnlineAt": "2026-04-27T08:33:07.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-Turbo_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1788,33 +1291,6 @@ "latestOnlineAt": "2026-04-28T02:40:03.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ3-mix_reference2video", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { @@ -1878,32 +1354,6 @@ "latestOnlineAt": "2026-04-27T12:09:25.000+00:00", "inferenceProvider": "vidu", "name": "ViduQ2-Pro-fast_img2video", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5, - "range": [ - 1, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json index b603e699..b0dd8943 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/virtualmodel-v2.json @@ -14,49 +14,19 @@ "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "virtualmodel-v2", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-25T15:18:50.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "虚拟模特V2", "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"virtualmodel-v2\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json index 943af02a..128b08cc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json +++ b/skills/bailian-docs-llm-wiki/models/groups/voice-enrollment.json @@ -14,9 +14,6 @@ "description": "大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "voice-enrollment", "qpmInfo": { "model-default-actual": { @@ -35,10 +32,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2024-09-12T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "大模型声音复刻及声音设计", - "docUrl": "https://help.aliyun.com/document_detail/2861519.html", - "predictConfig": [] + "docUrl": "https://help.aliyun.com/document_detail/2861519.html" } ] } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json index 582f6265..97ce5685 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json @@ -18,10 +18,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,43 +49,16 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-01T03:48:10.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-Image", "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3026980.html" } } @@ -102,10 +80,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-image-pro", + "prices": [ + { + "priceUnit": "每张", + "price": "0.5", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -128,43 +111,21 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-01T05:32:16.000+00:00", - "inferenceProvider": "bailian", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-Image-Pro", "docUrl": "https://help.aliyun.com/document_detail/3026980.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "2048*2048" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "生成数量", - "range": [ - 1, - 4 - ] - }, - { - "name": "组图生成", - "key": "enable_sequential", - "default": false - }, - { - "name": "智能改写", - "key": "thinking_mode", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", - "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", - "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data '{\n \"model\": \"wan2.7-image-pro\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"}\n ]\n }\n ]\n },\n \"parameters\": {\n \"enable_sequential\": true,\n \"n\": 4,\n \"size\": \"2K\"\n }\n}'\n", + "python": "import os\nimport dashscope\nfrom dashscope.aigc.image_generation import ImageGeneration\nfrom dashscope.api_entities.dashscope_response import Message\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nmessage = Message(\n role=\"user\",\n content=[\n {\n \"text\": \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\"\n }\n ]\n)\n\nprint(\"----sync call, please wait a moment----\")\nrsp = ImageGeneration.call(\n model='wan2.7-image-pro',\n api_key=api_key,\n messages=[message],\n enable_sequential=True,\n n=4,\n size=\"2K\"\n )\n\nprint(rsp)", + "java": "import com.alibaba.dashscope.aigc.imagegeneration.*;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.exception.UploadFileException;\nimport com.alibaba.dashscope.utils.Constants;\nimport com.alibaba.dashscope.utils.JsonUtils;\n\nimport java.util.Collections;\n\n\npublic class Main {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n \n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException {\n // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例）\n ImageGenerationMessage message = ImageGenerationMessage.builder()\n .role(\"user\")\n .content(Collections.singletonList(\n Collections.singletonMap(\"text\", \"电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。\")\n )).build();\n\n ImageGenerationParam param = ImageGenerationParam.builder()\n .apiKey(apiKey)\n .model(\"wan2.7-image-pro\")\n .messages(Collections.singletonList(message))\n .enableSequential(true)\n .n(4)\n .size(\"2K\")\n .build();\n\n ImageGeneration imageGeneration = new ImageGeneration();\n ImageGenerationResult result = null;\n try {\n System.out.println(\"----sync call, please wait a moment----\");\n result = imageGeneration.call(param);\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n public static void main(String[] args) {\n try {\n basicCall();\n } catch (ApiException | NoApiKeyException | UploadFileException e) {\n System.out.println(e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3026980.html" } } @@ -185,10 +146,15 @@ "collectionTag": "wan2.6", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-image", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -212,40 +178,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-15T11:55:32.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-Image", "docUrl": "https://help.aliyun.com/document_detail/3001143.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'X-DashScope-Async: enable' \\\n--data '{\n \"model\": \"wan2.6-image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"给我一个3张图辣椒炒肉教程\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"n\": 1,\n \"size\": \"1280*1280\",\n \"enable_interleave\":true\n }\n}'\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -266,10 +205,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-i2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -293,40 +237,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-23T13:38:14.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-I2I-Preview", "docUrl": "https://help.aliyun.com/document_detail/2982258.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n-H 'X-DashScope-Async: enable' \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H 'Content-Type: application/json' \\\n-d '{\n\"model\": \"wan2.5-i2i-preview\",\n\"input\": {\n\"prompt\": \"将花卉连衣裙换成一件复古风格的蕾丝长裙，领口和袖口有精致的刺绣细节。\",\n\"images\": [\n\"https://img.alicdn.com/imgextra/i3/O1CN01Z1BLz61dMGqxmijRd_!!6000000003721-2-tps-1080-1620.png\"\n]\n},\n\"parameters\": {\n\"size\": \"1280*1280\",\n\"n\": 1\n}\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -343,51 +260,29 @@ "description": "万相-通义图像编辑，支持预设编辑任务与指令式编辑，包含多种局部/全图编辑能力，如图像风格化、线稿生图、局部重绘、参考图生成、图像外扩、图像超分等。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-imageedit", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-03-25T07:22:03.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-ImageEdit", "docUrl": "https://help.aliyun.com/document_detail/2868981.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-imageedit\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-imageedit\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-imageedit\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n public static void main(String[] args) {\n syncCall();\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json index f06cfe9b..3c98f6f4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json @@ -19,10 +19,21 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -45,45 +56,19 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-03T03:21:23.000+00:00", - "inferenceProvider": "bailian", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-I2V", "docUrl": "https://help.aliyun.com/document_detail/3025059.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由rap构成，没有其他对话或杂音。\",\n \"media\": [\n {\n \"type\": \"first_frame\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\"\n },\n {\n \"type\": \"driving_audio\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n \n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"duration\": 10,\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3025059.html" } } @@ -107,10 +92,33 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-i2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.15", + "type": "720P_no_audio", + "priceName": "视频生成（720P 无声）" + }, + { + "priceUnit": "每秒", + "price": "0.25", + "type": "1080P_no_audio", + "priceName": "视频生成（1080P 无声）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -132,53 +140,14 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2026-01-15T07:18:54.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "wan2.6-I2V-flash", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", "category": "Wan", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" } } } @@ -201,10 +170,35 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-i2v", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "discount": 0.5, + "type": "720P_batch", + "priceName": "视频生成（720P Batch Chat）" + }, + { + "priceUnit": "每秒", + "price": "1", + "discount": 0.5, + "type": "1080P_batch", + "priceName": "视频生成（1080P Batch Chat）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -226,52 +220,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-03T13:03:01.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-I2V", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.6-i2v\",\n \"input\": {\n \"prompt\": \"一幅都市奇幻艺术的场景。一个充满动感的涂鸦艺术角色。一个由喷漆所画成的少年，正从一面混凝土墙上活过来。他一边用极快的语速演唱一首英文rap，一边摆着一个经典的、充满活力的说唱歌手姿势。场景设定在夜晚一个充满都市感的铁路桥下。灯光来自一盏孤零零的街灯，营造出电影般的氛围，充满高能量和惊人的细节。视频的音频部分完全由他的rap构成，没有其他对话或杂音。\",\n \"img_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/wpimhv/rap.png\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250925/ozwpvi/rap.mp3\"\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"duration\": 10,\n \"audio\": true,\n \"shot_type\":\"multi\"\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"\n" } } } @@ -295,10 +250,27 @@ "fine-tuning" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-i2v-preview", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.3", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -327,46 +299,13 @@ "lora" ] }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-I2V-Preview", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.5-i2v-preview',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -384,10 +323,21 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.14", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -409,41 +359,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-I2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -462,10 +384,27 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-kf2v-flash", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.1", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" + }, + { + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "0.48", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -494,40 +433,13 @@ "lora" ] }, - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-KF2V-Flash", "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wan2.2-kf2v-flash\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" } } } @@ -546,10 +458,21 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-animate-move", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.4", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -574,40 +497,13 @@ "modelAlias": "Wan2.2-Animate-Move", "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T04:02:30.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-Animate-Move", "docUrl": "https://help.aliyun.com/document_detail/2981852.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-move\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/adsyrp/move_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/kaakcn/move_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } @@ -626,10 +522,21 @@ "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-animate-mix", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio", + "priceName": "视频生成（std）" + }, + { + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_pro", + "priceName": "视频生成（pro）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -653,182 +560,93 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T04:02:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-Animate-Mix", "docUrl": "https://help.aliyun.com/document_detail/2982219.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n\"model\": \"wan2.2-animate-mix\",\n\"input\": {\n\"image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/bhkfor/mix_input_image.jpeg\",\n\"video_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250919/wqefue/mix_input_video.mp4\"\n},\n\"parameters\": {\n\"check_image\": true,\n\"mode\": \"wan-std\"\n}\n}'\n\ncurl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/13b1848b-5493-4c0e-8c44-xxxxxxxxxxxx \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } }, { "inferenceMetadata": { - "response_modality": [ - "Video" - ], + "response_modality": [], "request_modality": [ "Image" ] }, - "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", + "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-i2v-flash", - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 - }, - "model-default": { - "count_limit_period": 1, - "async_user_queue_limit": 500, - "count_limit": 2, - "async_task_timeout": 180, - "type": "model-default", - "async_user_concurrency_limit": 2 + "model": "wan2.2-s2v-detect", + "prices": [ + { + "priceUnit": "每张", + "price": "0.004", + "type": "image_detect_number", + "priceName": "图片检测" } - }, + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", - "trainingTypes": { - "sft": [ - "lora" - ] - }, - "inferenceProvider": "bailian", - "name": "Wan2.2-I2V-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-S2V-Detect", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "curl": "curl 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" } } } }, { "inferenceMetadata": { - "response_modality": [], + "response_modality": [ + "Video" + ], "request_modality": [ "Image" ] }, - "description": "wan2.2-s2v-detect 是 wan2.2-s2v 的辅助模型，用于确认输入的人物肖像图片是否符合 wan2.2-s2v 模型所需的人物肖像图片规范。wan2.2-s2v 模型基于 wan2.2-s2v-detect 检测通过的图片和人声音频文件进行视频生成。", + "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v-detect", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:09.000+00:00", - "inferenceProvider": "bailian", - "name": "Wan2.2-S2V-Detect", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978214.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, + "model": "wan2.2-s2v", + "prices": [ { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每秒", + "price": "0.5", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每秒", + "price": "0.9", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" } ], + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "通义万相2.2-数字人-S2V", + "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", "samples": { "dashscope": { "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/face-detect' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data-raw '{\n \"model\": \"wan2.2-s2v-detect\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\"\n }\n }'" + "curl": "curl 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" } } } @@ -842,53 +660,66 @@ "Image" ] }, - "description": "wan2.2-s2v 是一款视频生成模型，可基于人物图片和人声音频文件，生成高质量的人物说话/唱歌/表演动态视频。", + "description": "全新升级的万相2.2图生视频，生成速度更快。优化视频生成稳定性与成功率，更强大的指令遵循能力，稳定保持图片文字、人像和商品一致性，精准运镜控制。", "collectionTag": "wan2.2", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wan2.2-s2v", - "capabilities": [ - "VG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2025-08-25T11:54:37.000+00:00", - "inferenceProvider": "bailian", - "name": "通义万相2.2-数字人-S2V", - "docUrl": "https://help.aliyun.com/zh/document_detail/2978215.html", - "predictConfig": [ + "model": "wan2.2-i2v-flash", + "prices": [ { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" + "priceUnit": "每秒", + "price": "0.1", + "type": "video_ratio_480p", + "priceName": "视频生成（480P）" }, { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] + "priceUnit": "每秒", + "price": "0.2", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" }, { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] + "priceUnit": "每秒", + "price": "0.48", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" } ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + }, + "model-default": { + "count_limit_period": 1, + "async_user_queue_limit": 500, + "count_limit": 2, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 2 + } + }, + "capabilities": [ + "VG" + ], + "versionTag": "MAJOR", + "latestOnlineAt": "2025-08-11T03:53:00.000+00:00", + "trainingTypes": { + "sft": [ + "lora" + ] + }, + "inferenceProvider": "aliyun-bailian", + "name": "Wan2.2-I2V-Flash", + "docUrl": "https://help.aliyun.com/document_detail/2867393.html", "samples": { "dashscope": { "default": { - "curl": "curl 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis/' \\\n --header 'X-DashScope-Async: enable' \\\n --header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n --header 'Content-Type: application/json' \\\n --data '{\n \"model\": \"wan2.2-s2v\",\n \"input\": {\n \"image_url\": \"https://img.alicdn.com/imgextra/i3/O1CN011FObkp1T7Ttowoq4F_!!6000000002335-0-tps-1440-1797.jpg\",\n \"audio_url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250825/iaqpio/input_audio.MP3\"\n },\n \"parameters\": {\n \"resolution\": \"480P\"\n }\n }'" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wan2.2-i2v-flash',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -905,10 +736,15 @@ "description": "万相2.1-首尾帧-Plus，两张图片生成丝滑过度视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成画面细节更丰富。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-kf2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -930,40 +766,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-04-20T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-KF2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2880649.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nfirst_frame_url = \"https://wanx.alicdn.com/material/20250318/first_frame.png\"\nlast_frame_url = \"https://wanx.alicdn.com/material/20250318/last_frame.png\"\n\ndef sample_sync_call_kf2v():\n rsp = VideoSynthesis.call(api_key=api_key,\n model=\"wanx2.1-kf2v-plus\",\n prompt=\"写实风格，一只黑色小猫好奇地看向天空，镜头从平视逐渐上升，最后俯拍小猫好奇的眼神。\",\n first_frame_url=first_frame_url,\n last_frame_url=last_frame_url,\n resolution=\"720P\",\n prompt_extend=True)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_kf2v()" } } } @@ -980,10 +789,15 @@ "description": "万相2.1-图生视频-Plus，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，视频质量更高。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-i2v-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -1005,41 +819,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-20T03:30:02.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-I2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-plus',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } @@ -1056,50 +842,27 @@ "description": "万相2.1-图生视频-Turbo，让图片变为动态视频。支持大幅度复杂运动、物理规律遵循、丰富艺术风格和影视级画面质感，指令遵循能力进一步提升，生成速度更快。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-i2v-turbo", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.24", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "capabilities": [ "VG" ], "versionTag": "MAJOR", "latestOnlineAt": "2025-02-27T02:24:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-I2V-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2867393.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "720P" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import VideoSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API key）\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\nimg_url = \"https://cdn.translate.alibaba.com/r/wanx-demo-1.png\"\n\ndef sample_async_call_i2v():\n # call async api, will return the task information\n # you can get task status with the returned task id.\n rsp = VideoSynthesis.async_call(model='wanx2.1-i2v-turbo',\n prompt='一只猫在草地上奔跑',\n img_url=img_url)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(\"task_id: %s\" % rsp.output.task_id)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n \n # get the task information include the task status.\n status = VideoSynthesis.fetch(rsp)\n if status.status_code == HTTPStatus.OK:\n print(status.output.task_status) # check the task status\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (status.status_code, status.code, status.message))\n\n # wait the task complete, will call fetch interval, and check it's in finished status.\n rsp = VideoSynthesis.wait(rsp)\n print(rsp)\n if rsp.status_code == HTTPStatus.OK:\n print(rsp.output.video_url)\n else:\n print('Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_async_call_i2v()" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json index bbadbc33..f1e7799b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json @@ -58,43 +58,7 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-R2V", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 10 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { @@ -170,39 +134,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-R2V-Flash", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -266,34 +197,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-R2V", "docUrl": "https://help.aliyun.com/document_detail/3001146.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json index 71e37806..852b0911 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json @@ -17,10 +17,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.6-t2i", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -44,39 +49,9 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-12-15T08:05:15.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-T2I", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -100,10 +75,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.5-t2i-preview", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -127,43 +107,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-09-19T08:44:37.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-T2I-Preview", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*1280", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.5-t2i-preview\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -183,10 +133,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -208,43 +163,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2I-Plus", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -264,10 +189,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.2-t2i-flash", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -289,43 +219,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-07-27T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2I-Flash", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wan2.2-t2i-flash\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -344,10 +244,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-t2i-plus", + "prices": [ + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -369,43 +274,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-08T16:09:10.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2I-Plus", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-plus\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -424,10 +299,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.14", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -449,43 +329,13 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-08T16:12:34.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2I-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1280*720", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.1-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -504,116 +354,27 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-v1", - "capabilities": [ - "IG" - ], - "versionTag": "MAJOR", - "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", - "inferenceProvider": "bailian", - "name": "wanx-t2i", - "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "negative_prompt", - "key": "negativePrompt", - "tip": "通过指定用户不想看到的内容来优化模型输出，使模型产生更有针对性和理想的结果。" - }, + "prices": [ { - "name": "style", - "key": "style", - "default": "", - "tip": "输出风格" - }, - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "种子值", - "range": [ - 1, - 4294967289 - ] + "priceUnit": "每张", + "price": "0.16", + "type": "image_number", + "priceName": "图片生成" } ], - "samples": { - "dashscope": { - "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" - } - } - } - }, - { - "inferenceMetadata": { - "response_modality": [ - "Image" - ], - "request_modality": [ - "Text" - ] - }, - "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", - "features": [], - "provider": "wan", - "limit": { - "message": "model not exist" - }, - "model": "wanx-v1-0521", "capabilities": [ "IG" ], - "versionTag": "SNAPSHOT", - "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", - "inferenceProvider": "bailian", - "name": "万相-文本生成图像-2024-05-21", + "versionTag": "MAJOR", + "latestOnlineAt": "2024-01-05T08:26:20.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "wanx-t2i", "docUrl": "https://help.aliyun.com/document_detail/2712483.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } @@ -632,10 +393,15 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.0-t2i-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.04", + "type": "image_number", + "priceName": "图片生成" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -657,43 +423,42 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-01-20T03:29:28.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.0-T2I-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2862677.html", - "predictConfig": [ - { - "name": "size", - "key": "size", - "default": "1024*1024", - "tip": "输出分辨率" - }, - { - "name": "n", - "key": "n", - "default": 1, - "tip": "本次请求生成的图片数量" - }, - { - "name": "seed", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" + "samples": { + "dashscope": { + "default": { + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + } } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "Image" + ], + "request_modality": [ + "Text" + ] + }, + "description": "万相-文本生成图像大模型，支持中英文双语输入，重点风格包括但不限于水彩、油画、中国画、素描、扁平插画、二次元、3D卡通，本模型为通义万相的2024年5月21号的历史快照。", + "features": [], + "provider": "wan", + "model": "wanx-v1-0521", + "capabilities": [ + "IG" ], + "versionTag": "SNAPSHOT", + "latestOnlineAt": "2024-05-22T13:57:26.000+00:00", + "inferenceProvider": "aliyun-bailian", + "name": "万相-文本生成图像-2024-05-21", + "docUrl": "https://help.aliyun.com/document_detail/2712483.html", "samples": { "dashscope": { "default": { - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx2.0-t2i-turbo\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一间有着精致窗户的花店，漂亮的木质门，摆放着花朵\"\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"wanx-v1-0521\",\n prompt=prompt,\n n=1,\n size='1280*1280')\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n # 在当前目录下保存图片\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json index bdaeaa25..f4218289 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json @@ -58,43 +58,7 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-T2V", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长(秒)", - "key": "duration", - "default": 5, - "range": [ - 2, - 15 - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { @@ -174,45 +138,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.6-T2V", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "智能多镜", - "key": "shot_type", - "default": "single", - "tip": "开启后输出视频采用多分镜形式呈现" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -285,39 +210,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.5-T2V-Preview", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - }, - { - "name": "生成音频", - "key": "audio", - "default": true - } - ], "samples": { "dashscope": { "default": { @@ -378,34 +270,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.2-T2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1920*1080" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -459,34 +323,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2V-Plus", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { @@ -546,34 +382,6 @@ "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-T2V-Turbo", "docUrl": "https://help.aliyun.com/document_detail/2865250.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1280*720" - }, - { - "name": "视频时长", - "key": "duration", - "default": 5 - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": true, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json index 0d3e9cb0..77815faf 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json @@ -19,10 +19,21 @@ "model-experience" ], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wan2.7-videoedit", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.6", + "type": "video_ratio_720p", + "priceName": "视频生成（720P）" + }, + { + "priceUnit": "每秒", + "price": "1", + "type": "video_ratio_1080p", + "priceName": "视频生成（1080P）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -45,48 +56,14 @@ "modelAlias": "", "versionTag": "MAJOR", "latestOnlineAt": "2026-04-03T03:19:53.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.7-VideoEdit", "docUrl": "https://help.aliyun.com/document_detail/3021842.html", - "predictConfig": [ - { - "name": "清晰度", - "key": "resolution", - "default": "1080P" - }, - { - "name": "宽高比", - "key": "ratio", - "default": "16:9" - }, - { - "name": "视频时长", - "key": "duration", - "tip": "可以选择与输入视频时长相同，或者指定不大于输入视频的时长" - }, - { - "name": "声音设置", - "key": "audio_setting", - "tip": [ - "auto （默认）：模型根据 prompt 内容智能判断。若提示词涉及声音描述，可能重新生成音频；否则可能保留输入素材的原声。", - "origin：强制保留输入视频的原声，不重新生成。" - ] - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "tip": "随机数种子，用于控制模型生成内容的随机性", - "range": [ - 1, - 2147483647 - ] - } - ], + "category": "Wan", "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"wan2.7-videoedit\",\n \"input\": {\n \"prompt\": \"将视频中女孩的衣服替换为图片中的衣服\",\n \"media\": [\n {\n \"type\": \"video\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260403/nlspwm/T2VA_22.mp4\"\n },\n {\n \"type\": \"reference_image\",\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260402/fwjpqf/wan2.7-videoedit-change-clothes.png\"\n }\n ]\n },\n \"parameters\": {\n \"resolution\": \"720P\",\n \"prompt_extend\": true,\n \"watermark\": true\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3021842.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json index a4634e68..d3845269 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json @@ -14,49 +14,33 @@ "description": "图像背景生成可以基于输入的前景图像素材拓展生成背景信息，实现自然的光影融合效果，与细腻的写实画面生成。支持文本描述、图像引导等多种方式，同时支持对生成的图像智能添加文字内容。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-background-generation-v2", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-03-22T03:33:37.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "图像背景生成", "docUrl": "https://help.aliyun.com/document_detail/2712497.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-background-generation-v2\",\n \"input\": {\n \"base_image_url\": \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/main_images/new_main_img/a.png\",\n \"ref_image_url\": \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_images/c5e50d27be534709817b2ab080b0162f_0.jpg\",\n \"ref_prompt\": \"山脉和晚霞\",\n \"reference_edge\": {\n \"foreground_edge\": [\n \"https://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/huaban_soft_edge/6cdd13941cef1b11d885aea1717b983ae566b8efc9094-vcsvxa_fw658webp.png\",\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/2c36cc4b7da027279e87311dac48fc2d5d784b1e72c0e-x4f1wC_fw658webp.png\"\n ],\n \"background_edge\": [\n \"http://vision-poster.oss-cn-shanghai.aliyuncs.com/lllcho.lc/data/test_data/images/ref_edge/0718a9741e07c52ca5506e75c4f2b99e22fff68a4c7d3-P9WGLr_fw658webp.png\"\n ],\n \"foreground_edge_prompt\": [\n \"粉色桃花\",\n \"可爱小狗\"\n ],\n \"background_edge_prompt\": [\n \"树叶\"\n ]\n }\n },\n \"parameters\": {\n \"n\": 4,\n \"ref_prompt_weight\": 0.5,\n \"model_version\": \"v3\"\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json index e6b46d90..b741bf9a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-poster-generation-v1.json @@ -14,49 +14,19 @@ "description": "创意海报生成，您的创意海报魔法工厂！它能够根据你的要求自动生成海报的背景和文字排版，支持多种海报风格，从宣传到祝福，让每一张海报都成为你的个性宣言。无需设计基础，轻松制作出彩作品，让创意触手可及。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-poster-generation-v1", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-21T02:49:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "创意海报生成", "docUrl": "https://help.aliyun.com/document_detail/2807172.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\":\"wanx-poster-generation-v1\",\n \"input\": {\n \"title\":\"春节快乐\",\n \"sub_title\":\"家庭团聚，共享天伦之乐\",\n \"body_text\":\"春节是中国最重要的传统节日之一，它象征着新的开始和希望\",\n \"prompt_text_zh\":\"灯笼，小猫，梅花\",\n \"wh_ratios\":\"竖版\",\n \"lora_name\":\"童话油画\",\n \"lora_weight\":0.8,\n \"ctrl_ratio\":0.7,\n \"ctrl_step\":0.7,\n \"generate_mode\":\"generate\",\n \"generate_num\":1\n },\n \"parameters\":{}\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json index 96a6474d..3e0b63b6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json @@ -14,51 +14,29 @@ "description": "万相-涂鸦作画通过手绘任意内容加文字描述，即可生成精美的涂鸦绘画作品，作品中的内容在参考手绘线条的同时，兼顾创意性和趣味性。涂鸦作画支持扁平插画、油画、二次元、3D卡通和水彩5种风格，可用于创意娱乐、辅助设计、儿童教学等场景。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-sketch-to-image-lite", + "prices": [ + { + "priceUnit": "每张", + "price": "0.06", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-11T11:21:09.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "万相-涂鸦作画", "docUrl": "https://help.aliyun.com/document_detail/2712498.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-sketch-to-image-lite\",\n \"input\": {\n \"sketch_image_url\": \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\",\n \"prompt\": \"一棵参天大树\"\n },\n \"parameters\": {\n \"size\": \"768*768\",\n \"n\": 2,\n \"sketch_weight\": 3,\n \"style\": \"\"\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport os\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\nprompt = \"一棵参天大树\"\nsketch_image_url = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\"\nmodel = \"wanx-sketch-to-image-lite\"\ntask = \"image2image\"\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=model,\n prompt=prompt,\n n=1,\n style='',\n size='768*768',\n sketch_image_url=sketch_image_url,\n task=task)\nprint('response: %s' % rsp)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp.output)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String prompt = \"一棵参天大树\";\n String sketchImageUrl = \"https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg\";\n String model = \"wanx-sketch-to-image-lite\";\n ImageSynthesisParam param = ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"768*768\")\n .sketchImageUrl(sketchImageUrl)\n .style(\"\")\n .build();\n\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json index a1c224b1..2271a317 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json @@ -14,49 +14,27 @@ "description": "人像风格重绘可以将输入的人物图像进行多种风格化的重绘生成，使新生成的图像在兼顾原始人物相貌的同时，带来不同风格的绘画效果。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-style-repaint-v1", + "prices": [ + { + "priceUnit": "每张", + "price": "0.12", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-03-22T03:32:59.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "人像风格重绘", "docUrl": "https://help.aliyun.com/document_detail/2712493.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-style-repaint-v1\",\n \"input\": {\n \"image_url\": \"https://vigen-video.oss-cn-shanghai.aliyuncs.com/demo_image/image_demo_input.png\",\n \"style_index\": 3\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json index c8356a15..b1b06280 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-virtualmodel.json @@ -14,49 +14,19 @@ "description": "虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-virtualmodel", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-06-25T15:19:20.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "虚拟模特", "docUrl": "https://help.aliyun.com/document_detail/2796985.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/virtualmodel/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-virtualmodel\",\n \"input\": {\n \"base_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/%E7%9C%9F%E4%BA%BA%E6%A8%A1%E7%89%B9%E5%AE%9E%E6%8B%8D-%E5%A5%B3%20%281%29.jpeg\",\n \"mask_image_url\": \"https://huarong123.oss-cn-hangzhou.aliyuncs.com/image/image.jpg\",\n \"prompt\": \"一名年轻女子，身穿白色短裤，极简风格调色板，长镜头，双色效果（暗银色和浅粉色）\",\n \"face_prompt\": \"年轻女子，面容姣好，最高品质\"\n },\n \"parameters\": {\n \"short_side_size\": \"512\",\n \"n\": 1\n }\n}'\n\ncurl --location --request GET 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json index 564e557a..8e9f4d14 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-x-painting.json @@ -14,51 +14,21 @@ "description": "万相-图像局部重绘是基于自研的Composer组合生成框架的AI绘画创作大模型后置处理链路，能够根据用户输入的原始图片和意涂抹图中局部区域和prompt提示词文字内容，生成符合语义描述的多样化风格的局部重绘图像。通过知识重组与可变维度扩散模型，加速收敛并提升最终生成图片的效果, 布局自然、细节丰富、画面细腻、结果逼真。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx-x-painting", "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-05-28T10:45:39.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "万相-图像局部重绘", "docUrl": "https://help.aliyun.com/document_detail/2797051.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx-x-painting\",\n \"input\": {\n \"prompt\": \"一只狗戴着红色眼镜\",\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n },\n \"parameters\": {\n \"size\": \"1024*1024\",\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "from http import HTTPStatus\nfrom urllib.parse import urlparse, unquote\nfrom pathlib import PurePosixPath\nimport requests\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nprompt = \"一只狗戴着红色眼镜\"\nmodel = \"wanx-x-painting\"\ntask = \"image2image\"\nextra_input = {\n \"base_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\",\n \"mask_image_url\": \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\"\n}\n\n\nprint('----sync call, please wait a moment----')\nrsp = ImageSynthesis.call(model=model,\n prompt=prompt,\n n=1,\n size='1024*1024',\n task=task,\n extra_input=extra_input)\nif rsp.status_code == HTTPStatus.OK:\n print(rsp)\n # save file to current directory\n for result in rsp.output.results:\n file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1]\n with open('./%s' % file_name, 'wb+') as f:\n f.write(requests.get(result.url).content)\nelse:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public void syncCall() {\n String task = \"image2image\";\n ImageSynthesis imageSynthesis = new ImageSynthesis(task);\n ImageSynthesisParam param = genImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n\n private ImageSynthesisParam genImageSynthesis(){\n HashMap extraInputMap = new HashMap<>();\n extraInputMap.put(\"base_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/source3.jpg\");\n extraInputMap.put(\"mask_image_url\", \"http://synthesis-source.oss-accelerate.aliyuncs.com/lingji/validation/mask2img/demo/glasses.png\");\n String prompt = \"一只狗戴着红色眼镜\";\n String model = \"wanx-x-painting\";\n return ImageSynthesisParam.builder()\n .model(model)\n .prompt(prompt)\n .n(1)\n .size(\"1024*1024\")\n .extraInputs(extraInputMap)\n .build();\n }\n\n\n public static void main(String[] args){\n Main text2Image = new Main();\n text2Image.syncCall();\n }\n\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json index 5e744220..870443af 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json @@ -16,10 +16,15 @@ "description": "万相2.1-VACE-Plus，视频编辑统一模型。支持局部编辑、视频重绘、背景扩展、时长延展、图片参考等多种视频编辑与生成任务，支持文本、图像、视频等多模态条件控制。", "features": [], "provider": "wan", - "limit": { - "message": "model not exist" - }, "model": "wanx2.1-vace-plus", + "prices": [ + { + "priceUnit": "每秒", + "price": "0.7", + "type": "video_ratio", + "priceName": "视频生成（std）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -41,42 +46,15 @@ ], "versionTag": "MAJOR", "latestOnlineAt": "2025-05-13T16:00:00.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Wan2.1-VACE-Plus", "docUrl": "https://help.aliyun.com/document_detail/2922183.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", - "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", - "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"wanx2.1-vace-plus\",\n \"input\": {\n \"function\": \"stylization_all\",\n \"prompt\": \"转换成法国绘本风格\",\n \"base_image_url\": \"http://wanx.alicdn.com/material/20250318/stylization_all_1.jpeg\"\n },\n \"parameters\": {\n \"n\": 1\n }\n}'\n\ncurl -X GET https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\"", + "python": "import os\nfrom http import HTTPStatus\nfrom dashscope import ImageSynthesis\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\n# --- 准备工作：确保 API Key 已设置 ---\napi_key = os.getenv(\"DASHSCOPE_API_KEY\")\n\n\nmask_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\"\nbase_image_url = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\"\n\n\ndef sample_sync_call_imageedit():\n print('please wait...')\n rsp = ImageSynthesis.call(api_key=api_key,\n model=\"wanx2.1-vace-plus\",\n function=\"description_edit_with_mask\",\n prompt=\"陶瓷兔子拿着陶瓷小花\",\n mask_image_url=mask_image_url,\n base_image_url=base_image_url,\n n=1)\n assert rsp.status_code == HTTPStatus.OK\n\n print('response: %s' % rsp)\n if rsp.status_code == HTTPStatus.OK:\n for result in rsp.output.results:\n print(\"---------------------------\")\n print(result.url)\n else:\n print('sync_call Failed, status_code: %s, code: %s, message: %s' %\n (rsp.status_code, rsp.code, rsp.message))\n\n\nif __name__ == '__main__':\n sample_sync_call_imageedit()", + "java": "// Copyright (c) Alibaba, Inc. and its affiliates.\n\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam;\nimport com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.JsonUtils;\nimport com.alibaba.dashscope.utils.Constants;\nimport java.util.HashMap;\nimport java.util.Map;\n\n\npublic class ImageEditSync {\n static {\n Constants.baseHttpApiUrl = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n // 从环境变量中获取 DashScope API Key（即阿里云百炼平台 API 密钥）\n static String apiKey = System.getenv(\"DASHSCOPE_API_KEY\");\n\n\n static String maskImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3_mask.png\";\n static String baseImageUrl = \"http://wanx.alicdn.com/material/20250318/description_edit_with_mask_3.jpeg\";\n\n\n\n public static void syncCall() {\n // 设置parameters参数\n Map parameters = new HashMap<>();\n parameters.put(\"prompt_extend\", true);\n\n ImageSynthesisParam param =\n ImageSynthesisParam.builder()\n .apiKey(apiKey)\n .model(\"wanx2.1-vace-plus\")\n .function(ImageSynthesis.ImageEditFunction.DESCRIPTION_EDIT_WITH_MASK)\n .prompt(\"陶瓷兔子拿着陶瓷小花\")\n .maskImageUrl(maskImageUrl)\n .baseImageUrl(baseImageUrl)\n .n(1)\n .size(\"1024*1024\")\n .parameters(parameters)\n .build();\n\n ImageSynthesis imageSynthesis = new ImageSynthesis();\n ImageSynthesisResult result = null;\n try {\n System.out.println(\"---sync call, please wait a moment----\");\n result = imageSynthesis.call(param);\n } catch (ApiException | NoApiKeyException e){\n throw new RuntimeException(e.getMessage());\n }\n System.out.println(JsonUtils.toJson(result));\n }\n \n\n public static void main(String[] args) {\n syncCall();\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json index af883d86..0c1096c5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json @@ -14,49 +14,33 @@ "description": "WordArt锦书-文字变形可以对输入的文字边缘轮廓进行创意变形，根据提示词内容进行边缘变化，实现一种字体的更多种创意用法，返回带有文字内容的黑底白色mask图。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "wordart-semantic", + "prices": [ + { + "priceUnit": "每张", + "price": "0.24", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:30:26.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "WordArt锦书-文字变形", "docUrl": "https://help.aliyun.com/document_detail/2712513.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location --request POST 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" + "curl": "curl --location --request POST 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/semantic' \\\n--header 'X-DashScope-Async: enable' \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--data-raw '{\n \"model\": \"wordart-semantic\",\n \"input\": {\n \"text\": \"文字创意\",\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\"\n },\n \"parameters\": {\n \"steps\": 80,\n \"n\": 2,\n \"output_image_ratio\": \"1024x1024\",\n \"font_name\": \"dongfangdakai\"\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json index 540893be..2f4ee929 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json @@ -15,49 +15,33 @@ "description": "WordArt锦书-文字纹理生成可以对输入的文字内容或文字图片进行创意设计，根据提示词内容对文字添加材质和纹理，实现立体凸显或场景融合的效果，生成效果精美、风格多样的艺术字，结合背景可以直接作为文字海报使用。", "features": [], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "wordart-texture", + "prices": [ + { + "priceUnit": "每张", + "price": "0.08", + "type": "image_number", + "priceName": "图片生成" + } + ], "capabilities": [ "IG" ], "versionTag": "MAJOR", "latestOnlineAt": "2024-04-09T08:29:05.000+00:00", - "inferenceProvider": "bailian", + "offlineInfo": { + "inference": { + "announceUrl": "https://www.aliyun.com/notice/118434", + "offlineTime": "2026-10-10 00:00:00" + } + }, + "inferenceProvider": "aliyun-bailian", "name": "WordArt锦书-文字纹理生成", "docUrl": "https://help.aliyun.com/document_detail/2712510.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/wordart/texture' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--header 'Accept: application/json' \\\n--data '{\n \"model\": \"wordart-texture\",\n \"input\": {\n \"image\": \n {\n \"image_url\": \"https://dmshared-new.oss-cn-hangzhou.aliyuncs.com/junyan.hjy/wordart/lcy/example.png\"\n },\n \"prompt\": \"水果，蔬菜，温暖的色彩空间\",\n \"texture_style\": \"material\"\n },\n \"parameters\": \n {\n \"image_short_size\": 704,\n \"n\": 2,\n \"alpha_channel\": false\n }\n}'" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json index f2a2f453..b97bce31 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json @@ -18,10 +18,27 @@ "cache" ], "provider": "xiaomi", - "limit": { - "message": "model not exist" - }, "model": "xiaomi/mimo-v2.5-pro", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "7", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "21", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.4", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -53,34 +70,13 @@ "inferenceProvider": "xiaomi", "name": "xiaomi/mimo-v2.5-pro", "docUrl": "https://help.aliyun.com/document_detail/3033942.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { - "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", - "docUrl": "https://help.aliyun.com/document_detail/3021647.html" + "default": { + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"xiaomi/mimo-v2.5-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"xiaomi/mimo-v2.5-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// Initialize the OpenAI client\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // Read from the environment variable\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'xiaomi/mimo-v2.5-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + 'Full response' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "docUrl": "https://help.aliyun.com/document_detail/3033942.html" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json index 5ad27363..9a3083df 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json @@ -17,10 +17,21 @@ "model-experience" ], "provider": "qwen-domain-model", - "limit": { - "message": "model not exist" - }, "model": "z-image-turbo", + "prices": [ + { + "priceUnit": "每张", + "price": "0.1", + "type": "image_standard", + "priceName": "图片生成（标准）" + }, + { + "priceUnit": "每张", + "price": "0.2", + "type": "image_thinking", + "priceName": "图片生成（思考）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 1, @@ -38,32 +49,9 @@ ], "versionTag": "SNAPSHOT", "latestOnlineAt": "2025-12-18T06:43:39.000+00:00", - "inferenceProvider": "bailian", + "inferenceProvider": "aliyun-bailian", "name": "Z-Image-Turbo", "docUrl": "https://help.aliyun.com/document_detail/3002354.html", - "predictConfig": [ - { - "name": "分辨率", - "key": "size", - "default": "1024*1024", - "tip": "请先选择输出分辨率，再选择输出宽高比" - }, - { - "name": "随机种子", - "key": "seed", - "default": 1234, - "range": [ - 1, - 2147483647 - ] - }, - { - "name": "智能改写", - "key": "prompt_extend", - "default": false, - "tip": "开启后会使用大模型对输入prompt进行智能改写，仅对正向提示词有效。对于较短的输prompt生成效果提升明显，但会增加3-4秒耗时。" - } - ], "samples": { "dashscope": { "default": { diff --git a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json index fe07e72c..f3432ef9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json @@ -19,10 +19,27 @@ "cache" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5.2", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -54,37 +71,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5.2", "docUrl": "https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.2\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } @@ -107,10 +97,27 @@ "prefix-completion" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5.1", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "8", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "28", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "2", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -142,33 +149,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5.1", "docUrl": "https://help.aliyun.com/zh/model-studio/glm-zhipu", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5.1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } @@ -191,11 +175,28 @@ "prefix-completion" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "ZHIPU/GLM-5", "iconUrl": "", + "prices": [ + { + "priceUnit": "每百万tokens", + "price": "6", + "type": "input_token", + "priceName": "输入" + }, + { + "priceUnit": "每百万tokens", + "price": "22", + "type": "output_token", + "priceName": "输出" + }, + { + "priceUnit": "每百万tokens", + "price": "1.5", + "type": "input_token_cache", + "priceName": "输入（缓存命中）" + } + ], "qpmInfo": { "model-default-actual": { "count_limit_period": 60, @@ -228,33 +229,10 @@ "inferenceProvider": "zhipu-ai", "name": "ZHIPU/GLM-5", "docUrl": "https://help.aliyun.com/document_detail/3026315.html", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - }, - { - "name": "enable_thinking", - "key": "enable_thinking", - "default": true, - "tip": "推理模式" - } - ], "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://ws-nckitja1d28cec5v.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"ZHIPU/GLM-5\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True, \"reasoning_effort\": \"max\"},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3026315.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/index.json b/skills/bailian-docs-llm-wiki/models/index.json index c5024785..325c6806 100644 --- a/skills/bailian-docs-llm-wiki/models/index.json +++ b/skills/bailian-docs-llm-wiki/models/index.json @@ -1,10 +1,10 @@ { - "updatedAt": "2026-07-16", + "updatedAt": "2026-07-23", "totalFamilies": 170, - "totalModels": 385, + "totalModels": 386, "capabilityDistribution": { "TG": 35, - "IG": 30, + "IG": 31, "VG": 26, "TTS": 16, "Reasoning": 14, @@ -17,11 +17,10 @@ "ME": 2, "Realtime-Chatting": 2, "Realtime-Text-to-Speech": 2, - "TR": 2, - "3D-generation": 1 + "TR": 2 }, "providerDistribution": { - "qwen": 100, + "qwen": 102, "qwen-domain-model": 34, "wan": 13, "happyhorse": 4, @@ -33,7 +32,6 @@ "vidu": 2, "kling": 1, "stepfun": 1, - "tripo": 1, "xiaomi": 1 }, "families": [ @@ -635,20 +633,21 @@ "primaryCapability": "TG", "capabilities": [ "TG", - "Reasoning", - "VU" + "VU", + "Reasoning" ], "providers": [ "moonshot-ai" ], - "itemCount": 4, + "itemCount": 5, "items": [ "kimi/kimi-k2.5", "kimi/kimi-k2.6", "kimi/kimi-k2.7-code", - "kimi/kimi-k2.7-code-highspeed" + "kimi/kimi-k2.7-code-highspeed", + "kimi/kimi-k3" ], - "maxContextWindow": 262144 + "maxContextWindow": 1048576 }, { "slug": "kling-models-market-place", @@ -974,7 +973,7 @@ "items": [ "qwen-audio-3.0-realtime-flash" ], - "maxContextWindow": 8192 + "maxContextWindow": 40960 }, { "slug": "qwen-audio-realtime-plus", @@ -990,7 +989,7 @@ "items": [ "qwen-audio-3.0-realtime-plus" ], - "maxContextWindow": 8192 + "maxContextWindow": 40960 }, { "slug": "qwen-audio-tts", @@ -1080,17 +1079,20 @@ "TR" ], "providers": [ + "qwen", "qwen-domain-model" ], - "itemCount": 6, + "itemCount": 7, "items": [ + "qwen3.7-text-embedding", "text-embedding-async-v1", "text-embedding-async-v2", "text-embedding-v1", "text-embedding-v2", "text-embedding-v3", "text-embedding-v4" - ] + ], + "maxContextWindow": 131072 }, { "slug": "qwen-flash-character", @@ -1155,6 +1157,21 @@ "qwen-image-2.0" ] }, + { + "slug": "qwen-image-3.0-pro", + "name": "Qwen-Image-3.0-Pro", + "primaryCapability": "IG", + "capabilities": [ + "IG" + ], + "providers": [ + "qwen" + ], + "itemCount": 1, + "items": [ + "qwen-image-3.0-pro" + ] + }, { "slug": "qwen-image-edit-max", "name": "Qwen-Image-Edit-Max", @@ -1983,8 +2000,8 @@ "primaryCapability": "Reasoning", "capabilities": [ "Reasoning", - "VU", - "TG" + "TG", + "VU" ], "providers": [ "qwen" @@ -2403,7 +2420,8 @@ "name": "StepFun推理模型", "primaryCapability": "TG", "capabilities": [ - "TG" + "TG", + "VU" ], "providers": [ "stepfun" @@ -2462,22 +2480,6 @@ ], "maxContextWindow": 32768 }, - { - "slug": "tripo-models-market-place", - "name": "Tripo", - "primaryCapability": "3D-generation", - "capabilities": [ - "3D-generation" - ], - "providers": [ - "tripo" - ], - "itemCount": 2, - "items": [ - "Tripo/Tripo-H3.1", - "Tripo/Tripo-P1.0" - ] - }, { "slug": "vanchin-models-market-place", "name": "Vanchin DeepSeek", diff --git a/skills/bailian-docs-llm-wiki/models/index.md b/skills/bailian-docs-llm-wiki/models/index.md index 16b4073f..b2ba1b4c 100644 --- a/skills/bailian-docs-llm-wiki/models/index.md +++ b/skills/bailian-docs-llm-wiki/models/index.md @@ -1,6 +1,6 @@ # 百炼模型市场索引 -> 自动生成 · 共 170 个模型家族 · 385 个主干模型 · 更新于 2026-07-16 +> 自动生成 · 共 170 个模型家族 · 386 个主干模型 · 更新于 2026-07-23 **机器查询走结构化文件**： @@ -20,7 +20,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Kimi](groups/Kimi-K2.json) — Kimi是由月之暗面提供的开源模型，包含k2.7-code、k2.6、k2.5、k2-thinking、k2-instruct等多模态和大语言模型。 - 模型：`kimi-k2-thinking`, `kimi-k2.5`, `kimi-k2.6`, `kimi-k2.7-code`, `Moonshot-Kimi-K2-Instruct` - [Kimi](groups/kimi-models-market-place.json) — 由月之暗面提供的Kimi系列模型的API服务。 - - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed` + - 模型：`kimi/kimi-k2.5`, `kimi/kimi-k2.6`, `kimi/kimi-k2.7-code`, `kimi/kimi-k2.7-code-highspeed`, `kimi/kimi-k3` - [MiMo文本模型](groups/xiaomi-models-market-place.json) — 由小米MiMo提供的MiMo文本模型API服务 - 模型：`xiaomi/mimo-v2.5-pro` - [MiniMax文本模型](groups/minimax-models-market-place.json) — 由MiniMax提供的MiniMax-M系列文本模型API服务。 @@ -84,7 +84,7 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [通义法睿-Plus-32K](groups/farui-plus.json) — 通义法睿是以通义千问为基座经法律行业数据和知识专门训练的法律行业大模型产品，综合运用了模型精调、强化学习、 RAG检索增强、法律Agent技术，具有回答法律问题、推理法律适用、推荐裁判类案、辅助案情分… - 模型：`farui-plus` -## 图像生成 `IG` — 30 个家族 +## 图像生成 `IG` — 31 个家族 - [AI试衣-Plus版](groups/aitryon-plus.json) — aitryon-plus是一款效果出众的虚拟试衣图片生成模型，可基于服饰平拍图片以及人物正面全身照，输出服饰的人物试衣效果图片。 相较于aitryon模型，aitryon-plus模型在图片清晰度、服… - 模型：`aitryon-plus` @@ -102,6 +102,8 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - 模型：`qwen-image-2.0` - [Qwen-Image-2.0-Pro](groups/qwen-image-2.0-pro.json) — Qwen-Image-2.0系列满血版模型，实现了图片生成和图片编辑的融合；具备更专业的文字渲染1k token指令支持能力、更细腻的真实质感，细腻刻画写实场景、更强的语义遵循能力。满血版具备2.0系… - 模型：`qwen-image-2.0-pro` +- [Qwen-Image-3.0-Pro](groups/qwen-image-3.0-pro.json) — 内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。 细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实… + - 模型：`qwen-image-3.0-pro` - [Qwen-Image-Edit-Max](groups/qwen-image-edit-max.json) — 千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。 - 模型：`qwen-image-edit-max` - [Qwen-Image-Edit-Plus](groups/qwen-image-edit.json) — 千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。 @@ -390,11 +392,6 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. ## 翻译 `TR` — 2 个家族 - [Qwen-Embedding](groups/qwen-embedding.json) — 基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度… - - 模型：`text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` + - 模型：`qwen3.7-text-embedding`, `text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` - [Qwen-Rerank](groups/qwen-rerank.json) — 基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。 - 模型：`gte-rerank-v2`, `qwen3-rerank`, `qwen3-vl-rerank` - -## 3D 生成 `3D-generation` — 1 个家族 - -- [Tripo](groups/tripo-models-market-place.json) — AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。 - - 模型：`Tripo/Tripo-H3.1`, `Tripo/Tripo-P1.0` diff --git a/skills/bailian-docs-llm-wiki/models/models.jsonl b/skills/bailian-docs-llm-wiki/models/models.jsonl index 5e7d3d5e..2fb2a1a5 100644 --- a/skills/bailian-docs-llm-wiki/models/models.jsonl +++ b/skills/bailian-docs-llm-wiki/models/models.jsonl @@ -1,28 +1,28 @@ -{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} -{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} -{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} -{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} -{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} +{"model":"kimi-k2-thinking","name":"Kimi-K2-Thinking","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning"],"features":["model-experience","cache","function-calling"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.5","name":"Kimi-K2.5","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.6","name":"Kimi-K2.6","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["Reasoning","VU","TG"],"features":["cache","function-calling","model-experience"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"kimi-k2.7-code","name":"kimi-k2.7-code","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":229376,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"8.125"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.65"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"Moonshot-Kimi-K2-Instruct","name":"Moonshot-Kimi-K2-Instruct","family":"Kimi-K2","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG"],"features":["model-experience","cache","function-calling"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2948482.html","detailPath":"groups/Kimi-K2.json"} +{"model":"MiniMax-M2.1","name":"MiniMax-M2.1","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":172032,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax-M2.5","name":"MiniMax-M2.5","family":"MiniMax-M2.1","familyName":"MiniMax","provider":"mini-max","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","cache"],"contextWindow":204800,"maxInputTokens":196608,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3017140.html","detailPath":"groups/MiniMax-M2.1.json"} +{"model":"MiniMax/speech-02-hd","name":"speech-02-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-02-turbo","name":"speech-02-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-hd","name":"speech-2.8-hd","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"3.5"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"MiniMax/speech-2.8-turbo","name":"speech-2.8-turbo","family":"MiniMax-speech-market-place","familyName":"MiniMax-Speech系列语音模型","provider":"mini-max","capabilities":["TTS"],"features":[],"maxInputTokens":10000,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"9.9"},{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":60,"usage_limit":20000,"usage_limit_field":"characters","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/MiniMax-speech-market-place.json"} +{"model":"aitryon-parsing-v1","name":"AI试衣OutfitAnyone-图片分割","family":"aitryon-parsing-v1","familyName":"AI试衣OutfitAnyone-图片分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865249.html","detailPath":"groups/aitryon-parsing-v1.json"} +{"model":"aitryon-plus","name":"AI试衣-Plus版","family":"aitryon-plus","familyName":"AI试衣-Plus版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881846.html","detailPath":"groups/aitryon-plus.json"} {"model":"aitryon-refiner","name":"AI试衣OutfitAnyone-图片精修","family":"aitryon-refiner","familyName":"AI试衣OutfitAnyone-图片精修","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796663.html","detailPath":"groups/aitryon-refiner.json"} -{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} -{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} -{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} -{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} -{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} -{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} +{"model":"aitryon","name":"AI试衣-基础版","family":"aitryon","familyName":"AI试衣-基础版","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796626.html","detailPath":"groups/aitryon.json"} +{"model":"animate-anyone-detect-gen2","name":"舞动人像AnimateAnyone-detect","family":"animate-anyone-detect-gen2","familyName":"舞动人像AnimateAnyone-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786465.html","detailPath":"groups/animate-anyone-detect-gen2.json"} +{"model":"animate-anyone-gen2","name":"舞动人像AnimateAnyone","family":"animate-anyone-gen2","familyName":"舞动人像AnimateAnyone","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786464.html","detailPath":"groups/animate-anyone-gen2.json"} +{"model":"animate-anyone-template-gen2","name":"舞动人像AnimateAnyone-template","family":"animate-anyone-template-gen2","familyName":"舞动人像AnimateAnyone-template","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807955.html","detailPath":"groups/animate-anyone-template-gen2.json"} +{"model":"cosyvoice-clone-v1","name":"声音复刻CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v1","name":"语音合成CosyVoice大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v2","name":"语音生成cosyvoice-v2大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-flash","name":"语音生成CosyVoice-v3-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3-plus","name":"语音生成CosyVoice-v3-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen-domain-model","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2817551.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-flash","name":"语音生成CosyVoice-v3.5-flash大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} +{"model":"cosyvoice-v3.5-plus","name":"语音生成CosyVoice-v3.5-plus大模型","family":"cosyvoice","familyName":"CosyVoice大模型","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842586.html","detailPath":"groups/cosyvoice.json"} {"model":"deepseek-r1","name":"DeepSeek-R1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"},{"type":"input_token_batch","unit":"每百万tokens","price":"2"},{"type":"output_token_batch","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-0528","name":"DeepSeek-R1-0528","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-r1-distill-qwen-1.5b","name":"DeepSeek-R1-Distill-Qwen-1.5B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} @@ -31,183 +31,186 @@ {"model":"deepseek-r1-distill-qwen-7b","name":"DeepSeek-R1-Distill-Qwen-7B","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3","name":"DeepSeek-V3","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":65536,"maxInputTokens":57344,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3.1","name":"DeepSeek-V3.1","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} -{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","web-search","cache","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"2.5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"2"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":3750,"count_limit_period":15,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3750,"count_limit_period":15,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} +{"model":"deepseek-v3.2","name":"DeepSeek-V3.2","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","web-search","cache","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"1"},{"type":"output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"2.5"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"2"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":15000,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v3.2-exp","name":"Deepseek-V3.2-Exp","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","web-search"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-flash","name":"DeepSeek-V4-Flash","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} {"model":"deepseek-v4-pro","name":"DeepSeek-V4-Pro","family":"deepseek","familyName":"DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","web-search"],"contextWindow":1000000,"maxInputTokens":1000000,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":750,"count_limit_period":3,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2868565.html","detailPath":"groups/deepseek.json"} -{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} -{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} -{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} -{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} -{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} +{"model":"multimodal-embedding-v1","name":"通用多模态向量","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"contextWindow":0,"maxInputTokens":512,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.9"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-flash","name":"视觉向量-flash","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.15"},{"type":"embedding_token","unit":"每百万tokens","price":"0.15"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"tongyi-embedding-vision-plus","name":"视觉向量-plus","family":"embedding","familyName":"通义多模态向量","provider":"qwen-domain-model","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"0.5"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712517.html","detailPath":"groups/embedding.json"} +{"model":"emo-detect-v1","name":"悦动人像EMO-detect","family":"emo-detect-v1","familyName":"悦动人像EMO-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786463.html","detailPath":"groups/emo-detect-v1.json"} +{"model":"emo-v1","name":"悦动人像EMO","family":"emo-v1","familyName":"悦动人像EMO","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_duration_1-1","unit":"每秒","price":"0.08"},{"type":"video_duration_3-4","unit":"每秒","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2786461.html","detailPath":"groups/emo-v1.json"} +{"model":"emoji-detect-v1","name":"表情包Emoji-detect","family":"emoji-detect-v1","familyName":"表情包Emoji-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865371.html","detailPath":"groups/emoji-detect-v1.json"} +{"model":"emoji-v1","name":"表情包Emoji","family":"emoji-v1","familyName":"表情包Emoji","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865374.html","detailPath":"groups/emoji-v1.json"} {"model":"facechain-facedetect","name":"FaceChain人物图像检测","family":"facechain-facedetect","familyName":"FaceChain人物图像检测","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712507.html","detailPath":"groups/facechain-facedetect.json"} -{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} -{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} -{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} -{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} -{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} -{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} -{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"facechain-generation","name":"FaceChain人物写真生成","family":"facechain-generation","familyName":"FaceChain人物写真生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.18"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712501.html","detailPath":"groups/facechain-generation.json"} +{"model":"farui-plus","name":"通义法睿-Plus-32K","family":"farui-plus","familyName":"通义法睿-Plus-32K","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":12000,"maxInputTokens":12000,"maxOutputTokens":2000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"20"},{"type":"output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60},"model-default":{"count_limit":4,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"output_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2778998.html","detailPath":"groups/farui-plus.json"} +{"model":"fun-asr-flash-2026-06-15","name":"Fun-ASR-Flash-2026-06-15","family":"fun-asr-flash","familyName":"Fun-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/fun-asr-flash.json"} +{"model":"fun-asr-flash-8k-realtime","name":"Fun-ASR-Flash-8k实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr-realtime","name":"Fun-ASR实时语音识别","family":"fun-asr-realtime","familyName":"Fun-ASR实时语音识别","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842554.html","detailPath":"groups/fun-asr-realtime.json"} +{"model":"fun-asr","name":"Fun-ASR语音识别","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880903.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-asr-mtl","name":"Fun-ASR-MTL","family":"fun-asr","familyName":"Fun-ASR语音识别","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2978300.html","detailPath":"groups/fun-asr.json"} +{"model":"fun-music-preview","name":"音乐生成 Preview","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.005"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} +{"model":"fun-music-v1","name":"音乐生成","family":"fun-music","familyName":"音乐生成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.002"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030448.html","detailPath":"groups/fun-music.json"} {"model":"glm-4.5","name":"GLM-4.5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.5-air","name":"GLM-4.5-Air","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.6","name":"GLM-4.6","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["Reasoning","TG"],"features":["model-experience","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-4.7","name":"GLM-4.7","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-5","name":"GLM-5","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","cache"],"contextWindow":202752,"maxInputTokens":169984,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} {"model":"glm-5.1","name":"GLM-5.1","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":202745,"maxInputTokens":202745,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://www.alibabacloud.com/help/zh/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} -{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} -{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} -{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} -{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} -{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} -{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} -{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} -{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} +{"model":"glm-5.2","name":"GLM-5.2","family":"glm-4.5","familyName":"GLM","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["cache","function-calling","model-experience","structured-outputs"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-4.5.json"} +{"model":"glm-5.2-fast-preview","name":"GLM-5.2-Fast-Preview","family":"glm-fast","familyName":"GLM-5.2-Fast","provider":"zhipu-ai","capabilities":["TG"],"features":["cache","function-calling","prefix-completion","structured-outputs","web-search"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"16"},{"type":"output_token","unit":"每百万tokens","price":"56"},{"type":"input_token_cache","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2974045.html","detailPath":"groups/glm-fast.json"} +{"model":"gui-plus","name":"GUI-Plus","family":"gui-plus","familyName":"GUI-Plus","provider":"qwen-domain-model","capabilities":["VU"],"features":[],"contextWindow":256000,"maxInputTokens":254976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.5"},{"type":"output_token","unit":"每百万tokens","price":"4.5"}],"qpmInfo":{"model-default-actual":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":80,"count_limit_period":60,"usage_limit":540000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2997010.html","detailPath":"groups/gui-plus.json"} +{"model":"gummy-chat-v1","name":"一句话识别及翻译V1.0","family":"gummy-chat-v1","familyName":"一句话识别及翻译V1.0","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2866122.html","detailPath":"groups/gummy-chat-v1.json"} +{"model":"gummy-realtime-v1","name":"实时语音识别及翻译V1.0","family":"gummy-realtime-v1","familyName":"实时语音识别及翻译V1.0","provider":"qwen","capabilities":["Realtime-Audio-Translate"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00015"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865393.html","detailPath":"groups/gummy-realtime-v1.json"} +{"model":"happyhorse-1.0-i2v","name":"HappyHorse-1.0-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.1-i2v","name":"HappyHorse-1.1-I2V","family":"happyhorse-i2v","familyName":"HappyHorse-I2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029821.html","detailPath":"groups/happyhorse-i2v.json"} +{"model":"happyhorse-1.0-r2v","name":"HappyHorse-1.0-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} +{"model":"happyhorse-1.1-r2v","name":"HappyHorse-1.1-R2V","family":"happyhorse-r2v","familyName":"HappyHorse-R2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030778.html","detailPath":"groups/happyhorse-r2v.json"} +{"model":"happyhorse-1.0-t2v","name":"HappyHorse-1.0-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.1-t2v","name":"HappyHorse-1.1-T2V","family":"happyhorse-t2v","familyName":"HappyHorse-T2V","provider":"happyhorse","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3029820.html","detailPath":"groups/happyhorse-t2v.json"} +{"model":"happyhorse-1.0-video-edit","name":"HappyHorse-1.0-Video-Edit","family":"happyhorse-video-edit","familyName":"HappyHorse-Video-Edit","provider":"happyhorse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030779.html","detailPath":"groups/happyhorse-video-edit.json"} {"model":"image-erase-completion","name":"图像擦除补全","family":"image-erase-completion","familyName":"图像擦除补全","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840907.html","detailPath":"groups/image-erase-completion.json"} {"model":"image-instance-segmentation","name":"人物实例分割","family":"image-instance-segmentation","familyName":"人物实例分割","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2840906.html","detailPath":"groups/image-instance-segmentation.json"} {"model":"image-out-painting","name":"图像画面扩展","family":"image-out-painting","familyName":"图像画面扩展","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796845.html","detailPath":"groups/image-out-painting.json"} -{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} -{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} -{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} -{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} -{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} -{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} -{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} -{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} -{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} -{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} -{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} -{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} -{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} -{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} -{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} -{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} -{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"kimi/kimi-k2.5","name":"Kimi/Kimi K2.5","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.6","name":"Kimi/Kimi K2.6","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.1"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021620.html","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code","name":"kimi/kimi-k2.7-code","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6.5"},{"type":"output_token","unit":"每百万tokens","price":"27"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.3"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k2.7-code-highspeed","name":"kimi/kimi-k2.7-code-highspeed","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","Reasoning","VU"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"13"},{"type":"output_token","unit":"每百万tokens","price":"54"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.6"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kimi/kimi-k3","name":"kimi/kimi-k3","family":"kimi-models-market-place","familyName":"Kimi","provider":"moonshot-ai","capabilities":["TG","VU","Reasoning"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"20"},{"type":"output_token","unit":"每百万tokens","price":"100"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021620","detailPath":"groups/kimi-models-market-place.json"} +{"model":"kling/kling-v3-image-generation","name":"Kling Image 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-image-generation","name":"Kling Image 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.2"},{"type":"image_type_2k","unit":"每张","price":"0.2"},{"type":"image_type_4k","unit":"每秒","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-omni-video-generation","name":"Kling Video 3.0 Omni","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"720P_no_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_reference_video","unit":"每秒","price":"1.2"},{"type":"720P_no_audio_no_reference_video","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio_no_reference_video","unit":"每秒","price":"0.8"},{"type":"720P_no_audio_reference_video","unit":"每秒","price":"0.9"},{"type":"1080P_no_audio_reference_video","unit":"每秒","price":"1.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"kling/kling-v3-video-generation","name":"Kling Video 3.0","family":"kling-models-market-place","familyName":"可灵AI","provider":"kling","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.9"},{"type":"video_ratio_1080p","unit":"每秒","price":"1.2"},{"type":"720P_no_audio","unit":"每秒","price":"0.6"},{"type":"1080P_no_audio","unit":"每秒","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/kling-models-market-place.json"} +{"model":"liveportrait-detect","name":"灵动人像LivePortrait-detect","family":"liveportrait-detect","familyName":"灵动人像LivePortrait-detect","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856727.html","detailPath":"groups/liveportrait-detect.json"} +{"model":"liveportrait","name":"灵动人像LivePortrait","family":"liveportrait","familyName":"灵动人像LivePortrait","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.02"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2856730.html","detailPath":"groups/liveportrait.json"} +{"model":"MiniMax/MiniMax-M2.1","name":"MiniMax/MiniMax-M2.1","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.5","name":"MiniMax/MiniMax-M2.5","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.21"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":3}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M2.7","name":"MiniMax/MiniMax-M2.7","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.1"},{"type":"output_token","unit":"每百万tokens","price":"8.4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.42"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021647.html","detailPath":"groups/minimax-models-market-place.json"} +{"model":"MiniMax/MiniMax-M3","name":"MiniMax/MiniMax-M3","family":"minimax-models-market-place","familyName":"MiniMax文本模型","provider":"mini-max","capabilities":["TG","Reasoning","VU"],"features":["function-calling","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4.2"},{"type":"output_token","unit":"每百万tokens","price":"16.8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.84"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":60,"usage_limit":20000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3021647","detailPath":"groups/minimax-models-market-place.json"} +{"model":"paraformer-8k-v1","name":"Paraformer语音识别-8k-v1","family":"paraformer-8k-v1","familyName":"Paraformer语音识别-8k-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v1.json"} +{"model":"paraformer-8k-v2","name":"Paraformer语音识别-8k-v2","family":"paraformer-8k-v2","familyName":"Paraformer语音识别-8k-v2","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-8k-v2.json"} +{"model":"paraformer-mtl-v1","name":"Paraformer语音识别-mtl-v1","family":"paraformer-mtl-v1","familyName":"Paraformer语音识别-mtl-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-mtl-v1.json"} +{"model":"paraformer-realtime-8k-v1","name":"Paraformer实时语音识别-8k-v1","family":"paraformer-realtime-8k-v1","familyName":"Paraformer实时语音识别-8k-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v1.json"} +{"model":"paraformer-realtime-8k-v2","name":"Paraformer实时语音识别-8k-v2","family":"paraformer-realtime-8k-v2","familyName":"Paraformer实时语音识别-8k-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-8k-v2.json"} +{"model":"paraformer-realtime-v1","name":"Paraformer实时语音识别-v1","family":"paraformer-realtime-v1","familyName":"Paraformer实时语音识别-v1","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v1.json"} +{"model":"paraformer-realtime-v2","name":"Paraformer实时语音识别-v2","family":"paraformer-realtime-v2","familyName":"Paraformer实时语音识别-v2","provider":"qwen","capabilities":["Realtime-ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00024"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712536.html","detailPath":"groups/paraformer-realtime-v2.json"} +{"model":"paraformer-v1","name":"Paraformer语音识别-v1","family":"paraformer-v1","familyName":"Paraformer语音识别-v1","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v1.json"} +{"model":"paraformer-v2","name":"Paraformer语音识别-v2","family":"paraformer-v2","familyName":"Paraformer语音识别-v2","provider":"qwen","capabilities":["ASR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00008"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/paraformer-v2.json"} +{"model":"pixverse/pixverse-c1-it2v","name":"PixVerse-C1-it2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-kf2v","name":"PixVerse-C1-kf2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-r2v","name":"PixVerse-C1-r2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-c1-market-place.json"} +{"model":"pixverse/pixverse-c1-t2v","name":"PixVerse-C1-t2v","family":"pixverse-c1-market-place","familyName":"PixVerse C1","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.24"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.39"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.71"},{"type":"360P_no_audio","unit":"每秒","price":"0.18"},{"type":"540P_no_audio","unit":"每秒","price":"0.24"},{"type":"720P_no_audio","unit":"每秒","price":"0.3"},{"type":"1080P_no_audio","unit":"每秒","price":"0.56"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-c1-market-place.json"} {"model":"pixverse/pixverse-lipsync","name":"pixverse/pixverse-lipsync","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"video_generation","unit":"每秒","price":"0.12"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} {"model":"pixverse/pixverse-motioncontrol","name":"pixverse/pixverse-motioncontrol","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Video"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.27"},{"type":"video_ratio_540p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} {"model":"pixverse/pixverse-upscale","name":"pixverse/pixverse-upscale","family":"pixverse-capability-market-place","familyName":"PixVerse Capabilities","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_generation","unit":"每秒","price":"0.15"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-capability-market-place.json"} -{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} -{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} -{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} -{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} -{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} -{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":8192,"maxInputTokens":4096,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} +{"model":"pixverse/pixverse-v5.6-it2v","name":"PixVerse-V5.6-it2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-kf2v","name":"PixVerse-V5.6-kf2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-r2v","name":"PixVerse-V5.6-r2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v5.6-t2v","name":"PixVerse-V5.6-t2v","family":"pixverse-market-place","familyName":"PixVerse V5.6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.47"},{"type":"video_ratio_540p","unit":"每秒","price":"0.47"},{"type":"video_ratio_720p","unit":"每秒","price":"0.53"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"},{"type":"360P_no_audio","unit":"每秒","price":"0.21"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.44"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/pixverse-market-place.json"} +{"model":"pixverse/pixverse-v6-it2v","name":"PixVerse-V6-it2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025609.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-kf2v","name":"PixVerse-V6-kf2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025612.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-r2v","name":"PixVerse-V6-r2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/pixverse-reference-to-video-api-reference","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"pixverse/pixverse-v6-t2v","name":"PixVerse-V6-t2v","family":"pixverse-v6-market-place","familyName":"PixVerse V6","provider":"pixverse","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_360p","unit":"每秒","price":"0.21"},{"type":"video_ratio_540p","unit":"每秒","price":"0.27"},{"type":"video_ratio_720p","unit":"每秒","price":"0.36"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.68"},{"type":"360P_no_audio","unit":"每秒","price":"0.15"},{"type":"540P_no_audio","unit":"每秒","price":"0.21"},{"type":"720P_no_audio","unit":"每秒","price":"0.27"},{"type":"1080P_no_audio","unit":"每秒","price":"0.53"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"○ https://help.aliyun.com/document_detail/3025608.html","detailPath":"groups/pixverse-v6-market-place.json"} +{"model":"qvq-max","name":"QVQ-Max","family":"qvq-max","familyName":"QVQ-Max","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"32"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-max.json"} +{"model":"qvq-plus","name":"QVQ-Plus","family":"qvq-plus","familyName":"Qwen-QVQ-Plus","provider":"qwen","capabilities":["Reasoning","VU"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":106496,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2877996.html","detailPath":"groups/qvq-plus.json"} +{"model":"qwen-audio-3.0-realtime-flash","name":"千问实时语音对话大模型3.0（极速版）","family":"qwen-audio-realtime-flash","familyName":"Qwen-Audio-Realtime-Flash","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":40960,"maxInputTokens":16384,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"30"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"3"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"30"},{"type":"audio_output_token","unit":"每百万tokens","price":"100"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-flash.json"} +{"model":"qwen-audio-3.0-realtime-plus","name":"千问实时语音大模型 （标准版）","family":"qwen-audio-realtime-plus","familyName":"Qwen-Audio-Realtime-Plus","provider":"qwen","capabilities":["Realtime-Chatting"],"features":["function-calling"],"contextWindow":40960,"maxInputTokens":16384,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Text"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"40"},{"type":"audio_text_input_token","unit":"每百万tokens","price":"5"},{"type":"audio_text_output_token","unit":"每百万tokens","price":"40"},{"type":"audio_output_token","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3041584.html","detailPath":"groups/qwen-audio-realtime-plus.json"} {"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} {"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","family":"qwen-audio-tts","familyName":"Qwen-Audio-TTS","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-audio-tts.json"} -{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} -{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} -{"model":"qwen-deep-research","name":"qwen-deep-research","family":"qwen-deep-research","familyName":"qwen-deep-research","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975991.html","detailPath":"groups/qwen-deep-research.json"} -{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} -{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} -{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} -{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} +{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","family":"qwen-coder-plus","familyName":"Qwen-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3.5"},{"type":"output_token","unit":"每百万tokens","price":"7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-plus.json"} +{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","family":"qwen-coder-turbo","familyName":"Qwen-Coder-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen-coder-turbo.json"} +{"model":"qwen-deep-research","name":"qwen-deep-research","family":"qwen-deep-research","familyName":"qwen-deep-research","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"54"},{"type":"output_token","unit":"每百万tokens","price":"163"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975991.html","detailPath":"groups/qwen-deep-research.json"} +{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","family":"qwen-doc-turbo","familyName":"Qwen-Doc-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["cache"],"contextWindow":262144,"maxInputTokens":253952,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.12"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"0.75"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"0.06"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2948885.html","detailPath":"groups/qwen-doc-turbo.json"} +{"model":"qwen3.7-text-embedding","name":"Qwen3.7-通用文本向量","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen","capabilities":["TR"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2400,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":2400,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712516.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v1","name":"通用文本向量-v1","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v2","name":"通用文本向量-v2","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.35"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v3","name":"通用文本向量-v3","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712515.html","detailPath":"groups/qwen-embedding.json"} +{"model":"text-embedding-v4","name":"通用文本向量-v4","family":"qwen-embedding","familyName":"Qwen-Embedding","provider":"qwen-domain-model","capabilities":["TR"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":1,"usage_limit":1200000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-embedding.json"} +{"model":"qwen-flash-character","name":"Qwen-Flash-Character","family":"qwen-flash-character","familyName":"Qwen-Flash-Character","provider":"qwen","capabilities":["TG"],"features":["model-experience","cache","web-search"],"contextWindow":8192,"maxInputTokens":8000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.25"},{"type":"output_token","unit":"每百万tokens","price":"1.5"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.05"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-flash-character.json"} {"model":"qwen-flash","name":"Qwen-Flash","family":"qwen-flash","familyName":"Qwen-Flash","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":5000000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-flash.json"} -{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} -{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} -{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} -{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} -{"model":"qwen-image-max","name":"Qwen-Image-Max","family":"qwen-image-max","familyName":"Qwen-Image-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-max.json"} -{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} -{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} -{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} -{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} -{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} -{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} -{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} -{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} -{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} -{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} -{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} -{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} -{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} +{"model":"qwen-image-2.0-pro","name":"Qwen-Image-2.0-Pro","family":"qwen-image-2.0-pro","familyName":"Qwen-Image-2.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0-pro.json"} +{"model":"qwen-image-2.0","name":"Qwen-Image-2.0","family":"qwen-image-2.0","familyName":"Qwen-Image-2.0","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-2.0.json"} +{"model":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","family":"qwen-image-3.0-pro","familyName":"Qwen-Image-3.0-Pro","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3047054.html","detailPath":"groups/qwen-image-3.0-pro.json"} +{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","family":"qwen-image-edit-max","familyName":"Qwen-Image-Edit-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit-max.json"} +{"model":"qwen-image-edit","name":"Qwen-Image-Edit","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.3"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","family":"qwen-image-edit","familyName":"Qwen-Image-Edit-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2976416.html","detailPath":"groups/qwen-image-edit.json"} +{"model":"qwen-image-max","name":"Qwen-Image-Max","family":"qwen-image-max","familyName":"Qwen-Image-Max","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-max.json"} +{"model":"qwen-image","name":"Qwen-Image","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-image-plus","name":"Qwen-Image-Plus","family":"qwen-image-plus","familyName":"Qwen-Image-Plus","provider":"qwen","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":2,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975126.html","detailPath":"groups/qwen-image-plus.json"} +{"model":"qwen-long","name":"Qwen-Long","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-long-latest","name":"Qwen-Long-Latest","family":"qwen-long","familyName":"Qwen-Long","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":10000000,"maxInputTokens":10000000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.25"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":10000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2846146.html#72cee64e7ff13","detailPath":"groups/qwen-long.json"} +{"model":"qwen-math-plus","name":"Qwen-Math-Plus","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":6,"usage_limit":20000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","family":"qwen-math-plus","familyName":"Qwen-Math-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-plus.json"} +{"model":"qwen-math-turbo","name":"Qwen-Math-Turbo","family":"qwen-math-turbo","familyName":"Qwen-Math-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":4096,"maxInputTokens":3072,"maxOutputTokens":3072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2849934.html","detailPath":"groups/qwen-math-turbo.json"} +{"model":"qwen-max","name":"Qwen-Max","family":"qwen-max","familyName":"Qwen-Max","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"9.6"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.48"},{"type":"input_token_batch","unit":"每百万tokens","price":"1.2"},{"type":"output_token_batch","unit":"每百万tokens","price":"4.8"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-max.json"} +{"model":"qwen-mt-flash","name":"Qwen-MT-Flash","family":"qwen-mt-flash","familyName":"Qwen-MT-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-flash.json"} +{"model":"qwen-mt-image","name":"Qwen-MT-Image","family":"qwen-mt-image","familyName":"Qwen-MT-Image","provider":"qwen-domain-model","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.003"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2977163.html","detailPath":"groups/qwen-mt-image.json"} +{"model":"qwen-mt-lite","name":"Qwen-MT-Lite","family":"qwen-mt-lite","familyName":"Qwen-MT-Lite","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.6"},{"type":"output_token","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-lite.json"} +{"model":"qwen-mt-plus","name":"Qwen-MT-Plus","family":"qwen-mt-plus","familyName":"Qwen-MT-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"5.4"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":25000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-plus.json"} +{"model":"qwen-mt-turbo","name":"Qwen-MT-Turbo","family":"qwen-mt-turbo","familyName":"Qwen-MT-Turbo","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":16384,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.7"},{"type":"output_token","unit":"每百万tokens","price":"1.95"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":35000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860790.html","detailPath":"groups/qwen-mt-turbo.json"} +{"model":"qwen-omni-turbo-realtime","name":"Qwen-Omni-Turbo-Realtime","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo-realtime-latest","name":"Qwen-Omni-Turbo-Realtime-Latest","family":"qwen-omni-turbo-realtime","familyName":"Qwen-Omni-Turbo-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"6"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"18"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen-omni-turbo-realtime.json"} +{"model":"qwen-omni-turbo","name":"Qwen-Omni-Turbo","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","batch"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"},{"type":"vision_input_token_cache","unit":"每百万tokens","price":"0.3"},{"type":"text_input_token_cache","unit":"每百万tokens","price":"0.08"},{"type":"audio_input_token_cache","unit":"每百万tokens","price":"5"},{"type":"text_input_token_batch","unit":"每百万tokens","price":"0.2"},{"type":"audio_input_token_batch","unit":"每百万tokens","price":"12.5"},{"type":"vision_input_token_batch","unit":"每百万tokens","price":"0.75"},{"type":"multi_output_token_batch","unit":"每百万tokens","price":"25"},{"type":"multiin_text_output_token_batch","unit":"每百万tokens","price":"2.25"},{"type":"purein_text_output_token_batch","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-omni-turbo-latest","name":"Qwen-Omni-Turbo-Latest","family":"qwen-omni-turbo","familyName":"Qwen-Omni-Turbo","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.4"},{"type":"audio_input_token","unit":"每百万tokens","price":"25"},{"type":"vision_input_token","unit":"每百万tokens","price":"1.5"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"1.6"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"4.5"},{"type":"multi_output_token","unit":"每百万tokens","price":"50"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen-omni-turbo.json"} +{"model":"qwen-plus-character","name":"Qwen-Plus-Character","family":"qwen-plus-character","familyName":"Qwen-Plus-Character","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","web-search","cache","structured-outputs"],"contextWindow":32768,"maxInputTokens":32768,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.16"}],"qpmInfo":{"model-default-actual":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":120,"count_limit_period":60,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2874763.html","detailPath":"groups/qwen-plus-character.json"} {"model":"qwen-plus","name":"Qwen-Plus","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":15000,"count_limit_period":30,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-0112","name":"Qwen-Plus-2025-01-12","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} +{"model":"qwen-plus-1220","name":"Qwen-Plus-2024-12-20","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":5,"usage_limit":150000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} {"model":"qwen-plus-latest","name":"Qwen-Plus-Latest","family":"qwen-plus","familyName":"Qwen-Plus","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","batch"],"contextWindow":1000000,"maxInputTokens":995904,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10},"model-default":{"count_limit":7500,"count_limit_period":30,"usage_limit":200000,"usage_limit_field":"total_tokens","usage_limit_period":10}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-plus.json"} -{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} -{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} -{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} -{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} -{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} -{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} -{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} -{"model":"qwen-vl-plus","name":"QwenVL-Plus","family":"qwen-vl-plus","familyName":"Qwen-VL-Plus","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-plus.json"} -{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} -{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} -{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} -{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} -{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} -{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} +{"model":"gte-rerank-v2","name":"深度文本重排序","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen-domain-model","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1},"model-default":{"count_limit":84,"count_limit_period":1,"usage_limit":83000000,"usage_limit_field":"total_tokens","usage_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-rerank","name":"千问3-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"contextWindow":30000,"maxInputTokens":30000,"maxOutputTokens":0,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"embedding_token","unit":"每百万tokens","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":900,"count_limit_period":10,"usage_limit":5000000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen3-vl-rerank","name":"Qwen3-VL-Rerank","family":"qwen-rerank","familyName":"Qwen-Rerank","provider":"qwen","capabilities":["TR"],"features":[],"maxInputTokens":120000,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":9000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2780056.html","detailPath":"groups/qwen-rerank.json"} +{"model":"qwen-tts-realtime","name":"Qwen-TTS-Realtime","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts-realtime-latest","name":"Qwen-TTS-Realtime-Latest","family":"qwen-tts-realtime","familyName":"Qwen-TTS-Realtime","provider":"qwen","capabilities":["Realtime-Text-to-Speech"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2.4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen-tts-realtime.json"} +{"model":"qwen-tts","name":"Qwen-TTS","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-tts-latest","name":"Qwen-TTS-Latest","family":"qwen-tts","familyName":"Qwen-TTS","provider":"qwen","capabilities":["TTS"],"features":[],"contextWindow":8192,"maxInputTokens":512,"maxOutputTokens":7680,"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.6"},{"type":"qwen_tts_multi_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2881635.html#dc02de8e5earc","detailPath":"groups/qwen-tts.json"} +{"model":"qwen-turbo","name":"Qwen-Turbo","family":"qwen-turbo","familyName":"Qwen-Turbo","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","cache","structured-outputs"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.6"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.3"},{"type":"thinking_output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_input_token_cache","unit":"每百万tokens","price":"0.06"},{"type":"thinking_output_token_batch","unit":"每百万tokens","price":"1.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.3"},{"type":"thinking_input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12},"model-default":{"count_limit":300,"count_limit_period":15,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":12}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen-turbo.json"} +{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":60000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","family":"qwen-vl-embedding","familyName":"Qwen-VL-Embedding","provider":"qwen","capabilities":["ME"],"features":["model-experience"],"maxInputTokens":32000,"inferenceMetadata":{"response_modality":[],"request_modality":["Text","Image"]},"prices":[{"type":"embedding_image_token","unit":"每百万tokens","price":"1.8"},{"type":"embedding_token","unit":"每百万tokens","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":120000,"usage_limit_field":"total_usage","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2842587.html","detailPath":"groups/qwen-vl-embedding.json"} +{"model":"qwen-vl-max","name":"Qwen-VL-Max","family":"qwen-vl-max","familyName":"Qwen-VL-Max","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.32"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"150"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":5,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-max.json"} +{"model":"qwen-vl-ocr","name":"QwenVL-OCR","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience"],"contextWindow":34096,"maxInputTokens":30000,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"5"},{"type":"output_token","unit":"每百万tokens","price":"5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","family":"qwen-vl-ocr","familyName":"Qwen-VL-OCR","provider":"qwen-domain-model","capabilities":["VU"],"features":["model-experience","batch"],"contextWindow":38192,"maxInputTokens":30000,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.3"},{"type":"output_token","unit":"每百万tokens","price":"0.5"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.15"},{"type":"output_token_batch","unit":"每百万tokens","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"LATEST","docUrl":"https://help.aliyun.com/document_detail/2712576.html#f4595a6e1aa7h","detailPath":"groups/qwen-vl-ocr.json"} +{"model":"qwen-vl-plus","name":"QwenVL-Plus","family":"qwen-vl-plus","familyName":"Qwen-VL-Plus","provider":"qwen","capabilities":["VU"],"features":["model-experience","structured-outputs","prefix-completion","cache","batch","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.8"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.16"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.4"},{"type":"output_token_batch","unit":"每百万tokens","price":"1"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen-vl-plus.json"} +{"model":"qwen-voice-design","name":"Qwen-Voice-Design","family":"qwen-voice-design","familyName":"Qwen-声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_vc_model","unit":"每次","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3000986.html","detailPath":"groups/qwen-voice-design.json"} +{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","family":"qwen-voice-enrollment","familyName":"Qwen-声音复刻","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"tts_vc_model","unit":"次","price":"0.01"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2975034.html","detailPath":"groups/qwen-voice-enrollment.json"} +{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","family":"qwen2.5","familyName":"Qwen2.5-开源模型","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience"],"contextWindow":32768,"maxInputTokens":30720,"maxOutputTokens":2048,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"0.6"},{"type":"audio_input_token","unit":"每百万tokens","price":"38"},{"type":"vision_input_token","unit":"每百万tokens","price":"2"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"2.4"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"6"},{"type":"multi_output_token","unit":"每百万tokens","price":"76"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen2.5.json"} +{"model":"qwen3-asr-flash-filetrans","name":"Qwen3-ASR-Flash-Filetrans","family":"qwen3-asr-flash-filetrans","familyName":"Qwen3-ASR-Flash-Filetrans","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash-filetrans.json"} +{"model":"qwen3-asr-flash-realtime","name":"Qwen3-ASR-Flash-Realtime","family":"qwen3-asr-flash-realtime","familyName":"Qwen3-ASR-Flash-Realtime","provider":"qwen","capabilities":["Realtime-ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00033"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1},"model-default":{"count_limit":20,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2989727.html","detailPath":"groups/qwen3-asr-flash-realtime.json"} +{"model":"qwen3-asr-flash","name":"Qwen3-ASR-Flash","family":"qwen3-asr-flash","familyName":"Qwen3-ASR-Flash","provider":"qwen-domain-model","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"content_duration","unit":"每秒","price":"0.00022"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2979031.html","detailPath":"groups/qwen3-asr-flash.json"} {"model":"qwen3-coder-30b-a3b-instruct","name":"Qwen3-Coder-30B-A3B-Instruct","family":"qwen3-coder-30b-a3b-instruct","familyName":"Qwen3-Coder-30B-A3B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/zh/model-studio/qwen-coder#272bcaccea8ls","detailPath":"groups/qwen3-coder-30b-a3b-instruct.json"} {"model":"qwen3-coder-480b-a35b-instruct","name":"Qwen3-Coder-480B-A35B-Instruct","family":"qwen3-coder-480b-a35b-instruct","familyName":"Qwen3-Coder-480B-A35B-Instruct","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-480b-a35b-instruct.json"} {"model":"qwen3-coder-flash","name":"Qwen3-Coder-Flash","family":"qwen3-coder-flash","familyName":"Qwen3-Coder-Flash","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-flash.json"} {"model":"qwen3-coder-plus","name":"Qwen3-Coder-Plus","family":"qwen3-coder-plus","familyName":"Qwen3-Coder-Plus","provider":"qwen-domain-model","capabilities":["TG"],"features":["model-experience","function-calling","cache"],"contextWindow":1000000,"maxInputTokens":997952,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3-coder-plus.json"} -{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} -{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} +{"model":"qwen3-livetranslate-flash-realtime","name":"Qwen3-LiveTranslate-Flash-Realtime","family":"qwen3-livetranslate-flash-realtime","familyName":"Qwen3-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Image","Audio"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"64"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"8"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"64"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"240"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3-livetranslate-flash-realtime.json"} +{"model":"qwen3-livetranslate-flash","name":"Qwen3-LiveTranslate-Flash","family":"qwen3-livetranslate-flash","familyName":"Qwen3-LiveTranslate-Flash","provider":"qwen-domain-model","capabilities":["Realtime-ASR"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Audio","Video"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"4"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"10"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2999748.html","detailPath":"groups/qwen3-livetranslate-flash.json"} {"model":"qwen3-max","name":"Qwen3-Max","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} {"model":"qwen3-max-preview","name":"Qwen3-Max-Preview","family":"qwen3-max","familyName":"Qwen3-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["model-experience","cache"],"contextWindow":262144,"maxInputTokens":258048,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3-max.json"} -{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} -{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} -{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} -{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} -{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} -{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} -{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} -{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} -{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} -{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} -{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} +{"model":"qwen3-omni-30b-a3b-captioner","name":"Qwen3-Omni-30b-a3b-Captioner","family":"qwen3-omni-30b-a3b-captioner","familyName":"Qwen3-Omni-30b-a3b-Captioner","provider":"qwen","capabilities":["ASR"],"features":[],"contextWindow":65536,"maxInputTokens":32768,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"prices":[{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2980468.html","detailPath":"groups/qwen3-omni-30b-a3b-captioner.json"} +{"model":"qwen3-omni-flash-realtime","name":"Qwen3-Omni-Flash-Realtime","family":"qwen3-omni-flash-realtime","familyName":"Qwen3-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":[],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"2.2"},{"type":"audio_input_token","unit":"每百万tokens","price":"18.9"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.9"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"8.3"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"15.2"},{"type":"multi_output_token","unit":"每百万tokens","price":"75.1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3-omni-flash-realtime.json"} +{"model":"qwen3-omni-flash","name":"Qwen3-Omni-Flash","family":"qwen3-omni-flash","familyName":"Qwen3-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["model-experience","function-calling"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"multiin_text_output_token","unit":"每百万tokens","price":"12.7"},{"type":"multi_output_token","unit":"每百万tokens","price":"62.6"},{"type":"thinking_text_input_token","unit":"每百万tokens","price":"1.8"},{"type":"thinking_audio_input_token","unit":"每百万tokens","price":"15.8"},{"type":"thinking_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"thinking_purein_text_output_token","unit":"每百万tokens","price":"6.9"},{"type":"thinking_multiin_text_output_token","unit":"每百万tokens","price":"12.7"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3-omni-flash.json"} +{"model":"qwen3-tts-flash-realtime","name":"Qwen3-TTS-Flash-Realtime","family":"qwen3-tts-flash-realtime","familyName":"Qwen3-TTS-Flash-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-flash-realtime.json"} +{"model":"qwen3-tts-flash","name":"Qwen3-TTS-Flash","family":"qwen3-tts-flash","familyName":"Qwen3-TTS-Flash","provider":"qwen","capabilities":["TTS"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-flash.json"} +{"model":"qwen3-tts-instruct-flash-realtime","name":"qwen3-tts-instruct-flash-realtime","family":"qwen3-tts-instruct-flash-realtime","familyName":"qwen3-tts-instruct-flash-realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-instruct-flash-realtime.json"} +{"model":"qwen3-tts-instruct-flash","name":"Qwen3-TTS-Instruct-Flash","family":"qwen3-tts-instruct-flash","familyName":"Qwen3-TTS-Instruct-Flash","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-instruct-flash.json"} +{"model":"qwen3-tts-vc-realtime-2026-01-15","name":"qwen3-tts-vc-realtime-2026-01-15","family":"qwen3-tts-vc-realtime","familyName":"Qwen3-TTS-VC-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vc-realtime.json"} +{"model":"qwen3-tts-vc-2026-01-22","name":"Qwen3-TTS-VC-2026-01-22","family":"qwen3-tts-vc","familyName":"Qwen3-TTS-VC","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vc.json"} +{"model":"qwen3-tts-vd-realtime-2026-01-15","name":"qwen3-tts-vd-realtime-2026-01-15","family":"qwen3-tts-vd-realtime","familyName":"Qwen3-TTS-VD-Realtime","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2938790.html","detailPath":"groups/qwen3-tts-vd-realtime.json"} +{"model":"qwen3-tts-vd-2026-01-26","name":"Qwen3-TTS-VD-2026-01-26","family":"qwen3-tts-vd","familyName":"Qwen3-TTS-VD","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"cosy_tts_number","unit":"每万字符","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":3,"count_limit_period":1},"model-default":{"count_limit":3,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2879134.html","detailPath":"groups/qwen3-tts-vd.json"} {"model":"qwen3-vl-flash","name":"Qwen3-VL-Flash","family":"qwen3-vl-flash","familyName":"Qwen3-VL-Flash","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-flash.json"} {"model":"qwen3-vl-plus","name":"Qwen3-VL-Plus","family":"qwen3-vl-plus","familyName":"Qwen3-VL-Plus","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","cache","batch"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":250,"count_limit_period":5,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3-vl-plus.json"} -{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} -{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} -{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} -{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} -{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} -{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} -{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} +{"model":"qwen3.5-flash","name":"Qwen3.5-Flash","family":"qwen3.5-flash","familyName":"Qwen3.5-Flash","provider":"qwen","capabilities":["Reasoning","TG","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","batch","cache","prefix-completion"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-flash.json"} +{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"40"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"100"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"160"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} +{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} +{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"27"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"107"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"3.3"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} +{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"18"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"72"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"2.2"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"13.3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} +{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"80"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"300"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"60"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} +{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search","function-calling","batch"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"53"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"213"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"40"},{"type":"omni_audio_input_token_batch","unit":"每百万tokens","price":"26.5"},{"type":"omni_no_audio_input_token_batch","unit":"每百万tokens","price":"3.5"},{"type":"omni_no_audio_output_token_batch","unit":"每百万tokens","price":"20"},{"type":"omni_audio_input_token_batch_chat","unit":"每百万tokens","price":"53"},{"type":"omni_no_audio_input_token_batch_chat","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token_batch_chat","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} {"model":"qwen3.5-plus","name":"Qwen3.5-Plus","family":"qwen3.5-plus","familyName":"Qwen3.5-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-plus.json"} {"model":"qwen3.5-122b-a10b","name":"Qwen3.5-122B-A10B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} {"model":"qwen3.5-27b","name":"Qwen3.5-27B","family":"qwen3.5","familyName":"Qwen3.5开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","fine-tuning"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5.json"} @@ -216,95 +219,93 @@ {"model":"qwen3.6-flash","name":"Qwen3.6-Flash","family":"qwen3.6-flash","familyName":"Qwen3.6-Flash","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":10000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-flash.json"} {"model":"qwen3.6-max-preview","name":"Qwen3.6-Max-Preview","family":"qwen3.6-max","familyName":"Qwen3.6-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":245760,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-max.json"} {"model":"qwen3.6-plus","name":"Qwen3.6-Plus","family":"qwen3.6-plus","familyName":"Qwen3.6-Plus","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6-plus.json"} -{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} -{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} -{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.6-27b","name":"Qwen3.6-27B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["Reasoning","VU","TG"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"3"},{"type":"output_token","unit":"每百万tokens","price":"18"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.6-35b-a3b","name":"Qwen3.6-35B-A3B","family":"qwen3.6","familyName":"Qwen3.6开源模型","provider":"qwen","capabilities":["VU","TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","structured-outputs","web-search"],"contextWindow":262144,"maxInputTokens":260096,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.8"},{"type":"output_token","unit":"每百万tokens","price":"10.8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.6.json"} +{"model":"qwen3.7-max","name":"Qwen3.7-Max","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["Reasoning","TG"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"},{"type":"input_token_cache","unit":"每百万tokens","price":"2.4"},{"type":"input_token_batch","unit":"每百万tokens","price":"6"},{"type":"output_token_batch","unit":"每百万tokens","price":"18"},{"type":"input_token_cache_creation_5m","unit":"每百万tokens","price":"15"},{"type":"input_token_cache_read","unit":"每百万tokens","price":"1.2"},{"type":"input_token_batch_chat","unit":"每百万tokens","price":"12"},{"type":"output_token_batch_chat","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} +{"model":"qwen3.7-max-preview","name":"Qwen3.7-Max-Preview","family":"qwen3.7-max","familyName":"Qwen3.7-Max","provider":"qwen","capabilities":["TG","Reasoning"],"features":["function-calling","model-experience","prefix-completion","web-search"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"36"}],"qpmInfo":{"model-default-actual":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":6,"count_limit_period":6,"usage_limit":500000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-max.json"} {"model":"qwen3.7-plus","name":"Qwen3.7-Plus","family":"qwen3.7-plus","familyName":"Qwen3.7-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["batch","cache","function-calling","model-experience","prefix-completion","web-search","structured-outputs"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":3000,"count_limit_period":6,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.7-plus.json"} -{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-14b","name":"Qwen3-14B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"},{"type":"ft","unit":"每百万tokens","price":"30"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b","name":"Qwen3-235B-A22B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-instruct-2507","name":"Qwen3-235B-A22B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-235b-a22b-thinking-2507","name":"Qwen3-235B-A22B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b","name":"Qwen3-30B-A3B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":30,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-instruct-2507","name":"Qwen3-30B-A3B-Instruct-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-30b-a3b-thinking-2507","name":"Qwen3-30B-A3B-Thinking-2507","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience"],"contextWindow":81920,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-32b","name":"Qwen3-32B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"},{"type":"ft","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":40,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-8b","name":"Qwen3-8B","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"6"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} {"model":"qwen3-coder-next","name":"通义千问3-Coder-Next","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":["model-experience"],"contextWindow":262144,"maxInputTokens":204800,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2850166.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} -{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} -{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} -{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"qwen3-next-80b-a3b-instruct","name":"Qwen3-Next-80B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-next-80b-a3b-thinking","name":"Qwen3-Next-80B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["Reasoning","TG"],"features":[],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"1"},{"type":"thinking_output_token","unit":"每百万tokens","price":"10"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-instruct","name":"Qwen3-VL-235B-A22B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-235b-a22b-thinking","name":"Qwen3-VL-235B-A22B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":1,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-instruct","name":"Qwen3-VL-30B-A3B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.75"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-30b-a3b-thinking","name":"Qwen3-VL-30B-A3B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.75"},{"type":"thinking_output_token","unit":"每百万tokens","price":"7.5"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-instruct","name":"Qwen3-VL-32B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-32b-thinking","name":"Qwen3-VL-32B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion"],"contextWindow":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"2"},{"type":"thinking_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-instruct","name":"Qwen3-VL-8B-Instruct","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":129024,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwen3-vl-8b-thinking","name":"Qwen3-VL-8B-Thinking","family":"qwen3","familyName":"Qwen3开源模型","provider":"qwen","capabilities":["VU","Reasoning"],"features":["model-experience","function-calling","structured-outputs","prefix-completion","fine-tuning"],"contextWindow":131072,"maxInputTokens":126976,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"thinking_input_token","unit":"每百万tokens","price":"0.5"},{"type":"thinking_output_token","unit":"每百万tokens","price":"5"},{"type":"ft","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2845871.html","detailPath":"groups/qwen3.json"} +{"model":"qwq-plus","name":"QwQ-Plus","family":"qwq-plus","familyName":"Qwen-QwQ-Plus","provider":"qwen","capabilities":["Reasoning"],"features":["model-experience","function-calling","web-search","batch"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.6"},{"type":"output_token","unit":"每百万tokens","price":"4"},{"type":"input_token_batch","unit":"每百万tokens","price":"0.8"},{"type":"output_token_batch","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":10,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2870973.html","detailPath":"groups/qwq-plus.json"} +{"model":"sambert-beth-v1","name":"Sambert语音合成-Beth","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-betty-v1","name":"Sambert语音合成-Betty","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-brian-v1","name":"Sambert语音合成-Brian","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cally-v1","name":"Sambert语音合成-Cally","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-camila-v1","name":"Sambert语音合成-Camila","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-cindy-v1","name":"Sambert语音合成-Cindy","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-clara-v1","name":"Sambert语音合成-Clara","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-donna-v1","name":"Sambert语音合成-Donna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-eva-v1","name":"Sambert语音合成-Eva","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-hanna-v1","name":"Sambert语音合成-Hanna","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-indah-v1","name":"Sambert语音合成-Indah","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-perla-v1","name":"Sambert语音合成-Perla","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-waan-v1","name":"Sambert语音合成-Waan","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhichu-v1","name":"Sambert语音合成-知厨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhida-v1","name":"Sambert语音合成-知达","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhide-v1","name":"Sambert语音合成-知德","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhifei-v1","name":"Sambert语音合成-知飞","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhigui-v1","name":"Sambert语音合成-知柜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhihao-v1","name":"Sambert语音合成-知浩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijia-v1","name":"Sambert语音合成-知佳","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhijing-v1","name":"Sambert语音合成-知婧","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhilun-v1","name":"Sambert语音合成-知伦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimao-v1","name":"Sambert语音合成-知猫","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimiao-emo-v1","name":"Sambert语音合成-知妙（多情感）","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiming-v1","name":"Sambert语音合成-知茗","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhimo-v1","name":"Sambert语音合成-知墨","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhina-v1","name":"Sambert语音合成-知娜","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhinan-v1","name":"Sambert语音合成-知楠","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqi-v1","name":"Sambert语音合成-知琪","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiqian-v1","name":"Sambert语音合成-知倩","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiru-v1","name":"Sambert语音合成-知茹","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishu-v1","name":"Sambert语音合成-知树","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhishuo-v1","name":"Sambert语音合成-知硕","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhistella-v1","name":"Sambert语音合成-知莎","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiting-v1","name":"Sambert语音合成-知婷","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiwei-v1","name":"Sambert语音合成-知薇","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiang-v1","name":"Sambert语音合成-知祥","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhixiao-v1","name":"Sambert语音合成-知笑","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiya-v1","name":"Sambert语音合成-知雅","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiye-v1","name":"Sambert语音合成-知晔","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiying-v1","name":"Sambert语音合成-知颖","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyuan-v1","name":"Sambert语音合成-知媛","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} +{"model":"sambert-zhiyue-v1","name":"Sambert语音合成-知悦","family":"sambert","familyName":"Sambert语音合成","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Text"]},"prices":[{"type":"tts_text_number","unit":"每万字符","price":"1"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712458.html","detailPath":"groups/sambert.json"} {"model":"shoemodel-v1","name":"鞋靴模特","family":"shoemodel-v1","familyName":"鞋靴模特","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2804662.html","detailPath":"groups/shoemodel-v1.json"} -{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} -{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-r1-0528","name":"SiliconFlow DeepSeek-R1-0528","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3-0324","name":"SiliconFlow DeepSeek-V3-0324","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":163840,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.1-terminus","name":"SiliconFlow DeepSeek-V3.1-Terminus","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} +{"model":"siliconflow/deepseek-v3.2","name":"SiliconFlow DeepSeek-V3.2","family":"siliconflow-models","familyName":"SiliconFlow DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","prefix-completion"],"contextWindow":163840,"maxInputTokens":163840,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":50000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3014912.html","detailPath":"groups/siliconflow-models.json"} {"model":"speech-biasing","name":"语音识别热词","family":"speech-biasing","familyName":"语音识别热词","provider":"qwen","capabilities":["ASR"],"features":[],"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712535.html","detailPath":"groups/speech-biasing.json"} -{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} -{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} -{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} -{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} -{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} -{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"stepfun/step-3.7-flash","name":"stepfun/step-3.7-flash","family":"stepfun-models-market-place","familyName":"StepFun推理模型","provider":"stepfun","capabilities":["TG","VU"],"features":["function-calling","structured-outputs","cache","batch"],"contextWindow":262144,"maxInputTokens":262144,"maxOutputTokens":262144,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1.35"},{"type":"output_token","unit":"每百万tokens","price":"8.1"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.27"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":2000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3036697.html","detailPath":"groups/stepfun-models-market-place.json"} +{"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.4"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} +{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.2"},{"type":"output_token","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} +{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2.7"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} {"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["VU","TG"],"features":["structured-outputs"],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.216"},{"type":"output_token","unit":"每百万tokens","price":"0.216"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["structured-outputs","prefix-completion","function-calling","cache"],"contextWindow":131072,"maxInputTokens":131072,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG","Reasoning"],"features":["function-calling","cache"],"contextWindow":131072,"maxInputTokens":32768,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"12"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"3"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":30,"count_limit_period":60,"usage_limit":600000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":393216,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"12"},{"type":"output_token","unit":"每百万tokens","price":"24"},{"type":"input_token_cache","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":300000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} -{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} -{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} +{"model":"video-style-transform","name":"视频风格重绘","family":"video-style-transform","familyName":"视频风格重绘","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.2"},{"type":"video_ratio_720p","unit":"每秒","price":"0.5"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2846319.html","detailPath":"groups/video-style-transform.json"} +{"model":"videoretalk","name":"声动人像VideoRetalk","family":"videoretalk","familyName":"声动人像VideoRetalk","provider":"qwen","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Audio"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860466.html","detailPath":"groups/videoretalk.json"} {"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.625"},{"type":"image_type_2k","unit":"每张","price":"1"},{"type":"image_type_4k","unit":"每张","price":"1.46875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} {"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.28125"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} {"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","family":"vidu-image-models-market-place","familyName":"Vidu AI生图","provider":"vidu","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_type_1k","unit":"每张","price":"0.9375"},{"type":"image_type_2k","unit":"每张","price":"0.9375"},{"type":"image_type_4k","unit":"每张","price":"1.71875"}],"qpmInfo":{"model-default-actual":{"count_limit":300,"count_limit_period":60},"model-default":{"count_limit":300,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3045893.html","detailPath":"groups/vidu-image-models-market-place.json"} @@ -331,55 +332,55 @@ {"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","family":"vidu-models-market-place","familyName":"Vidu AI生视频","provider":"vidu","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_540p","unit":"每秒","price":"0.25"},{"type":"video_ratio_720p","unit":"每秒","price":"0.375"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.4375"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","detailPath":"groups/vidu-models-market-place.json"} {"model":"virtualmodel-v2","name":"虚拟模特V2","family":"virtualmodel-v2","familyName":"虚拟模特V2","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/virtualmodel-v2.json"} {"model":"voice-enrollment","name":"大模型声音复刻及声音设计","family":"voice-enrollment","familyName":"大模型声音复刻及声音设计","provider":"qwen","capabilities":["TTS"],"features":[],"inferenceMetadata":{"response_modality":["Audio"],"request_modality":["Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":1},"model-default":{"count_limit":10,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861519.html","detailPath":"groups/voice-enrollment.json"} -{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} -{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} -{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982258.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.6-image","name":"Wan2.6-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image","Text"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001143.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image","name":"Wan2.7-Image","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image","Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.5"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026980.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","family":"wan-image-edit","familyName":"Wan-Image","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2868981.html","detailPath":"groups/wan-image-edit.json"} +{"model":"wan2.2-animate-mix","name":"Wan2.2-Animate-Mix","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.6"},{"type":"video_ratio_pro","unit":"每秒","price":"0.9"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2982219.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-animate-move","name":"Wan2.2-Animate-Move","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Video","Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.4"},{"type":"video_ratio_pro","unit":"每秒","price":"0.6"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2981852.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-flash","name":"Wan2.2-I2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-i2v-plus","name":"Wan2.2-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-kf2v-flash","name":"Wan2.2-KF2V-Flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.1"},{"type":"video_ratio_720p","unit":"每秒","price":"0.2"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.48"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v","name":"通义万相2.2-数字人-S2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.5"},{"type":"video_ratio_720p","unit":"每秒","price":"0.9"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978215.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.2-s2v-detect","name":"Wan2.2-S2V-Detect","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":[],"request_modality":["Image"]},"prices":[{"type":"image_detect_number","unit":"每张","price":"0.004"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/document_detail/2978214.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.5-i2v-preview","name":"Wan2.5-I2V-Preview","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience","fine-tuning"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Image","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v","name":"Wan2.6-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.6-i2v-flash","name":"wan2.6-I2V-flash","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wan2.7-i2v","name":"Wan2.7-I2V","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3025059.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-plus","name":"Wan2.1-I2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-i2v-turbo","name":"Wan2.1-I2V-Turbo","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867393.html","detailPath":"groups/wan-image-to-video.json"} +{"model":"wanx2.1-kf2v-plus","name":"Wan2.1-KF2V-Plus","family":"wan-image-to-video","familyName":"Wan-I2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880649.html","detailPath":"groups/wan-image-to-video.json"} {"model":"wan2.6-r2v","name":"Wan2.6-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} {"model":"wan2.6-r2v-flash","name":"Wan2.6-R2V-Flash","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Image","Video","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.3"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.5"},{"type":"720P_no_audio","unit":"每秒","price":"0.15"},{"type":"1080P_no_audio","unit":"每秒","price":"0.25"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} {"model":"wan2.7-r2v","name":"Wan2.7-R2V","family":"wan-reference-to-video","familyName":"Wan-R2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3001146.html","detailPath":"groups/wan-reference-to-video.json"} -{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2i-flash","name":"Wan2.2-T2I-Flash","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.2-t2i-plus","name":"Wan2.2-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.5-t2i-preview","name":"Wan2.5-T2I-Preview","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wan2.6-t2i","name":"Wan2.6-T2I","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":1,"count_limit_period":1},"model-default":{"count_limit":1,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx-v1","name":"wanx-t2i","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.16"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} {"model":"wanx-v1-0521","name":"万相-文本生成图像-2024-05-21","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/2712483.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} -{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.0-t2i-turbo","name":"Wan2.0-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.04"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-plus","name":"Wan2.1-T2I-Plus","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} +{"model":"wanx2.1-t2i-turbo","name":"Wan2.1-T2I-Turbo","family":"wan-text-to-image","familyName":"Wan-T2I","provider":"wan","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.14"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2862677.html","detailPath":"groups/wan-text-to-image.json"} {"model":"wan2.2-t2v-plus","name":"Wan2.2-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.14"},{"type":"video_ratio_1080p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.5-t2v-preview","name":"Wan2.5-T2V-Preview","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.3"},{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.6-t2v","name":"Wan2.6-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video","Audio"],"request_modality":["Text","Audio"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"},{"type":"720P_batch","unit":"每秒","price":"0.6"},{"type":"1080P_batch","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wan2.7-t2v","name":"Wan2.7-T2V","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Audio","Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wanx2.1-t2v-plus","name":"Wan2.1-T2V-Plus","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} {"model":"wanx2.1-t2v-turbo","name":"Wan2.1-T2V-Turbo","family":"wan-text-to-video","familyName":"Wan-T2V","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text"]},"prices":[{"type":"video_ratio_480p","unit":"每秒","price":"0.24"},{"type":"video_ratio_720p","unit":"每秒","price":"0.24"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2865250.html","detailPath":"groups/wan-text-to-video.json"} -{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} -{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} +{"model":"wan2.7-videoedit","name":"Wan2.7-VideoEdit","family":"wan-video-edit","familyName":"Wan-VideoEdit","provider":"wan","capabilities":["VG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Image","Text","Video"]},"prices":[{"type":"video_ratio_720p","unit":"每秒","price":"0.6"},{"type":"video_ratio_1080p","unit":"每秒","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":1},"model-default":{"count_limit":5,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3021842.html","detailPath":"groups/wan-video-edit.json"} +{"model":"wanx-background-generation-v2","name":"图像背景生成","family":"wanx-background-generation-v2","familyName":"图像背景生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712497.html","detailPath":"groups/wanx-background-generation-v2.json"} {"model":"wanx-poster-generation-v1","name":"创意海报生成","family":"wanx-poster-generation-v1","familyName":"创意海报生成","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2807172.html","detailPath":"groups/wanx-poster-generation-v1.json"} -{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} -{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} +{"model":"wanx-sketch-to-image-lite","name":"万相-涂鸦作画","family":"wanx-sketch-to-image-lite","familyName":"万相-涂鸦作画","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.06"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712498.html","detailPath":"groups/wanx-sketch-to-image-lite.json"} +{"model":"wanx-style-repaint-v1","name":"人像风格重绘","family":"wanx-style-repaint-v1","familyName":"人像风格重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.12"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712493.html","detailPath":"groups/wanx-style-repaint-v1.json"} {"model":"wanx-virtualmodel","name":"虚拟模特","family":"wanx-virtualmodel","familyName":"虚拟模特","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2796985.html","detailPath":"groups/wanx-virtualmodel.json"} {"model":"wanx-x-painting","name":"万相-图像局部重绘","family":"wanx-x-painting","familyName":"万相-图像局部重绘","provider":"wan","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2797051.html","detailPath":"groups/wanx-x-painting.json"} -{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} -{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} -{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} -{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} -{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} -{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} -{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"wanx2.1-vace-plus","name":"Wan2.1-VACE-Plus","family":"wanx2.1-vace-plus","familyName":"Wan2.1-VACE-Plus","provider":"wan","capabilities":["VG"],"features":[],"inferenceMetadata":{"response_modality":["Video"],"request_modality":["Text","Image","Video"]},"prices":[{"type":"video_ratio","unit":"每秒","price":"0.7"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2922183.html","detailPath":"groups/wanx2.1-vace-plus.json"} +{"model":"wordart-semantic","name":"WordArt锦书-文字变形","family":"wordart-semantic","familyName":"WordArt锦书-文字变形","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text"]},"prices":[{"type":"image_number","unit":"每张","price":"0.24"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712513.html","detailPath":"groups/wordart-semantic.json"} +{"model":"wordart-texture","name":"WordArt锦书-文字纹理生成","family":"wordart-texture","familyName":"WordArt锦书-文字纹理生成","provider":"qwen","capabilities":["IG"],"features":[],"inferenceMetadata":{"response_modality":["Image"],"request_modality":["Text","Image"]},"prices":[{"type":"image_number","unit":"每张","price":"0.08"}],"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712510.html","detailPath":"groups/wordart-texture.json"} +{"model":"xiaomi/mimo-v2.5-pro","name":"xiaomi/mimo-v2.5-pro","family":"xiaomi-models-market-place","familyName":"MiMo文本模型","provider":"xiaomi","capabilities":["TG"],"features":["function-calling","structured-outputs","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"7"},{"type":"output_token","unit":"每百万tokens","price":"21"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.4"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3033942.html","detailPath":"groups/xiaomi-models-market-place.json"} +{"model":"z-image-turbo","name":"Z-Image-Turbo","family":"z-image-turbo","familyName":"Z-Image-Turbo","provider":"qwen-domain-model","capabilities":["IG"],"features":["model-experience"],"inferenceMetadata":{"response_modality":["Text","Image"],"request_modality":["Text"]},"prices":[{"type":"image_standard","unit":"每张","price":"0.1"},{"type":"image_thinking","unit":"每张","price":"0.2"}],"qpmInfo":{"model-default-actual":{"count_limit":2,"count_limit_period":1},"model-default":{"count_limit":2,"count_limit_period":1}},"versionTag":"SNAPSHOT","docUrl":"https://help.aliyun.com/document_detail/3002354.html","detailPath":"groups/z-image-turbo.json"} +{"model":"ZHIPU/GLM-5","name":"ZHIPU/GLM-5","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","cache","structured-outputs","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"6"},{"type":"output_token","unit":"每百万tokens","price":"22"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.5"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3026315.html","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.1","name":"ZHIPU/GLM-5.1","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG"],"features":["function-calling","structured-outputs","cache","prefix-completion"],"contextWindow":204800,"maxInputTokens":204800,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/zh/model-studio/glm-zhipu","detailPath":"groups/zhipu-models-market-place.json"} +{"model":"ZHIPU/GLM-5.2","name":"ZHIPU/GLM-5.2","family":"zhipu-models-market-place","familyName":"智谱GLM系列文本模型","provider":"zhipu-ai","capabilities":["TG","Reasoning"],"features":["function-calling","structured-outputs","prefix-completion","cache"],"contextWindow":1048576,"maxInputTokens":1048576,"maxOutputTokens":131072,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"8"},{"type":"output_token","unit":"每百万tokens","price":"28"},{"type":"input_token_cache","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":200,"count_limit_period":60,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://bailian.console.aliyun.com/cn-beijing?tab=doc#/doc/?type=model&url=3026315","detailPath":"groups/zhipu-models-market-place.json"} diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md index 3b1ff1ad..04d60432 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md @@ -163,6 +163,8 @@ string 默认值为空，此时使用 text-embedding-v2 模型。 +text-embedding-v4 + RerankModelName string @@ -322,7 +324,7 @@ array 否 -创建知识库时可同步导入文件。此处通过指定类目 ID，可导入对应类目下的所有文件（建议导入不超过 10000 个。如有剩余文件，后续可调用 **SubmitIndexAddDocumentsJob** 接口继续导入）。 +创建知识库时可同步导入文件。此处通过指定类目 ID，可导入对应类目下的所有文件（建议导入不超过 500 个。如有剩余文件，后续可调用 **SubmitIndexAddDocumentsJob** 接口继续导入）。 string diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md new file mode 100644 index 00000000..807c00f2 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md @@ -0,0 +1,233 @@ +# AddChunk - 新增切片 + +使用此API可为文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库添加切片。 + +## 接口说明 + +- 对于文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库，本接口可向指定知识库中添加切片内容；目前尚不支持对音视频搜索类（multimedia）知识库进行相关操作。仅当数据源为表格连接器(excel)时，对数据查询与图片问答类型知识库的操作方可生效。 + +- RAM 用户（子账号）需要首先获取阿里云百炼的 [API 权限](https://help.aliyun.com/zh/model-studio/grant-data-access-permission-to-ram-user)（需要`AliyunBailianDataFullAccess`，已包括 sfm:ChunkList 权限点），并[加入一个业务空间](https://help.aliyun.com/zh/model-studio/grant-the-business-space-permission-to-ram-users)后，方可调用本接口。阿里云账号（主账号）可直接调用无须授权。建议您通过最新版[阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)[阿里云百炼 SDK](https://api.alibabacloud.com/api-tools/sdk/bailian?version=2023-12-29)来调用本接口。 + +- 调用本接口前，请确保您的知识库已经创建完成且未被删除（即知识库 ID`IndexId`有效）。 + +- 本接口具有幂等性。 + + +**限流说明：** 本接口频繁调用会被限流，频率请勿超过 10 次/秒。如遇限流，请稍后重试。 + +## 调试 + +[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/AddChunk) + + [![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png) 调试](https://api.aliyun.com/api/bailian/2023-12-29/AddChunk) + +## **授权信息** + +当前API暂无授权信息透出。 + +## 请求语法 + +``` +POST /{WorkspaceId}/chunk/create HTTP/1.1 +``` + +## 路径参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +WorkspaceId + +string + +是 + +工作区标识 + +llm-19hxxxxx7htdf9lh + +## 请求参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +PipelineId + +string + +是 + +知识库 id + +79c0alxxxx + +dataId + +string + +否 + +文件 id + +doc\_xxx + +field + +object + +否 + +插入的切片内容信息，以键值对形式传入。文档搜索类知识库使用固定 key 列表： + +- content（**String**）：**必填**，切片正文内容 + +- title（**String**）**选填**，切片标题 + +- image\_urls（**Array**）：**选填**，切片包含的图片链接，最多 10 张 + + +数据查询类、图片问答类知识库 key 不固定，由该知识库的数据源表格决定：key 为 Excel 列标题，value 为对应列的值。 + +{ "content": "The Bailian platform supports parsing multiple document formats including PDF, Word, and PPT.", "title": "Document Parsing and Chunking", "image\_urls": \[ "https://example.com/images/chunk-flow.png", "https://example.com/images/parsing-result.png" \] } + +any + +否 + +插入切片的表头字段信息，仅数据查询类与图片问答类知识库支持。需要参与检索或参与回复的表头为必填。各类型取值要求： + +- **String 类型** ：最大长度 6000 + +- **时间 类型**：13 位时间戳（毫秒） + +- **Long 类型**：整数，最大 2147483647 + +- **Double 类型**：支持小数 + +- **image\_url 类型**：最多 5 张，多张用英文逗号拼接为一个字符串 + + +{"Product Name": "Wireless Bluetooth Headphones", "Publish Time": 1752624000000, "Stock Quantity": 1580, "Unit Price": 299.99, "image\_url":"https://example.com/images/headphones-front.jpg,https://example.com/images/headphones-side.jpg,https://example.com/images/headphones-package.jpg" } + +## **返回参数** + +**名称** + +**类型** + +**描述** + +**示例值** + +object + +Schema of Response + +RequestId + +string + +请求 id + +35A267BF-xxxx-54DB-8394-AA3B0742D833 + +Code + +string + +错误状态码 + +Index.InvalidParameter + +Message + +string + +错误信息 + +Required parameter(%s) missing or invalid, please check the request parameters. + +Success + +boolean + +接口调用是否成功 + +**枚举值：** + +- true : + + true + +- false : + + false + + +true + +Data + +boolean + +请求成功返回的业务数据 + +**枚举值：** + +- true : + + true + +- false : + + false + + +true + +Status + +string + +接口返回的状态码 + +200 + +## 示例 + +正常返回示例 + +`JSON`格式 + +``` +{ + "RequestId": "35A267BF-xxxx-54DB-8394-AA3B0742D833", + "Code": "Index.InvalidParameter", + "Message": "Required parameter(%s) missing or invalid, please check the request parameters.", + "Success": true, + "Data": true, + "Status": "200" +} +``` + +## 错误码 + +访问[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)查看更多错误码。 + +## **变更历史** + +更多信息，参考[变更详情](https://api.aliyun.com/document/bailian/2023-12-29/AddChunk#workbench-doc-change-demo)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md index 70c29639..8d6507eb 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md @@ -373,3 +373,9 @@ API概述 申请临时文件上传许可 该接口用于高代码部署，其他场景暂不支持。用于申请临时文件上传许可，之后需要自己完成文件上传动作。 + +[AddChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addchunk) + +新增切片 + +使用此API可为文档搜索类（document）、数据查询类（table）、图片问答类（image）知识库添加切片。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md index fa10b5cc..ef5ee20d 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md @@ -78,11 +78,11 @@ **关联操作** -sfm:ChangeParseSetting +sfm:ListCategory -[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) +[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) -update +list \*全部资源 @@ -92,11 +92,11 @@ update 无 -sfm:UpdateFileTag +sfm:GetIndexJobStatus -[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag) +[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) -update +get \*全部资源 @@ -106,11 +106,11 @@ update 无 -sfm:DeleteCategory +sfm:AddCategory -[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) +[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) -delete +create \*全部资源 @@ -120,11 +120,11 @@ delete 无 -sfm:UpdatePromptTemplate +sfm:GetAlipayUrl -[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) +[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) -update +none \*全部资源 @@ -134,11 +134,11 @@ update 无 -sfm:SubmitIndexJob +sfm:DeleteMemoryNode -[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) +[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) -create +delete \*全部资源 @@ -162,11 +162,11 @@ get 无 -sfm:DeleteChunk +sfm:SubmitIndexJob -[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) +[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) -delete +create \*全部资源 @@ -176,11 +176,11 @@ delete 无 -sfm:GetAlipayTransferStatus +sfm:DeleteCategory -[GetAlipayTransferStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipaytransferstatus) +[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) -none +delete \*全部资源 @@ -190,11 +190,11 @@ none 无 -sfm:DeletePromptTemplate +sfm:ListIndexFileDetails -[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) +[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) -delete +list \*全部资源 @@ -204,11 +204,11 @@ delete 无 -sfm:ListIndex +sfm:UpdateConnector -[ListIndices](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindices) +[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) -list +update \*全部资源 @@ -218,11 +218,11 @@ list 无 -sfm:CreateMemoryNode +sfm:ChangeParseSetting -[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) +[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) -create +update \*全部资源 @@ -232,11 +232,11 @@ create 无 -sfm:UpdateMemory +sfm:GetMemory -[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) +[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) -update +get \*全部资源 @@ -246,11 +246,11 @@ update 无 -sfm:ChunkList +sfm:GetAvailableParserTypes -[ListChunks](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listchunks) +[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) -list +get \*全部资源 @@ -260,11 +260,11 @@ list 无 -sfm:ListIndexFileDetails +sfm:UpdateMemory -[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) +[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) -list +update \*全部资源 @@ -274,9 +274,9 @@ list 无 -sfm:SubmitIndexAddDocumentsJob +sfm:CreateMemory -[SubmitIndexAddDocumentsJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexadddocumentsjob) +[CreateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememory) create @@ -288,11 +288,11 @@ create 无 -sfm:UpdateIndex +sfm:ApplyFileUploadLease -[UpdateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateindex) +[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) -update +none \*全部资源 @@ -316,25 +316,11 @@ list 无 -sfm:Retrieve - -[Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) - -none - -\*全部资源 - -`*****` - -无 - -无 - -sfm:AddConnector +sfm:DeleteConnector -[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) +DeleteConnector -create +delete \*全部资源 @@ -344,11 +330,11 @@ create 无 -sfm:ListMemories +sfm:BatchUpdateFileTag -[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) +[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) -list +update \*全部资源 @@ -372,11 +358,11 @@ create 无 -sfm:UpdateConnector +sfm:DeleteChunk -[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) +[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) -update +delete \*全部资源 @@ -386,9 +372,9 @@ update 无 -sfm:DeleteFile +sfm:DeleteMemory -[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) +[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) delete @@ -400,11 +386,11 @@ delete 无 -sfm:AddCategory +sfm:DeletePromptTemplate -[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) +[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) -create +delete \*全部资源 @@ -414,11 +400,11 @@ create 无 -sfm:BatchUpdateFileTag +sfm:ListMemories -[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) +[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) -update +list \*全部资源 @@ -428,11 +414,11 @@ update 无 -sfm:ListIndexFiles +sfm:AddConnector -[ListIndexDocuments](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexdocuments) +[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) -list +create \*全部资源 @@ -442,11 +428,11 @@ list 无 -sfm:ListMemoryNodes +sfm:GetConnector -[ListMemoryNodes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemorynodes) +[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) -list +get \*全部资源 @@ -456,11 +442,11 @@ list 无 -sfm:DescribeFile +sfm:UpdatePromptTemplate -[DescribeFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile) +[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) -none +update \*全部资源 @@ -470,11 +456,11 @@ none 无 -sfm:CreateIndex +sfm:DeleteFiles -[CreateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createindex) +[DeleteFiles](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefiles) -create +delete \*全部资源 @@ -484,11 +470,11 @@ create 无 -sfm:ListCategory +sfm:GetMemoryNode -[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) +[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) -list +get \*全部资源 @@ -498,11 +484,11 @@ list 无 -sfm:AddTable +sfm:ApplyTempStorageLease -[AddTable](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addtable) +[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) -create +none \*全部资源 @@ -512,11 +498,11 @@ create 无 -sfm:UpdateMemoryNode +sfm:DeleteFile -[UpdateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememorynode) +[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) -update +delete \*全部资源 @@ -526,11 +512,11 @@ update 无 -sfm:GetAvailableParserTypes +sfm:AddFilesFromAuthorizedOss -[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) +[AddFilesFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfilesfromauthorizedoss) -get +create \*全部资源 @@ -540,11 +526,11 @@ get 无 -sfm:DeleteMemory +sfm:CreateMemoryNode -[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) +[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) -delete +create \*全部资源 @@ -554,11 +540,11 @@ delete 无 -sfm:DeleteConnector +sfm:UpdateChunk -DeleteConnector +[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) -delete +update \*全部资源 @@ -568,11 +554,11 @@ delete 无 -sfm:UpdateTableFromAuthorizedOss +sfm:ListCategory -[UpdateTableFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatetablefromauthorizedoss) +[ListCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listcategory) -update +list \*全部资源 @@ -582,11 +568,11 @@ update 无 -sfm:ApplyTempStorageLease +sfm:GetIndexJobStatus -[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) +[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) -none +get \*全部资源 @@ -596,9 +582,9 @@ none 无 -sfm:GetConnector +sfm:GetParseSettings -[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) +[GetParseSettings](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getparsesettings) get @@ -610,11 +596,11 @@ get 无 -sfm:DeleteMemoryNode +sfm:GetAlipayUrl -[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) +[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) -delete +none \*全部资源 @@ -624,11 +610,11 @@ delete 无 -sfm:GetMemory +sfm:AddCategory -[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) +[AddCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addcategory) -get +create \*全部资源 @@ -638,11 +624,11 @@ get 无 -sfm:DeleteIndex +sfm:SubmitIndexJob -[DeleteIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindex) +[SubmitIndexJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexjob) -none +create \*全部资源 @@ -652,9 +638,9 @@ none 无 -sfm:UpdateChunk +sfm:ChangeParseSetting -[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) +[ChangeParseSetting](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-changeparsesetting) update @@ -666,9 +652,9 @@ update 无 -sfm:DeleteIndexDocument +sfm:DeleteMemoryNode -[DeleteIndexDocument](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindexdocument) +[DeleteMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememorynode) delete @@ -680,11 +666,11 @@ delete 无 -sfm:GetMemoryNode +sfm:UpdateConnector -[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) +[UpdateConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateconnector) -get +update \*全部资源 @@ -694,9 +680,9 @@ get 无 -sfm:GetIndexJobStatus +sfm:GetMemory -[GetIndexJobStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexjobstatus) +[GetMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemory) get @@ -708,11 +694,11 @@ get 无 -sfm:ApplyFileUploadLease +sfm:ListIndexFileDetails -[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) +[ListIndexFileDetails](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexfiledetails) -none +list \*全部资源 @@ -722,11 +708,25 @@ none 无 -sfm:GetPromptTemplate +sfm:DeleteConnector -[GetPromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getprompttemplate) +DeleteConnector -get +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateMemory + +[UpdateMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememory) + +update \*全部资源 @@ -750,9 +750,79 @@ create 无 -sfm:GetAlipayUrl +sfm:ListFile -[GetAlipayUrl](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipayurl) +[ListFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listfile) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteChunk + +[DeleteChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletechunk) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetAvailableParserTypes + +[GetAvailableParserTypes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getavailableparsertypes) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteCategory + +[DeleteCategory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletecategory) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListMemories + +[ListMemories](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemories) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ApplyFileUploadLease + +[ApplyFileUploadLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applyfileuploadlease) none @@ -764,11 +834,11 @@ none 无 -sfm:CreatePromptTemplate +sfm:BatchUpdateFileTag -[CreatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createprompttemplate) +[BatchUpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-batchupdatefiletag) -create +update \*全部资源 @@ -778,9 +848,9 @@ create 无 -sfm:GetIndexMonitor +sfm:GetConnector -[GetIndexMonitor](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexmonitor) +[GetConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getconnector) get @@ -792,6 +862,48 @@ get 无 +sfm:GetMemoryNode + +[GetMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getmemorynode) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddFile + +[AddFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfile) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddConnector + +[AddConnector](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addconnector) + +create + +\*全部资源 + +`*****` + +无 + +无 + sfm:DeleteFiles [DeleteFiles](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefiles) @@ -806,6 +918,48 @@ delete 无 +sfm:DeleteFile + +[DeleteFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletefile) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeletePromptTemplate + +[DeletePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteprompttemplate) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdatePromptTemplate + +[UpdatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateprompttemplate) + +update + +\*全部资源 + +`*****` + +无 + +无 + sfm:AddFilesFromAuthorizedOss [AddFilesFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addfilesfromauthorizedoss) @@ -820,6 +974,174 @@ create 无 +sfm:CreateMemoryNode + +[CreateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-creatememorynode) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteMemory + +[DeleteMemory](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deletememory) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateChunk + +[UpdateChunk](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatechunk) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ApplyTempStorageLease + +[ApplyTempStorageLease](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-applytempstoragelease) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetIndexMonitor + +[GetIndexMonitor](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getindexmonitor) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:Retrieve + +[Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DeleteIndex + +[DeleteIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindex) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListMemoryNodes + +[ListMemoryNodes](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listmemorynodes) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListIndex + +[ListIndices](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindices) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateMemoryNode + +[UpdateMemoryNode](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatememorynode) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetPromptTemplate + +[GetPromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getprompttemplate) + +get + +\*全部资源 + +`*****` + +无 + +无 + +sfm:CreateIndex + +[CreateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createindex) + +create + +\*全部资源 + +`*****` + +无 + +无 + sfm:ListPromptTemplates [ListPromptTemplates](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listprompttemplates) @@ -834,6 +1156,160 @@ list 无 +sfm:DeleteIndexDocument + +[DeleteIndexDocument](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-deleteindexdocument) + +delete + +\*全部资源 + +`*****` + +无 + +无 + +sfm:SubmitIndexAddDocumentsJob + +[SubmitIndexAddDocumentsJob](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-submitindexadddocumentsjob) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateFileTag + +[UpdateFileTag](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatefiletag) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateTableFromAuthorizedOss + +[UpdateTableFromAuthorizedOss](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updatetablefromauthorizedoss) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ListIndexFiles + +[ListIndexDocuments](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listindexdocuments) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:UpdateIndex + +[UpdateIndex](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-updateindex) + +update + +\*全部资源 + +`*****` + +无 + +无 + +sfm:DescribeFile + +[DescribeFile](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-describefile) + +none + +\*全部资源 + +`*****` + +无 + +无 + +sfm:ChunkList + +[ListChunks](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-listchunks) + +list + +\*全部资源 + +`*****` + +无 + +无 + +sfm:AddTable + +[AddTable](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-addtable) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:CreatePromptTemplate + +[CreatePromptTemplate](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-createprompttemplate) + +create + +\*全部资源 + +`*****` + +无 + +无 + +sfm:GetAlipayTransferStatus + +[GetAlipayTransferStatus](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-getalipaytransferstatus) + +none + +\*全部资源 + +`*****` + +无 + +无 + ## 资源（Resource） 下表是_大模型服务平台百炼_定义的资源，这些资源可以在 RAM 权限策略语句的`Resource`元素中使用，用来授予对该资源执行具体操作的权限。 其中，资源 ARN 是资源在阿里云上的唯一标识。具体说明如下： diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md index 587f5b5c..0d68fb2b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/multimodal-products/product-billing.md @@ -171,7 +171,7 @@ Fun-ASR、通义千问3-ASR-Flash-Realtime 语音合成 -CosyVoice-v3-Plus、通义千问3-TTS 系列 +Qwen-Audio-3.0-TTS-Plus、Qwen-Audio-3.0-TTS-Flash、CosyVoice-v3.5-Plus、CosyVoice-v3.5-Flash、CosyVoice-v3-Plus、通义千问3-TTS 系列 3x diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-createvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-deletevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-getvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-listvocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-hot-word-management/api-contactcenterai-2024-06-03-updatevocab.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md index 3270fff8..8109bacb 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md index 625c3b4e..7a8bebc2 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md index 98330913..dba5d89b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md @@ -33,7 +33,7 @@ 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取Workspace ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md index ed3dbcea..ac8dae76 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md @@ -13,7 +13,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取workspaceId和appId** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md index 6c60f1c8..20bf2df3 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md @@ -11,7 +11,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取**Workspace **ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md index ed7f704a..681142a9 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md @@ -11,7 +11,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 ## **获取**Workspace **ID和App ID** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md index 4d31a8c1..df528f55 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md @@ -8,7 +8,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md index 22d62806..30ba2ffd 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md @@ -6,7 +6,7 @@ - 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures)。 +- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md index 23e8f390..36a8a71c 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md @@ -6,7 +6,7 @@ **重要** -使用子账号开通服务时若出现报错提示，是需要主账号对其子账号进行授予其相应RAM权限。具体权限以及相关操作请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/lingque-ccai-dialogue-analytics-ram-subaccount-usage-and-authorization-procedures) +使用子账号开通服务时若出现报错提示，是需要主账号对其子账号进行授予其相应RAM权限。具体权限以及相关操作请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis) - 路径：进入[应用广场](https://bailian.console.aliyun.com/#/app-market)→点击[全部应用](https://bailian.console.aliyun.com/cn-beijing#/app-market/all)→选择**伶鹊CCAI-对话分析AIO** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/technology-integration-scheme.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceengines.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-updatevoiceaccessprofile.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-createvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-updatevariable.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/product-0verview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md index f5d39958..82fecbfe 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-changeset.md @@ -36,7 +36,7 @@ OpenAPI 错误码发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更。 @@ -70,7 +70,7 @@ OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 [查看API文档](https://api.aliyun.com/document/AnyTrans/2025-07-07/BatchTranslate) -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) OpenAPI 错误码发生变更、OpenAPI 返回结构发生变更。 @@ -230,7 +230,7 @@ OpenAPI 名称 操作 -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) 新增 OpenAPI。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md index 811877bb..fb6d78c0 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-ram.md @@ -136,7 +136,7 @@ none anytrans:BatchTranslateForHtml -[BatchTranslateForHtml](https://help.aliyun.com/zh/model-studio/api-anytrans-2025-07-07-batchtranslateforhtml) +[BatchTranslateForHtml](https://help.aliyun.com/zh/document_detail/3037877.html) none diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md deleted file mode 100644 index ae184fa3..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/guidelines-for-use.md +++ /dev/null @@ -1,179 +0,0 @@ -# 使用指南 - -本篇文档主要介绍通义晓蜜对话Agent中对话Agent构建和可视化流程技能编排的使用指南。 - -## **1\. 开通产品** - -- 路径：[阿里云百炼-应用广场-通义晓蜜对话Agent](https://bailian.console.aliyun.com/?spm=5176.29619931.J__Z58Z6CX7MY__Ll8p1ZOR.1.74cd521cS5IvE8#/app/app-market/beebot)。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935960.png) - -- 进入晓蜜对话Agent应用操作控制台，在右上角点击**免费开通**。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935961.png) - -- 进入通义晓蜜对话Agent的开通界面，开通后，按实际调用量收费，即按日生成账单在阿里云账户余额中扣除。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932056.png) - -## **2\.** Agent构建 - -### **2.1 创建应用** - -点击**创建应用**，输入应用名称和应用描述，即可完成基础的Agent创建。 - -- 应用名称：按照业务需求填写Agent的名称； - -- 应用描述：按照业务应用场景添加Agent的描述。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0220563471/p935962.png) - -创建完成后的操作控制台： - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6538449471/p965131.png) - -### **2.2 配置界面** - -#### **1\. 人设** - -- 定义：在机器人空间，每个机器人可以添加人设，即通过提示词的方式调试机器人，可以在人设中对机器人的角色、工作流、技能、限制要求等内容在人设中作对应的定义要求。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932089.png) - -- 选择模板：您可在**选择模板**窗口，插入通用模板，按照模板要求输入对应的人设限制。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932111.png) - -- 优化：当您编辑完人设后，可以点击**优化**，模型会根据您的输入，优化人设内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932123.png) - - -#### **2\. 机器人开场白** - -- 定义：指您打开机器人对话框时，机器人发起对话的开场白。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932121.png) - - -#### **3\. 知识库（流程技能、高频问答知识）** - -- 定义：可绑定当前Agent对应的流程技能和高频问答知识。流程技能的具体的配置步骤可参考《[3\. 可视化流程技能编排](#a7e7986fc4k9q)》，高频问答知识的配置步骤可参考《[4\. 高频问答知识的配置](#d0b14121834k6)》。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965135.png) - - -#### **4\. 更多设置** - -- 模型：当前内置**通义晓蜜大模型**，其中您可更改温度系数和上下文轮次来修改模型 - -- 通识知识： - -- 回复语言：指机器人回复的语言 - - - 与用户语言相同：指机器人回复的语言根据您提问的语言进行回复； - - - 仅中文：指无论您用什么语言提问，机器人都用中文回复； - - - 仅英文：指无论您用什么语言提问，机器人都用英文回复。 - -- 安全： - - - 安全拦截：当系统检查到输入和输出内容涉及到“答案敏感词”时，自动回复“敏感词话术”； - - - 安全预设话术：用户问题包含“用户敏感词”或模型生成回复包含敏感内容时，机器人自动使用敏感回复话术进行回复。默认安全预设话术为：“您说的这个问题我不能回答，您可以尝试询问其他问题”。 - -- 模型生成异常：当系统发生异常或服务超时情况下，机器人兜底回复话术。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932146.png) - -### **2.3 调用量界面** - -在调用量界面可以查看token的消耗量。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932315.png) - -### **2.4 API界面** - -对话Agent的调试信息会同步到API接口中，您可复制当前代码进行调用。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932320.png) - -## **3\.** 可视化流程技能编排 - -### **3.1 新建流程** - -1. 点击**知识库管理>流程**或者**+绑定流程**，打开流程列表窗口。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932331.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932343.png) - -2. 点击**新建流程**，打开新建流程创建。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932344.png) - -3. 按照要求填写对应流程**名称**和**描述**，点击**确认**，即可创建空白流程 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932352.png) - -4. 找到新建的流程，点击**编辑**，进入流程编排页面。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932353.png) - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2963872471/p932356.png) - -5. 新建开启分支，进入流程画布，在开始节点后，添加“分支”设置触发进入流程后继分支的条件。具体的操作步骤可查看下面的小视频： - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931138.gif) - -6. 根据对话逻辑选择节点编排对话流程。 - - ![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931139.png) - - -**重要** - -流程编排完成后，点击**测试**，机器人将自动进行流程完整性检测，如果错误，请根据错误提示优化流程。具体情况可参考如下小视频： - -![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931140.gif) - -### **3.2 流程调试** - -1. 调试前需对环境进行设置 - - -- 服务模拟：流程内的 API 插件会直接使用 mock 值进行返回，适用于 API 还没有准备好的情况。 - -- 随路参数：在用户发送问题时，同时带给机器人的外部参数，如电话接通时，可以将用户呼入号码以随路参数传递给机器人，后续在 API 插件调用时可以使用该参数。 - - -![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931141.png) - -2. 对话调试 - - 直接进行对话，机器人回复后，可以点击**生成完成**查看机器人输出的内容，针对参数收集可以查看到机器人收集到的具体参数信息。 - - ![image.gif](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8133542471/p931142.gif) - - -## **4**. 高频问答知识的配置 - -### **4.1 新建高频问答知识** - -1. 点击**知识库管理>高频问答**或者**+绑定高频问答**，打开高频问答列表窗口。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965149.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965150.png) - -2. 点击**创建高频问答库**，打开创建窗口，填写高频问答库名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965152.png) - -3. 找到新建的高频问答库，点击编辑，进入添加高频问答知识的页面。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965154.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965155.png) - -4. 点击**新增高频问答**或者**导入高频问答**，在高频问答库中添加对应的高频问题，按照要求填写问题、答案类型、问题答案、生效时间、相似问法，点击**提交**，即可新增成功。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965159.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965160.png) - - 新增成功后：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965161.png) - - -### **4.2 绑定高频问答知识库及测试效果** - -1. 点击**绑定**，即可将对应知识库绑定到机器人上。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965162.png) - - -绑定成功展示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965165.png) - -2. 在左侧对话测试窗测试效果，如下图所示：![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5538449471/p965166.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md deleted file mode 100644 index 7ef2855d..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md +++ /dev/null @@ -1,117 +0,0 @@ -# API概览 - -## **API标准及多语言预置SDK** - -本产品（`ContactCenterAI/2024-06-03`）的OpenAPI采用[ROA](https://help.aliyun.com/zh/sdk/product-overview/roa-mechanism)签名风格。我们已经为开发者封装了常见编程语言的SDK，开发者可通过[下载SDK](https://api.aliyun.com/api-tools/sdk/ContactCenterAI?version=2024-06-03)直接调用本产品OpenAPI而无需关心技术细节。如果现有SDK不能满足使用需求，可通过签名机制进行自签名对接。由于自签名细节非常复杂，需花费 5个工作日左右。因此建议加入我们的服务钉钉群（147535001692），在专家指导下进行签名对接。 - -在使用API前，您需要准备好身份账号及访问密钥（AccessKey），才能有效通过客户端工具（SDK、CLI等）访问API。细节请参见[获取AccessKey](https://help.aliyun.com/zh/ram/user-guide/create-an-accesskey-pair)。 - -## **自定义签名场景** - -若您的业务场景有特殊需求，需通过自签名方式对接 API，建议优先咨询我们的技术支持团队（服务钉钉群：147535001692），获取专业指导以确保高效接入。 - -## **账号与安全准备** - -阿里云账号具备对所有资源的完全管理权限。一旦 AccessKey 泄露，所有相关资源都将面临未经授权访问的风险。为确保安全，建议创建一个仅具备 API 访问权限的[RAM用户](https://help.aliyun.com/zh/ram/user-guide/create-a-ram-user)并配置其 AccessKey，同时基于最小权限原则 (PoLP) 配置 RAM 策略。仅在明确需要阿里云账号权限的特定场景下，才使用阿里云账号。 - -## API目录 - -API - -标题 - -API概述 - -[RunCompletion](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletion) - -通过模版ID调用通义晓蜜CCAI-对话分析AIO应用 - -支持调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[RunCompletionMessage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-runcompletionmessage) - -使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -支持以Message协议格式调用通义晓蜜CCAI-对话分析AIO应用获取对话摘要、关键信息抽取、质检结果、对话分析结果，应用调用支持 HTTP 调用来完成客户的响应，目前提供普通 HTTP 和 HTTP SSE 两种协议，您可根据自己的需求自行选择。 - -[AnalyzeConversation](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation) - -通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -获取对话摘要、标题生成、关键词、字段信息抽取、问题及解决方案、服务质检、代办事项、满意度、情绪检测、QA抽取、用户画像、标签分类等对话分析结果，应用调用支持 HTTP 调用来完成客户的响应。 - -[GetTaskResult](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-gettaskresult) - -通过任务ID获取离线任务分析结果 - -通过任务ID获取离线任务对话分析结果。应用调用支持 HTTPS调用来完成客户的响应。 - -[CreateTask](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask) - -通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -通过创建离线异步任务，进行对话分析。应用调用支持 HTTP 调用来完成客户的响应。 - -[AnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeimage) - -图片内容分析 - -通过通义晓蜜CCAI-对话分析AIO应用进行图片内容分析。具体包括以下场景：水印检测。应用调用支持 HTTP 调用来完成客户的响应。 - -[GeneralAnalyzeImage](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-generalanalyzeimage) - -通用图片分析 - -通用图片分析。 - -## 热词管理 - -API - -标题 - -API概述 - -[CreateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createvocab) - -创建热词 - -将一组语音热词上传到服务端，并获取返回热词ID。 - -[UpdateVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-updatevocab) - -修改热词 - -根据词表的ID可以更新对应的词表信息，包括词表名称、词表描述信息、词表的词和权重。 - -[ListVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-listvocab) - -获取热词列表 - -列举指定业务空间下的热词列表信息。 - -[DeleteVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-deletevocab) - -删除热词 - -根据词表的ID删除对应的词表。 - -[GetVocab](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-getvocab) - -获取热词 - -根据词表的ID获取对应的词表信息。 - -## 不推荐或白名单开放 - -API - -标题 - -API概述 - -[AnalyzeAudioSync](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeaudiosync) - -语音文件实时分析 - -对进行语音文件进行实时对话分析。应用调用支持 HTTPS 调用来完成客户的响应。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md deleted file mode 100644 index fe169264..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md +++ /dev/null @@ -1,141 +0,0 @@ -# 字段信息抽取最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行字段信息抽取的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的信息抽取能力，对客服和用户的对话记录（文本、录音文件）进行理解、识别、抽取，如抽取客服工单中的字段信息，如客户所在的地区信息、年龄、日期时间、办理事项等，提升工单填写效率。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行字段信息抽取，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和属性描述，进行属性抽取。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4924970771/CAEQURiBgICV77fjnRkiIDlmNTk4MWEyZDJhYzQyZjBhNjY0ZjVjYmRlNjc1MzA54811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - - - -com.alibaba - -fastjson - -2.0.58 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // 抽取字段名称和字段描述 - List fieldList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field1 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field1.setName("问题类型"); - field1.setDesc("客户咨询的问题类型"); - fieldList.add(field1); - AnalyzeConversationRequest.AnalyzeConversationRequestFields field2 = new AnalyzeConversationRequest.AnalyzeConversationRequestFields(); - field2.setName("公司名称"); - field2.setDesc("客服所属的保险公司名称"); - fieldList.add(field2); - - request.setFields(fieldList); - - // fields表示属性抽取任务 - request.setResultTypes(Arrays.asList("fields")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md deleted file mode 100644 index ac8dcf19..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md +++ /dev/null @@ -1,138 +0,0 @@ -# 客服服务质检最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行客服服务质检的最佳实践。 - -## **应用场景** - -通过通义晓蜜CCAI-AIO的服务质检能力，分析客服和用户的对话记录（文本、录音文件）、发现客服的服务质量问题，进而提升客服服务效率、服务规范，提升客户体验。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行服务质检，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息，对话内容和质检项，进行对话分析。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0924970771/CAEQURiBgIDOp7TjnRkiIDUzYjVkNGQ4OTdjZDQ1MjliYWVmMjk5MjEzNzczYTNm4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - AnalyzeConversationRequest request = new AnalyzeConversationRequest(); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection serviceInspection = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspection(); - List inspectionContents = new ArrayList<>(); - - // 质检项定义 - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content1 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content1.setTitle("客服是否过度承诺"); - content1.setContent("客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。"); - inspectionContents.add(content1); - - AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents content2 = new AnalyzeConversationRequest.AnalyzeConversationRequestServiceInspectionInspectionContents(); - content2.setTitle("客户情绪是否正向"); - content2.setContent("分析对话内容，输出用户在对话中表现出的情绪，详细要求：a. 当客户表现出负面情绪时，判定为消极；b. 当客户表现中积极情绪时，判定为积极；c. 如果客户文本没有明显的消极或积极情感色彩，则判定为中性。"); - inspectionContents.add(content2); - - - serviceInspection.setInspectionContents(inspectionContents); - serviceInspection.setInspectionIntroduction("请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等"); - serviceInspection.setSceneIntroduction("保险销售场景"); - - request.setServiceInspection(serviceInspection); - // service_inspection表示服务质检任务 - request.setResultTypes(Arrays.asList("service_inspection")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md deleted file mode 100644 index fabe48ad..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md +++ /dev/null @@ -1,118 +0,0 @@ -# 摘要生成（含摘要/标题/关键词）最佳实践 - -本文向您介绍一个通过通义晓蜜CCAI-AIO对话分析进行摘要总结的最佳实践。 - -## 应用场景 - -通过通义晓蜜CCAI-AIO的总结摘要能力自动提取文档中的重要信息，如对话摘要总结、标题生成、关键词等，从而提高工作效率，减少人工处理成本。 - -## **方案概览** - -使用通义晓蜜CCAI-AIO对话分析进行摘要总结，只需几步： - -1. 开通阿里云百炼服务：首先我们需要开通阿里云百炼服务，开通调用服务后才能测试模型体验、调用模型或应用体验服务。 - -2. 开通并创建通义晓蜜CCAI-AIO对话分析应用：通过阿里云百炼创建一个通义晓蜜CCAI-AIO对话分析应用，并获取调用通义晓蜜CCAI-AIO对话分析应用 API 的相关凭证。 - -3. 基于API实现对话分析：安装SDK，填充API中应用信息和对话内容，进行摘要总结。 - - -## **方案架构** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8924970771/CAEQURiBgMCMwbTpnRkiIGY5ODI0YzFiZjA5MTQ2MjhhNGUzYTVmNThjYWQyODdl4811491_20241206164318.181.svg) - -## **开通阿里云百炼服务** - -开通阿里云百炼服务：请参考[产品开通](https://help.aliyun.com/zh/model-studio/activate-alibaba-cloud-model-studio)。 - -## **开通并创建通义晓蜜CCAI-AIO对话分析应用** - -开通并创建通义晓蜜CCAI-AIO对话分析并创建应用，请参考[使用指南](https://help.aliyun.com/zh/model-studio/tongyi-xiaomi-ccai-aio-user-guide/)。 - -## **获取AccessKeyID和AccessKeySecret** - -如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -## **获取Workspace ID和App ID** - -获取Workspace ID和App ID，请参考[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)。 - -## 安装SDK - -## Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception { - //建议用户配置env防止ak泄漏 - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - // 对话内容 - List sentenceList = new ArrayList<>(); - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("您好，这里是xxx保险公司，请问有什么可以帮您"); - sentenceList.add(sentences1); - - AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("嗯，我想办理一个健康险，帮我介绍下有哪些"); - sentenceList.add(sentences2); - - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-adslsddxxxx"); - request.setDialogue(dialogue); - - // summary 表示总结摘要任务 - // keywords 表示关键词抽取 - // title 表示抽取标题 - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md deleted file mode 100644 index 592b3229..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md +++ /dev/null @@ -1,563 +0,0 @@ -# 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用 - -本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - -- 关于任务类型，请参见[通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-analyzeconversation)中resultTypes字段的描述。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取workspaceId和appId** - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935587.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6398853471/p935588.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么领取五险一金呢").build()); - - List resultTypes=new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue=AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.analyzeConversationWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - AnalyzeConversationResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## 同步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest request = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest(); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue dialogue = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogue(); - List sentenceList = new ArrayList<>(); - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences1 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问您想咨询五险一金哪方面的问题呢"); - sentenceList.add(sentences1); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences sentences2 = new com.aliyun.contactcenterai20240603.models.AnalyzeConversationRequest.AnalyzeConversationRequestDialogueSentences(); - sentences2.setRole("user"); - sentences2.setText("怎么查询账户呢"); - sentenceList.add(sentences2); - dialogue.setSentences(sentenceList); - dialogue.setSessionId("session-1111"); - request.setDialogue(dialogue); - - request.setSceneName("中国移动"); - request.setResultTypes(Arrays.asList("summary")); - request.setStream(false); - - com.aliyun.contactcenterai20240603.models.AnalyzeConversationResponse response = client.analyzeConversation(workspaceId, appId, request); - System.out.println(JSONObject.toJSONString(response)); - } - -} -``` - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": false, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - events, err := util.ReadAsString(resp["body"].(io.ReadCloser)) - - if err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - - fmt.Println(tea.StringValue(events)) - - return nil -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` - -## 异步非流式调用 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - - List messageList = new ArrayList<>(); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("agent").text("请问您想咨询五险一金哪方面的问题呢").build()); - messageList.add(AnalyzeConversationRequest.Sentences.builder().role("user").text("怎么查询账户呢").build()); - - List resultTypes = new ArrayList<>(); - resultTypes.add("summary"); - - AnalyzeConversationRequest.Dialogue dialogue = AnalyzeConversationRequest.Dialogue.builder().sessionId("session-01") - .sentences(messageList).build(); - - AnalyzeConversationRequest completionParam = AnalyzeConversationRequest.builder().modelCode("tyxmPlus").resultTypes(resultTypes) - .workspaceId(workspaceId).appId(appId).dialogue(dialogue).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture x = client.analyzeConversation(completionParam); - AnalyzeConversationResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL***********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - - } -} -``` - -## 同步流式调用 - -Go - -``` -// 示例代码，其中阿里云AK、SK，CCAI的业务空间ID（workspaceId）和应用ID（appId），替换为用户当前的。 -// tea-utils使用这个版本，go get github.com/alibabacloud-go/tea-utils/v2@v2.0.5-0.20240708091240-f3d7eca052de - -// This file is auto-generated, don't edit it. Thanks. -package main - -import ( - "fmt" - "io" - "os" - - openapi "github.com/alibabacloud-go/darabonba-openapi/v2/client" - openapiutil "github.com/alibabacloud-go/openapi-util/service" - util "github.com/alibabacloud-go/tea-utils/v2/service" - "github.com/alibabacloud-go/tea/tea" -) - -/** - * API 相关 - * @param path params - * @return OpenApi.Params - */ -func CreateApiInfo() (_result *openapi.Params) { - params := &openapi.Params{ - // 接口名称 - Action: tea.String("AnalyzeConversation"), - // 接口版本 - Version: tea.String("2024-06-03"), - // 接口协议 - Protocol: tea.String("HTTPS"), - // 接口 HTTP 方法 - Method: tea.String("POST"), - AuthType: tea.String("AK"), - Style: tea.String("ROA"), - // 接口 PATH - Pathname: tea.String("/YOUR_CCAI_WORKSPACEID/ccai/app/YOUR_CCAI_APP_ID/analyze_conversation"), - // 接口请求体内容格式 - ReqBodyType: tea.String("json"), - // 接口响应体内容格式，注意一定得是binary格式，CallApi才会透传出response body进行ReadAsSSE - BodyType: tea.String("binary"), - } - _result = params - return _result -} - -func _main(args []*string) (_err error) { - // 工程代码泄露可能会导致 AccessKey 泄露，并威胁账号下所有资源的安全性。以下代码示例仅供参考。 - // 建议使用更安全的 STS 方式，更多鉴权访问方式请参见：https://help.aliyun.com/document_detail/378661.html。 - config := &openapi.Config{ - AccessKeyId: tea.String("YOUR_ALIYUN_AK"), - AccessKeySecret: tea.String("YOUR_ALIYUN_SK"), - } - config.Endpoint = tea.String("contactcenterai.cn-shanghai.aliyuncs.com") - client, err := openapi.NewClient(config) - if err != nil { - return err - } - - params := CreateApiInfo() - // query params - queries := map[string]interface{}{} - queries["workspaceId"] = tea.String("YOUR_CCAI_WORKSPACEID") - queries["appId"] = tea.String("YOUR_CCAI_APP_ID") - - body := map[string]interface{}{ - "dialogue": map[string]interface{}{ - "sentences": []map[string]*string{map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("您好，请问有什么问题需要解决"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("怎么领取游戏币呢"), - }, map[string]*string{ - "role": tea.String("agent"), - "text": tea.String("请登录个人账号，在账号下，看看是否有推荐的待领取的游戏币呢，如果有会推送给您的"), - }, map[string]*string{ - "role": tea.String("user"), - "text": tea.String("好，谢谢"), - }}, - "sessionId": "2323", - }, - "modelCode": "tyxmPlus", - "resultTypes": []*string{tea.String("question_solution")}, - "stream": true, - } - - // runtime options - runtime := &util.RuntimeOptions{} - request := &openapi.OpenApiRequest{ - Query: openapiutil.Query(queries), - Body: body, - //Body: tea.String("{\n \"stream\": false,\n \"modelCode\": \"tyxmPlus\",\n \"dialogue\": {\n \"sentences\": [\n {\n \"role\": \"user\",\n \"text\": \"号主开挂了模拟宇宙无线祝福\\n联系方式: ******\\n图片上传:\\n视频上传:\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"乘客您好，欢迎登录本次星穹列车帕~麻烦您提供一下以下信息：*游戏项目：\\n*被举报角色UID：\"\n },\n {\n \"role\": \"user\",\n \"text\": \"通行证id******\\n\\n2024-06-10 18:25:52 [玩家] ***:\\nuid******\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题我们之前已经记录反馈了，会进行核实的~如有结果我们会在服务进度或在线服务中告知，您可以留意相关提示。十分抱歉给您带来不便\\n\"\n },\n {\n \"role\": \"agent\",\n \"text\": \"您的问题咨询完成啦，那客服娘贴心提示，不要忘记消耗开拓力哦~祝愿您在完成探索的途中获得美好的回忆哦~希望您抽空也记得给客服娘进行下评价，挥挥~~\\n\"\n }\n ],\n \"sessionId\": \"ss01\"\n },\n \"resultTypes\": [\n \"question_solution\"\n ],\n \"serviceInspection\": {\n \"inspectionIntroduction\": \"请检测客服是否存在服务不当的行为，包括：过度承诺、故意套取客户隐私信息等\",\n \"sceneIntroduction\": \"保险销售场景\",\n \"inspectionContents\": [\n {\n \"title\": \"客服是否过度承诺\",\n \"content\": \"客服在服务客户过程中，基于已有的服务标准是否存在过度承诺的行为，如：最快到货时间是12小时，无法给客户承诺更快的到货时间。\"\n }\n ]\n },\n \"fields\": [\n {\n \"code\": \"name\",\n \"name\": \"姓名\",\n \"desc\": \"用户的姓名\"\n },\n {\n \"code\": \"question\",\n \"name\": \"问题\",\n \"desc\": \"用户的问题\"\n }\n ]\n}"), - } - - // 复制代码运行请自行打印 API 的返回值 - // 返回值为 Map 类型，可从 Map 中获得三类数据：响应体 body、响应头 headers、HTTP 返回的状态码 statusCode。 - resp, err := client.CallApi(params, request, runtime) - if err != nil { - return err - } - - fmt.Println(resp["headers"]) - fmt.Println(resp["statusCode"]) - - // 迭代读取SSE内容 - eventChan, errChan := util.ReadAsSSE(resp["body"].(io.ReadCloser)) - for { - select { - case event, ok := <-eventChan: - if !ok { - return nil - } - fmt.Println("-------------------------------------") - fmt.Printf("Event ID: %s, Event name: %s, Data: %s\n", event.ID, event.Event, event.Data) - case err, ok := <-errChan: - if ok && err != nil { - fmt.Printf("Error: %v\n", err) - return err - } - } - } - -} - -func main() { - err := _main(tea.StringSlice(os.Args[1:])) - if err != nil { - panic(err) - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md deleted file mode 100644 index ed6cafa1..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md +++ /dev/null @@ -1,225 +0,0 @@ -# 通过通义晓蜜CCAI-对话分析AIO应用进行图片分析 - -本文向您介绍一个通过通义晓蜜CCAI-对话分析AIO应用进行图片分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 同步非流失调用 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - AnalyzeImageRequest request = new AnalyzeImageRequest(); - request.setStream(false); - request.setResultTypes(Arrays.asList("watermark")); - List imageList = new ArrayList<>(); - imageList.add("http://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - request.setImageUrls(imageList); - AnalyzeImageResponse response=client.analyzeImage(workspaceId,appId,request); - System.out.println(response); - } -} -``` - -## 异步非流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(false).imageUrls(imageList).build(); - CompletableFuture x = client.analyzeImage(request); - AnalyzeImageResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println("ALL**********************"); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -## 异步流式调用 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - List imageList = new ArrayList<>(); - imageList.add("https://img.alicdn.com/imgextra/i3/O1CN01sRvtsv1WKi6WlKiiP_!!6000000002770-0-tps-1024-1024.jpg"); - AnalyzeImageRequest request = AnalyzeImageRequest.builder().appId(appId).workspaceId(workspaceId) - .resultTypes(Arrays.asList("watermark")).stream(true).imageUrls(imageList).build(); - ResponseIterable x = client.analyzeImageWithResponseIterable(request); - ResponseIterator iterator = x.iterator(); - String lastTxt = ""; - while (iterator.hasNext()) { - AnalyzeImageResponseBody event = iterator.next(); - lastTxt = event.getText(); - System.out.println(JSON.toJSONString(event)); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - } -} -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md deleted file mode 100644 index 7dcbd086..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md +++ /dev/null @@ -1,197 +0,0 @@ -# 通过上传离线任务数据进行通义晓蜜CCAI-对话分析 - -本文向您介绍一个通过上传离线任务数据进行通义晓蜜CCAI-对话分析的最佳实践。 - -- 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 关于各API的详细出入参说明，请参见[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 前提条件 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - - -## **获取**Workspace **ID和App ID** - -### **Workspace ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935590.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 在业务空间管理列表中获取的Workspace ID为入参中workspaceId。 - - -### **App ID** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1229853471/p935591.png) - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**页面，点击**应用实践**。 - -2. 在应用实践列表中找到点击通义晓蜜CCAI-对话分析AIO的**立即查看**。 - -3. 点击上方**我的应用**，展示应用卡片列表。 - -4. 每个卡片上的应用ID即为需要获取的App ID。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 代码示例 - -**说明** - -请用已获取的Workspace ID替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，App ID替换示例中的YOUR\_APPID，代码才能正常运行。为防止密钥泄露，建议将AccessKeyID和AccessKeySecret设置为环境变量。 - -## 创建语音任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - - request.setTaskType("audio"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskRequest.CreateTaskRequestTranscription transcription=new CreateTaskRequest.CreateTaskRequestTranscription(); - transcription.setFileName("***.mkv"); - transcription.setVoiceFileUrl("https://***.oss-cn-beijing.aliyuncs.com/****/***.mkv"); - request.setTranscription(transcription); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - -} -``` - -## 创建文本任务 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - String workspaceId = "YOUR_WORKSPACEID"; - String appId = "YOUR_APPID"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - CreateTaskRequest request=new CreateTaskRequest(); - - CreateTaskRequest.CreateTaskRequestDialogue dialogue = new CreateTaskRequest.CreateTaskRequestDialogue(); - - List sentences = new ArrayList<>(); - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences1 = new CreateTaskRequest.CreateTaskRequestDialogueSentences(); - sentences1.setRole("agent"); - sentences1.setText("请问有什么事，你什么性别,胖不胖"); - sentences.add(sentences1); - - CreateTaskRequest.CreateTaskRequestDialogueSentences sentences2 = new CreateTaskRequest.CreateTaskRequestDialogueSentences (); - sentences2.setRole("user"); - sentences2.setText("我要买保险，我是男的，很瘦"); - sentences.add(sentences2); - dialogue.setSentences(sentences); - dialogue.setSessionId("sessionId-01"); - - request.setDialogue(dialogue); - - request.setTaskType("text"); - request.setResultTypes(Arrays.asList("summary")); - request.setModelCode("tyxmPlus"); - - CreateTaskResponse response=client.createTask(workspaceId,appId,request); - System.out.println(JSONObject.toJSONString(response.getBody())); - } - - -} -``` - -## 获取任务结果 - -``` -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import lombok.extern.slf4j.Slf4j; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; - -public class CcaiPaasTest { - - public static void main(String[] args) throws Exception{ - String accessKeyId = "YOUR_ACCESS_KEY_ID"; - String accessKeySecret = "YOUR_ACCESS_KEY_SECRET"; - - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - String taskId = "*****-****-****-*****-****"; - GetTaskResultRequest request = new GetTaskResultRequest(); - request.setTaskId(taskId); - GetTaskResultResponse response = client.getTaskResult(request); - System.out.println(JSONObject.toJSONString(response)); - } -} -``` - -## **相关文档** - -关于任务类型，请参[通过上传离线任务数据进行通义晓蜜CCAI-对话分析](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-createtask)见中resultTypes字段的描述。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 75449a68..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,385 +0,0 @@ -# 通过原生Prompt调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用Java SDK的安装、使用及注意事项。 - - 关于Java SDK的更多说明，请参见[开始使用](https://help.aliyun.com/zh/sdk/developer-reference/get-started-with-alibaba-cloud-classic-sdk-for-java)。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## 接口入参位置 - -### **workspaceId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935585.png) - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9998853471/p935586.png) - -1. 访问[应用广场应用实践](https://bailian.console.aliyun.com/#/app-market/lightApplication)页面，选择**通义晓蜜CCAI-对话分析AIO**，单击**立即查看**。 - -2. 单击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -## 安装SDK - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## **异步流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - - ResponseIterable x = client.runCompletionMessageWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionMessageResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; - -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - - public static void main(String[] args) throws Exception{ - //Prompt - List messageList = new ArrayList<>(); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("system").content("You are a helpful assistant.").build()); - messageList.add(RunCompletionMessageRequest.Messages.builder().role("user").content("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。").build()); - - RunCompletionMessageRequest completionParam = RunCompletionMessageRequest.builder() - .workspaceId(workspaceId).appId(appId).messages(messageList).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - - CompletableFuture generateCompletionResponseCompletableFuture = client.runCompletionMessage(completionParam); - RunCompletionMessageResponse generateCompletionResponse = generateCompletionResponseCompletableFuture.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getText())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getFinishReason())); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody().getRequestId())); - } -} -``` - -Python - -``` -import asyncio - -from alibabacloud_contactcenterai20240603.client import Client - -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import (RunCompletionRequestDialogueSentences, - RunCompletionRequestDialogue, - RunCompletionRequestFields, - RunCompletionRequest) -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -async def run_async_sse(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = await client.run_completion_message_async(workSpace, appId, request) - body = response.body - print(body) - -if __name__ == '__main__': - asyncio.run(run_async_sse()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; - -public class CcaiPaasTest { - - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String workspaceId="YOUR_WORKSPACEID"; - private static String appId="YOUR_APPID"; - - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - - Client client = new Client(config); - - //Prompt - RunCompletionMessageRequest request = new RunCompletionMessageRequest(); - List messageList = new ArrayList<>(); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message1 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message1.setRole("system").setContent("You are a helpful assistant."); - RunCompletionMessageRequest.RunCompletionMessageRequestMessages message2 = new RunCompletionMessageRequest.RunCompletionMessageRequestMessages(); - message2.setRole("user").setContent("请阅读以下对话内容，按照要求执行指令任务。\\n```\\n\\n客服:你自己的话\\n客户:你自己要号手机\\n客户:31\\n客服:他是1公分\\n客户:那年后，我看\\n客户:后来点开看\\n客户:嗯，要不是以前的\\n客户:这不看你看\\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\\n客服:你在这跟到后面来\\n客服:这里不是大多少天都会员了吗\\n客户:那你不是大于多少天都结了吗\\n客服:嗯\\n客户:嗯\\n客服:我们有1个\\n客户:就是你这个\\n客服:他说你这里面也按照\\n客户:他有多少是蒙细别的吗\\n客户:这3天之后\\n客服:当前执行的\\n客服:这些，其实你这个也是\\n客户:那你这个颜色\\n客服:按照他给了预期了，是不是结多少天\\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\\n客服:那是什么问题\\n客服:你3天的时候关内\\n客户:你要是原来那个\\n客户:你基本上就那你要看\\n客服:你要是原来的，你，你基本上都要不要\\n客服:5天\\n客户:对的\\n客户:啊，关于\\n客户:15新啊\\n客户:上面录的\\n\\n```\\n任务指令如下：\\n```\\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\\n```\\n请依据上述指令，确保准确无误地完成任务。"); - messageList.add(message1); - messageList.add(message2); - - request.setMessages(messageList); - request.setStream(false); - request.setModelCode("tyxmTurbo"); - - RunCompletionMessageResponse responseBody = client.runCompletionMessage(workspaceId, appId, request); - RunCompletionMessageResponseBody body = responseBody.getBody(); - System.out.println(JSON.toJSONString(body)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client - -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages - -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" - -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - - client = Client(config) - - # Prompt - role = "system" - content = "You are a helpful assistant." - requestMessage1 = RunCompletionMessageRequestMessages(content, role) - role = "user" - content = "请阅读以下对话内容，按照要求执行指令任务。\n```\n\n客服:你自己的话\n客户:你自己要号手机\n客户:31\n客服:他是1公分\n客户:那年后，我看\n客户:后来点开看\n客户:嗯，要不是以前的\n客户:这不看你看\n客服:那年后我是按后来改成按执行的时间延后了，要不以前的时间延后，你根本就没法延\n客服:你在这跟到后面来\n客服:这里不是大多少天都会员了吗\n客户:那你不是大于多少天都结了吗\n客服:嗯\n客户:嗯\n客服:我们有1个\n客户:就是你这个\n客服:他说你这里面也按照\n客户:他有多少是蒙细别的吗\n客户:这3天之后\n客服:当前执行的\n客服:这些，其实你这个也是\n客户:那你这个颜色\n客服:按照他给了预期了，是不是结多少天\n客户:按照他给的利息嘛，是不是延延延多少天啊，没\n客服:那是什么问题\n客服:你3天的时候关内\n客户:你要是原来那个\n客户:你基本上就那你要看\n客服:你要是原来的，你，你基本上都要不要\n客服:5天\n客户:对的\n客户:啊，关于\n客户:15新啊\n客户:上面录的\n\n```\n任务指令如下：\n```\n你是一个智能办公助理，可以对对话分析内容生成简单的摘要\n```\n请依据上述指令，确保准确无误地完成任务。" - requestMessage2 = RunCompletionMessageRequestMessages(content, role) - - listRequestMessage = [requestMessage1, requestMessage2] - - request = RunCompletionMessageRequest() - request.messages = listRequestMessage - request.model_code = "tyxmTurbo" - request.stream = False - - # 发送请求 - response = client.run_completion_message(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md deleted file mode 100644 index 9fa17554..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md +++ /dev/null @@ -1,404 +0,0 @@ -# 通过模板ID调用通义晓蜜CCAI-对话分析AIO应用 - -## **前提条件** - -- 本文向您介绍通义晓蜜CCAI-对话分析AIO应用SDK的安装、使用及注意事项。 - -- 如果您还未创建AccessKeyID和AccessKeySecret，请参考[获取 AccessKey 与 AgentKey](https://help.aliyun.com/zh/model-studio/get-accesskey-appid-and-agentkey)。 - -- 如果您使用子账号调用接口，请参考[通义晓蜜CCAI-对话分析RAM子账号使用方式和授权操作](https://help.aliyun.com/zh/model-studio/use-and-authorize-ram-users-for-ccai-dialogue-analysis)。 - -- 各API详细出入参说明请查看左侧目录中的[API目录](https://help.aliyun.com/zh/model-studio/api-contactcenterai-2024-06-03-dir/)。 - - -## **接口入参位置** - -### **workspaceId** - -1. 访问[**业务空间管理**](https://bailian.console.aliyun.com/?admin=1#/efm/business_management)页面。 - -2. 业务空间管理列表中Workspace ID为入参中workspaceId。 - - -### **appId** - -1. 访问**应用广场**页面，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 每个卡片上的应用ID即为接口参数中appId。 - - -### **templateIds** - -1. 访问**[应用广场](https://bailian.console.aliyun.com/#/app-market)**，点击通义晓蜜CCAI-对话分析AIO的**查看详情**。 - -2. 点击上方**我的应用**，展示应用卡片列表。 - -3. 点击**管理**进入对应的应用卡片。 - -4. 点击**自定义指令**模板，切换为**专业构建模式**。 - -5. 点击右上方**指令模板管理**。 - -6. **自定义模板**列表中**模板ID**为入参中templateIds。 - -7. 如果还未创建自定义模板，请直接点击右上角**保存指令模板**按钮。 - - -## **安装SDK** - -## 同步Java - - - -com.aliyun - -contactcenterai20240603 - -3.6.4 - - - -## 异步Java - - - -com.aliyun - -alibabacloud-contactcenterai20240603 - -3.0.12 - - - -## Python - -pip install alibabacloud\_contactcenterai20240603 - -## 异步流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).modelCode("tyxmTurbo").dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(true).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - ResponseIterable x = client.runCompletionWithResponseIterable(completionParam); - ResponseIterator iterator = x.iterator(); - String lastTxt=""; - while (iterator.hasNext()) { - RunCompletionResponseBody event = iterator.next(); - //System.out.println(event.getText()); - //System.out.println(event.getFinishReason()); - //System.out.println(event.getRequestId()); - lastTxt=event.getText(); - } - System.out.println("ALL***********************"); - System.out.println(lastTxt); - System.out.println("请求成功的请求头值："); - System.out.println(x.getStatusCode()); - System.out.println(x.getHeaders()); - } -} -``` - -## **异步非流式调用** - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.auth.credentials.Credential; -import com.aliyun.auth.credentials.provider.StaticCredentialProvider; -import com.aliyun.core.http.HttpMethod; -import com.aliyun.sdk.gateway.pop.Configuration; -import com.aliyun.sdk.gateway.pop.auth.SignatureAlgorithm; -import com.aliyun.sdk.gateway.pop.auth.SignatureVersion; -import com.aliyun.sdk.service.contactcenterai20240603.AsyncClient; -import com.aliyun.sdk.service.contactcenterai20240603.models.*; -import darabonba.core.RequestConfiguration; -import darabonba.core.ResponseIterable; -import darabonba.core.ResponseIterator; -import darabonba.core.client.ClientOverrideConfiguration; -import java.util.ArrayList; -import java.util.Arrays; -import java.util.List; -import java.util.concurrent.CompletableFuture; -import java.util.concurrent.ExecutionException; -import java.util.concurrent.TimeUnit; -import java.util.concurrent.TimeoutException; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - private static StaticCredentialProvider provider = StaticCredentialProvider.create( - Credential.builder() - .accessKeyId(accessKeyId) - .accessKeySecret(accessKeySecret) - .build() - ); - private static AsyncClient client = AsyncClient.builder() - .region("cn-shanghai") - .credentialsProvider(provider) - .serviceConfiguration(Configuration.create() - .setSignatureVersion(SignatureVersion.V3) - .setSignatureAlgorithmV3(SignatureAlgorithm.ACS3_HMAC_SHA256) - ).overrideConfiguration( - ClientOverrideConfiguration.create() - .setProtocol("HTTPS") - .setEndpointOverride("contactcenterai.cn-shanghai.aliyuncs.com") - ).build(); - public static void main(String[] args) throws Exception{ - //对话内容 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.Sentences sentenceDto1 = RunCompletionRequest.Sentences.builder().role("user").text("我要办理信用卡").build(); - RunCompletionRequest.Sentences sentenceDto2 = RunCompletionRequest.Sentences.builder().role("agent").text("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息").build(); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - RunCompletionRequest.Dialogue dialogue = RunCompletionRequest.Dialogue.builder().sessionId("session_01_asdfasdfasd") - .sentences(sentenceDTOList).build(); - //属性信息 - List fieldList = new ArrayList<>(); - RunCompletionRequest.Fields field1 = RunCompletionRequest.Fields.builder().name("姓名").desc("用户的名称").build(); - RunCompletionRequest.Fields field2 = RunCompletionRequest.Fields.builder().name("信用卡号").desc("用户的信用卡号").build(); - fieldList.add(field1); - fieldList.add(field2); - //构建请求参数 - RunCompletionRequest completionParam = RunCompletionRequest.builder() - .workspaceId(workspaceId).appId(appId).requestConfiguration(RequestConfiguration.create() - .setHttpMethod(HttpMethod.POST)).dialogue(dialogue).fields(fieldList).templateIds(Arrays.asList(templateId)).stream(false).build(); - System.out.println(JSON.toJSONString(completionParam)); - //发送请求 - CompletableFuture x = client.runCompletion(completionParam); - RunCompletionResponse generateCompletionResponse = x.get(10, TimeUnit.SECONDS); - System.out.println(JSON.toJSONString(generateCompletionResponse.getBody())); - System.out.println(generateCompletionResponse.getBody().getText()); - System.out.println(generateCompletionResponse.getBody().getRequestId()); - } -} -``` - -Python - -``` -import asyncio -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -async def run_async(): - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", - "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = await client.run_completion_async(workSpace, appId, request) - body = response.body - print(body) -if __name__ == '__main__': - asyncio.run(run_async()) -``` - -## 同步非流式调用 - -**说明** - -请将workspaceId替换示例中的YOUR\_WORKSPACEID，AccessKeyID替换示例中的YOUR\_ACCESS\_KEY\_ID，AccessKeySecret替换示例中的YOUR\_ACCESS\_KEY\_SECRET，appId替换示例中的YOUR\_APPID，templateIds替换示例中的YOUR\_TEMPLATE，代码才能正常运行。 - -Java - -``` -import com.alibaba.fastjson.JSON; -import com.alibaba.fastjson.JSONArray; -import com.alibaba.fastjson.JSONObject; -import com.aliyun.contactcenterai20240603.Client; -import com.aliyun.contactcenterai20240603.models.*; -import com.aliyun.teaopenapi.models.Config; -import java.util.*; -public class CcaiPaasTest { - private static String workspaceId="YOUR_WORKSPACEID"; - private static String accessKeyId="YOUR_ACCESS_KEY_ID"; - private static String accessKeySecret="YOUR_ACCESS_KEY_SECRET"; - private static String appId="YOUR_APPID"; - private static Long templateId=1L; //替换为您在模板管理中的模板ID - public static void main(String[] args) throws Exception{ - Config config = new Config(); - config.setAccessKeyId(accessKeyId).setAccessKeySecret(accessKeySecret).setEndpoint("contactcenterai.cn-shanghai.aliyuncs.com") - .setRegionId("cn-shanghai").setProtocol("HTTPS"); - Client client = new Client(config); - RunCompletionRequest request = new RunCompletionRequest(); - //对话信息 - List sentenceDTOList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto1 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto1.setRole("user").setText("我要办理信用卡"); - RunCompletionRequest.RunCompletionRequestDialogueSentences sentenceDto2 = new RunCompletionRequest.RunCompletionRequestDialogueSentences(); - sentenceDto2.setRole("agent").setText("好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息"); - sentenceDTOList.add(sentenceDto1); - sentenceDTOList.add(sentenceDto2); - //属性填充 - List fieldList = new ArrayList<>(); - RunCompletionRequest.RunCompletionRequestFields field1 = new RunCompletionRequest.RunCompletionRequestFields(); - field1.setName("姓名").setDesc("用户的名称"); - RunCompletionRequest.RunCompletionRequestFields field2 = new RunCompletionRequest.RunCompletionRequestFields(); - field2.setName("信用卡号").setDesc("用户的信用卡号"); - fieldList.add(field1); - fieldList.add(field2); - RunCompletionRequest.RunCompletionRequestDialogue dialogue = new RunCompletionRequest.RunCompletionRequestDialogue(); - dialogue.setSessionId("session_01_asdfasdfasd").setSentences(sentenceDTOList); - //构建请求参数 - request.setDialogue(dialogue).setStream(false).setModelCode("tyxmTurbo").setFields(fieldList).setTemplateIds(Arrays.asList(templateId)); - RunCompletionResponse runCompletionResponse = client.runCompletion(workspaceId, appId, request); - RunCompletionResponseBody responseBody = runCompletionResponse.getBody(); - System.out.println(JSON.toJSONString(responseBody)); - } -} -``` - -Python - -``` -from alibabacloud_contactcenterai20240603.client import Client -import alibabacloud_contactcenterai20240603 -import alibabacloud_tea_openapi -from alibabacloud_tea_openapi.models import Config -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest, \ - RunCompletionRequestDialogueSentences, RunCompletionRequestDialogue, RunCompletionRequestFields, \ - RunCompletionRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequest -from alibabacloud_contactcenterai20240603.models import RunCompletionMessageRequestMessages -ak = "YOUR_ACCESS_KEY_ID" -sk = "YOUR_ACCESS_KEY_SECRET" -workSpace = "YOUR_WORKSPACEID" -appId = "YOUR_APPID" -templateId = "YOUR_TEMPLATE" -if __name__ == '__main__': - config = Config() - config.access_key_id = ak - config.access_key_secret = sk - config.endpoint = "contactcenterai.cn-shanghai.aliyuncs.com" - config.region_id = "cn-shanghai" - client = Client(config) - # 对话 - sentence1 = RunCompletionRequestDialogueSentences("chat01", "user", "我要办理信用卡") - sentence2 = RunCompletionRequestDialogueSentences("chat02", "agent", "好的，稍等10分钟，我现在为您办理，请先提供相关的个人信息") - sentenceList = [sentence1, sentence2] - dialogue = RunCompletionRequestDialogue(sentenceList, "session_01_asdfasdfasd") - # 属性填充 - fields1 = RunCompletionRequestFields("", "用户的名称", None, "姓名") - fields2 = RunCompletionRequestFields("", "用户的信用卡号", None, "信用卡号") - fieldsList = [fields1, fields2] - # 构建请求参数 - templateIds = [templateId] - request = RunCompletionRequest() - request.dialogue = dialogue - request.fields = fieldsList - request.model_code = "tyxmTurbo" - request.stream = False - request.template_ids = templateIds - response = client.run_completion(workSpace, appId, request) - body = response.body - print(body) -``` diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md deleted file mode 100644 index 3f99a056..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/product-overview-1.md +++ /dev/null @@ -1,67 +0,0 @@ -# 产品概述 - -本文档介绍通义晓蜜CCAI-对话分析AIO产品能力及产品优势。 - -## **什么是通义晓蜜CCAI-对话分析AIO** - -对话分析AIO，即对话分析all-in-one API，是基于深度调优的对话大模型， 为营销服类产品提供智能化升级所需的生成式摘要总结、质检、分析等能力的官方应用。 - -- **面向对象：**开发者、自研企业、传统呼叫中心采购者及友商等。 - -- **核心能力：**为客户提供多种规格专属模型（支持切换），并提供多指令精准执行、自带最佳实践的API及贴合业务场景的被集成方式。 - -- **服务形式：**通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式。 - -- **核心价值：**免去企业SFT、探索prompt 、应用最佳实践开发的成本，使企业专注在业务逻辑的实现。 - -- **付费模式：**按调用次数后付费。 - - -## **产品能力** - -1. **多指令精准执行** - - -- **总结摘要：**根据对话内容，记录通话中最核心的信息。 - -- **信息抽取：**根据指令，抽取并组织关键信息。 - -- **质检分析：**按照指令进行检测，例如情绪、敏感词等 - -- **多指令任务：**生成标题、生成关键词、生成摘要、维度检测、信息抽取等。 - - -2. **多模型选择** - - -- 模型配置：平台提供通义晓蜜-Turbo、通义晓蜜-Plus两种模型规格。 - - -3. **多指令模板** - - -- **指令模板：**平台提供官方预置常用指令模板并支持管理自定义模板。 - - -4. **调试信息** - - -- SaaS管理：提供SaaS调试窗，用于验证效果并最终提供API服务。 - -- API服务：通过API服务输出给客户 - - -**5.数据报表** - -- 调用数据：提供按时、按日调用数据报表。 - -- 账单数据：请前往[费用与成本](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)查询。 - - -## **产品优势** - -- 结合大模型特性打造新能力，将过去多个业务场景凝聚成一个个通用、轻量、易集成的能力。 - -- 企业对于一次服务记录/工单所需要的所有质检、分析、处理的动作，可通过一次调用完成，得到结构化的结果。 - -- 提供通话工单分类（支持级联分类）、自定义信息抽取、自定义质检规则并检出能力。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md deleted file mode 100644 index 8b9c7fd7..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/application-management.md +++ /dev/null @@ -1,38 +0,0 @@ -# 如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量 - -本文档介绍如何对应用进行编辑、删除等管理，如何进行API调用、如何查看调用量。 - -## **我的应用** - -路径：[应用广场](https://bailian.console.aliyun.com/#/app-market)→ 点击[应用实践](https://bailian.console.aliyun.com/?tab=app#/app-market/lightApplication)→ 找到**通义晓蜜CCAI-对话分析AIO**，选择[查看详情](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)，即可进入**我的应用** **。** 在**应用广场**页面，找到**通义晓蜜CCAI-对话分析AIO**应用卡片，单击**立即查看**。 - -### 3.1 应用复制、删除、修改 - -- **应用复制：**在**我的应用**中，可以点击应用右上角选择复制应用，对该应用进行复制。应用复制：在**我的应用**页面，单击目标应用卡片右上角的菜单图标，选择**复制应用**，即可复制该应用。应用复制成功后，在**我的应用**列表中可看到新增的复制应用卡片，其名称自动追加"副本"后缀。每张应用卡片底部提供**管理**、**调用量**、**API调用**三个操作入口。 - -- **应用修改：**同时可以点击应用右上角选择修改应用，对该应用名称进行修改。应用修改：在**我的应用**中，点击应用右上角的三点菜单，选择**修改应用**，即可对该应用信息进行修改。应用修改：可以在**编辑应用名称**对话框中修改**应用名称**，名称最多支持50个字符，修改完成后单击**确定**保存。 - -- **应用删除：**如果需要删除该应用，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。应用删除：同样可以点击应用右上角选择**删除应用**，对该应用进行删除。应用删除：点击应用右上角选择删除应用，系统弹出**确认删除**对话框，提示"删除已发布应用可能会对您的线上业务产生影响，您确定删除吗？"，单击**确认删除**完成删除操作，单击**取消**可取消操作。 - - -### 3.2 API调用 - -- 应用API：应用提供标准化API接口，用于向客户系统输出对话分析结果。支持集成官方预置Agent模板或用户自定义模板，客户可通过API获取结构化响应，并自主设计前端展示样式。 - - - 当通过**对话分析Agent创建方式**创建应用时，使用自定义指令的API调用： - - 在**对话分析Agent**的自定义指令页面右上角，单击**API示例**按钮，页面右侧展开API调用信息面板，依次包含：获取AccessKeyID和AccessKeySecret、获取Workspace ID和App ID、安装SDK（支持Java、Python、Go，Java通过Maven dependency引入）以及代码示例（支持Java、Curl、Python、Go）。代码示例中展示了`CcAiPassSync`类的调用逻辑，包括设置`workspaceId`与`appId`常量、构造`RunCompletionRequest`请求及组装对话内容。左侧**指令信息**区域可引入`${dialogue}`和`${fields}`两个变量，**模型配置**选择所需模型后，单击底部**测试（command + enter）**按钮即可在中间**效果测试**区域查看调用结果。 - - - 通过**自定义创建方式**的API调用：单击页面顶部的**API示例**按钮，右侧面板展示API调用所需信息。单击**获取API-KEY、APP-ID和Workspace ID**蓝色链接获取AccessKey ID、AccessKey Secret、Workspace ID和App ID。安装SDK时，在Maven项目的pom.xml中添加依赖`com.aliyun:alibabacloud-ccai-passthrough-sync`（版本3.3.3）。代码示例使用Java类`CcAiPassSync`，设置`workspaceId`和`appId`参数，构造`RunCompletionRequest`对象，传入包含user和agent角色的对话句子列表进行调用。 - - -### **3.3 调用量** - -- 调用次数按输入和输出token总数计量，以2000 tokens 为一个计量单位，输入与输出总token数小于等于2000 tokens为1次调用，大于2000小于等于4000 tokens 为2次调用，向上取整，以此类推。 - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -在**调用量**页签中，可通过**所属空间**、**选择模型**、**应用**、**数据来源**下拉筛选条件，以及时间范围（**昨天**、**近3天**、**近七天**、**近15天**或自定义日期范围）查看对应的调用次数统计折线图。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md deleted file mode 100644 index b33e5bc8..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md +++ /dev/null @@ -1,60 +0,0 @@ -# 如何基于自定义方式创建应用 - -本文档介绍了如何通过自定义方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**，再点击**创建应用**。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入应用调试界面。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**自定义创建**：通过编写自定义指令（Prompt）来构建应用逻辑，用户可使用内置或自定义指令模板进行测试，适合有一定经验的人员。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -## **应用配置** - -进入已经创建完成的应用中进行配置。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009398.png) - -- **模型配置：**通义晓蜜Plus 和 Turbo 模型仅支持语音与文本分析；当分析对象为图片时，系统将默认使用通义晓蜜VL模型，该模型不可手动选择。 - -- **指令信息：**通过编写指令信息来配置对应的任务、格式、要求等，来完成对应分析任务。 - - - **变量配置：**若需要在对话过程中引用更多变量可以在此配置，在指令编辑器中输入 `/` 可触发变量自动补全，选择后插入对应变量引用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009509.png) - - - **选择指令模板：**同时可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009498.png) - - - **保存指令模板：**在编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。选择‘覆盖已有指令模板’，需从下拉列表中选择目标模板名称。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009505.png) - - - **指令优化：**对编写完成的指令信息进行AI优化。![指令优化示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009511.png) - -- **分析对象类型** - - 分析对象类型可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3624933771/p1059909.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - -- **字段信息：**当需要获取到对话内容的字段信息时，可以使用“信息抽取预置模板”、“多指令模板”创建指令任务，同时也需要引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![字段信息示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - -- 点击**“测试”**按钮**，**查看测试结果![测试结果示例图](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6379940671/p1009528.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md deleted file mode 100644 index 85bc7021..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md +++ /dev/null @@ -1,140 +0,0 @@ -# 如何进行基于对话分析Agent方式创建应用 - -本文档介绍了如何通过对话分析Agent方式创建应用 - -## **创建应用** - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935525.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用创建方式：** - - - 这里选择**基于对话分析Agent创建**：通过预置最佳实践示例或上传对话数据，体验大模型生成式摘要、总结、服务质检等全场景应用能力。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7209853471/p935528.png) - - -### 基于对话分析Agent创建 - -**说明** - -注意：当选择基于对话分析Agent创建方式，只有选择自**定义指令**\-**专业构建模式**可以用图片分析，其他方式无法对图片进行分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006405.png) - -进入已经创建完成的应用中后，可以选择**对话分析Agent**、**自定义指令**方式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914644.png) - -1. #### **选择对话分析Agent方式**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914645.png) - - **对话分析维度：**可以根据实际业务需求的需要通过对话分析Agent来选择对应的维度。 - - - **标准指令：**已预置标准prompt，可快速生成理想结果。可选值为【标题、摘要、关键词、Q&A、问题解决方案】，至少选择一个选项。 - - - **高级指令：**服务质检、标签分类等高级指令有一定的业务属性，该部分指令在示例通话中已预置标准prompt，若自行上传通话数据且对结果有一定要求，建议在预置指令模板→专业模式中编辑自定义指令进行调试。可选值为【服务质检、关键信息、标签分类】。 - - - **分析对象类型：**根据自己业务需要分析的数据类型选择【纯文本、语音】。当选择**纯文本**时，可以上传不超过15000字的文本内容，同时我们还提供了行业对话示例来进行测试；当选择**语音**时，支持单个不超过40MB的WAV或MP3格式文件，上传完成后会自动将其转译为文本信息内容。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9479940671/p1008496.png) - - **信息内容：**需要按照以下格式编写对话信息，可以通过使用已经提供的行业对话示例。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可直接插入内置的行业对话示例文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914647.png) - - - **点击“测试”按钮，查看输出结果：**测试出来的结果生成的指令与我选择的对话分析维度相对应。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914651.png) - -2. #### **选择自定义模板方式** - - 可以选择简单构建模式&专业构建模式。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914696.png) - - - ##### **简单构建模式说明** ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914697.png) - - - **模型配置**：通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **指令类型：**可以根据实际业务需求分析的全部维度，根据选择的标准指令与高级指令，在指令信息中编辑指令的prompt。 - - - 标准指令：可以选择【标题、摘要、关键词、Q&A、问题解决方案】，并在下方指令信息中展示出系统内置prompt，可以对其进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850108.png) - - - 高级指令：可以选择【服务质检、关键信息、标签分类】，并在下面指令信息中展示对应配置，进行自定义修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057646.png) - - - **服务质检：**对通话中存在的客户情绪、敏感词、服务质量等内容进行检测。可根据业务需要质检的场景和质检项。 - - - **质检项（名称+描述）：**设置服务质检中需要质检的名称与描述。 - - - **关键信息（名称+类型）：**提取通话中配置的关键信息，通过名称与类型配置。 - - - **标签分类（名称+描述）：**对通话内容进行分类定义，标签长度不超过10个字。 - - - **分析对象类型：**需要按照以下格式编写对话信息，可选择纯文本、语音两种方式。 - - - 对话信息建议按如下格式填写： - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx - - - 可以选择行业对话示例进行插入内置对话文本。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850123.png) - - - **点击“测试”按钮，查看测试结果。**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0986579371/p914710.png) - - - ##### 专业构建模式说明![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914716.png) - - **说明** - - 在模式切换时，当前模式所编辑的内容不再生效，对话内容将会重置。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057648.png) - - - **模型配置：**通义晓蜜-Plus、通义晓蜜-Turbo。 - - - **选择指令模板：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了总结摘要、信息抽取、服务质检、标签分类、多指令任务，共五类模板。同时支持自定义指令模板，或在官方预置模板基础上自定义修改的指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006412.png) - - - **保存指令模板：**在自定义编写指令信息或者自定义修改系统行业示例后，可以点击“保存指令模板”按钮，进行保存，选择指令模板保存方式，可选【新增指令模板、覆盖已有指令模板】，在“指令模板管理”中可以查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006414.png) - - - **指令模板管理：**可以查看预置模板和自定义保存的模板指令。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914728.png) - - - **变量配置：**除了内置${field}、${dialogue}两个变量以外，如需在分析过程中引用更多变量，可以完成变量配置。测试数据可以作为测试过程中的模拟数据，临时使用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939566.png) - - - 在编辑完成后，即可在指令信息中插入，使用“/”进行插入保存的变量。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939584.png) - - - **分析对象类型：**可以分为三种，纯文本、语音、图片，同时支持添加知识库进行辅助分析，支持添加热词组有助于提升语音转译准确性。 - - - **知识库：**开启后可添加文档、表格、图片等类型知识用于辅助分析，当分析对象类型选择为图片时无法使用知识库。具体介绍可参考文档：[知识库的使用](https://help.aliyun.com/zh/model-studio/using-the-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6344413771/p1057649.png) - - - **选择文本时**：需要按照以下格式编写对话信息，同时也可以通过使用已经提供的行业对话示例。 - - 客户：xxx - - 客服：xxx - - 客户：xxx - - 客服：xxx![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006434.png) - - - **选择语音时：**自定义上传一个不超过40MB、WAV、MP3格式的文件，同时可以选择添加/新建热词组，提升语音转译效果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006437.png) - - 上传完成后将自动识别语音内容，并可以设置客户/客服先发言顺序。 - - - **选择图片识别后**：可点击上传一张不超过10MB、JPEG/JPG/PNG等常见图片格式，上传成功后可以通过指令信息对图片进行检测分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3501767571/p1006443.png) - - - **字段信息：**当使用“信息抽取预置模板”、“多指令模板”创建指令任务，可引入变量${fields}填写字段信息。 - - 填写格式为：字段名：字段描述。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0159666271/p850538.png) - - - 点击**“测试”**按钮**，**查看输出结果![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7898424471/p939588.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md deleted file mode 100644 index b3b43117..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/hot-phrase-management.md +++ /dev/null @@ -1,48 +0,0 @@ -# 热词组配置管理与使用 - -为提升语音转译的准确性，您可以在语音质检分析场景中使用热词组。本文档将介绍其配置与使用方法。 - -## **热词配置** - -热词组仅对离线/实时语音质检分析场景生效，用于提升语音转译的准确性。 - -### **1.热词组管理** - -- 进入热词组管理的路径： - - - 路径1：进入[**通义晓蜜CCAI-对话分析AIO**](https://bailian.console.aliyun.com/?tab=app#/app/app-market/ccai)后，点击我的应用，可在界面中看到**热词组管理**按钮。 - - - 路径2：通过进入具体应用的配置页面，点击**选择热词组**时进行添加对应的热词组。 - - **说明** - - 目前热词组支持通过自定义创建的应用；通过Agent创建的应用，需在‘专业构建模式’下才能找到并配置热词组。 - - 点击后右侧弹出 **选择热词组** 侧边栏，顶部提示"热词组可以提高语音转译准确性，使用时仅对语音质检分析生效"。可通过搜索框查找已有热词组，或单击 **新建热词组** 创建新的热词组，列表中勾选所需热词组即可完成关联。 - -- 在热词组管理界面，点击‘**新建热词组**’，填写名称和热词后，点击‘**确定**’完成创建。单击**新建热词组**按钮打开新建弹窗，在弹窗中填写热词组名称并通过**添加热词**逐条添加热词。 - - - 热词组名称：不超过10个字。 - - - 热词：每个热词组不超过128个热词。 - - - 热词：填入对应需要提高准确率的热词，重要限制：当前版本热词仅支持纯汉字，不支持英文、数字及特殊字符。包含非汉字的热词将无法生效。 - - - 权重：取值范围为1到5之间的整数，默认值：2；如果效果不明显可以适当增加权重，但是当权重较大时可能会引起负面效果，导致其他词语识别不准确。 - - - 操作：删除已添加的热词。 - -- 创建成功后，新的热词组将显示在列表中，并支持编辑或删除。在已创建的热词组名称右侧，单击齿轮图标，可在下拉菜单中选择**编辑**或**删除**该热词组。 - - -### **2.热词组在应用中的使用** - -进入具体应用的设置，选择并绑定所需的热词组。 - -**说明** - -每个应用仅支持绑定一个热词组。 - -1. 选择对应的热词组：单击**选择热词组**按钮，在弹出的浮层中单击目标热词组对应的**选用**按钮即可完成绑定。 - -2. 选择完成后可以查看绑定情况。在**自定义指令**配置页面下方的**热词组**区域，单击**选择热词组**按钮，将已创建的热词组添加到当前指令配置中。已添加的热词组会以标签形式展示，可单击删除图标移除。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md deleted file mode 100644 index c64ccbc5..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-aio-user-guide/using-the-knowledge-base.md +++ /dev/null @@ -1,74 +0,0 @@ -# 知识库的使用 - -采用检索增强生成（RAG）技术，根据用户上传的外部信息源检索相关信息，然后将检索到的内容整合到用户输入中，从而帮助大模型生成更准确的回答。 - -## **知识库** - -知识库仅对自定义指令的专业模式或者自定义创建的应用中的纯文本和语音对象分析场景生效，用于提升大模型生成回答的准确性。 - -### **1.知识库配置** - -1. **知识库配置的开启** - - 在自定义创建应用中开启**知识**开关后可以添加文档、表格、图片类型知识用于辅助分析。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059009.png) - -2. **知识库的调用方式** - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059025.png) - - 1. **必定调用：**每一轮的用户输入都会进入知识库检索与联网搜索，适用于高频知识问答场景。 - - 2. **智能调用：**智能判断用户的输入信息是否进入知识库检索或者联网搜索，适合灵活的对话场景。 - -3. **可使用的知识说明**![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059033.png) - - 1. **文档知识：**使用知识库文件类数据构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 2. **表格：**使用数据中心表格类、数据库类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - - 3. **图片：**使用数据中心图片类构建文档搜索知识库后，对话分析Agent能够引用知识库内容判断分析结果。 - -4. **知识库其他配置：** - - 1. **知识库过滤：**开启后，将通过配置判断prompt的方式引入大模型判断，对文档和表格召回结果进行二次智能过滤。 - - 2. **对话摘要总结：**开启后，可通过大模型对其对话内容进行概括总结。 - -5. **知识库的选择** - - 开启知识库后，可以在对应的文档、表格、图片知识类目中点击“+”按钮，右侧弹出对应知识库列表，可以选择已存的知识库进行的进行添加。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059059.png) - -6. **知识库的相似度阈值与权重** - - 选择对应的知识库后，支持配置相似度阈值与权重来控制筛选检索结果和知识库召回顺序。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059337.png) - - 1. **知识库的相似度阈值：** - - 阈值范围在0.01～1，默认为0.2，可以根据场景需要进行调整，注意：只有语义相似度得分高于此值的文本才会被召回。若此值设置得过高，将导致知识库丢弃所有相关的文本。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059377.png) - - 2. **知识库的权重：** - - 权重范围在0.5～2.默认为1，可以根据场景需要进行调整，在多路召回时，若多个知识库召回的文本切片相似度分数相同，系统将优先返回权重更高的知识库中的文本切片。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059385.png) - -7. **知识库的创建** - - 点击**创建新知识库**按钮，跳转到知识库创建的页面中，具体操作可参考[创建和使用知识库](https://help.aliyun.com/zh/model-studio/rag-knowledge-base)。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059090.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059088.png) - - **说明** - - 目前通义晓蜜CCAI-AIO产品所绑定的知识库仅支持文档搜索、数据查询、图片问答，不支持音视频搜索。 - - -### **2.知识库在应用中的使用** - -- 根据两种创建应用的方式区分： - - - 在**基于对话分析Agent创建方式**中的使用：仅支持在自定义指令中的**专业构建模式**中使用，**分析对象类型**需要选择纯文本、语音。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0461133771/p1059324.png) - - - 在**自定义创建方式**中的使用：**分析对象类型**选择纯文本、语音即可使用。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md deleted file mode 100644 index 3e231b8b..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-voice-dialogue-robot/operation-guide.md +++ /dev/null @@ -1,85 +0,0 @@ -# 语音对话机器人操作指南 - -本文档介绍通义晓蜜CCAI-语音对话机器人在阿里云百炼控制台如何操作。 - -## **1\. 创建应用** - -- 路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -- 第一步：首先点击**我的应用**按钮，再点击**创建应用**按钮；![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000694.png) - -- 第二步：进入新建应用弹窗，编辑应用名称与应用创建方式。点击**确定**。进入调试窗口。 - - - **应用名称：**根据实际业务需要修改应用名称。 - - - **应用描述：**自定义填入应用实际使用描述说明。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000698.png) - - -## **2.机器人配置** - -机器人配置目前使用的为prompt构建模式。 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9886579371/p914777.png) - -- **模型选择：**通义晓蜜-Plus、通义晓蜜-Max、通义晓蜜-Turbo。 - -- **指令信息：**选择指令模板，可以选择直接使用官方预置模板，当前线上提供了通用场景、服务满意度调研、家电上门安装预约、游戏福利推送介绍四种模板。同时支持自定义指令模板。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000715.png) - -- **变量配置：**可在指令信息中通过**${xxx}**样式进行插入。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000733.png) - -- **指令配置：**如果当前参考的流程话术中存在特殊指令#\[...\]，请在回复中添加#\[...\]，目前支持传入挂机指令#\[HangUp\]。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000747.png) - -- **语音配置：**进行音色选择配置，配置完成后可在机器人呼叫时运用。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000739.png) - - - 音色模板：可选择大模型音色，如：龙小夏V2、龙小夏等。 - - - 音量：范围为0～100，值越大声音越响亮。 - - - 语速：范围为-500～500，值越大语速越快。 - - - 音调：范围为-500～500，值越大音调越高昂。 - -- **高级配置：**对机器人的其他能力进行配置。 - - - 静默超时：自定义配置时长，范围在1～60秒，当对话过程中用户回复超过配置的静默时长后播报静默话术。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000740.png) - -- **点击“呼叫”按钮，查看测试结果。** - - **说明** - - 目前通过控制台测试时不支持变量传入。 - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000746.png) - -- **发布机器人：**点击**发布**按钮，当页面提出**发布成功**，即表示为成功发布。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000757.png) - - -## **3.我的应用** - -路径：**[应用广场](https://bailian.console.aliyun.com/#/app-market)**\-应用实践-通义晓蜜CCAI-语音对话机器人-立即查看。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000691.png) - -### **3.1 调用量** - -- 调用次数按小时进行计量上报，查询当天时，折线图展示每小时调用量曲线。 - -- 查询某个日期区间数据时，折线图展示按天调用量曲线。 - - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000755.png) - -### 3.2 应用修改、删除 - -- **应用修改：**进入我的应用后，可以点击应用右上角选择修改应用，对该应用名称进行修改。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000750.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000751.png) - -- **删除应用：**进入我的应用后，可以点击应用的右上角选择删除应用，对该应用进行删除，弹出“确认删除”二次确认框，选择确认删除，将成功删除。![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000752.png)![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000753.png) - - -### **3.3API调用** - -应用API：通过API服务输出给客户，方便客户进行集成和使用官方预置模板或自定义模板，客户自定义前端样式 - -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9719585571/p1000756.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-tongyi-xiaomi-ccai-chatbot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md index d03c071a..cc6f0110 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/light-application-best-practices/best-practices-for-workflow-integration-video-understanding.md @@ -19,7 +19,7 @@ ### **步骤一：创建函数（FC）** -1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/fc/user-guide/creating-an-event-function)。 +1. 在[函数计算](https://fcnext.console.aliyun.com/cn-hangzhou/functions)控制台创建事件函数。具体操作，请参见[创建事件函数](https://help.aliyun.com/zh/functioncompute/creating-an-event-function)。 - 创建视频理解提交任务函数（建议函数名称：SubmitVideoAnalysisTask）。 @@ -49,7 +49,7 @@ pip3 install alibabacloud_endpoint_util alibabacloud_tea_openapi alibabacloud_quanmiaolightapp20240801 -t . ``` - - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/fc/user-guide/create-a-custom-layer-1)。 + - 方案二：以“层”的方式安装。相关文档，请参见[创建自定义层](https://help.aliyun.com/zh/functioncompute/fc/create-a-custom-layer-1)。 3. 您可自行优化调整代码中的入参：比如`prompt`模板、`modelId`等。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-cancelaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-queryaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitaudittask.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md index 8f161902..d2d7873d 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md @@ -462,24 +462,6 @@ API API概述 -[QueryAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-queryaudittask) - -查询审核结果 - -查询审核结果。 - -[SubmitAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitaudittask) - -提交审核任务 - -提交审核任务 - -[CancelAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-cancelaudittask) - -取消审核任务 - -取消审核任务 - [SubmitSmartAudit](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitsmartaudit) 提交智能审校任务 @@ -1607,3 +1589,21 @@ SubmitParseDocumentLayoutTask 获取排版任务结果 获取排版任务结果 + +[CancelAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-cancelaudittask) + +取消审核任务 + +取消审核任务 + +[QueryAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-queryaudittask) + +查询审核结果 + +查询审核结果。 + +[SubmitAuditTask](https://help.aliyun.com/zh/model-studio/api-aimiaobi-2023-08-01-submitaudittask) + +提交审核任务 + +提交审核任务 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md new file mode 100644 index 00000000..54844525 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-after-sales-service-scope.md @@ -0,0 +1,57 @@ +# 阿里云百炼平台售后服务范围说明 + +## **阿里云百炼平台售后服务范围说明** + +欢迎您使用阿里云百炼。本《**阿里云百炼平台售后服务范围说明》是对您使用阿里云百炼相关产品和服务时适用的售后服务范围的说明。** + +1\. 在您购买的服务期限内，我们将为您提供如下售后基础服务，即通过官网、电话及阿里云APP提供7×24的电话咨询（95187、400电话）、智能在线和标准工单支持。支持范围包括： + +（1）关于阿里云百炼模型服务与产品功能、架构的咨询； + +（2）使用、配置阿里云百炼模型服务的最佳实践； + +（3）阿里云百炼模型服务的使用咨询、技术问题及故障诊断； + +（4）阿里云百炼API及阿里云百炼官方SDK问题的故障诊断； + +（5）与阿里云百炼管理控制台相关的问题； + +（6）与阿里云相关的账号问题咨询支持； + +（7）与阿里云相关的财务、合同及计费问题的咨询支持。 + +2\. 阿里云百炼将以阿里云官网页面公布的[客户服务权益](https://www.aliyun.com/service/customer-service-benefits?spm=5176.support-home.J_3451238410.1.12d1156fPBBxO0)标准向您提供相应的售后服务支持。 + +3\. 阿里云百炼同时提供付费版的售后增值服务（包括支持计划等），该等服务需在阿里云官网订购后生效使用。您还可通过阿里云获得其他付费的售后服务，具体详见阿里云的网站相关页面的收费售后服务内容。如您的项目需要更深度的技术支持（如业务代码编写指导、定制化集成方案等），建议联系阿里云商务经理沟通定制化服务方案。 + +4\. 为了方便您的生产或使用，如您选择将阿里云百炼服务与外部（非阿里云百炼平台上的）第三方工具或产品进行对接，阿里云将尽商业上合理的努力为您提供第三方工具在接入阿里云百炼模型推理服务过程中的方向性建议，但针对非阿里云百炼服务相关的问题，我们无法提供专业意见。 + +（1）我们提供建议的范围包括： + +（i）确认阿里云百炼API接口及服务端的可用状态； + +（ii）阿里云百炼官方API调用示例及SDK使用说明参考； + +（iii）协助核查阿里云百炼服务端调用明细和计费记录； + +（iv）基本的连通性测试建议（如通过curl等标准工具测试阿里云百炼服务地址的可达性）。 + +**（2）我们提供建议的范围不包括：** + +（i）第三方工具（如Cursor、Windsurf、Cline、OpenClaw等）的安装、部署、配置、升级及日常使用指导； + +（ii）第三方工具的产品功能、交互逻辑及内部实现问题的排查； + +（iii）其他云厂商、企业或社区提供的产品或服务的配置与运维； + +（iv）用户业务代码的编写、调试与实现； + +（v）用户本地环境（含内网、代理、VPN、防火墙、操作系统等）导致的连通性或兼容性问题的排查； + +（vi）第三方工具内部显示的Token数量、费用预估值或调用统计与阿里云计费数据之间的差异解释； + +（vii）所有第三方工具的安装、补丁更新、测试、故障诊断、优化等日常运维服务； + +（viii）基于阿里云百炼模型服务原生能力之上的第三方自建业务相关支持。 + +5\. 但请您注意，除我们另有书面说明外，第三方工具不构成我们的代理、受托或联合服务主体，**我们不对外部第三方工具的任何陈述、承诺或行为承担责任。**您知悉并确认，阿里云仅负责阿里云百炼平台自身的运营维护，即百炼服务端的技术架构、API接口、计量计费系统、控制台功能等；**第三方工具的运行维护（如AI编程工具的安装配置、开源代理框架的部署调优等）由您及相应工具提供方负责。**当您的问题出现在百炼模型服务的使用过程中，但其原因、责任范围或依赖关系已超出阿里云百炼平台本身可直接提供支持和保障的范围时——通常涉及您侧系统、外部第三方服务、网络环境、账号权限、业务流程或非标集成，阿里云将协助进行初步排查。若经排查确认问题来源于非阿里云侧，阿里云将给予方向性建议并引导您联系相应的服务主体。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md index 941388e9..87bc4bba 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-support/application-related-agreements.md @@ -2,6 +2,6 @@ - [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=5176.28197581.0.0.16e829a4HTC9FE) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) +- [阿里云百炼体验功能特别说明](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20260716114753386/20260716114753386.html) - [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md index 29f6ac10..36597cef 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md @@ -271,7 +271,45 @@ ### **2.2 模型调用费用** -在创建、更新、检索、命中测试知识库时，会调用向量模型（用于内容向量化）和排序模型（Rerank，用于重排序），这些调用会产生费用。 +在创建、更新、检索知识库以及使用知识问答服务时，会调用以下模型，这些调用会产生独立于规格费用之外的模型调用费用： + +**模型类别** + +**模型名称** + +**用途** + +**向量模型** + +[text-embedding-v4](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4)等 + +文档类知识库的文本向量化 + +[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding) + +图片问答类、音视频搜索类知识库的多模态向量化 + +**排序模型** + +[qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank) + +文档类知识库检索结果的二次排序（可选） + +[qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank) + +图片问答类、音视频搜索类知识库检索结果的二次排序（可选） + +**路由模型** + +[qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) + +开启知识库路由时，系统调用 qwen-plus 判断查询应路由至哪些知识库 + +**问答模型** + +[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3.7-plus?serviceSite=asia-pacific-china) 等 + +知识问答服务中生成回答的大语言模型，由用户在应用中自行选择 **重要** @@ -281,20 +319,28 @@ **多个知识库计费规则：**阿里云百炼应用挂载了多个知识库时，会在多个知识库内执行检索，Token 消耗量（Query 向量化和 Rerank 排序）**按知识库数量倍数增加**（N 个知识库则消耗量 × N）。 -#### **2.2.1 创建/更新知识库** +#### **2.2.1 知识管理（创建与更新知识库）** - **调用场景：**上传新文件或增量更新时，调用向量模型对文本内容进行向量化处理。 - **计费说明：按新增内容的 Token 数量计费。**删除文件不产生模型调用费用。 +- **调用的模型：** + + - 文档搜索类知识库：[text-embedding-v4](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4) 或 [text-embedding-v3](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v3)（文本向量模型）。 + + - 图片问答类、音视频搜索类知识库：[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding)（多模态向量模型）。 + -#### **2.2.2 检索知识库** +#### **2.2.2 知识检索** - **调用场景** - 1. 调用向量模型，对用户的查询（Query）进行向量化。 + 1. **向量化：**调用向量模型，对用户的查询（Query）进行向量化。 + + 2. **知识库路由（可选）：**若应用关联了多个知识库并开启了知识库路由功能，系统会调用 [qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) 判断用户查询应路由至哪些知识库，该调用按 qwen-plus 的 Token 用量计费。 - 2. 调用排序模型（Rerank），对初步检索到的结果进行重新排序，以提升最终答案的精准度。 + 3. **排序（可选）：**调用排序模型对初步检索到的结果进行重新排序，以提升最终答案的精准度。文档搜索类知识库使用 [qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank)，图片问答类和音视频搜索类知识库使用 [qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank)。 - **计费说明** @@ -304,7 +350,7 @@ - **检索流程与计费关系详解** - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0647879771/CAEQaxiBgMCFrtjd3BkiIDA2ZWRiNzYxYzZiNzRkNGM5Mzg4NGQ5ZjhlODBlOWZj6139615_20260107153729.136.svg) + ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6004524871/CAEQaxiBgMCFrtjd3BkiIDA2ZWRiNzYxYzZiNzRkNGM5Mzg4NGQ5ZjhlODBlOWZj6139615_20260107153729.136.svg) 1. **初步召回** 系统根据以下参数从知识库中召回文本切片： @@ -328,7 +374,22 @@ Rerank 模型排序后，系统会根据**最终召回最大数量**参数（例如 5）返回相应数量的切片。 -#### **2.2.3 费用优化建议** +#### **2.2.3 知识问答** + +通过百炼应用（智能体应用、工作流应用）使用知识库进行问答时，除了检索阶段的模型费用外，还会产生以下模型调用费用： + +- **问答生成模型：**系统根据您在应用中选择的问答模型（如 qwen-plus 等）生成回答，按该模型的 Token 用量计费。具体价格以[模型计费标准](https://help.aliyun.com/zh/model-studio/model-pricing)为准。 + +- **预文件解析（可选）：**当用户在对话中上传文件并开启预文件解析功能时，系统会调用 [qwen3-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank) 对文件内容进行排序处理，按排序模型的 Token 用量计费。 + +- **知识库路由（可选）：**若应用关联了多个知识库并开启了路由功能，系统会调用 [qwen-plus](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen-plus-latest) 进行路由判断（详见 [2.2.2 知识检索](#c868ef3a653qx)）。 + + +**重要** + +知识问答服务的完整费用 = **规格费用**（知识库运行时长）+ **检索阶段的模型费用**（向量化 + 排序 + 路由）+ **问答阶段的模型费用**（问答生成 + 预文件解析）。各模型费用按实际 Token 消耗量独立计算，请关注[模型计费标准](https://help.aliyun.com/zh/model-studio/model-pricing)了解各模型的单价。 + +#### **2.2.4 费用优化建议** 有以下两种方式： @@ -366,6 +427,19 @@ 点击知识库卡片上的**命中测试**，进入配置调试页面进行测试，会产生相应的模型（向量模型、排序模型）调用计费。 +#### **2.2.5 节省计划抵扣说明** + +知识库使用的向量模型（如 text-embedding-v4）和排序模型（如 qwen3-rerank）属于百炼平台 A 类模型，其调用费用支持通过以下节省计划抵扣： + +- **AI 通用型节省计划**（推荐）：覆盖 A 类全部模型（含文本向量、多模态向量、排序模型），按月承诺消费享阶梯折扣。详情请参见[节省计划与资源包](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 + +- **向量及排序模型节省计划**：专门针对向量和排序模型的节省计划，一次性购买固定金额。详情请参见[节省计划与资源包 > 向量及排序模型节省计划](https://help.aliyun.com/zh/model-studio/savings-plan-and-resource-package)。 + + +**说明** + +节省计划仅可抵扣模型调用费用，不可抵扣知识库的规格费用（运行时长费用）。规格费用的优化请参见[资源包](#a06c023507qq3)。 + ## 3\. 计费示例 ### 3.1 连续运行 1 天 @@ -396,7 +470,7 @@ ### 3.2 创建、更新与检索知识库 -基于 **text-embedding-v4**（向量模型）与 **qwen3-rerank**（排序模型），价格均为 **0.0005 元/千 Token**。 +以下示例基于文档搜索类知识库，使用 [**text-embedding-v4**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/text-embedding-v4)（向量模型）与 [**qwen3-rerank**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-rerank)（排序模型），价格均为 **0.0005 元/千 Token**。图片问答类和音视频搜索类知识库使用的多模态模型（[qwen3-vl-embedding](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-embedding)、[qwen3-vl-rerank](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3-vl-rerank)）价格请参见对应模型详情页。 **计费逻辑**：费用 = Token 消耗量（以“千 Token”为单位） × 模型单价 @@ -550,7 +624,7 @@ 5. **为什么我的排序（Rerank）费用特别高？如何降低模型调用费用？** - 排序（Rerank）模型的费用与您最终返回的结果数量无关，而是由**初步召回**的文本切片总数决定的。降低模型调用费用详见本文[2.2.3 费用优化建议](#9a9e30ecc3pbe)内容。 + 排序（Rerank）模型的费用与您最终返回的结果数量无关，而是由**初步召回**的文本切片总数决定的。降低模型调用费用详见本文[2.2.4 费用优化建议](#9a9e30ecc3pbe)内容。 6. **如何彻底停止知识库的计费？删除库内文件可以吗？** diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md index c5cdc98d..73ce26f5 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/knowledge-base/rag-knowledge-base.md @@ -62,7 +62,7 @@ 2. 填写**知识库名称**和**知识库描述**，其余设置保持默认，点击**下一步**。 -3. 选择**默认类目**，上传[阿里云百炼系列手机产品介绍.docx](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250603/duuuxk/%E9%98%BF%E9%87%8C%E4%BA%91%E7%99%BE%E7%82%BC%E7%B3%BB%E5%88%97%E6%89%8B%E6%9C%BA%E4%BA%A7%E5%93%81%E4%BB%8B%E7%BB%8D.docx)文件。点击**下一步**，然后点击**完成**。 +3. 选择**配置类目**，上传[阿里云百炼系列手机产品介绍.docx](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250603/duuuxk/%E9%98%BF%E9%87%8C%E4%BA%91%E7%99%BE%E7%82%BC%E7%B3%BB%E5%88%97%E6%89%8B%E6%9C%BA%E4%BA%A7%E5%93%81%E4%BB%8B%E7%BB%8D.docx)文件。点击**下一步**，然后点击**完成**。 ### 2\. 集成到业务应用 @@ -184,12 +184,10 @@ 2. **填写基础信息** - 根据应用场景选择合适的**知识库类型**（单一知识库不支持同时选择多个类型）。选择**文档搜索**类型后，还需选择**使用场景**（基础文档问答、图文并茂回复、视觉理解（富文本文档）或极速问答）： + 根据应用场景选择合适的**知识库类型**（单一知识库不支持同时选择多个类型）。选择**文档搜索**类型后，还需选择**使用场景**（基础文档问答、视觉理解（富文本文档）或极速问答）： - **基础文档问答：**适用于纯文本文档的语义检索。 - - **图文并茂回复：**适用于需要返回图文混排内容的场景。 - - **视觉理解（富文本文档）：**使用多模态向量模型对 PDF、图片等富文本文档进行视觉级理解和索引，保留原始版面信息。适合含有复杂排版、图表、公式的文档，支持文字、图片和图文组合三种命中测试模式。 - **极速问答：**针对检索速度进行优化，适合高度结构化或简单文档类型（如 FAQ、产品参数表等），提供极速低延时的问答体验。索引配置与基础文档问答一致，差异在于后端检索策略针对低延迟场景进行了专项优化。仅支持文本查询，不支持图片输入。 @@ -466,7 +464,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ## **相似度阈值** @@ -534,7 +532,7 @@ > 文档搜索类知识库无法实现此效果。 - - 支持导入多个Excel文件，但要求各文件的**表结构完全一致**。 + - 支持导入单份xlsx、xls格式的文档，文件大小限制20MB以内。 - **选择连接器：**选择指定的数据连接器。如尚未创建数据连接器，请参阅[数据连接](https://help.aliyun.com/zh/model-studio/data-connection)。 @@ -609,7 +607,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -721,7 +719,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -865,7 +863,7 @@ - **知识库类型限制：**仅适用于文档搜索类、数据查询类、音视频搜索类知识库。**图片问答类知识库不支持**。 - - **使用场景限制：**仅「基础文档问答」和「图文并茂回复」两种使用场景支持。「视觉理解（富文本文档）」和「极速问答」**不支持**。 + - **使用场景限制：**仅**基础文档问答**使用场景支持。**视觉理解（富文本文档）**和**极速问答**不支持。 ### **相似度阈值** @@ -894,7 +892,7 @@ - **使用场景**可根据需求选择**基础文档问答**、**图文并茂回复**、**视觉理解（富文本文档）**或**极速问答**（适用于高度结构化或简单文档类型，任务明确，提供极低延迟的问答体验）。 + **使用场景**可根据需求选择**基础文档问答**、**视觉理解（富文本文档）**或**极速问答**（适用于高度结构化或简单文档类型，任务明确，提供极低延迟的问答体验）。 在请求高峰时段，创建过程可能需要数小时（取决于数据量），请耐心等待。 @@ -905,11 +903,15 @@ ## **文档搜索类知识库** -- **自动更新（推荐）** +1. **自动更新（推荐）** - 通过对象存储OSS管理文件，借助函数计算 FC 监听文件变更事件，自动同步更新至知识库，实现知识的实时更新。详见[告别手动操作，让AI知识库自动更新](https://www.aliyun.com/solution/tech-solution/auto-updated-knowledge-base)。 + 通过对象存储OSS类型连接器监听OSS内数据变动，自动同步更新至知识库，实现知识的实时更新。操作步骤如下： -- **手动更新** + 1. **创建连接器：**前往[创建连接器](https://bailian.console.aliyun.com/cn-beijing?tab=app#/connector/create)，创建**OSS**类型的连接器并选择进行监听的Bucket。 + + 2. **创建知识库：**创建一个知识库，在**选择数据**步骤的**数据来源**选项选择创建完成的OSS类型的连接器。 + +2. **手动更新** 在[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)页面，找到目标知识库，单击卡片上的**查看详情**。 @@ -1230,22 +1232,12 @@ **方式一（仅适用智能体应用）** - 1. 在[构建知识库](#c0fa1080aerzp)时，知识库类型选择**文档搜索**，使用场景选择**图文并茂回复**。 - - > 选择图文并茂回复后，知识库将从文件插图中提取摘要，大模型根据摘要与问题的相关性自主决定是否插入图片。 + 1. 在[构建知识库](#c0fa1080aerzp)时，知识库类型选择**文档搜索**，使用场景选择**视觉理解（富文本文档）**。 - **重要** - - 上传文档时不能选择**电子文档解析**，否则无法获取图片内容。电子文档解析不识别文档中的图片，会导致图文并茂回复功能无法正常使用。 + > 选择视觉理解（富文本文档）回复后，知识库将从文件插图中提取摘要，大模型根据摘要与问题的相关性自主决定是否插入图片。 2. 创建或编辑智能体应用时，选择**千问-Plus**或**千问-Plus-Latest**模型（经测试，两款模型效果最佳）。点击**文档知识库**右侧的**+**按钮，添加上一步构建的知识库。 - **说明** - - 召回长度须小于文档实际长度。若召回长度超过文档实际长度，系统将直接返回完整文档内容，不执行图文并茂的逻辑判断。 - - > 注意：当前"图文并茂回复"与"展示回答来源"功能暂不支持同时开启。 - 3. 实际问答效果： ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1699676371/p903021.png) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md index 0e578a2d..2a13317b 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/application-introduction.md @@ -45,12 +45,15 @@ Python 编码（专业代码） AI 自主决策、动态规划 由大模型根据提示词自主规划任务步骤 + 由预定义流程精确控制 每一步都由预设的节点定义，逻辑确定 + 完全由代码控制 所有逻辑和执行路径由代码定义 + 适合人群 @@ -74,9 +77,9 @@ AI 工程师、开发者 - **智能体应用** - - 如何创建和配置？请参考：[新版智能体应用](https://help.aliyun.com/zh/model-studio/new-single-agent-application)（推荐）、[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)。 + - 如何创建和配置？请参考：[新版智能体应用（Agent 2.0）](https://help.aliyun.com/zh/model-studio/new-single-agent-application)（推荐）、[智能体应用（Agent 1.0）](https://help.aliyun.com/zh/model-studio/single-agent-application)。 - - 如何快速上手实际案例？请参考：[创建智能问答 AI 电商客服助手](https://help.aliyun.com/zh/document_detail/2878136.html#0a9fbaf6a71q7)、[集成高德 MCP 的旅行规划智能体](https://help.aliyun.com/zh/document_detail/2880695.html)。 + - 如何快速上手实际案例？请参考：[创建智能问答 AI 电商客服助手](https://help.aliyun.com/zh/document_detail/2878136.html#0a9fbaf6a71q7)、[集成高德 MCP 的旅行规划智能体](https://help.aliyun.com/zh/model-studio/use-cases/integrate-amap-mcp-travel-planning-agent)。 - 如何通过 API 调用？请参考：[新版智能体应用 API](https://help.aliyun.com/zh/model-studio/new-agent-application-api-reference)、[调用智能体应用](https://help.aliyun.com/zh/model-studio/call-single-agent-application/)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md index ede0fcdc..48466f2f 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md +++ b/skills/bailian-docs-llm-wiki/raw/application-user-guide/llm-application/new-single-agent-application.md @@ -1,4 +1,4 @@ -# 新版智能体应用（Agent 2.0） +# 新版智能体应用 新版智能体应用（Agent 2.0）将知识库、MCP 等多种能力统一为工具，并通过自主思考和规划来调用，以解决复杂任务。 @@ -53,7 +53,7 @@ - **最长回复长度**：模型生成的长度限制，不包含提示词。 - - **temperature**：控制生成随机性和多样性，数值越高随机性越强。 + - **温度系数**：控制生成随机性和多样性，数值越高随机性越强。 - **enable\_thinking**：是否开启思考模式。开启思考模式有助于提升智能体的反思效果。不支持思考模式的模型无法配置 enable\_thinking 参数。 @@ -245,3 +245,15 @@ - 意图与技能的相关性：请评估问题的表述是否清晰，其意图是否能明确指向特定技能。如果意图模糊或与技能功能不相关，模型可能选择不调用。 - 执行轮次限制：请检查是否达到了 ReAct 轮次上限。智能体可能已规划调用该技能，但在执行到该步骤前因轮次耗尽而被强制终止。 + + +### 智能体应用是否支持上下文缓存？ + +支持**隐式缓存**，暂不支持在应用中配置**显式缓存**。 + +- **隐式缓存**：智能体在调用支持隐式缓存的模型时会自动生效，无需任何配置、也无法关闭。系统会自动识别并缓存请求的公共前缀（如相同的系统提示词、多轮对话历史、知识库召回内容等），命中缓存的输入 Token 按标准输入单价的 20% 计费，可相应降低模型调用成本。 + +- **显式缓存**：需要在模型调用请求中主动为指定内容创建缓存标记。智能体应用由平台统一构造模型请求，暂不支持配置显式缓存。 + + +上下文缓存的工作模式、支持的模型及计费详情，请参见[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md index 62ca0940..382676af 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md @@ -21,7 +21,9 @@ Tripo 3D模型生成支持**文生3D模型**、**单图生3D模型和多图生3D ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -35,7 +37,8 @@ Tripo 3D模型生成支持**文生3D模型**、**单图生3D模型和多图生3D ## 文生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -53,7 +56,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 单图生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -71,7 +75,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 多图生3D模型（有贴图） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -96,7 +101,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 图片顺序为前、左、后、右，不需要的视角传入空对象`{}`即可。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -121,7 +127,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > 需同时将`texture`和`pbr`设为`false`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -324,7 +331,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -343,7 +350,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md index 601bfe7e..b7ba1534 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md @@ -39,8 +39,6 @@ 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/iwaouc/asr_example.wav)。 - ``` from http import HTTPStatus import dashscope @@ -58,7 +56,7 @@ recognition = Recognition(model='fun-asr-realtime', format='wav', sample_rate=16000, callback=None) -result = recognition.call('asr_example.wav') +result = recognition.call('{YOUR_AUDIO_FILE}') if result.status_code == HTTPStatus.OK: print('识别结果：') print(result.get_sentence()) @@ -221,8 +219,6 @@ if __name__ == '__main__': ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/acoict/asr_example.wav)。 - ``` import os import time @@ -270,8 +266,8 @@ recognition = Recognition(model='fun-asr-realtime', try: audio_data: bytes = None - f = open("asr_example.wav", 'rb') - if os.path.getsize("asr_example.wav"): + f = open("{YOUR_AUDIO_FILE}", 'rb') + if os.path.getsize("{YOUR_AUDIO_FILE}"): # 一次性将文件数据全部读入buffer file_buffer = f.read() f.close() diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md index 10ee8202..3a99c0b6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Android SDK +# Fun-ASR非实时语音识别Android SDK -本文档提供了Fun-ASR录音文件识别Android SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Fun-ASR非实时语音识别Android SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 @@ -13,7 +13,7 @@ - [下载最新SDK整合包](https://help.aliyun.com/zh/isi/sdk-selection-and-download)。 - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 - 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 - 用 Android Studio 打开工程。示例代码位于`DashFunAsrFileTranscriberActivity.java`，替换 API Key 后体验功能。 @@ -99,7 +99,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -170,7 +170,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -596,7 +596,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md index 2dbd2ce9..2ccfc352 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md @@ -1,12 +1,10 @@ -# Fun-ASR录音文件识别HTTP API参考 +# Fun-ASR非实时语音识别HTTP API参考 -本文介绍Fun-ASR录音文件识别HTTP API的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别HTTP API的参数和接口细节。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 -## **DashScope异步调用（Fun-ASR）** - -### **流程说明** +## **流程说明** 与DashScope同步调用（一次请求、立即返回结果）不同，异步调用专为处理长音频文件或耗时较长的任务设计，该模式采用“提交-轮询”的两步式流程，避免了因长时间等待而导致的请求超时： @@ -23,7 +21,7 @@ - 当任务处理完成后，结果查询接口将返回最终的识别结果。 -### **服务端点** +## **服务端点** ## 华北2（北京） @@ -56,7 +54,7 @@ 使用新版域名（`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`）提交任务时，请求体中必须包含`parameters`对象。即使无需设置任何参数，也必须传入空对象`{}`，否则任务可正常提交，但识别将失败。 -### **请求头** +## **请求头** **参数** @@ -90,11 +88,11 @@ string 异步任务标识。仅提交任务接口需要传入，固定为`enable`，请勿遗漏，否则无法提交任务。 -### **提交任务接口** +## **提交任务接口** 提交语音识别任务。该接口异步返回，业务侧需结合[查询任务接口](#480630e0582sb)轮询任务状态。 -#### **请求体** +### **请求体** 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 @@ -107,7 +105,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi "model": "fun-asr", "input": { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ] }, "parameters": { @@ -289,7 +287,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi - en: 英文 -#### **返回体** +### **返回体** ``` { @@ -319,11 +317,11 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 任务状态。提交成功时返回`PENDING`。 -### **查询任务接口** +## **查询任务接口** 查询语音识别任务的执行情况和结果。建议轮询调用直至任务终态。 -#### **请求体** +### **请求体** 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 @@ -340,7 +338,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks 查询任务需指定其ID，该ID为[提交任务接口](#418f2ac8ecxm4)被调用后返回的`task_id`。 -#### **返回体** +### **返回体** ## 正常示例 @@ -355,7 +353,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks "end_time": "2024-09-12 15:11:40.903", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url": "{YOUR_AUDIO_URL}", "transcription_url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/pre/filetrans-16k/20240912/15%3A11/409a4b92-445b-4dd8-8c1d-f110954d82d8-1.json?Expires=1726211500&OSSAccessKeyId=yourOSSAccessKeyId&Signature=v5Owy5qoAfT7mzGmQgH0g8C****%3D", "subtask_status": "SUCCEEDED" } @@ -383,7 +381,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "FILE_DOWNLOAD_FAILED", "message": "FILE_DOWNLOAD_FAILED", "subtask_status": "FAILED" @@ -485,11 +483,11 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks 失败的子任务数。 -### **其他接口：批量查询任务状态/取消任务** +## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 -### **识别结果说明** +## **识别结果说明** 识别结果保存为JSON文件。 @@ -497,7 +495,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -637,887 +635,3 @@ punctuation string 预测出的词之后的标点符号（如有）。 - -## **DashScope同步调用（Fun-ASR-Flash）** - -**重要** - -该功能不支持SDK调用。 - -### **服务端点** - -## 华北2（北京） - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -## 新加坡 - -`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -**重要** - -阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议迁移至新域名： - -- 华北2（北京）地域：从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - -- 新加坡地域：从 `dashscope-intl.aliyuncs.com` 迁移至 `{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - -`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 - -### **请求头** - -**参数** - -**类型** - -**是否必选** - -**说明** - -Authorization - -string - -是 - -鉴权令牌，格式为`Bearer `，使用时将"``"替换为实际的API Key。 - -Content-Type - -string - -是 - -请求体的媒体类型，固定为`application/json`。 - -X-DashScope-SSE - -string - -是 - -用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 - -### **请求体** - -以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - -## 非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 携带上下文-非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_text", - "text": "你好啊" - } - ] - }, - { - "role": "assistant", - "content": [ - { - "type": "text", - "text": "你好啊，我是通义千问，有什么可以帮助你的？" - } - ] - }, - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## 携带上下文-流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_text", - "text": "你好啊" - } - ] - }, - { - "role": "assistant", - "content": [ - { - "type": "text", - "text": "你好啊，我是通义千问，有什么可以帮助你的？" - } - ] - }, - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" - } - } - ] - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000" - } -}' -``` - -## Base64 - -可输入Base64编码数据（[Data URL](https://www.rfc-editor.org/rfc/rfc2397)），格式为：`data:;base64,`。 - -- ``：MIME类型 - - 因音频格式而异，例如： - - - WAV：`audio/wav` - - - MP3：`audio/mpeg` - -- ``：音频转成的Base64编码的字符串 - - Base64编码会增大体积，请控制原文件大小，确保编码后仍符合输入音频大小限制（10MB） - -- 示例：`data:audio/wav;base64,SUQzBAAAAAAAI1RTU0UAAAAPAAADTGF2ZjU4LjI5LjEwMAAAAAAAAAAAAAAA//PAxABQ/BXRbMPe4IQAhl9` - - **点击查看示例代码** - - Python - - ``` - import base64, pathlib - - # input.mp3为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - file_path = pathlib.Path("input.mp3") - base64_str = base64.b64encode(file_path.read_bytes()).decode() - data_uri = f"data:audio/mpeg;base64,{base64_str}" - ``` - - Java - - ``` - import java.nio.file.*; - import java.util.Base64; - - public class Main { - /** - * filePath为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - */ - public static String toDataUrl(String filePath) throws Exception { - byte[] bytes = Files.readAllBytes(Paths.get(filePath)); - String encoded = Base64.getEncoder().encodeToString(bytes); - return "data:audio/mpeg;base64," + encoded; - } - - public static void main(String[] args) throws Exception { - System.out.println(toDataUrl("input.mp3")); - } - } - ``` - - -``` -import base64, pathlib -import os -import requests - -# input.wav为待识别的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 -file_path = pathlib.Path("input.wav") -base64_str = base64.b64encode(file_path.read_bytes()).decode() -data_uri = f"data:audio/wav;base64,{base64_str}" - -url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" - -headers = { - "Authorization": f"Bearer {os.environ['DASHSCOPE_API_KEY']}", - "Content-Type": "application/json", - "X-DashScope-SSE": "disable", -} - -payload = { - "model": "fun-asr-flash-2026-06-15", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "type": "input_audio", - "input_audio": { - "data": data_uri, - }, - } - ], - } - ] - }, - "parameters": { - "format": "wav", - "sample_rate": "16000", - }, -} - -response = requests.post(url, headers=headers, json=payload) -print(response.status_code) -print(response.json()) -``` - -**model** `_string_` **（必选）** - -模型名称，固定为`fun-asr-flash-2026-06-15`。 - -**input** `_object_` **（必选）** - -输入信息。 - -**属性** - -**messages** `_array(object)_` **（必选）** - -消息列表。包含当前待识别的音频，以及可选的对话上下文（用于提升识别效果）。 - -**重要** - -上下文功能用于提升专有词汇的识别准确率，使用方法详见[快速开始](https://help.aliyun.com/zh/model-studio/improve-asr-accuracy#ctx-quickstart-sec)。约束：上下文消息（`input_text` 和 `text` 类型）各最多 5 条，超出时保留最近的 5 条。每轮上下文文本总长度（`user` 和 `assistant` 的 `text` 字段长度之和）不超过 400 个字符（按字符数计算，每个字符计为 1），超出部分从末尾截断。 - -**重要** - -携带上下文时，`messages` 中的消息顺序有要求：上下文消息必须按对话轮次排列，每轮中 `user`（`input_text` 类型）必须在对应的 `assistant`（`text` 类型）之前；包含 `input_audio` 的 `user` 消息必须放在 `messages` 数组的最后。 - -**属性** - -**role** `_string_` **（必选）** - -消息角色。取值范围： - -- `user`（必选）：用户消息。type为`input_audio`时表示当前待识别的音频；type为`input_text`时表示前几轮的识别结果或领域相关的词表（可选，上下文）。 - -- `assistant`（可选，上下文）：前几轮大语言模型的回复内容。 - - -**content** `_array(object)_` **（必选）** - -消息内容列表。 - -**属性** - -**type** `_string_` **（必选）** - -内容类型。每个请求至少需要一条`input_audio`类型的消息。取值范围： - -- `input_audio`（必选）：当前待识别的音频输入（role为user），需同时传入`input_audio`对象。 - -- `input_text`（可选，上下文）：前几轮用户语音的识别结果或领域相关的词表（role为user），需同时传入`text`字段。 - -- `text`（可选，上下文）：前几轮大语言模型的回复内容（role为assistant），需同时传入`text`字段。 - - -**input\_audio** `_object_` **（条件必选）** - -当`type`为`input_audio`时必填。 - -**属性** - -**data** `_string_` **（必选）** - -待识别音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。支持以下两种方式： - -- **音频文件URL**：直接传入可公开访问的音频文件地址。 - -- **Base64 Data URI**：采用Data URI格式传入Base64编码的音频数据，值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 - - -示例（URL方式）：`https://example.com/audio/sample.wav` - -示例（Base64方式）：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` - -**text** `_string_` **（条件必选）** - -当`type`为`input_text`时，填入前几轮用户语音的识别结果或领域相关的词表；当`type`为`text`时，填入前几轮大语言模型的回复内容。文本按字符数计算，每个字符计为 1。每轮上下文中所有消息的 `text` 字段长度之和不超过 400 个字符，超出部分从末尾截断。 - -**parameters** `_object_` **（必选）** - -模型参数。 - -**属性** - -**format** `_string_` **（必选）** - -音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**sample\_rate** `_string_` （可选） - -音频采样率，单位Hz。例如`16000`表示16kHz采样率。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -### **返回体** - -## 非流式 - -``` -{ - "output": { - "sentence": { - "begin_time": 760, - "channel_id": 0, - "end_time": 3800, - "sentence_end": true, - "sentence_id": 1, - "text": "Hello World，这里是阿里巴巴语音实验室。", - "words": [ - {"begin_time": 760, "end_time": 1040, "fixed": true, "punctuation": "", "text": "Hello"}, - {"begin_time": 1040, "end_time": 1240, "fixed": true, "punctuation": "，", "text": " World"}, - {"begin_time": 1360, "end_time": 1880, "fixed": true, "punctuation": "", "text": "这里是"}, - {"begin_time": 1880, "end_time": 2520, "fixed": true, "punctuation": "", "text": "阿里巴巴"}, - {"begin_time": 2520, "end_time": 2840, "fixed": true, "punctuation": "", "text": "语音"}, - {"begin_time": 2840, "end_time": 3800, "fixed": true, "punctuation": "。", "text": "实验室"} - ] - }, - "text": "Hello World，这里是阿里巴巴语音实验室。" - }, - "usage": { - "duration": 4 - }, - "request_id": "40e0734d-096f-9ae3-86c1-a8c013287561" -} -``` - -## 流式 - -设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议返回识别结果。SSE事件格式如下： - -``` -id:{序列号} -event:result -:HTTP_STATUS/200 -data:{JSON数据} -``` - -返回示例： - -``` -id:1 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":3800,"words":[{"end_time":1040,"punctuation":"","begin_time":760,"fixed":true,"text":"Hello"},{"end_time":1240,"punctuation":"，","begin_time":1040,"fixed":true,"text":" World"},{"end_time":1880,"punctuation":"","begin_time":1360,"fixed":true,"text":"这里是"},{"end_time":2520,"punctuation":"","begin_time":1880,"fixed":true,"text":"阿里巴巴"},{"end_time":2840,"punctuation":"","begin_time":2520,"fixed":true,"text":"语音"},{"end_time":3800,"punctuation":"。","begin_time":2840,"fixed":true,"text":"实验室"}],"begin_time":760,"text":"Hello World，这里是阿里巴巴语音实验室。","channel_id":0},"text":"Hello World，这里是阿里巴巴语音实验室。"},"usage":{"duration":4},"request_id":"fc1582e4-935c-9fc2-a482-a98bf43daa69"} -``` - -**request\_id** `_string_` - -本次请求的唯一标识。 - -**output** `_object_` - -输出结果。 - -**属性** - -**text** `_string_` - -当前累积的完整识别文本。 - -**sentence** `_object_` - -当前句子的详细信息。 - -**属性** - -**sentence\_id** `_integer_` - -句子编号，从1开始。 - -**sentence\_end** `_boolean_` - -是否为该句的最终结果。为`true`时表示该句识别完成。 - -**begin\_time** `_integer_` - -句子开始时间，单位毫秒。 - -**end\_time** `_integer_` - -句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 - -**text** `_string_` - -当前句子的识别文本。 - -**channel\_id** `_integer_` - -声道编号，从0开始。 - -**words** `_array_` - -词级别时间戳列表。 - -**属性** - -**text** `_string_` - -词文本。 - -**begin\_time** `_integer_` - -词开始时间，单位毫秒。 - -**end\_time** `_integer_` - -词结束时间，单位毫秒。 - -**punctuation** `_string_` - -词后的标点符号。无标点时为空字符串。 - -**fixed** `_boolean_` - -词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 - -**usage** `_object_` - -用量信息。仅在`sentence_end`为`true`时返回。 - -**属性** - -**duration** `_integer_` - -已处理的音频时长，单位秒。 - -### **SSE 流式结果处理逻辑** - -在流式模式下，客户端需关注以下处理要点： - -1. 每收到一个SSE事件，解析`data`字段中的JSON。 - -2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 - -3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 - - -## **DashScope同步调用（Fun-ASR-R**ealtime**）** - -**重要** - -- 该功能只支持北京地域。 - -- 不支持SDK调用。 - - -### **服务端点** - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -**重要** - -阿里云百炼为华北2（北京）地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`。 - -`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 - -### **请求头** - -**参数** - -**类型** - -**是否必选** - -**说明** - -Authorization - -string - -是 - -鉴权令牌，格式为`Bearer `，使用时将“``”替换为实际的API Key。 - -Content-Type - -string - -是 - -请求体的媒体类型，固定为`application/json`。 - -X-DashScope-SSE - -string - -是 - -用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 - -### **请求体** - -## 非流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: disable" \ - --data '{ - "model": "fun-asr-realtime", - "input": { - "messages": [] - }, - "parameters": { - "audio_address": "https://example.com/audio/sample.mp3", - "format": "mp3" - }, - "resources": [] -}' -``` - -## 流式 - -``` -curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --header "X-DashScope-SSE: enable" \ - --data '{ - "model": "fun-asr-realtime", - "input": { - "messages": [] - }, - "parameters": { - "audio_address": "https://example.com/audio/sample.mp3", - "format": "mp3" - }, - "resources": [] -}' -``` - -**model** `_string_` **（必选）** - -模型名称。 - -取值范围： - -- `fun-asr-realtime`（稳定版模型） - -- `fun-asr-realtime-2026-02-28`（即 Fun-Realtime-ASR-preview） - - -**input** `_object_` **（条件必选）** - -输入信息。与输入音频文件URL方式（通过`parameters.audio_address`传入）二选一，使用Base64方式上传音频时需要填写。 - -**属性** - -**messages** `_array(object)_` **（必选）** - -消息列表。 - -**属性** - -**content** `_array(object)_` **（必选）** - -用户消息的内容。仅允许设置一组消息。 - -**属性** - -**audio** `_string_` **（必选）** - -待识别音频，采用Data URI格式传入Base64编码的音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -使用Base64方式上传音频时需要填写。 - -值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 - -示例：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` - -**role** `_string_` **（必选）** - -用户消息的角色，固定为`user`。使用Base64方式上传音频时需要填写。 - -**parameters** `_object_` **（必选）** - -模型参数。 - -**属性** - -**audio\_address** `_string_` **（条件必选）** - -音频文件URL地址。与Base64方式（通过`input.messages`传入）二选一，使用URL方式时必填。需为可公开访问的地址。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**format** `_string_` **（必选）** - -音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 - -**vad\_enabled** `_boolean_` （可选） - -是否启用端点检测（VAD）。 - -默认为`true`（启用 VAD 检测）。 - -设为 `false` 时关闭 VAD 检测；但当音频时长超过 1 分钟时，系统将自动启用 VAD 检测，忽略此参数设置。 - -**resources** `_array_` （可选） - -资源列表，预留字段，当前可传空数组`[]`。 - -### **返回体** - -## 非流式 - -``` -{ - "output": { - "sentence": { - "begin_time": 160, - "channel_id": 0, - "end_time": 1680, - "sentence_end": true, - "sentence_id": 1, - "text": "欢迎使用阿里云。", - "words": [ - {"begin_time": 160, "end_time": 520, "fixed": true, "punctuation": "", "text": "欢迎"}, - {"begin_time": 520, "end_time": 880, "fixed": true, "punctuation": "", "text": "使用"}, - {"begin_time": 880, "end_time": 1280, "fixed": true, "punctuation": "", "text": "阿里"}, - {"begin_time": 1280, "end_time": 1680, "fixed": true, "punctuation": "。", "text": "云"} - ] - }, - "text": "欢迎使用阿里云。" - }, - "usage": { - "duration": 2 - }, - "request_id": "eff4c092-2289-9b43-a4cd-80e591fa90f5" -} -``` - -## 流式 - -设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议分多次返回中间识别结果和最终结果。每个SSE事件格式如下： - -``` -id:{序列号} -event:result -:HTTP_STATUS/200 -data:{JSON数据} -``` - -中间结果（句子开始、词逐步增长）： - -``` -id:1 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":false,"sentence_begin":true,"words":[],"begin_time":0,"text":"","channel_id":0},"text":""},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} - -id:2 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"}],"begin_time":160,"text":"欢迎","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} - -id:3 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":false,"text":"使用"}],"begin_time":160,"text":"欢迎使用","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎使用"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} -``` - -最终结果（句子结束，包含`usage`）： - -``` -id:4 -event:result -:HTTP_STATUS/200 -data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":1680,"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":true,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":true,"text":"使用"},{"end_time":1280,"punctuation":"","begin_time":880,"fixed":true,"text":"阿里"},{"end_time":1680,"punctuation":"。","begin_time":1280,"fixed":true,"text":"云"}],"begin_time":160,"text":"欢迎使用阿里云。","channel_id":0},"text":"欢迎使用阿里云。"},"usage":{"duration":2},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} -``` - -**request\_id** `_string_` - -本次请求的唯一标识。 - -**output** `_object_` - -输出结果。 - -**属性** - -**text** `_string_` - -当前累积的完整识别文本。 - -**sentence** `_object_` - -当前句子的详细信息。 - -**属性** - -**sentence\_id** `_integer_` - -句子编号，从1开始。 - -**sentence\_end** `_boolean_` - -是否为该句的最终结果。为`true`时表示该句识别完成。 - -**begin\_time** `_integer_` - -句子开始时间，单位毫秒。 - -**end\_time** `_integer_` - -句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 - -**text** `_string_` - -当前句子的识别文本。 - -**channel\_id** `_integer_` - -声道编号，从0开始。 - -**words** `_array_` - -词级别时间戳列表。 - -**属性** - -**text** `_string_` - -词文本。 - -**begin\_time** `_integer_` - -词开始时间，单位毫秒。 - -**end\_time** `_integer_` - -词结束时间，单位毫秒。 - -**punctuation** `_string_` - -词后的标点符号。无标点时为空字符串。 - -**fixed** `_boolean_` - -词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 - -**usage** `_object_` - -用量信息。仅在`sentence_end`为`true`时返回。 - -**属性** - -**duration** `_integer_` - -已处理的音频时长，单位秒。 - -### **SSE 流式结果处理逻辑** - -在流式模式下，客户端需关注以下处理要点： - -1. 每收到一个SSE事件，解析`data`字段中的JSON。 - -2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 - -3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md index e59f1695..928864d9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别iOS SDK +# Fun-ASR非实时语音识别iOS SDK -本文档提供了Fun-ASR录音文件识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Fun-ASR非实时语音识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 @@ -102,7 +102,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -173,7 +173,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -590,7 +590,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md index 8008ce57..47e93c0f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Java SDK +# Fun-ASR非实时语音识别Java SDK -本文介绍Fun-ASR录音文件识别Java SDK的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别Java SDK的参数和接口细节。 **重要** @@ -32,7 +32,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -41,7 +41,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4867892871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6278074871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -85,7 +85,7 @@ public class Main { .model("fun-asr") // 此处以fun-asr为例，可按需更换模型名称。模型列表：https://help.aliyun.com/zh/model-studio/models .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -107,7 +107,7 @@ public class Main { ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4867892871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6278074871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -152,7 +152,7 @@ public class Main { .model("fun-asr") // 此处以fun-asr为例，可按需更换模型名称。模型列表：https://help.aliyun.com/zh/model-studio/models .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -188,7 +188,7 @@ TranscriptionParam param = TranscriptionParam.builder() .model("fun-asr") .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); ``` @@ -507,7 +507,7 @@ public JsonObject getOutput() "end_time":"2025-02-13 16:12:10.189", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/16%3A12/3baafe5f-d09d-46c6-8b01-724927670edb-1.json?Expires=1739520730&OSSAccessKeyId=yourOSSAccessKeyId&Signature=BF7vPxlsJN9hkJlY%2BLReezxOwK8%3D", "subtask_status":"SUCCEEDED" } @@ -533,7 +533,7 @@ public JsonObject getOutput() "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -615,7 +615,7 @@ public String getMessage() ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -775,7 +775,7 @@ TranscriptionParam param = .model("fun-asr") .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -854,7 +854,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -873,7 +873,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md index 97a0b5b3..051845e2 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md @@ -1,6 +1,6 @@ -# Fun-ASR录音文件识别Python SDK +# Fun-ASR非实时语音识别Python SDK -本文介绍Fun-ASR录音文件识别Python SDK的参数和接口细节。 +本文介绍Fun-ASR非实时语音识别Python SDK的参数和接口细节。 **重要** @@ -32,7 +32,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -41,7 +41,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5594892871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3178074871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -77,7 +77,7 @@ dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") task_response = Transcription.async_call( model='fun-asr', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'] + file_urls=['{YOUR_AUDIO_URL}'] ) transcribe_response = Transcription.wait(task=task_response.output.task_id) @@ -88,7 +88,7 @@ if transcribe_response.status_code == HTTPStatus.OK: ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5594892871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3178074871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -124,7 +124,7 @@ dashscope.api_key = os.getenv("DASHSCOPE_API_KEY") transcribe_response = Transcription.async_call( model='fun-asr', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'] + file_urls=['{YOUR_AUDIO_URL}'] ) while True: @@ -407,7 +407,7 @@ list\[str\] "end_time":"2025-02-13 17:31:21.867", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A31/20ee4e4f-0404-4806-b617-c7d4c62eed19-1.json?Expires=1739525481&OSSAccessKeyId=yourOSSAccessKeyId&Signature=3q%2B1uQmRwltd7FPn5HQM2mBKw74%3D", "subtask_status":"SUCCEEDED" } @@ -440,7 +440,7 @@ list\[str\] "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -563,7 +563,7 @@ transcription\_url "end_time":"2025-02-13 17:59:28.828", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A59/70e737cc-bf8c-418b-b0c8-83fab192a0fa-1.json?Expires=1739527168&OSSAccessKeyId=yourOSSAccessKeyId&Signature=AtGjIKI%2BdgbzjJIu%2BHsr1R5nSAY%3D", "subtask_status":"SUCCEEDED" } @@ -589,7 +589,7 @@ transcription\_url "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -657,7 +657,7 @@ transcription\_url ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -857,7 +857,7 @@ def fetch(cls, ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -876,7 +876,7 @@ def fetch(cls, "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md new file mode 100644 index 00000000..ecf7e674 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md @@ -0,0 +1,561 @@ +# 非实时语音识别（Fun-ASR-Flash）API参考 + +本文介绍Fun-ASR-Flash非实时语音识别HTTP API的参数和接口细节。 + +**用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**重要** + +该功能不支持SDK调用。 + +## **服务端点** + +## 华北2（北京） + +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +## 新加坡 + +`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +**重要** + +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议迁移至新域名： + +- 华北2（北京）地域：从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `dashscope-intl.aliyuncs.com` 迁移至 `{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 + +## **请求头** + +**参数** + +**类型** + +**是否必选** + +**说明** + +Authorization + +string + +是 + +鉴权令牌，格式为`Bearer `，使用时将"``"替换为实际的API Key。 + +Content-Type + +string + +是 + +请求体的媒体类型，固定为`application/json`。 + +X-DashScope-SSE + +string + +是 + +用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 + +## **请求体** + +以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 + +## 非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 携带上下文-非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_text", + "text": "你好啊" + } + ] + }, + { + "role": "assistant", + "content": [ + { + "type": "text", + "text": "你好啊，我是通义千问，有什么可以帮助你的？" + } + ] + }, + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## 携带上下文-流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_text", + "text": "你好啊" + } + ] + }, + { + "role": "assistant", + "content": [ + { + "type": "text", + "text": "你好啊，我是通义千问，有什么可以帮助你的？" + } + ] + }, + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": "{YOUR_AUDIO_URL}" + } + } + ] + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000" + } +}' +``` + +## Base64 + +可输入Base64编码数据（[Data URL](https://www.rfc-editor.org/rfc/rfc2397)），格式为：`data:;base64,`。 + +- ``：MIME类型 + + 因音频格式而异，例如： + + - WAV：`audio/wav` + + - MP3：`audio/mpeg` + +- ``：音频转成的Base64编码的字符串 + + Base64编码会增大体积，请控制原文件大小，确保编码后仍符合输入音频大小限制（10MB） + +- 示例：`data:audio/wav;base64,SUQzBAAAAAAAI1RTU0UAAAAPAAADTGF2ZjU4LjI5LjEwMAAAAAAAAAAAAAAA//PAxABQ/BXRbMPe4IQAhl9` + + **点击查看示例代码** + + Python + + ``` + import base64, pathlib + + # 请替换为自己的音频文件路径，确保其符合音频要求 + file_path = pathlib.Path("{YOUR_AUDIO_FILE}") + base64_str = base64.b64encode(file_path.read_bytes()).decode() + data_uri = f"data:audio/mpeg;base64,{base64_str}" + ``` + + Java + + ``` + import java.nio.file.*; + import java.util.Base64; + + public class Main { + /** + * 请替换为自己的音频文件路径，确保其符合音频要求 + */ + public static String toDataUrl(String filePath) throws Exception { + byte[] bytes = Files.readAllBytes(Paths.get(filePath)); + String encoded = Base64.getEncoder().encodeToString(bytes); + return "data:audio/mpeg;base64," + encoded; + } + + public static void main(String[] args) throws Exception { + System.out.println(toDataUrl("{YOUR_AUDIO_FILE}")); + } + } + ``` + + +``` +import base64, pathlib +import os +import requests + +# 请替换为自己的音频文件路径，确保其符合音频要求 +file_path = pathlib.Path("{YOUR_AUDIO_FILE}") +base64_str = base64.b64encode(file_path.read_bytes()).decode() +data_uri = f"data:audio/wav;base64,{base64_str}" + +url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation" + +headers = { + "Authorization": f"Bearer {os.environ['DASHSCOPE_API_KEY']}", + "Content-Type": "application/json", + "X-DashScope-SSE": "disable", +} + +payload = { + "model": "fun-asr-flash-2026-06-15", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "type": "input_audio", + "input_audio": { + "data": data_uri, + }, + } + ], + } + ] + }, + "parameters": { + "format": "wav", + "sample_rate": "16000", + }, +} + +response = requests.post(url, headers=headers, json=payload) +print(response.status_code) +print(response.json()) +``` + +**model** `_string_` **（必选）** + +模型名称，固定为`fun-asr-flash-2026-06-15`。 + +**input** `_object_` **（必选）** + +输入信息。 + +**属性** + +**messages** `_array(object)_` **（必选）** + +消息列表。包含当前待识别的音频，以及可选的对话上下文（用于提升识别效果）。 + +**重要** + +上下文功能用于提升专有词汇的识别准确率，使用方法详见[快速开始](https://help.aliyun.com/zh/model-studio/improve-asr-accuracy#ctx-quickstart-sec)。约束：上下文消息（`input_text` 和 `text` 类型）各最多 5 条，超出时保留最近的 5 条。每轮上下文文本总长度（`user` 和 `assistant` 的 `text` 字段长度之和）不超过 400 个字符（按字符数计算，每个字符计为 1），超出部分从末尾截断。 + +**重要** + +携带上下文时，`messages` 中的消息顺序有要求：上下文消息必须按对话轮次排列，每轮中 `user`（`input_text` 类型）必须在对应的 `assistant`（`text` 类型）之前；包含 `input_audio` 的 `user` 消息必须放在 `messages` 数组的最后。 + +**属性** + +**role** `_string_` **（必选）** + +消息角色。取值范围： + +- `user`（必选）：用户消息。type为`input_audio`时表示当前待识别的音频；type为`input_text`时表示前几轮的识别结果或领域相关的词表（可选，上下文）。 + +- `assistant`（可选，上下文）：前几轮大语言模型的回复内容。 + + +**content** `_array(object)_` **（必选）** + +消息内容列表。 + +**属性** + +**type** `_string_` **（必选）** + +内容类型。每个请求至少需要一条`input_audio`类型的消息。取值范围： + +- `input_audio`（必选）：当前待识别的音频输入（role为user），需同时传入`input_audio`对象。 + +- `input_text`（可选，上下文）：前几轮用户语音的识别结果或领域相关的词表（role为user），需同时传入`text`字段。 + +- `text`（可选，上下文）：前几轮大语言模型的回复内容（role为assistant），需同时传入`text`字段。 + + +**input\_audio** `_object_` **（条件必选）** + +当`type`为`input_audio`时必填。 + +**属性** + +**data** `_string_` **（必选）** + +待识别音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。支持以下两种方式： + +- **音频文件URL**：直接传入可公开访问的音频文件地址。 + +- **Base64 Data URI**：采用Data URI格式传入Base64编码的音频数据，值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 + + +示例（URL方式）：`https://example.com/audio/sample.wav` + +示例（Base64方式）：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` + +**text** `_string_` **（条件必选）** + +当`type`为`input_text`时，填入前几轮用户语音的识别结果或领域相关的词表；当`type`为`text`时，填入前几轮大语言模型的回复内容。文本按字符数计算，每个字符计为 1。每轮上下文中所有消息的 `text` 字段长度之和不超过 400 个字符，超出部分从末尾截断。 + +**parameters** `_object_` **（必选）** + +模型参数。 + +**属性** + +**format** `_string_` **（必选）** + +音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**sample\_rate** `_string_` （可选） + +音频采样率，单位Hz。例如`16000`表示16kHz采样率。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +## **返回体** + +## 非流式 + +``` +{ + "output": { + "sentence": { + "begin_time": 760, + "channel_id": 0, + "end_time": 3800, + "sentence_end": true, + "sentence_id": 1, + "text": "Hello World，这里是阿里巴巴语音实验室。", + "words": [ + {"begin_time": 760, "end_time": 1040, "fixed": true, "punctuation": "", "text": "Hello"}, + {"begin_time": 1040, "end_time": 1240, "fixed": true, "punctuation": "，", "text": " World"}, + {"begin_time": 1360, "end_time": 1880, "fixed": true, "punctuation": "", "text": "这里是"}, + {"begin_time": 1880, "end_time": 2520, "fixed": true, "punctuation": "", "text": "阿里巴巴"}, + {"begin_time": 2520, "end_time": 2840, "fixed": true, "punctuation": "", "text": "语音"}, + {"begin_time": 2840, "end_time": 3800, "fixed": true, "punctuation": "。", "text": "实验室"} + ] + }, + "text": "Hello World，这里是阿里巴巴语音实验室。" + }, + "usage": { + "duration": 4 + }, + "request_id": "40e0734d-096f-9ae3-86c1-a8c013287561" +} +``` + +## 流式 + +设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议返回识别结果。SSE事件格式如下： + +``` +id:{序列号} +event:result +:HTTP_STATUS/200 +data:{JSON数据} +``` + +返回示例： + +``` +id:1 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":3800,"words":[{"end_time":1040,"punctuation":"","begin_time":760,"fixed":true,"text":"Hello"},{"end_time":1240,"punctuation":"，","begin_time":1040,"fixed":true,"text":" World"},{"end_time":1880,"punctuation":"","begin_time":1360,"fixed":true,"text":"这里是"},{"end_time":2520,"punctuation":"","begin_time":1880,"fixed":true,"text":"阿里巴巴"},{"end_time":2840,"punctuation":"","begin_time":2520,"fixed":true,"text":"语音"},{"end_time":3800,"punctuation":"。","begin_time":2840,"fixed":true,"text":"实验室"}],"begin_time":760,"text":"Hello World，这里是阿里巴巴语音实验室。","channel_id":0},"text":"Hello World，这里是阿里巴巴语音实验室。"},"usage":{"duration":4},"request_id":"fc1582e4-935c-9fc2-a482-a98bf43daa69"} +``` + +**request\_id** `_string_` + +本次请求的唯一标识。 + +**output** `_object_` + +输出结果。 + +**属性** + +**text** `_string_` + +当前累积的完整识别文本。 + +**sentence** `_object_` + +当前句子的详细信息。 + +**属性** + +**sentence\_id** `_integer_` + +句子编号，从1开始。 + +**sentence\_end** `_boolean_` + +是否为该句的最终结果。为`true`时表示该句识别完成。 + +**begin\_time** `_integer_` + +句子开始时间，单位毫秒。 + +**end\_time** `_integer_` + +句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 + +**text** `_string_` + +当前句子的识别文本。 + +**channel\_id** `_integer_` + +声道编号，从0开始。 + +**words** `_array_` + +词级别时间戳列表。 + +**属性** + +**text** `_string_` + +词文本。 + +**begin\_time** `_integer_` + +词开始时间，单位毫秒。 + +**end\_time** `_integer_` + +词结束时间，单位毫秒。 + +**punctuation** `_string_` + +词后的标点符号。无标点时为空字符串。 + +**fixed** `_boolean_` + +词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 + +**usage** `_object_` + +用量信息。仅在`sentence_end`为`true`时返回。 + +**属性** + +**duration** `_integer_` + +已处理的音频时长，单位秒。 + +## **SSE 流式结果处理逻辑** + +在流式模式下，客户端需关注以下处理要点： + +1. 每收到一个SSE事件，解析`data`字段中的JSON。 + +2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 + +3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md new file mode 100644 index 00000000..3c19542d --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md @@ -0,0 +1,328 @@ +# 非实时语音识别（Fun-ASR-Realtime）API参考 + +本文介绍Fun-ASR-Realtime非实时语音识别HTTP API的参数和接口细节。 + +**用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide)。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**重要** + +- 该功能只支持北京地域。 + +- 不支持SDK调用。 + + +## **服务端点** + +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +**重要** + +阿里云百炼为华北2（北京）地域推出了业务空间专属域名，能够为推理请求提供卓越的性能和更高的稳定性，建议从 `dashscope.aliyuncs.com` 迁移至 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`。 + +`{WorkspaceId}`需要替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。现有域名仍可正常使用。 + +## **请求头** + +**参数** + +**类型** + +**是否必选** + +**说明** + +Authorization + +string + +是 + +鉴权令牌，格式为`Bearer `，使用时将“``”替换为实际的API Key。 + +Content-Type + +string + +是 + +请求体的媒体类型，固定为`application/json`。 + +X-DashScope-SSE + +string + +是 + +用于控制是否以SSE流式方式返回结果。设置为`enable`时开启SSE流式返回模式，服务端会分多次返回中间识别结果和最终结果；设置为`disable`或不传该参数则仅返回最终结果。 + +## **请求体** + +## 非流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: disable" \ + --data '{ + "model": "fun-asr-realtime", + "input": { + "messages": [] + }, + "parameters": { + "audio_address": "https://example.com/audio/sample.mp3", + "format": "mp3" + }, + "resources": [] +}' +``` + +## 流式 + +``` +curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ + --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --header "Content-Type: application/json" \ + --header "X-DashScope-SSE: enable" \ + --data '{ + "model": "fun-asr-realtime", + "input": { + "messages": [] + }, + "parameters": { + "audio_address": "https://example.com/audio/sample.mp3", + "format": "mp3" + }, + "resources": [] +}' +``` + +**model** `_string_` **（必选）** + +模型名称。 + +取值范围： + +- `fun-asr-realtime`（稳定版模型） + +- `fun-asr-realtime-2026-02-28`（即 Fun-Realtime-ASR-preview） + + +**input** `_object_` **（条件必选）** + +输入信息。与输入音频文件URL方式（通过`parameters.audio_address`传入）二选一，使用Base64方式上传音频时需要填写。 + +**属性** + +**messages** `_array(object)_` **（必选）** + +消息列表。 + +**属性** + +**content** `_array(object)_` **（必选）** + +用户消息的内容。仅允许设置一组消息。 + +**属性** + +**audio** `_string_` **（必选）** + +待识别音频，采用Data URI格式传入Base64编码的音频数据。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +使用Base64方式上传音频时需要填写。 + +值由`data:{MIME_TYPE};base64,`前缀与Base64编码的音频数据拼接而成。支持的MIME类型包括`audio/wav`、`audio/mp3`等。 + +示例：`data:audio/wav;base64,{BASE64_ENCODED_DATA}` + +**role** `_string_` **（必选）** + +用户消息的角色，固定为`user`。使用Base64方式上传音频时需要填写。 + +**parameters** `_object_` **（必选）** + +模型参数。 + +**属性** + +**audio\_address** `_string_` **（条件必选）** + +音频文件URL地址。与Base64方式（通过`input.messages`传入）二选一，使用URL方式时必填。需为可公开访问的地址。关于支持的音频格式、文件大小限制、时长限制等输入要求，请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**format** `_string_` **（必选）** + +音频格式。根据实际音频格式填写，支持`wav`、`mp3`、`opus`等。详情请参见[音频规格](https://help.aliyun.com/zh/model-studio/asr-model/#asr-audio-spec02)。 + +**vad\_enabled** `_boolean_` （可选） + +是否启用端点检测（VAD）。 + +默认为`true`（启用 VAD 检测）。 + +设为 `false` 时关闭 VAD 检测；但当音频时长超过 1 分钟时，系统将自动启用 VAD 检测，忽略此参数设置。 + +**resources** `_array_` （可选） + +资源列表，预留字段，当前可传空数组`[]`。 + +## **返回体** + +## 非流式 + +``` +{ + "output": { + "sentence": { + "begin_time": 160, + "channel_id": 0, + "end_time": 1680, + "sentence_end": true, + "sentence_id": 1, + "text": "欢迎使用阿里云。", + "words": [ + {"begin_time": 160, "end_time": 520, "fixed": true, "punctuation": "", "text": "欢迎"}, + {"begin_time": 520, "end_time": 880, "fixed": true, "punctuation": "", "text": "使用"}, + {"begin_time": 880, "end_time": 1280, "fixed": true, "punctuation": "", "text": "阿里"}, + {"begin_time": 1280, "end_time": 1680, "fixed": true, "punctuation": "。", "text": "云"} + ] + }, + "text": "欢迎使用阿里云。" + }, + "usage": { + "duration": 2 + }, + "request_id": "eff4c092-2289-9b43-a4cd-80e591fa90f5" +} +``` + +## 流式 + +设置`X-DashScope-SSE: enable`时，服务端以Server-Sent Events协议分多次返回中间识别结果和最终结果。每个SSE事件格式如下： + +``` +id:{序列号} +event:result +:HTTP_STATUS/200 +data:{JSON数据} +``` + +中间结果（句子开始、词逐步增长）： + +``` +id:1 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":false,"sentence_begin":true,"words":[],"begin_time":0,"text":"","channel_id":0},"text":""},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} + +id:2 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"}],"begin_time":160,"text":"欢迎","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} + +id:3 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":false,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":false,"text":"使用"}],"begin_time":160,"text":"欢迎使用","channel_id":0,"sentence_id":1,"sentence_end":false},"text":"欢迎使用"},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} +``` + +最终结果（句子结束，包含`usage`）： + +``` +id:4 +event:result +:HTTP_STATUS/200 +data:{"output":{"sentence":{"sentence_id":1,"sentence_end":true,"end_time":1680,"words":[{"end_time":520,"punctuation":"","begin_time":160,"fixed":true,"text":"欢迎"},{"end_time":880,"punctuation":"","begin_time":520,"fixed":true,"text":"使用"},{"end_time":1280,"punctuation":"","begin_time":880,"fixed":true,"text":"阿里"},{"end_time":1680,"punctuation":"。","begin_time":1280,"fixed":true,"text":"云"}],"begin_time":160,"text":"欢迎使用阿里云。","channel_id":0},"text":"欢迎使用阿里云。"},"usage":{"duration":2},"request_id":"372d19b3-993f-9288-adf0-a99f7606bd30"} +``` + +**request\_id** `_string_` + +本次请求的唯一标识。 + +**output** `_object_` + +输出结果。 + +**属性** + +**text** `_string_` + +当前累积的完整识别文本。 + +**sentence** `_object_` + +当前句子的详细信息。 + +**属性** + +**sentence\_id** `_integer_` + +句子编号，从1开始。 + +**sentence\_end** `_boolean_` + +是否为该句的最终结果。为`true`时表示该句识别完成。 + +**begin\_time** `_integer_` + +句子开始时间，单位毫秒。 + +**end\_time** `_integer_` + +句子结束时间，单位毫秒。仅在`sentence_end`为`true`时返回。 + +**text** `_string_` + +当前句子的识别文本。 + +**channel\_id** `_integer_` + +声道编号，从0开始。 + +**words** `_array_` + +词级别时间戳列表。 + +**属性** + +**text** `_string_` + +词文本。 + +**begin\_time** `_integer_` + +词开始时间，单位毫秒。 + +**end\_time** `_integer_` + +词结束时间，单位毫秒。 + +**punctuation** `_string_` + +词后的标点符号。无标点时为空字符串。 + +**fixed** `_boolean_` + +词是否已稳定。`false`表示后续事件中该词的时间戳可能调整。 + +**usage** `_object_` + +用量信息。仅在`sentence_end`为`true`时返回。 + +**属性** + +**duration** `_integer_` + +已处理的音频时长，单位秒。 + +## **SSE 流式结果处理逻辑** + +在流式模式下，客户端需关注以下处理要点： + +1. 每收到一个SSE事件，解析`data`字段中的JSON。 + +2. 通过`output.sentence.sentence_end`判断当前句子是否结束：当该值为`true`时，该句识别完成，词级时间戳已稳定，可作为最终结果使用；当该值为`false`时，识别仍在进行中，文本和时间戳可能在后续事件中更新。 + +3. `usage`信息仅在句子结束事件中返回，可用于计量音频处理时长。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md index 2872f1a4..b9a5a871 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md @@ -143,14 +143,12 @@ 提交单个语音实时转写任务，通过传入本地文件的方式同步阻塞地拿到转写结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9365892871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3558074871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) 实例化[Recognition类](#adcb5e9bddbyq)，调用`call`方法绑定[请求参数](#d72d661a1brzp)和待识别文件，进行识别并最终获取识别结果。 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/elouas/asr_example.wav)。 - ``` import com.alibaba.dashscope.audio.asr.recognition.Recognition; import com.alibaba.dashscope.audio.asr.recognition.RecognitionParam; @@ -161,7 +159,7 @@ import java.io.File; public class Main { public static void main(String[] args) { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; // 创建Recognition实例 Recognition recognizer = new Recognition(); // 创建RecognitionParam @@ -177,7 +175,7 @@ public class Main { .build(); try { - System.out.println("识别结果：" + recognizer.call(param, new File("asr_example.wav"))); + System.out.println("识别结果：" + recognizer.call(param, new File("{YOUR_AUDIO_FILE}"))); } catch (Exception e) { e.printStackTrace(); } finally { @@ -200,7 +198,7 @@ public class Main { 提交单个语音实时转写任务，通过实现回调接口的方式流式输出实时识别结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9365892871/CAEQURiBgID1ooWUpxkiIDcyOTEyYjZiZmUxNzRkZjVhMTNhYmNkYjI2NzYzYTMy4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3558074871/CAEQURiBgID1ooWUpxkiIDcyOTEyYjZiZmUxNzRkZjVhMTNhYmNkYjI2NzYzYTMy4709861_20241015153444.149.svg) 1. 启动流式语音识别 @@ -244,7 +242,7 @@ import java.util.concurrent.TimeUnit; public class Main { public static void main(String[] args) throws InterruptedException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; ExecutorService executorService = Executors.newSingleThreadExecutor(); executorService.submit(new RealtimeRecognitionTask()); executorService.shutdown(); @@ -332,8 +330,6 @@ class RealtimeRecognitionTask implements Runnable { ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/oiydrd/asr_example.wav)。 - ``` import com.alibaba.dashscope.audio.asr.recognition.Recognition; import com.alibaba.dashscope.audio.asr.recognition.RecognitionParam; @@ -364,9 +360,9 @@ class TimeUtils { public class Main { public static void main(String[] args) throws InterruptedException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; ExecutorService executorService = Executors.newSingleThreadExecutor(); - executorService.submit(new RealtimeRecognitionTask(Paths.get(System.getProperty("user.dir"), "asr_example.wav"))); + executorService.submit(new RealtimeRecognitionTask(Paths.get(System.getProperty("user.dir"), "{YOUR_AUDIO_FILE}"))); executorService.shutdown(); // wait for all tasks to complete @@ -503,7 +499,7 @@ import java.nio.ByteBuffer; public class Main { public static void main(String[] args) throws NoApiKeyException { // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 - Constants.baseWebsocketApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + Constants.baseWebsocketApiUrl = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference"; // 创建一个Flowable Flowable audioSource = Flowable.create( diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md index bba41596..38973852 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md @@ -143,19 +143,17 @@ 提交单个语音实时转写任务，通过传入本地文件的方式同步阻塞地拿到转写结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6455892871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4858074871/CAEQURiBgMDS0c2RpxkiIDNmYjBlMTE3ODQxYTQ3Nzk4MGMxNTc5MjY3OWVjZjlj4709861_20241015153444.149.svg) 实例化[Recognition类](#d6bc1f133f871)绑定[请求参数](#555007db2033f)，调用`call`进行识别/翻译并最终获取[识别结果（RecognitionResult）](#bc3e1a43d6hhy)。 点击查看完整示例 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/iwaouc/asr_example.wav)。 - ``` from http import HTTPStatus from dashscope.audio.asr import Recognition # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" # 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key # import dashscope @@ -167,13 +165,15 @@ recognition = Recognition(model='paraformer-realtime-v2', # “language_hints”只支持paraformer-realtime-v2模型 language_hints=['zh', 'en'], callback=None) -result = recognition.call('asr_example.wav') +result = recognition.call('{YOUR_AUDIO_FILE}') if result.status_code == HTTPStatus.OK: print('识别结果：') - print(result.get_sentence()) + sentences = result.get_sentence() + for sentence in sentences: + print(sentence['text']) else: print('Error: ', result.message) - + print( '[Metric] requestId: {}, first package delay ms: {}, last package delay ms: {}' .format( @@ -183,11 +183,15 @@ print( )) ``` +`result.get_sentence()`在非流式调用（`call`）中返回**句子列表**（`List[Dict]`），每个元素为`Dict[str, Any]`，包含`text`（识别文本）、`begin_time` / `end_time`（时间戳）、`words`（字时间戳）等字段。如需获取识别文本，需遍历列表并通过`sentence['text']`提取。 + +在流式回调（`on_event`）中，`result.get_sentence()`返回**单句信息**（`Dict[str, Any]`），可直接使用`sentence['text']`获取识别文本。两种调用模式的返回值类型不同，详见[识别结果（RecognitionResult）](#bc3e1a43d6hhy)中的`get_sentence`方法说明。 + ### **双向流式调用** 提交单个语音实时转写任务，通过实现回调接口的方式流式输出实时识别结果。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6455892871/CAEQURiBgIDvi..2pxkiIGE4NTc3Njg4ZGM2YzQ2NzVhZGI3MzE2YWUwYTA3OGEy4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4858074871/CAEQURiBgIDvi..2pxkiIGE4NTc3Njg4ZGM2YzQ2NzVhZGI3MzE2YWUwYTA3OGEy4709861_20241015153444.149.svg) 1. 启动流式语音识别 @@ -221,7 +225,7 @@ import dashscope import pyaudio from dashscope.audio.asr import * # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" mic = None stream = None @@ -342,14 +346,12 @@ if __name__ == '__main__': ## 识别本地语音文件 -示例中用到的音频为：[asr\_example.wav](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250210/acoict/asr_example.wav)。 - ``` import os import time from dashscope.audio.asr import * # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 -dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" +dashscope.base_websocket_api_url = "wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/inference" # 若没有将API Key配置到环境变量中，需将下面这行代码注释放开，并将apiKey替换为自己的API Key # import dashscope @@ -393,8 +395,8 @@ recognition.start() try: audio_data: bytes = None - f = open("asr_example.wav", 'rb') - if os.path.getsize("asr_example.wav"): + f = open("{YOUR_AUDIO_FILE}", 'rb') + if os.path.getsize("{YOUR_AUDIO_FILE}"): while True: audio_data = f.read(3200) if not audio_data: diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md index ff800b01..697261be 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Android SDK +# Paraformer非实时语音识别Android SDK -本文档提供了Paraformer录音文件识别Android SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Paraformer非实时语音识别Android SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) @@ -17,7 +17,10 @@ - [下载最新SDK整合包](https://help.aliyun.com/zh/isi/sdk-selection-and-download)。 - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 - 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + + + - 用 Android Studio 打开工程。示例代码位于`DashParaformerFileTranscriberActivity.java`，替换 API Key 后体验功能。 @@ -103,7 +106,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -174,7 +177,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -617,7 +620,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md index 2147dda6..0ecf2ea8 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别iOS SDK +# Paraformer非实时语音识别iOS SDK -本文档提供了Paraformer录音文件识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 +本文档提供了Paraformer非实时语音识别iOS SDK的详细使用指南，帮助您将语音转换为文本。 **用户指南：**[非实时语音识别](https://help.aliyun.com/zh/model-studio/non-realtime-speech-recognition-user-guide) @@ -106,7 +106,7 @@ 是 - 运行模式。录音文件识别固定为 `"1"`。 + 运行模式。非实时语音识别固定为 `"1"`。 `device_id` @@ -177,7 +177,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { @@ -611,7 +611,7 @@ ``` { "file_urls": [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ], "async_request": false, "nls_config": { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md index b3d42809..65269d08 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Java SDK +# Paraformer非实时语音识别Java SDK -本文介绍Paraformer录音文件识别Java SDK的参数和接口细节。 +本文介绍Paraformer非实时语音识别Java SDK的参数和接口细节。 **重要** @@ -27,7 +27,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -36,7 +36,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7695892871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0509074871/CAEQURiBgMCO2_fRpxkiIDBlNzI4YmMyNTU3ODRlM2Y4NjUxZWU4YmUxNjliMmFl4709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -80,7 +80,7 @@ public class Main { .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -102,7 +102,7 @@ public class Main { ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7695892871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0509074871/CAEQURiBgIDnxvjRpxkiIGI1NjJjOTgyNTVhMTRiMjM4OWVjYzFmZTExNGZjYzE14709861_20241015153444.149.svg) 1. 配置[请求参数](#48ea212b1d08r)。 @@ -147,7 +147,7 @@ public class Main { .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -185,7 +185,7 @@ TranscriptionParam param = TranscriptionParam.builder() .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); ``` @@ -537,7 +537,7 @@ public JsonObject getOutput() "end_time":"2025-02-13 16:12:10.189", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/16%3A12/3baafe5f-d09d-46c6-8b01-724927670edb-1.json?Expires=1739520730&OSSAccessKeyId=yourOSSAccessKeyId&Signature=BF7vPxlsJN9hkJlY%2BLReezxOwK8%3D", "subtask_status":"SUCCEEDED" } @@ -563,7 +563,7 @@ public JsonObject getOutput() "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -645,7 +645,7 @@ public String getMessage() ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -807,7 +807,7 @@ TranscriptionParam param = .parameter("language_hints", new String[]{"zh", "en"}) .fileUrls( Arrays.asList( - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav")) + "{YOUR_AUDIO_URL}")) .build(); try { Transcription transcription = new Transcription(); @@ -886,7 +886,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -907,7 +907,7 @@ public TranscriptionResult fetch(TranscriptionQueryParam queryParam) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md index e3ffe0b6..175befce 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别Python SDK +# Paraformer非实时语音识别Python SDK -本文介绍Paraformer录音文件识别Python SDK的参数和接口细节。 +本文介绍Paraformer非实时语音识别Python SDK的参数和接口细节。 **重要** @@ -27,7 +27,7 @@ ## **快速开始** -[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行录音文件识别： +[核心类（Transcription）](#adcb5e9bddbyq)提供了异步提交任务、同步等待任务结束和异步查询任务执行结果的接口。可通过如下两种调用方式进行非实时语音识别： - 异步提交任务+同步等待任务结束：提交任务后，阻塞当前线程直到任务结束并获取识别结果。 @@ -36,7 +36,7 @@ ### **异步提交任务+同步等待任务结束** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2057892871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1609074871/CAEQURiBgMCvo5zjpxkiIDQyNzUwZjVjMWM3MjQ5Nzg4ODBjNDRjNzE1ZGFiOGFj4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -69,7 +69,7 @@ dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co task_response = Transcription.async_call( model='paraformer-v2', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'], + file_urls=['{YOUR_AUDIO_URL}'], language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型 ) @@ -81,7 +81,7 @@ if transcribe_response.status_code == HTTPStatus.OK: ### **异步提交任务+异步查询任务执行结果** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2057892871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2609074871/CAEQURiBgMCN3qzkpxkiIGE0YmU4YTdjMWNiNzRmYjJhMjFlMWZkZmFmOWQ1NmEx4709861_20241015153444.149.svg) 1. 调用[核心类（Transcription）](#adcb5e9bddbyq)的`async_call`方法并设置[请求参数](#340f6879fci7d)。 @@ -114,7 +114,7 @@ dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co transcribe_response = Transcription.async_call( model='paraformer-v2', - file_urls=['https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav'], + file_urls=['{YOUR_AUDIO_URL}'], language_hints=['zh', 'en'] # “language_hints”只支持paraformer-v2模型 ) @@ -451,7 +451,7 @@ int "end_time":"2025-02-13 17:31:21.867", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A31/20ee4e4f-0404-4806-b617-c7d4c62eed19-1.json?Expires=1739525481&OSSAccessKeyId=yourOSSAccessKeyId&Signature=3q%2B1uQmRwltd7FPn5HQM2mBKw74%3D", "subtask_status":"SUCCEEDED" } @@ -484,7 +484,7 @@ int "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -607,7 +607,7 @@ transcription\_url "end_time":"2025-02-13 17:59:28.828", "results":[ { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "transcription_url":"https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/prod/paraformer-v2/20250213/17%3A59/70e737cc-bf8c-418b-b0c8-83fab192a0fa-1.json?Expires=1739527168&OSSAccessKeyId=yourOSSAccessKeyId&Signature=AtGjIKI%2BdgbzjJIu%2BHsr1R5nSAY%3D", "subtask_status":"SUCCEEDED" } @@ -633,7 +633,7 @@ transcription\_url "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -701,7 +701,7 @@ transcription\_url ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -903,7 +903,7 @@ def fetch(cls, ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **错误码** @@ -924,7 +924,7 @@ def fetch(cls, "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md index 971b8184..2fb481c4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-restful-api.md @@ -1,6 +1,6 @@ -# Paraformer录音文件识别RESTful API +# Paraformer非实时语音识别HTTP API -本文介绍Paraformer录音文件识别RESTful API的参数和接口细节。 +本文介绍Paraformer非实时语音识别HTTP API的参数和接口细节。 **重要** @@ -59,7 +59,7 @@ X-DashScope-Async: enable // 请勿遗漏该请求头，否则无法提交任务 "model":"paraformer-v2", //模型名，必选 "input":{ "file_urls":[ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav" + "{YOUR_AUDIO_URL}" ] //待识别文件，必选 } "parameters":{ @@ -90,7 +90,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header "Content-Type: application/json" \ --header "X-DashScope-Async: enable" \ - --data '{"model":"paraformer-v2","input":{"file_urls":["https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav"]},"parameters":{"channel_id":[0]}}' + --data '{"model":"paraformer-v2","input":{"file_urls":["{YOUR_AUDIO_URL}"]},"parameters":{"channel_id":[0]}}' ``` **参数** @@ -450,7 +450,7 @@ string "end_time": "2024-09-12 15:11:40.903", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url": "{YOUR_AUDIO_URL}", "transcription_url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/pre/filetrans-16k/20240912/15%3A11/409a4b92-445b-4dd8-8c1d-f110954d82d8-1.json?Expires=1726211500&OSSAccessKeyId=yourOSSAccessKeyId&Signature=v5Owy5qoAfT7mzGmQgH0g8C****%3D", "subtask_status": "SUCCEEDED" } @@ -480,7 +480,7 @@ string "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" @@ -546,7 +546,7 @@ JSON数据中各字段含义请参见[识别结果说明](#a9021178ccl7s)。 ``` { - "file_url":"https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "file_url":"{YOUR_AUDIO_URL}", "properties":{ "audio_format":"pcm_s16le", "channels":[ @@ -689,7 +689,7 @@ string ## **其他接口：批量查询任务状态/取消任务** -详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的录音文件识别任务，同时支持取消`PENDING`（排队）状态的任务。 +详情请参见[管理异步任务](https://help.aliyun.com/zh/model-studio/manage-asynchronous-tasks)：支持批量查询24小时内提交的非实时语音识别任务，同时支持取消`PENDING`（排队）状态的任务。 ## **完整示例** @@ -704,7 +704,7 @@ import time api_key = "your-dashscope-api-key" # 在此处替换为您的API Key file_urls = [ - "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/paraformer/hello_world_female2.wav", + "{YOUR_AUDIO_URL}", ] language_hints = ["zh", "en"] @@ -797,7 +797,7 @@ print("transcription result: ", result) "end_time": "2024-12-16 16:31:02.375", "results": [ { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/samples/audio/sensevoice/rich_text_exaple_1.wav", + "file_url": "{YOUR_AUDIO_URL}", "code": "InvalidFile.DownloadFailed", "message": "The audio file cannot be downloaded.", "subtask_status": "FAILED" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md index 4a80c542..e847488f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md @@ -1,4 +1,4 @@ -# 录音文件识别（Qwen-ASR）API参考 +# 非实时语音识别（Qwen-ASR）API参考 本文介绍 Qwen-ASR 模型的输入与输出参数。可通过OpenAI 兼容或DashScope协议调用 API。 @@ -70,7 +70,7 @@ try: # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key = "sk-xxx", api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -84,7 +84,7 @@ try: { "type": "input_audio", "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "data": "{YOUR_AUDIO_URL}" } } ], @@ -130,7 +130,7 @@ const client = new OpenAI({ // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey: "sk-xxx", apiKey: process.env.DASHSCOPE_API_KEY, - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); @@ -146,7 +146,7 @@ async function main() { { type: "input_audio", input_audio: { - data: "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + data: "{YOUR_AUDIO_URL}" } } ] @@ -157,11 +157,9 @@ async function main() { // stream_options: { // "include_usage": true // }, - extra_body: { - asr_options: { - // language: "zh", - enable_itn: false - } + asr_options: { + // language: "zh", + enable_itn: false } }); @@ -191,7 +189,7 @@ main(); #### cURL -以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +以下为华北2（北京）地域的配置，调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)，各地域的配置不同。 ``` curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions' \ @@ -205,7 +203,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode { "type": "input_audio", "input_audio": { - "data": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "data": "{YOUR_AUDIO_URL}" } } ], @@ -245,7 +243,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode import base64, pathlib # input.mp3为用于声音复刻的本地音频文件，请替换为自己的音频文件路径，确保其符合音频要求 - file_path = pathlib.Path("input.mp3") + file_path = pathlib.Path("{YOUR_AUDIO_FILE}") base64_str = base64.b64encode(file_path.read_bytes()).decode() data_uri = f"data:audio/mpeg;base64,{base64_str}" ``` @@ -268,7 +266,7 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode // 使用示例 public static void main(String[] args) throws Exception { - System.out.println(toDataUrl("input.mp3")); + System.out.println(toDataUrl("{YOUR_AUDIO_FILE}")); } } ``` @@ -276,8 +274,6 @@ curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode #### Python SDK -示例中用到的音频文件为：[welcome.mp3](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260105/wotsae/welcome.mp3)。 - ``` import base64 from openai import OpenAI @@ -286,7 +282,7 @@ import pathlib try: # 请替换为实际的音频文件路径 - file_path = "welcome.mp3" + file_path = "{YOUR_AUDIO_FILE}" # 请替换为实际的音频文件MIME类型 audio_mime_type = "audio/mpeg" @@ -301,7 +297,7 @@ try: # 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key # 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：api_key = "sk-xxx", api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + # 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", ) @@ -349,8 +345,6 @@ except Exception as e: #### Node.js SDK -示例中用到的音频文件为：[welcome.mp3](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260105/wotsae/welcome.mp3)。 - ``` // 运行前的准备工作: // Windows/Mac/Linux 通用: @@ -364,8 +358,8 @@ const client = new OpenAI({ // 新加坡和北京地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：apiKey: "sk-xxx", apiKey: process.env.DASHSCOPE_API_KEY, - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 + baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", }); const encodeAudioFile = (audioFilePath) => { @@ -374,7 +368,7 @@ const encodeAudioFile = (audioFilePath) => { }; // 请替换为实际的音频文件路径 -const dataUri = `data:audio/mpeg;base64,${encodeAudioFile("welcome.mp3")}`; +const dataUri = `data:audio/mpeg;base64,${encodeAudioFile("{YOUR_AUDIO_FILE}")}`; async function main() { try { @@ -399,11 +393,9 @@ async function main() { // stream_options: { // "include_usage": true // }, - extra_body: { - asr_options: { - // language: "zh", - enable_itn: false - } + asr_options: { + // language: "zh", + enable_itn: false } }); @@ -443,7 +435,7 @@ main(); System Message `_object_`（可选） -模型的目标或角色。如果设置系统消息，请放在messages列表的第一位。 +用于为语音识别提供上下文（Context），如背景文本和实体词表等参考信息，不支持设置模型角色等传统系统提示词。如果设置系统消息，请放在messages列表的第一位。 **属性** @@ -912,15 +904,7 @@ curl -X POST "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services { "content": [ { - "text": "" - } - ], - "role": "system" - }, - { - "content": [ - { - "audio": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "audio": "{YOUR_AUDIO_URL}" } ], "role": "user" @@ -961,7 +945,7 @@ public class Main { MultiModalMessage userMessage = MultiModalMessage.builder() .role(Role.USER.getValue()) .content(Arrays.asList( - Collections.singletonMap("audio", "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3"))) + Collections.singletonMap("audio", "{YOUR_AUDIO_URL}"))) .build(); Map asrOptions = new HashMap<>(); @@ -981,7 +965,7 @@ public class Main { } public static void main(String[] args) { try { - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 + // 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; simpleMultiModalConversationCall(); } catch (ApiException | NoApiKeyException | UploadFileException e) { @@ -998,11 +982,11 @@ public class Main { import os import dashscope -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +# 以下为华北2（北京）地域的配置，调用时请将"{WorkspaceId}"替换为真实的业务空间ID，各地域的配置不同。 dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' messages = [ - {"role": "user", "content": [{"audio": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3"}]} + {"role": "user", "content": [{"audio": "{YOUR_AUDIO_URL}"}]} ] response = dashscope.MultiModalConversation.call( @@ -1035,7 +1019,7 @@ print(response) System Message `_object_`（可选） -模型的目标或角色。如果设置系统消息，请放在messages列表的第一位。 +用于为语音识别提供上下文（Context），如背景文本和实体词表等参考信息，不支持设置模型角色等传统系统提示词。如果设置系统消息，请放在messages列表的第一位。 仅千问3-ASR-Flash支持该参数。 @@ -1460,7 +1444,7 @@ curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.c --data '{ "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id":[ @@ -1498,7 +1482,7 @@ public class Main { { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -1514,7 +1498,7 @@ public class Main { { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -1595,7 +1579,7 @@ headers = { payload = { "model": "qwen3-asr-flash-filetrans", "input": { - "file_url": "https://dashscope.oss-cn-beijing.aliyuncs.com/audios/welcome.mp3" + "file_url": "{YOUR_AUDIO_URL}" }, "parameters": { "channel_id": [0], @@ -2010,6 +1994,14 @@ print(response.json()) + + + + + + + + 详情参见[异步调用识别结果说明](#2c27ad3e80p4y)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md index 9839c136..d5bc3a81 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md @@ -100,7 +100,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 **启用方式：**配置客户端`[session.update](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#af43722339yva)`事件的`session.turn_detection`参数。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1592892871/CAEQaxiBgICW9b6H3RkiIGM5MDgwMTNkMjBjMDRlNTNiOGZlODNjZGJhNDQ3NGJm5812623_20251022102739.334.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7151064871/CAEQaxiBgICW9b6H3RkiIGM5MDgwMTNkMjBjMDRlNTNiOGZlODNjZGJhNDQ3NGJm5812623_20251022102739.334.svg) - 客户端通过发送`[input_audio_buffer.append](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#a42f8e9111n72)`事件将音频追加到缓冲区。 @@ -112,6 +112,10 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 - 客户端在音频提交完后，发送`[session.finish](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#147ce70052d4z)`事件通知服务端结束当前会话。 + **警告** + + 在 VAD 模式下，推完音频后必须先发送`[session.finish](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#147ce70052d4z)`事件再关闭连接。如果客户端直接关闭 WebSocket 连接而未发送该事件，服务端将丢弃当前 in\_progress item，`conversation.item.input_audio_transcription.completed` 等事件将不会到达。建议在调用 `ws.Close()` 之前，先发送 `{"type":"session.finish"}`，并等待收到`[session.finished](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#6eaa77339djdv)`事件后再关闭连接。 + - 服务端在检测到语音结束时返回`[input_audio_buffer.speech_stopped](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#3d73b074cak7k)`事件。 - 服务端返回`[input_audio_buffer.committed](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-server-events#1108a3764an0e)`事件。 @@ -131,7 +135,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 **启用方式：**将客户端`[session.update](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#af43722339yva)`事件的`session.turn_detection`设为null。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1592892871/CAEQaxiBgMDUp8qH3RkiIGEyYTc0NTI1ZmQ1OTQ5NjliNWE0OTYwYTAwMDBlMjBm5812623_20251022102739.334.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8151064871/CAEQaxiBgMDUp8qH3RkiIGEyYTc0NTI1ZmQ1OTQ5NjliNWE0OTYwYTAwMDBlMjBm5812623_20251022102739.334.svg) - 客户端通过发送`[input_audio_buffer.append](https://help.aliyun.com/zh/model-studio/qwen-asr-realtime-client-events#a42f8e9111n72)`事件将音频追加到缓冲区。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md index 3d4ec241..2c798bb3 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md @@ -277,7 +277,7 @@ API Key。建议使用时效性短、安全性更高的[临时API Key](https://h 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -394,7 +394,7 @@ SSML 的使用限制（支持的模型、音色和接口），请参见[使用 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 > 时间戳结果在[INativeStreamInputTtsCallback](#secstreamcallback)的all\_response中。 @@ -440,7 +440,7 @@ cosyvoice-v1不支持该参数。 - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -460,9 +460,15 @@ cosyvoice-v1不支持该参数。 - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 `instruction` @@ -834,6 +840,7 @@ STREAM\_INPUT\_TTS\_EVENT\_TASK\_FAILED - 解压 ZIP 包。在 `app/libs` 目录中获取 AAR 格式 SDK，并添加到项目依赖。 需要 Android CPP 接入时，使用 ZIP 包内的 `android_libs` 与 `android_include` 获取动态库和头文件。 + - 用 Android Studio 打开工程。示例代码位于`DashCosyVoiceStreamTtsActivity.java`，替换 API Key 后体验功能。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md index dabe4c85..69c32225 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md @@ -91,7 +91,7 @@ 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -173,7 +173,7 @@ SSML 的使用限制（支持的模型、音色和接口），请参见[使用 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **seed** `_integer_` （可选） @@ -209,7 +209,7 @@ cosyvoice-v1不支持该参数。 - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -229,9 +229,15 @@ cosyvoice-v1不支持该参数。 - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 **instruction** `_string_` （可选） @@ -365,7 +371,7 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 ## **finish-task** -**说明**：通知服务端文本发送完毕，请求结束任务。 +**说明**：通知服务端文本发送完毕，请求结束任务。如需取消当前轮次的语音合成任务，可在 `input` 中设置 `directive` 为 `cancel`。 **发送时机**：所有文本发送完毕后立即发送。 @@ -400,10 +406,41 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 } ``` +**取消任务示例**： + +``` +{ + "header": { + "action": "finish-task", + "task_id": "2bf83b9a-baeb-4fda-8d9a-xxxxxxxxxxxx", + "streaming": "duplex" + }, + "payload": { + "input": { + "directive": "cancel" + } + } +} +``` + **payload** `_object_` **（必选）** **属性** **input** `_object_` **（必选）** -固定为 `{}`。 +任务输入。为空对象 `{}` 时表示正常结束任务；包含 `directive` 时可用于取消当前轮次的语音合成任务。 + +**directive** `_string_` （可选） + +控制任务结束行为。当前仅支持取值为 `cancel`，表示取消当前轮次的语音合成任务，服务端会立即返回 `task-finished` 事件，且不会输出后续音频。 + +取消后，可在当前 WebSocket 连接上重新发起语音合成任务（发送新的 `run-task` 事件），无需重新建立连接。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md index 5a534022..6631da6e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md @@ -243,7 +243,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 语音合成所使用的音色。 - - **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) + - **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -360,7 +360,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 默认值：false。 - 仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 + 仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 > 时间戳结果在[onStreamInputTtsEventCallback](#bea29bbafcosq)的all\_response中。 @@ -406,7 +406,7 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -426,9 +426,15 @@ Qwen-Audio-TTS/CosyVoice 支持一次性输入和流式输入两种调用方式 - vi：越南语 + - es：西班牙语 + - it：意大利语 - - ms：马来语 + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 `instruction` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md index f3e4feda..d8dec276 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md @@ -121,6 +121,29 @@ public void streamingComplete() 结束双向流式调用，通知服务端所有文本已发送完毕。 +### **streamingCancel() - 取消双向流式调用** + +**方法签名**： + +``` +public void streamingCancel() +``` + +**说明**：取消当前轮次的双向流式语音合成任务。调用后，SDK 会立即结束当前任务。取消后可在当前连接上继续发起新的合成任务，无需重新初始化 `SpeechSynthesizer` 实例。 + +**重要** + +**版本要求**：使用该功能需要 Java SDK 版本不低于 2.22.26。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 + + ### **callAsFlowable() - 单向流式合成（响应式）** **方法签名**： @@ -242,7 +265,7 @@ public long getFirstPackageDelay() ``` SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 模型 - .voice("longanlingxi") // 音色 + .voice("longanhuan_v3.6") // 音色 .format(SpeechSynthesisAudioFormat.WAV_8000HZ_MONO_16BIT) // 音频编码格式、采样率 .volume(50) // 音量，取值范围：[0, 100] .speechRate(1.0f) // 语速，取值范围：[0.5, 2] @@ -278,7 +301,7 @@ String 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -345,7 +368,7 @@ boolean 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 `seed(int)` @@ -391,7 +414,7 @@ List - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -411,9 +434,15 @@ List - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 `instruction(String)` @@ -458,7 +487,7 @@ paramHotFix.setReplace(replaceItems); SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 模型 - .voice("your_voice") // 替换成qwen-audio-3.0-tts-flash复刻音色 + .voice("longanhuan_v3.6") // 音色 .hotFix(paramHotFix) .build(); ``` @@ -488,7 +517,7 @@ Map ``` SpeechSynthesisParam param = SpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_markdown_filter", true) .build(); ``` @@ -842,7 +871,7 @@ SDK提供了语音合成的关键接口，支持以下几种调用方式： ### **非流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQURiBgIDHpsn4phkiIDQ0ZGE2OTk3NmY5NTRhNDVhZDQwNWE3ZGZiMzk4Yjk54709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6538354871/CAEQURiBgIDHpsn4phkiIDQ0ZGE2OTk3NmY5NTRhNDVhZDQwNWE3ZGZiMzk4Yjk54709861_20241015153444.149.svg) 发送的文本长度不得超过20000字符。 @@ -864,7 +893,7 @@ public class Main { // 模型 private static String model = "qwen-audio-3.0-tts-flash"; // 音色 - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() { // 请求参数 @@ -917,7 +946,7 @@ public class Main { ### **单向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQVRiBgMCfo..hrBkiIGEyMjNkZjVlMWZiYzRhZDU4ZjEyZjdjMmMzYjM1YzMz4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7538354871/CAEQVRiBgMCfo..hrBkiIGEyMjNkZjVlMWZiYzRhZDU4ZjEyZjdjMmMzYjM1YzMz4709861_20241015153444.149.svg) 发送的文本长度不得超过20000字符。 @@ -949,7 +978,7 @@ public class Main { // 模型 private static String model = "qwen-audio-3.0-tts-flash"; // 音色 - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() { CountDownLatch latch = new CountDownLatch(1); @@ -1024,7 +1053,7 @@ public class Main { ### 双向流式调用 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1094204871/CAEQVRiBgICHxPGhrBkiIGE3ZTVmMzY0YzI3NzQxYTFiYWE2MmU2NTBhMDgzZGM14709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7538354871/CAEQVRiBgICHxPGhrBkiIGE3ZTVmMzY0YzI3NzQxYTFiYWE2MmU2NTBhMDgzZGM14709861_20241015153444.149.svg) 单次发送文本长度不得超过 20000 字符，且累计发送文本总长度不得超过 20 万字符。 @@ -1076,7 +1105,7 @@ public class Main { "减少了用户等待时间。", "适用于调用大规模", "语言模型（LLM），以", "流式输入文本的方式", "进行语音合成的场景。"}; private static String model = "qwen-audio-3.0-tts-flash"; // 模型 - private static String voice = "longanlingxi"; // 音色 + private static String voice = "longanhuan_v3.6"; // 音色 public static void streamAudioDataToSpeaker() { // 配置回调函数 @@ -1177,7 +1206,7 @@ class TimeUtils { public class Main { private static String model = "qwen-audio-3.0-tts-flash"; // 模型 - private static String voice = "longanlingxi"; // 音色 + private static String voice = "longanhuan_v3.6"; // 音色 public static void streamAudioDataToSpeaker() throws NoApiKeyException { // 请求参数 @@ -1252,7 +1281,7 @@ public class Main { "减少了用户等待时间。", "适用于调用大规模", "语言模型（LLM），以", "流式输入文本的方式", "进行语音合成的场景。"}; private static String model = "qwen-audio-3.0-tts-flash"; - private static String voice = "longanlingxi"; + private static String voice = "longanhuan_v3.6"; public static void streamAudioDataToSpeaker() throws NoApiKeyException { // 模拟流式输入 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md index 8e94f9c5..4d00d326 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md @@ -125,6 +125,47 @@ def streaming_complete(self) -> None **说明**：通知服务端所有文本已发送完毕，阻塞当前线程直到剩余文本合成完成并返回所有音频数据。未调用此方法可能导致尾部文本无法转换为语音。 +### **streaming\_cancel() - 取消流式合成** + +**方法签名**： + +``` +def streaming_cancel(self, complete_timeout_millis: int = 10000) -> None +``` + +**参数说明**： + +**参数** + +**类型** + +**必填** + +**说明** + +complete\_timeout\_millis + +int + +否 + +等待服务端返回 task-finished 事件的超时时间，单位毫秒。默认值：10000。 + +**说明**：取消当前轮次的流式语音合成任务。调用后，SDK 会立即结束当前任务。取消后可在当前连接上继续发起新的合成任务，无需重新初始化 `SpeechSynthesizer` 实例。 + +**重要** + +**版本要求**：使用该功能需要 Python SDK 版本不低于 1.26.4。 + +**重要** + +**模型限制**： + +- 华北2（北京）地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型仅 v2 及以上版本支持该功能。 + +- 新加坡地域：Qwen-Audio-TTS 系列模型的所有模型都支持该功能；CosyVoice 系列模型不支持该功能。 + + ### **get\_last\_request\_id() - 获取请求ID** **方法签名**： @@ -185,7 +226,7 @@ str 语音合成所使用的音色。 -- **系统音色**：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- **系统音色**：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - **复刻音色**：通过声音复刻功能定制 @@ -263,7 +304,7 @@ int ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="longanlingxi", + voice="longanhuan_v3.6", additional_params={"bit_rate": 128000} ) ``` @@ -278,7 +319,7 @@ bool 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **说明** @@ -287,7 +328,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", + voice="longanhuan_v3.6", additional_params={"word_timestamp_enabled": True} ) ``` @@ -334,7 +375,7 @@ list\[str\] - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -354,9 +395,15 @@ list\[str\] - vi：越南语 +- es：西班牙语 + - it：意大利语 -- ms：马来语 +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 instruction @@ -388,7 +435,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="longanlingxi", + voice="longanhuan_v3.6", additional_params={ "enable_aigc_tag": True, "aigc_propagator": "your_propagator", @@ -447,7 +494,7 @@ qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v2、cosyvoice-v1 ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", # 替换成qwen-audio-3.0-tts-flash复刻音色 + voice="longanhuan_v3.6", # 音色 hot_fix={ "pronunciation": [{"天气": "tian1 qi4"}], "replace": [{"今天": "金天"}] @@ -483,7 +530,7 @@ bool ``` synthesizer = SpeechSynthesizer( model="qwen-audio-3.0-tts-flash", - voice="your_voice", # 替换成qwen-audio-3.0-tts-flash复刻音色 + voice="longanhuan_v3.6", # 音色 additional_params={"enable_markdown_filter": True} ) ``` @@ -693,7 +740,7 @@ SDK提供了语音合成的关键接口，支持以下几种调用方式： ### **非流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQURiBgMDRr9T4phkiIGNmYzBiZjFkZjQ4MDQzZGU4NDIyZDU2NWJjYjkyZTQ04709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQURiBgMDRr9T4phkiIGNmYzBiZjFkZjQ4MDQzZGU4NDIyZDU2NWJjYjkyZTQ04709861_20241015153444.149.svg) 单次调用发送的文本长度不得超过20000字符，超出限制将返回错误。 @@ -718,7 +765,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 实例化SpeechSynthesizer，并在构造方法中传入模型（model）、音色（voice）等请求参数 synthesizer = SpeechSynthesizer(model=model, voice=voice) @@ -736,7 +783,7 @@ with open('output.mp3', 'wb') as f: ### **单向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQVRiBgIDv9fShrBkiIDhmNTk5YmQ1ZDgwNzRjZjRiN2VlMTU5YzI1ZGMwMTlm4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQVRiBgIDv9fShrBkiIDhmNTk5YmQ1ZDgwNzRjZjRiN2VlMTU5YzI1ZGMwMTlm4709861_20241015153444.149.svg) 单次调用发送的文本长度不得超过20000字符，超出限制将返回错误。 @@ -769,7 +816,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 定义回调接口 class Callback(ResultCallback): @@ -823,7 +870,7 @@ synthesizer.call("今天天气怎么样？") ### **双向流式调用** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/1194204871/CAEQVRiBgMDb7PahrBkiIDVkNjEwOTMxYjEwOTRmOWFhMmI1OTRiY2Q3ZDgzZmE54709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0658354871/CAEQVRiBgMDb7PahrBkiIDVkNjEwOTMxYjEwOTRmOWFhMmI1OTRiY2Q3ZDgzZmE54709861_20241015153444.149.svg) 单次发送文本长度不得超过 20000 字符，且累计发送文本总长度不得超过 20 万字符。 @@ -888,7 +935,7 @@ dashscope.base_websocket_api_url='wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.c # 模型 model = "qwen-audio-3.0-tts-flash" # 音色 -voice = "longanlingxi" +voice = "longanhuan_v3.6" # 定义回调接口 class Callback(ResultCallback): diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md index 90389819..56124c1b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md @@ -87,7 +87,7 @@ Authorization 鉴权在 WebSocket 握手阶段验证。如果 API Key 无效或 ## 交互流程 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8684204871/CAEQaxiBgID50pCW3hkiIDVlOWNkODdhOGYyYjQ2ZDFiMzgyYjNmMmUzOGZkNGVh4709861_20241015153444.149.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9336814871/CAEQaxiBgID50pCW3hkiIDVlOWNkODdhOGYyYjQ2ZDFiMzgyYjNmMmUzOGZkNGVh4709861_20241015153444.149.svg) 客户端事件和服务端事件的详细说明，请参见[客户端事件](https://help.aliyun.com/zh/model-studio/cosyvoice-client-events)和[服务端事件](https://help.aliyun.com/zh/model-studio/cosyvoice-server-events)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md index e78e4fb1..a25c2588 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md @@ -66,7 +66,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "model": "qwen-audio-3.0-tts-flash", "input": { "text": "我家的后面有一个很大的花园。", - "voice": "longanlingxi", + "voice": "longanhuan_v3.6", "format": "wav", "sample_rate": 24000 } @@ -84,7 +84,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ "model": "qwen-audio-3.0-tts-flash", "input": { "text": "我家的后面有一个很大的花园。", - "voice": "longanlingxi", + "voice": "longanhuan_v3.6", "format": "wav", "sample_rate": 24000 } @@ -135,7 +135,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -211,7 +211,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **seed** `_integer_` （可选） @@ -245,7 +245,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -265,6 +265,16 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + **instruction** `_string_` （可选） diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md index 7e77c178..1c261549 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md @@ -207,7 +207,7 @@ String 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -301,7 +301,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("你好") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_ssml", true) .build(); ``` @@ -312,7 +312,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("你好") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_ssml", true)) .build(); ``` @@ -327,7 +327,7 @@ boolean 默认值：false。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 **说明** @@ -339,7 +339,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("word_timestamp_enabled", true) .build(); ``` @@ -350,7 +350,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("word_timestamp_enabled", true)) .build(); ``` @@ -377,7 +377,7 @@ int HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("seed", 1234) .build(); ``` @@ -388,7 +388,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("seed", 1234)) .build(); ``` @@ -421,7 +421,7 @@ List - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -441,6 +441,16 @@ List - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + **说明** @@ -452,7 +462,7 @@ List HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("language_hints", Arrays.asList("zh")) .build(); ``` @@ -463,7 +473,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("language_hints", Arrays.asList("zh"))) .build(); ``` @@ -488,7 +498,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("instruction", "请用非常开心的语气说话。") .build(); ``` @@ -499,7 +509,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("instruction", "请用非常开心的语气说话。")) .build(); ``` @@ -530,7 +540,7 @@ int HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .format("opus") .parameter("bit_rate", 32) .build(); @@ -542,7 +552,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .format("opus") .parameters(Collections.singletonMap("bit_rate", 32)) .build(); @@ -570,7 +580,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .build(); ``` @@ -581,7 +591,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_aigc_tag", true)) .build(); ``` @@ -608,7 +618,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .parameter("aigc_propagator", "xxxx") .build(); @@ -624,7 +634,7 @@ map.put("aigc_propagator", "xxxx"); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(map) .build(); ``` @@ -651,7 +661,7 @@ String HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_aigc_tag", true) .parameter("aigc_propagate_id", "xxxx") .build(); @@ -667,7 +677,7 @@ map.put("aigc_propagate_id", "xxxx"); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(map) .build(); ``` @@ -726,7 +736,7 @@ hotFix.put("replace", replace); HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("今天天气真好。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("hot_fix", hotFix) .build(); ``` @@ -738,7 +748,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("今天天气真好。") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("hot_fix", hotFix)) .build(); ``` @@ -774,7 +784,7 @@ boolean HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("# 标题\n正文内容") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameter("enable_markdown_filter", true) .build(); ``` @@ -785,7 +795,7 @@ HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() HttpSpeechSynthesisParam param = HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") .text("# 标题\n正文内容") - .voice("longanlingxi") + .voice("longanhuan_v3.6") .parameters(Collections.singletonMap("enable_markdown_filter", true)) .build(); ``` @@ -837,7 +847,7 @@ public class CosyVoiceSyncExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key @@ -877,7 +887,7 @@ public class CosyVoiceSyncExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("我家的后面有一个很大的花园。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key @@ -938,7 +948,7 @@ public class CosyVoiceStreamExample { HttpSpeechSynthesisParam.builder() .model("qwen-audio-3.0-tts-flash") // 更换模型时，需同步更换为对应版本的音色 .text("今天天气真好，适合出去玩。") - .voice("longanlingxi") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 + .voice("longanhuan_v3.6") // 该音色适用于qwen-audio-3.0-tts系列，cosyvoice-v3请使用longanyang等v3音色，cosyvoice-v2请使用longxiaochun_v2等v2音色 .format("wav") .sampleRate(24000) // 未配置环境变量时，将下行替换为：apiKey("sk-xxx")，即替换为实际的API Key diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md index 7143911a..a3f50903 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md @@ -109,7 +109,7 @@ str 取值范围： -- 系统音色：参见[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) +- 系统音色：参见[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list) - 声音复刻音色：如何创建音色请参见[CosyVoice声音复刻/设计API](https://help.aliyun.com/zh/model-studio/cosyvoice-clone-design-api) @@ -222,7 +222,7 @@ bool - False：关闭。 -仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 +仅在流式输出模式下可用。支持的音色范围：cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash、cosyvoice-v3-plus和cosyvoice-v2模型的复刻音色，以及[Qwen-Audio-TTS音色列表](https://help.aliyun.com/zh/model-studio/qwen-audio-tts-voice-list)、[CosyVoice音色列表](https://help.aliyun.com/zh/model-studio/cosyvoice-voice-list)中标记为支持的系统音色。qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash及其他模型的复刻音色不支持此功能。 seed @@ -264,7 +264,7 @@ list - zh：中文 -- en：英文 +- en：英语 - fr：法语 @@ -284,6 +284,16 @@ list - vi：越南语 +- es：西班牙语 + +- it：意大利语 + +- ms：马来西亚语 + +- fil：菲律宾语 + +- ar：阿拉伯语 + instruction @@ -443,7 +453,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co result = HttpSpeechSynthesizer.call( model="qwen-audio-3.0-tts-flash", # 更换模型时，需同步更换为对应版本的音色 text="今天是个好日子，适合构建人们喜爱的产品！", - voice="longanhuan", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 + voice="longanhuan_v3.6", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 format="wav", sample_rate=24000, stream=False, @@ -478,7 +488,7 @@ dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.co stream_result = HttpSpeechSynthesizer.call( model="qwen-audio-3.0-tts-flash", # 更换模型时，需同步更换为对应版本的音色 text="今天是个好日子，适合构建人们喜爱的产品！", - voice="longanhuan", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 + voice="longanhuan_v3.6", # 该音色适用于cosyvoice-v3系列，cosyvoice-v2请使用longxiaochun_v2等v2音色 format="wav", sample_rate=24000, stream=True, diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md index 2673fe86..a2402160 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md @@ -181,7 +181,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -189,11 +189,11 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -215,7 +215,13 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ - it：意大利语 - - ms：马来语 + - es：西班牙语 + + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 - cosyvoice-v3-plus： @@ -297,7 +303,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -307,7 +313,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md index be92e944..36aa6d87 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md @@ -121,7 +121,7 @@ customParam 否 -自定义参数，可指定languageHints、maxPromptAudioLength等。 +自定义参数，可通过 parameter() 方法指定 language\_hints、max\_prompt\_audio\_length 等参数。 **返回值**：`Voice` 对象，通过 `getVoiceId()` 方法获取音色ID。 @@ -283,108 +283,11 @@ String 声音复刻模型，固定为"voice-enrollment"。 -languageHints(List) - -List - -**重要** - -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 - -辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 - -此参数为数组，但当前版本仅处理第一个元素。 - -取值范围（因模型而异）： - -- qwen-audio-3.0-tts-flash： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - - - pt：葡萄牙语 - - - th：泰语 - - - id：印尼语 - - - vi：越南语 - - - it：意大利语 - - - ms：马来语 - -- cosyvoice-v3-plus： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - -- cosyvoice-v3.5-plus、cosyvoice-v3.5-flash、cosyvoice-v3-flash： - - - zh：中文 - - - en：英文 - - - fr：法语 - - - de：德语 - - - ja：日语 - - - ko：韩语 - - - ru：俄语 - - - pt：葡萄牙语 - - - th：泰语 - - - id：印尼语 - - - vi：越南语 - - -默认值：\["zh"\]。 - -maxPromptAudioLength(Float) - -Float - -**重要** - -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 - -音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 - -默认值：10.0。 - parameter(String, Object) Object -设置[扩展参数](#a18b66cba924j)，如 parameter("enable\_preprocess", false)。 +设置自定义参数，如 parameter("language\_hints", Arrays.asList("zh"))、parameter("max\_prompt\_audio\_length", 10.0f)、parameter("enable\_preprocess", false)。 ### **扩展参数** @@ -404,7 +307,7 @@ boolean **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 @@ -422,7 +325,7 @@ import com.alibaba.dashscope.utils.Constants; import org.slf4j.Logger; import org.slf4j.LoggerFactory; -import java.util.Collections; +import java.util.Arrays; public class Main { private static final Logger logger = LoggerFactory.getLogger(Main.class); @@ -444,8 +347,8 @@ public class Main { fileUrl, VoiceEnrollmentParam.builder() .model(cloneModelName) - .languageHints(Collections.singletonList("zh")) - // .maxPromptAudioLength(10.0f) + .parameter("language_hints", Arrays.asList("zh")) + // .parameter("max_prompt_audio_length", 10.0f) // .parameter("enable_preprocess", false) .build()); diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md index 14b73cd2..42410c27 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md @@ -112,7 +112,7 @@ List\[str\] **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -120,11 +120,11 @@ List\[str\] 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 - - en：英文 + - en：英语 - fr：法语 @@ -146,7 +146,13 @@ List\[str\] - it：意大利语 - - ms：马来语 + - es：西班牙语 + + - ms：马来西亚语 + + - fil：菲律宾语 + + - ar：阿拉伯语 - cosyvoice-v3-plus： @@ -199,7 +205,7 @@ float **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -213,7 +219,7 @@ bool **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md index 80482553..7dfaaf7e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-client-events.md @@ -2,7 +2,7 @@ Qwen-Audio Realtime API的客户端事件参考。 -**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-websocket-api)。 +**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/fun-audiochat-realtime-websocket-api)。 ## **session.update** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md similarity index 67% rename from skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md rename to skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md index 25294168..75868574 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-websocket-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/fun-audiochat-realtime-websocket-api.md @@ -127,7 +127,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 server\_vad 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 语音输入 loop \[用户说话中\] loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088432.svg) 按时间顺序，客户端与服务端的交互流程如下： @@ -148,7 +148,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 模型播报期间，若 VAD 检测到用户开始说话，服务端会取消当前响应（返回 `response.done`，状态为 `cancelled`），随后开始新一轮语音输入和响应。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio.delta input\_audio\_buffer.append（用户开始说话） response.done（status=cancelled） input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088435.svg) ### **smart\_turn 模式** @@ -160,7 +160,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 smart\_turn 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 loop \[用户说话中\] 无效语音（可能出现 0~N 次） loop \[用户说话中\] loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append conversation.item.ambient\_audio\_transcription.delta conversation.item.ambient\_audio\_transcription.completed input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped conversation.item.input\_audio\_transcription.completed commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088441.svg) 与 server\_vad 模式的主要区别： @@ -175,13 +175,13 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 与 server\_vad 模式的打断处理基本一致。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio.delta input\_audio\_buffer.append（用户开始说话） response.done（status=cancelled） input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.speech\_stopped commit audio buffer input\_audio\_buffer.committed conversation.item.created response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088443.svg) ## **无效轮次** 已判定有效的语音可能被撤回（`input_audio_buffer.speech_stopped` 返回 `reason=turn_invalid`），此时不触发推理，客户端应继续发送音频等待下一轮有效语音。下图展示了无效轮次的交互时序： -服务端 客户端 服务端 客户端 loop \[用户说话中\] loop \[用户说话中\] 轮次无效 继续发送音频，等待下一轮 input\_audio\_buffer.append input\_audio\_buffer.speech\_started input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.speech\_stopped（reason=turn\_invalid） +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088444.svg) ### **说话人增强配置流程** @@ -216,7 +216,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 下图展示了 push-to-talk 模式下的典型交互时序： -服务端 客户端 服务端 客户端 会话初始化 语音输入 loop \[用户说话中\] 响应生成 loop \[流式输出\] loop \[多轮交互\] connect session.created session.update session.updated input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.commit（用户松开按键） conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.committed conversation.item.created response.create（手动触发推理） response.created response.output\_item.added conversation.item.created response.content\_part.added response.audio\_transcript.delta response.audio.delta response.audio\_transcript.done response.audio.done response.content\_part.done response.output\_item.done response.done +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088447.svg) 按时间顺序，客户端与服务端的交互流程如下： @@ -233,7 +233,7 @@ Qwen-Audio Realtime API 支持三种交互模式，通过 `session.update` 事 客户端发送 `response.cancel` 取消当前响应，服务端返回 `response.done`（状态为 `cancelled`，原因为 `client_cancelled`）。下图展示了用户打断的交互时序： -服务端 客户端 服务端 客户端 服务端正在流式输出 loop \[流式输出中\] 用户打断 新一轮语音输入 loop \[用户说话中\] 新一轮推理 response.audio\_transcript.delta response.audio.delta response.cancel response.done（status=cancelled, reason=client\_cancelled） input\_audio\_buffer.append conversation.item.input\_audio\_transcription.delta input\_audio\_buffer.commit conversation.item.input\_audio\_transcription.completed input\_audio\_buffer.committed conversation.item.created response.create response.created +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7268354871/p1088449.svg) ## **各模式操作约束** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md index d74b004c..5cf16168 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/voice-conversation-api-references/real-time-voice-conversation-api-references/qwen-audio-realtime-server-events.md @@ -2,7 +2,7 @@ Qwen-Audio Realtime API 的服务端事件参考。所有服务端事件均包含 `event_id`（服务端自动生成）和 `type`（事件类型）公共字段。 -**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-websocket-api)。 +**用户指南**：[实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides)。如需了解事件交互时序，请参见[WebSocket API](https://help.aliyun.com/zh/model-studio/fun-audiochat-realtime-websocket-api)。 ## **error** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md index 0a687e48..96d7ae74 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md @@ -80,7 +80,6 @@ curl --location --request POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.c "generate_mode":"generate", "generate_num":1 }, - "auxiliary_parameters": "WMq4SC4......", "parameters":{} }' ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md index 720eeeec..825a4dde 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md @@ -71,14 +71,16 @@ wanx-background-generation-v2 ### **步骤1：创建任务获取任务ID** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/background-generation/generation/` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation` #### **请求头（Headers）** ## 图像背景生成 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/background-generation/generation/' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/background-generation/generation' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -396,7 +398,7 @@ foreground\_edge图像列表和background\_edge图像列表之和不得超过10 ### **步骤2：根据任务ID查询结果** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` #### **请求头（Headers）** @@ -404,10 +406,10 @@ foreground\_edge图像列表和background\_edge图像列表之和不得超过10 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md new file mode 100644 index 00000000..1537ef67 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md @@ -0,0 +1,459 @@ +# 千问-图像生成与编辑3.0 API参考 + +千问-图像生成与编辑3.0模型同时支持文生图（T2I）和图生图/图像编辑（I2I），可根据文本提示词直接生成图像，也可基于1-3张参考图结合编辑指令进行精确编辑。 + +**重要** + +该模型目前处于邀测阶段，您需要前往模型广场申请开通后方可使用。 + +## **模型概览** + +**模型名称** + +**模型简介** + +**输出图像规格** + +qwen-image-3.0-pro + +千问图像生成与编辑3.0模型，同时支持文生图（T2I）和图生图/图像编辑（I2I）。 + +图像分辨率： + +- **文生图（T2I）**：总像素需在512\*512至2048\*2048之间。 + +- **图生图（I2I）**：总像素需在512\*512至2048\*2048之间。 + +- **默认**：不指定`size`时，模型根据提示词自动推荐分辨率。 + + +图像格式：png + +## **前提条件** + +在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 + +**重要** + +华北2（北京）和新加坡地域拥有独立的 **API Key** 与**请求地址**，不可混用，跨地域调用将导致鉴权失败或服务报错。 + +**重要** + +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： + +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 + +## HTTP调用 + +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +**新加坡地域**：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` + +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +#### 请求参数 + +## 文生图（T2I） + +``` +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ +--header 'Content-Type: application/json' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--data '{ + "model": "qwen-image-3.0-pro", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "text": "画面是一张竖幅户外人像摄影，整体从上到下呈现温暖的午后街景氛围。顶部左侧到上方大面积被深绿色藤蔓和橙色小花覆盖，花叶从建筑檐口自然垂落，受阳光照射的叶片呈黄绿色高光，阴影处则偏深绿，形成浓密而柔和的背景层次。左上至中上区域是一块深蓝色横向招牌，招牌表面较暗、略带磨砂质感，上面以白色哥特体大字写着 Il Messaggero，文字位于画面左侧偏上，部分被前景花叶轻微遮挡，字体高对比、带装饰性尖角和粗细变化。招牌下方是报刊亭或书报摊的玻璃展示窗，黑色金属框架将橱窗分隔成多个矩形区域，内部陈列着许多报纸、杂志和书刊封面，但大多因景深虚化和光线反射而难以辨读，形成浅色纸张与深色边框交错的背景纹理。画面右上方是强烈的逆光区域，阳光从街道尽头照入，背景建筑被虚化成米灰色块面，边缘柔和，呈现明显的浅景深效果。画面中部偏右是一名年轻成年女性的半身至膝上人像，她回头面向镜头微笑，身体略向右转，肩背朝向观者，姿态自然放松。她有长而浓密的黑色波浪卷发，发丝被逆光勾勒出金色轮廓光，发梢在右侧向外散开，显得轻盈蓬松。她肤色白皙，脸型柔和偏鹅蛋形，眉形细致，眼睛明亮，眼妆清透，睫毛明显，面部带有自然高光，唇部为柔和珊瑚红色，笑容露齿，表情亲切明朗。她佩戴小巧耳饰，身穿黑色细肩带露背连衣裙，面料颜色深黑、轮廓简洁，细肩带从肩部向背部延伸，背部线条清晰。画面下部偏左到中部，她双手抱着一束玫瑰花，花束体积较大，主要由橙色、杏色、粉色和浅桃色玫瑰组成，花瓣层层卷曲，边缘被阳光照亮，绿色叶片和长花茎从花束下方垂出，花束与黑色裙装形成鲜明色彩对比。右侧背景是一条被阳光照亮的城市街道，地面呈暖灰与金黄色调，远处建筑、街边设施和一个模糊的红色圆形交通标志位于右下远景，均因焦外虚化而只保留色块和轮廓。整张照片采用暖色胶片感处理，带有细腻颗粒、柔和对比和明显逆光边缘光，人物位于视觉焦点，背景报刊亭、花藤、街道和阳光共同营造出浪漫、明亮、都市漫步式的氛围。" + } + ] + } + ] + }, + "parameters": { + "prompt_extend": true + } +}' +``` + +## 图生图/图像编辑（I2I） + +``` +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ +--header 'Content-Type: application/json' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--data '{ + "model": "qwen-image-3.0-pro", + "input": { + "messages": [ + { + "role": "user", + "content": [ + { + "image": "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png" + }, + { + "text": "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物脱下原本的米色针织上衣，换上一套彰显高雅气质的都市职场穿搭，身穿一件质感垂顺的香槟色真丝衬衫，外搭一件剪裁利落的深灰色休闲西装外套，下身搭配同色系的高腰阔腿裤，整体造型既干练又富有女人味。场景设定在一家装修现代简约的高端咖啡店内，背景是通透的落地玻璃窗，窗外隐约可见繁华的城市街景，室内摆放着深色实木长桌和舒适的皮质座椅，桌面上放置着一台打开的银色笔记本电脑、一份文件和一杯热气腾腾的美式咖啡。人物呈现出慵懒而放松的办公姿态，身体微微后仰倚靠在椅背上，一只手臂自然搭在扶手上，另一只手轻轻握着咖啡杯置于桌边，头部微侧，眼神清澈从容且带有一丝慵懒地直视镜头，嘴角挂着一抹优雅自信的微笑。人物化着精致得体的正式场合妆容，底妆清透干净，眉眼线条清晰利落，唇部涂抹着显气色的豆沙色口红，展现出成熟知性的魅力。光线采用午后柔和的自然光，从侧面透过落地窗洒入，在人物的面部轮廓和衣物褶皱上留下细腻的光影过渡，背景呈现自然的景深虚化效果，色彩以大地色、灰色和暖白色为主调，营造出宁静、高级且充满故事感的都市办公氛围，构图采用经典的竖幅七分身人像视角，人物位于画面视觉中心略偏右，比例协调，画质清晰细腻。" + } + ] + } + ] + }, + "parameters": { + "prompt_extend": true + } +}' +``` + +##### 请求头（Headers） + +**Content-Type** `_string_` **（必选）** + +请求内容类型。此参数必须设置为`application/json`。 + +**Authorization** `_string_`**（必选）** + +请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 + +##### 请求体（Request Body） + +**model** `_string_` **（必选）** + +模型名称，当前可用模型为`qwen-image-3.0-pro`。 + +**input** `_object_` **（必选）** + +输入参数对象，包含以下字段： + +**属性** + +**messages** `_array_` **（必选）** + +请求内容数组。**当前仅支持单轮对话**，因此数组内**有且只有一个对象**，该对象包含`role`和`content`两个属性。 + +**属性** + +**role** `_string_` **（必选）** + +消息发送者角色，必须设置为`user`。 + +**content** `_array_` **（必选）** + +消息内容数组，根据使用场景有不同的组合方式： + +- **文生图（T2I）**：仅包含一个`{"text": "..."}`对象。 + +- **图生图（I2I）**：包含1-3个`{"image": "..."}`对象和1个`{"text": "..."}`对象。 + + +**属性** + +**image** `_string_` （可选） + +输入图像的 URL 或 Base64 编码数据。I2I场景下支持传入1-3张图像。多图输入时，按照数组顺序定义图像顺序。 + +**图像要求：** + +- 图像格式：JPG、JPEG、PNG、BMP、TIFF、WEBP和GIF。 + +- 图像分辨率：建议图像的宽和高均在384像素至2048像素之间。 + +- 图像大小：不超过10MB。 + + +**支持的输入格式** + +1. 公网URL：支持 HTTP 和 HTTPS 协议。您也可在此[获取临时公网URL](https://help.aliyun.com/zh/model-studio/get-temporary-file-url)。 + +2. Base64 编码：格式为`data:{MIME_type};base64,{base64_data}`。 + + +**text** `_string_` **（必选）** + +正向提示词，用于描述您期望生成或编辑的图像内容、风格和构图。支持中英文。 + +**注意**：仅支持传入一个text，不传或传入多个将报错。 + +**parameters** `_object_` （可选） + +控制图像生成的附加参数。 + +**属性** + +**prompt\_extend** `_boolean_` （可选） + +是否开启提示词智能改写，默认值为 `true`（建议开启）。开启后，模型会优化正向提示词，对描述较简单的提示词效果提升明显。 + +**n** `_integer_` （可选） + +输出图像的数量，支持输出1-6张图片，默认值为1。 + +**size** `_string_` （可选） + +设置输出图像的分辨率，格式为`宽*高`，例如`"1024*1024"`。未指定时由模型根据提示词自动推荐分辨率。 + +- **文生图（T2I）**：像素范围512\*512至2048\*2048。 + +- **图生图（I2I）**：像素范围512\*512至2048\*2048。 + + +**negative\_prompt** `_string_` （可选） + +反向提示词，用来描述不希望在画面中看到的内容，可以对画面进行限制。 + +**seed** `_integer_` （可选） + +随机数种子，取值范围`[0, 2147483647]`。固定种子可使生成结果相对稳定。 + +**watermark** `_boolean_` （可选） + +是否添加水印，默认值为 `false`。 + +#### 响应参数 + +## 任务执行成功 + +任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 + +``` +{ + "output": { + "choices": [ + { + "finish_reason": "stop", + "message": { + "content": [ + { + "image": "https://dashscope-result-sz.oss-cn-shenzhen.aliyuncs.com/xxx.png?Expires=xxx" + } + ], + "role": "assistant" + } + } + ] + }, + "usage": { + "width": 1024, + "height": 1024, + "image_count": 1 + }, + "request_id": "571ae02f-5c9d-436c-83c2-f221e6df0xxx" +} +``` + +## 任务执行异常 + +如果因为某种原因导致任务执行失败，将返回相关信息，可以通过code和message字段明确指示错误原因。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +``` +{ + "request_id": "31f808fd-8eef-9004-xxxxx", + "code": "InvalidApiKey", + "message": "Invalid API-key provided." +} +``` + +**output** `_object_` + +包含模型生成结果。 + +**属性** + +**choices** `_array_` + +结果选项列表。 + +**属性** + +**finish\_reason** `_string_` + +任务停止原因，自然停止时为`stop`。 + +**message** `_object_` + +模型返回的消息。 + +**属性** + +**role** `_string_` + +消息的角色，固定为`assistant`。 + +**content** `_array_` + +消息内容，包含生成的图像信息。 + +**属性** + +**image** `_string_` + +生成图像的 URL，格式为PNG。**链接有效期为24小时**，请及时下载并保存图像。 + +**usage** `_object_` + +本次调用的资源使用情况，仅调用成功时返回。 + +**属性** + +**width** `_integer_` + +生成图像的宽度（像素）。 + +**height** `_integer_` + +生成图像的高度（像素）。 + +**image\_count** `_integer_` + +生成图像的张数。 + +**request\_id** `_string_` + +请求唯一标识。可用于请求明细溯源和问题排查。 + +**code** `_string_` + +请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +**message** `_string_` + +请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +## SDK调用 + +以下以图生图/图像编辑（I2I）为示例，展示Python和Java SDK的调用方式。 + +## Python + +``` +import os +import base64 +import mimetypes +import dashscope +from dashscope import MultiModalConversation + +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' + +def encode_file(file_path): + mime_type, _ = mimetypes.guess_type(file_path) + if not mime_type or not mime_type.startswith("image/"): + raise ValueError("Unsupported or unrecognized image format") + with open(file_path, "rb") as image_file: + encoded_string = base64.b64encode(image_file.read()).decode('utf-8') + return f"data:{mime_type};base64,{encoded_string}" + +# [方法一] 使用公网图像URL +image_url = "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png" + +# [方法二] 使用Base64编码图像 +# image_url = encode_file("./your_image.png") + +response = MultiModalConversation.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + model="qwen-image-3.0-pro", + messages=[{ + "role": "user", + "content": [ + {"image": image_url}, + {"text": "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物换上一套彰显高雅气质的都市职场穿搭，场景设定在一家装修现代简约的高端咖啡店内。"} + ] + }], + prompt_extend=True +) + +print(response) +if response.status_code == 200: + url = response.output.choices[0].message.content[0]["image"] + print(f"Generated image URL: {url}") +else: + print(f"Error: {response.code} - {response.message}") +``` + +## Java + +``` +import java.util.Arrays; +import java.util.Base64; +import java.util.Collections; +import java.io.IOException; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; +import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; +import com.alibaba.dashscope.common.MultiModalMessage; +import com.alibaba.dashscope.common.Role; +import com.alibaba.dashscope.utils.Constants; + +public class ImageEditExample { + public static void main(String[] args) { + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; + + // [方法一] 使用公网图像URL + String imageUrl = "https://alidocs.oss-cn-zhangjiakou.aliyuncs.com/res/yBRq1ZPYEaXdyOdv/img/33a80a19-7ac7-4c64-b0fa-7d685b7046a0.png"; + + // [方法二] 使用Base64编码图像 + // String imageUrl = encodeFile("/path/to/your/image.png"); + + MultiModalConversation conv = new MultiModalConversation(); + MultiModalMessage userMessage = MultiModalMessage.builder() + .role(Role.USER.getValue()) + .content(Arrays.asList( + Collections.singletonMap("image", imageUrl), + Collections.singletonMap("text", "帮我生成一张充满高级感的都市风格女性写真，画面中人物完美保留输入图片中这位年轻女性的面部特征与一头柔顺的黑色长发。人物换上一套彰显高雅气质的都市职场穿搭，场景设定在一家装修现代简约的高端咖啡店内。") + )) + .build(); + MultiModalConversationParam param = MultiModalConversationParam.builder() + .apiKey(System.getenv("DASHSCOPE_API_KEY")) + .model("qwen-image-3.0-pro") + .messages(Arrays.asList(userMessage)) + .parameter("prompt_extend", true) + .build(); + try { + MultiModalConversationResult result = conv.call(param); + System.out.println(result); + } catch (Exception e) { + e.printStackTrace(); + } + } + + public static String encodeFile(String filePath) { + Path path = Paths.get(filePath); + if (!Files.exists(path)) { + throw new IllegalArgumentException("File does not exist: " + filePath); + } + String mimeType = null; + try { + mimeType = Files.probeContentType(path); + } catch (IOException e) { + throw new IllegalArgumentException("Cannot detect file type: " + filePath); + } + if (mimeType == null || !mimeType.startsWith("image/")) { + throw new IllegalArgumentException("Unsupported or unrecognized image format"); + } + byte[] fileBytes = null; + try { + fileBytes = Files.readAllBytes(path); + } catch (IOException e) { + throw new IllegalArgumentException("Cannot read file content: " + filePath); + } + String encodedString = Base64.getEncoder().encodeToString(fileBytes); + return "data:" + mimeType + ";base64," + encodedString; + } +} +``` + +## **错误码** + +如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md index bb4d9f20..0441479a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/client-events.md @@ -89,7 +89,7 @@ Qwen-Omni-Realtime API的客户端事件参考。 默认音色： -- Qwen3.5-Omni-Realtime 系列：`Tina` +- Qwen3.5-Omni-Realtime 系列模型：`Tina` - Qwen3-Omni-Flash-Realtime：`Cherry` @@ -135,7 +135,7 @@ VAD 类型，可选值： - `server_vad`（默认值）：基于声学特征检测语音结束。 -- `semantic_vad`：基于语义有效性检测语音结束，可过滤回应语、背景音等无意义声音。仅 `qwen3.5-omni-realtime` 模型支持。 +- `semantic_vad`：基于语义有效性检测语音结束，可过滤回应语、背景音等无意义声音。仅 `qwen3.5-omni-realtime` 系列模型支持。 **threshold** `_float_`（可选） @@ -160,7 +160,7 @@ VAD 灵敏度。值越低，VAD 越灵敏，越容易将微弱声音（包括背 **enable\_search** `_boolean_`（可选） -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索。设为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要联网搜索来回答用户问题。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md index 0aece341..5d4f89d4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md @@ -176,7 +176,7 @@ VAD类型，取值如下： - `server_vad`（默认值）：基于声学特征检测用户语音结束。 -- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`qwen3.5-omni-realtime`模型支持。 +- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`qwen3.5-omni-realtime`系列模型支持。 turnDetectionThreshold @@ -210,7 +210,7 @@ enable\_search Boolean -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索功能。设置为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要搜索来回应用户的即时问题。 @@ -230,7 +230,7 @@ tools List> -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 工具定义列表。启用后，模型可自主判断是否需要调用外部工具来回应用户的问题。命中工具调用时，模型不生成音频，仅返回工具调用参数。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md index 869f563c..cd0f1593 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md @@ -95,7 +95,7 @@ str 默认音色： -- `Qwen3.5-Omni-Realtime`: `"Tina"` +- `Qwen3.5-Omni-Realtime`系列模型: `"Tina"` - `Qwen3-Omni-Flash-Realtime`: `"Cherry"` @@ -157,7 +157,7 @@ VAD类型，取值如下： - `server_vad`（默认值）：基于声学特征检测用户语音结束。 -- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`Qwen3.5-Omni-Realtime`模型支持。 +- `semantic_vad`：基于语义有效性检测用户语音结束，可过滤无意义语音（如回应语、背景音）。仅`Qwen3.5-Omni-Realtime`系列模型支持。 turn\_detection\_threshold @@ -191,7 +191,7 @@ enable\_search bool -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 是否启用联网搜索功能。设置为 `true` 启用，默认为 `false`。启用后，模型可自主判断是否需要搜索来回应用户的即时问题。 @@ -209,7 +209,7 @@ tools list\[dict\] -**仅在使用 Qwen3.5-Omni-Realtime 模型时生效。** +**仅在使用 Qwen3.5-Omni-Realtime 系列模型时生效。** 工具定义列表。启用后，模型可自主判断是否需要调用外部工具来回应用户的问题。命中工具调用时，模型不生成音频，仅返回工具调用参数。 @@ -244,7 +244,7 @@ temperature越高，生成的内容更多样，反之，生成的内容更确定 由于temperature与top\_p均可以控制生成内容的多样性，因此建议只设置其中一个值。 -- `Qwen3.5-Omni-Realtime`系列：0.7 +- `Qwen3.5-Omni-Realtime`系列模型：0.7 - `Qwen3-Omni-Flash-Realtime`系列：0.9 @@ -267,7 +267,7 @@ top\_p越高，生成的内容更多样。反之，生成的内容更确定。 top\_p默认值： -- `Qwen3.5-Omni-Realtime`系列：0.8 +- `Qwen3.5-Omni-Realtime`系列模型：0.8 - `Qwen3-Omni-Flash-Realtime`系列：1.0 @@ -286,7 +286,7 @@ integer top\_k默认值： -- `Qwen3.5-Omni-Realtime`系列：20 +- `Qwen3.5-Omni-Realtime`系列模型：20 - `Qwen3-Omni-Flash-Realtime`系列：50 @@ -317,7 +317,7 @@ float repetition\_penalty默认值： -- `Qwen3.5-Omni-Realtime`系列：1.0 +- `Qwen3.5-Omni-Realtime`系列模型：1.0 - 其他模型：1.05 @@ -334,7 +334,7 @@ float presence\_penalty默认值： -- `Qwen3.5-Omni-Realtime`系列：1.5 +- `Qwen3.5-Omni-Realtime`系列模型：1.5 - 其他模型：0.0 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md index 1b9a731f..98d52374 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/server-events.md @@ -162,7 +162,7 @@ VAD检测阈值。 **enable\_search** `_boolean_` -是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 模型支持。 +是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 系列模型支持。 **search\_options** `_object_` @@ -308,7 +308,7 @@ VAD检测阈值。 **enable\_search** `_boolean_`（可选） -是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 模型支持。 +是否启用联网搜索功能。仅 Qwen3.5-Omni-Realtime 系列模型支持。 **search\_options** `_object_`（可选） diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md index 59f0dda7..d2ab3b8a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/error-code.md @@ -767,13 +767,13 @@ AI 助理准确分析出原因，并给出解决方案： ### **request timeout after 23 seconds.** -**原因：** 超过23秒未向服务发送数据。该报错信息在使用[实时语音合成（Sambert）](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)、[语音识别（Paraformer）](https://help.aliyun.com/zh/model-studio/paraformer-speech-recognition)和[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时产生。 +**原因：** 超过23秒未向服务发送数据。该报错信息在使用[实时语音合成（Sambert）](https://help.aliyun.com/zh/model-studio/sambert-speech-synthesis/)、[语音识别（Paraformer）](https://help.aliyun.com/zh/model-studio/paraformer-speech-recognition)和[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时产生。 **解决方案：** 请检查为什么长时间未向服务器发送数据。如果长时间（超过23秒）不向服务端发送消息，请及时结束任务。 ### **Please ensure input text is valid.** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是由于未发送待合成文本引起的。可能原因包括：参数遗漏（未为 `text` 参数赋值）或代码异常（导致对 `text` 参数的赋值失败）。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是由于未发送待合成文本引起的。可能原因包括：参数遗漏（未为 `text` 参数赋值）或代码异常（导致对 `text` 参数的赋值失败）。 **解决方案：** 请排查代码，确保 `text` 参数被正确赋值并发送。 @@ -816,7 +816,7 @@ AI 助理准确分析出原因，并给出解决方案： ### **\[tts:\]Engine return error code: 418** -**原因：** 使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，请求参数 `voice`（音色）不正确，或 `model`（模型）与 `voice`（音色）版本不匹配。 +**原因：** 使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，请求参数 `voice`（音色）不正确，或 `model`（模型）与 `voice`（音色）版本不匹配。 **解决方案：** @@ -831,7 +831,7 @@ AI 助理准确分析出原因，并给出解决方案： ### **Request voice is invalid!** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是因为未设置音色。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，此错误通常是因为未设置音色。 **解决方案：** 请检查是否对`voice`参数赋值。若您使用[WebSocket API参考](https://help.aliyun.com/zh/model-studio/cosyvoice-websocket-api)，请参照API文档按照正确JSON格式配置参数。 @@ -1015,7 +1015,7 @@ AI 助理准确分析出原因，并给出解决方案： **解决方案：**前往[费用与成本](https://usercenter2.aliyun.com/home)查看是否欠费： -- 未欠费：请确认该 API Key 是否属于当前账号； +- 未欠费：请确认该 API Key 是否属于当前账号。如果账号不存在欠费的情况，可能账户出现异常，详情请联系客服进一步排查。 - 欠费：请及时充值。充值后，系统余额可能存在延迟，请稍等后重试。 @@ -1745,7 +1745,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - **填写错误**：阿里云百炼的 API Key 以 `sk-` 开头，请确认未误填其他模型提供商的密钥，且复制时未包含多余空格或换行符。 -- **套餐专属 API Key（Coding Plan / Token Plan 团队版）**：Coding Plan 与 Token Plan 团队版均提供以 `sk-sp-` 开头的专属 API Key，**必须配合各自的专属 Base URL 使用**，不可与通用 API Key/Base URL 混用（混用会返回本鉴权错误）。其中 Coding Plan 的专属地址为 https://coding.dashscope.aliyuncs.com/v1；Token Plan 团队版的专属 Base URL 可在控制台**我的订阅**的 API Key 区域查看。请确认同时更新了 API Key 和 Base URL，具体配置方法请分别参见[接入AI工具](https://help.aliyun.com/zh/model-studio/use-coding-plan-in-ai-tools/)与[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-quickstart)。 +- **套餐专属 API Key（Coding Plan / Token Plan 团队版）**：Coding Plan 与 Token Plan 团队版均提供以 `sk-sp-` 开头的专属 API Key，**必须配合各自的专属 Base URL 使用**，不可与通用 API Key/Base URL 混用（混用会返回本鉴权错误）。其中 Coding Plan 的专属地址为 https://coding.dashscope.aliyuncs.com/v1；Token Plan 团队版的专属 Base URL 可在控制台**我的订阅**的 API Key 区域查看。请确认同时更新了 API Key 和 Base URL，具体配置方法请分别参见[接入AI工具](https://help.aliyun.com/zh/model-studio/use-coding-plan-in-ai-tools/)与[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-team-quickstart)。 - **地域不匹配**：API Key 和 Base URL 属于不同的地域，例如使用了华北2（北京）地域的 API Key 和新加坡地域的 Base URL。请确认您使用的 API Key 位于[北京](https://bailian.console.aliyun.com/?tab=globalset#/efm/api_key)地域页面或[新加坡](https://modelstudio.console.aliyun.com/?tab=globalset#/efm/api_key)地域页面，或[美国](https://modelstudio.console.aliyun.com/us-east-1)地域页面。各地域对应的 Base URL 如下： @@ -1958,6 +1958,8 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 - 请前往模型广场开通模型服务。 +- 如果您通过国际站 API 端点（如 `dashscope-us.aliyuncs.com`）发起调用，请注意不同地域可用的模型列表不同。调用前请确认目标模型是否在该地域可用，部分模型在美国地域需使用带 `-us` 后缀的模型名称（如 `qwen-max-us`）。 + ## **404-**model\_not\_supported @@ -2186,7 +2188,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 **原因：** 语音合成中使用的音色不存在。 -**解决方案：** 请检查`voice`参数，确保指定了正确的音色名称。可用音色请参见[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)。 +**解决方案：** 请检查`voice`参数，确保指定了正确的音色名称。可用音色请参见[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)。 ## **500-**InternalError.FileUpload @@ -2438,7 +2440,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 ### **Cannot resolve symbol 'ttsv2'** -**原因：** 若您使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，出现该问题的原因是DashScope SDK版本过低。 +**原因：** 若您使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)，出现该问题的原因是DashScope SDK版本过低。 **解决方案：** 请[安装最新版 DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk#f80a232bb24v7)。 @@ -2458,7 +2460,7 @@ A：请核对资源包的可抵扣范围。以qwen-plus/qwen-plus-latest系列 ### **InputRequiredException: Parameter invalid: text is null** -**原因**：使用[实时语音合成（CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时未发送待合成文本。 +**原因**：使用[实时语音合成（Qwen-Audio-TTS/CosyVoice）](https://help.aliyun.com/zh/model-studio/cosyvoice-large-model-for-speech-synthesis/)时未发送待合成文本。 **解决方案：**调用语音合成接口时为 `text` 参数赋值。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md index 3c29a67e..1b74ba58 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/preparations/get-api-key.md @@ -4,7 +4,7 @@ **说明** -本文介绍的是百炼按量付费的 API Key。如果您使用的是 Token Plan 或 Coding Plan，请使用对应的专属 API Key（以`sk-sp-`开头），获取方式请参见[Token Plan API Key](https://help.aliyun.com/zh/model-studio/token-plan-quickstart#tp04-h-step2)和[Coding Plan 的 API Key](https://help.aliyun.com/zh/model-studio/coding-plan#2531c37fd64f9)。 +本文介绍的是百炼按量付费的 API Key。如果您使用的是 Token Plan 或 Coding Plan，请使用对应的专属 API Key（以`sk-sp-`开头），获取方式请参见[Token Plan API Key](https://help.aliyun.com/zh/model-studio/token-plan-team-quickstart#tp04-h-step2)和[Coding Plan 的 API Key](https://help.aliyun.com/zh/model-studio/coding-plan#2531c37fd64f9)。 ## **获取API Key** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md new file mode 100644 index 00000000..8eaec477 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md @@ -0,0 +1,17 @@ +# AOQ SDK简介 + +阿里云AOQ SDK面向实时多模态场景，帮助开发者快速搭建基于阿里云实时多模态模型的解决方案。 + +欢迎使用AOQ SDK来实现您的业务需求，阿里云AOQ基于阿里云多年以来服务全球客户的深厚技术沉淀，以面向实时多模态场景向全球开发者开放的产品，致力于帮助全球开发者快速高效的搭建基于阿里云的实时多模态模型的解决方案。 + +## **API设计与回调机制** + +阿里云AOQ SDK的API设计遵循状态式API设计原则。客户只需要把想要的状态通过API传至SDK，SDK内部会进行状态判断，并在适当时机达成用户想要的目标状态，同时在执行时通过回调机制及时向客户反馈SDK当前执行状态，使客户使用我们的SDK时享受到无忧的体验。您只需设置期望的状态，SDK便会自动处理并通过回调通知您，无需重复调用，也不需要关心具体的调用时机和场景。 + +## **异常处理机制** + +遇到异常情况时，AOQ SDK会优先尝试内部解决，仅在内部无法恢复时，才会提示客户介入处理。目前这些无法恢复的内部错误主要分为物理限制（如网络问题、音频设备故障、视频设备故障、CPU或内存资源限制）和外部因素（token无效）两类。 + +## **下载指南** + +阿里云AOQ SDK是阿里云自研产品，请通过阿里云官网进行下载，详情请参见[SDK下载](https://help.aliyun.com/zh/model-studio/realtime-sdk-download)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md new file mode 100644 index 00000000..f675adaa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md @@ -0,0 +1,1179 @@ +# 音频常用功能介绍 + +AOQ Client SDK 提供了完整的音频能力，覆盖音频采集、播放、编解码配置、扬声器管理、文件混音、外部音频流注入、音频帧数据回调等核心场景。本文档基于 Android（Java）、iOS（Objective-C）、Ohos（ArkTS）三个平台的公开 API，对音频常用功能进行统一介绍。 + +## **音频采集** + +音频采集用于打开设备麦克风，将实时音频数据送入 SDK 编码推流管线。SDK 支持两种采集模式： + +- **内部采集**（默认）：SDK 自动管理麦克风设备的打开、录音和关闭。 + +- **外部采集**：由应用自行管理麦克风，采集到的 PCM 数据通过外部音频流接口输入 SDK。 + + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +isExternal + +bool + +false + +是否使用外部采集模式 + +isVoipMode + +bool + +false + +是否启用 VoIP 模式（硬件 AEC），移动端有效，采集播放参数先到为准 + +channel + +int + +1 + +采集通道数，支持 1（单声道）/ 2（立体声） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开启采集 + +`startAudioCapture(config)` + +`startAudioCapture:config:` + +`startAudioCapture(config)` + +关闭采集 + +`stopAudioCapture()` + +`stopAudioCapture` + +`stopAudioCapture()` + +静音/取消静音 + +`muteAudioCapture(mute)` + +`muteAudioCapture:` + +`muteAudioCapture(mute)` + +### 使用示例 + +**Android** + +``` +AoqAudioCaptureConfig config = new AoqAudioCaptureConfig(); +config.isVoipMode = true; +config.channel = 1; +engine.startAudioCapture(config); +``` + +**iOS** + +``` +AoqAudioCaptureConfig *config = [[AoqAudioCaptureConfig alloc] init]; +config.isVoipMode = YES; +config.channel = 1; +[engine startAudioCapture:config]; +``` + +**Ohos** + +``` +const config: AoqAudioCaptureConfig = { isVoipMode: true, channel: 1 }; +engine.startAudioCapture(config); +``` + +## **音频播放** + +音频播放用于将接收到的远端音频数据渲染到本地扬声器或耳机。SDK 支持播放暂停/恢复（带淡入淡出）、打断当前轮音频通话等高级控制。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +isVoipMode + +bool + +false + +是否启用 VoIP 模式（硬件AEC），移动端有效，采集播放参数先到为准 + +isDefaultSpeaker + +bool + +true + +是否默认使用扬声器（移动端有效，非VoIP时无效） + +isExternal + +bool + +false + +是否使用外部播放模式 + +channel + +int + +1 + +播放通道数，支持 1（单声道）/ 2（立体声） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开始播放 + +`startAudioPlayer(config)` + +`startAudioPlayer:config:` + +`startAudioPlayer(config)` + +停止播放 + +`stopAudioPlayer()` + +`stopAudioPlayer` + +`stopAudioPlayer()` + +暂停播放 + +`pauseAudioPlayer(fadeMs)` + +`pauseAudioPlayer:` + +`pauseAudioPlayer(fadeMs)` + +恢复播放 + +`resumeAudioPlayer(fadeMs)` + +`resumeAudioPlayer:` + +`resumeAudioPlayer(fadeMs)` + +打断通话 + +`interruptAudioPlayer(trackType, fadeMs)` + +`interruptAudioPlayer:fadeMs:` + +`interruptAudioPlayer(trackType, fadeMs)` + +**说明** + +**fadeMs 参数**：暂停和恢复播放时的淡入/淡出时长（毫秒），设为 0 则立即切换。 + +## **扬声器管理** + +控制音频输出设备在扬声器和听筒之间切换。 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +切换扬声器 + +`enableSpeakerphone(enable)` + +`enableSpeakerphone:` + +`enableSpeakerphone(enable)` + +查询扬声器状态 + +`isSpeakerphoneEnabled()` + +`isSpeakerphoneEnabled` + +`isSpeakerphoneEnabled()` + +**说明** + +需要在 VoIP 模式下才允许切换，非 VoIP 时，`enableSpeakerphone` 调用有 OnError(AoqECAudioDeviceEarpieceRequiresVoipMode) 错误通知。 + +**说明** + +**iOS 特殊行为**：iPad 设备只有扬声器模式；当 AVAudioSession 不是 PlayAndRecord 类别时，也始终返回 YES。 + +## **音频编解码配置** + +设置音频上行（编码器）和下行（解码器）的编码格式、采样率、声道数和码率。表示推流/拉流的格式。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Audio + +音频轨道类型，当前只支持一条音频流 + +codecType + +AoqEncoderType + +AudioPCM + +编码类型：AudioPCM(1) 或 AudioOpus(2) + +sampleRate + +int + +48000 + +采样率，Opus 支持 8K/16K/48K，PCM 支持 8K/16K/32K/48K + +channel + +int + +1 + +声道数，支持 1（单声道）/ 2（立体声） + +bitrate + +int + +32000 + +码率（bps） + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置编码参数 + +`setAudioEncoderConfig(config)` + +`setAudioEncoderConfig:` + +`setAudioEncoderConfig(config)` + +设置解码参数 + +`setAudioDecoderConfig(config)` + +`setAudioDecoderConfig:` + +`setAudioDecoderConfig(config)` + +### 支持的编码格式 + +**枚举值** + +**数值** + +**说明** + +AoqEncoderTypeAudioPCM + +1 + +PCM 裸音频 + +AoqEncoderTypeAudioOpus + +2 + +Opus 编码 + +## **音频文件混音** + +支持将本地音频文件混入当前音频流中一起推流和/或本地播放。每个音频文件通过业务自分配的 `fileId` 标识，可同时管理多个文件实例。 + +### 混音配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +fileName + +String + +\- + +音频文件路径（含文件名） + +cycles + +int + +\-1 + +循环次数，-1 表示无限循环 + +startPosMs + +long + +0 + +起始播放位置（毫秒） + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +playoutVolume + +int + +100 + +本地播放音量 \[0-100\] + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开始播放 + +`startAudioFile(fileId, config)` + +`startAudioFile:config:` + +`startAudioFile(fileId, config)` + +停止播放 + +`stopAudioFile(fileId)` + +`stopAudioFile:` + +`stopAudioFile(fileId)` + +暂停 + +`pauseAudioFile(fileId)` + +`pauseAudioFile:` + +`pauseAudioFile(fileId)` + +恢复 + +`resumeAudioFile(fileId)` + +`resumeAudioFile:` + +`resumeAudioFile(fileId)` + +获取文件时长 + +`getAudioFileDuration(fileId)` + +`getAudioFileDuration:` + +`getAudioFileDuration(fileId)` + +获取当前位置 + +`getAudioFileCurrentPosition(fileId)` + +`getAudioFileCurrentPosition:` + +`getAudioFileCurrentPosition(fileId)` + +设置播放位置 + +`setAudioFilePositionMillis(fileId, pos)` + +`setAudioFilePositionMillis:positionMillis:` + +`setAudioFilePositionMillis(fileId, pos)` + +设置音量 + +`setAudioFileVolume(fileId, type, vol)` + +`setAudioFileVolume:type:volume:` + +`setAudioFileVolume(fileId, type, vol)` + +获取音量 + +`getAudioFileVolume(fileId, type)` + +`getAudioFileVolume:type:` + +`getAudioFileVolume(fileId, type)` + +**说明** + +**音量方向（type）**：`AoqAudioStreamPublish(0)` 控制推流音量；`AoqAudioStreamPlayout(1)` 控制本地播放音量。 + +### 状态回调 + +**状态码** + +**数值** + +**说明** + +AoqAudioFileNone + +0 + +初始状态 + +AoqAudioFileStarted + +1 + +已开始播放 + +AoqAudioFileStopped + +2 + +已停止 + +AoqAudioFilePaused + +3 + +已暂停 + +AoqAudioFileResumed + +4 + +已恢复 + +AoqAudioFileEnded + +5 + +播放结束 + +AoqAudioFileBuffering + +6 + +缓冲中 + +AoqAudioFileBufferingEnd + +7 + +缓冲结束 + +AoqAudioFileFailed + +8 + +播放失败 + +## **外部音频流** + +外部音频流允许将应用生成的 PCM 音频数据注入到 SDK 的音频管线中，支持推流和/或本地播放。典型场景包括 TTS 语音合成输出、AI 模型音频输出、背景音效等。每个外部音频流通过业务自分配的 `streamId` 标识。 + +### 配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Audio + +音频轨道类型 + +codecType + +AoqEncoderType + +AudioPCM + +音频流格式 + +channels + +int + +1 + +声道数 + +sampleRate + +int + +48000 + +采样率，支持 8/12/16/24/32/44.1/48/64/88.2/96/176.4/192K + +playoutVolume + +int + +100 + +本地播放音量 \[0-100\] + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +maxBufferDuration + +int + +600000 + +最大缓冲时长（毫秒），取值范围 \[100, ~\]，超过时 Push 失败 + +enable3A + +bool + +false + +输入 PCM 是否经过 3A 处理 + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +新增外部音频流 + +`addAudioExternalStream(streamId, config)` + +`addAudioExternalStream:config:` + +`addAudioExternalStream(streamId, config)` + +输入音频数据 + +`pushAudioExternalStreamData(streamId, data)` + +`pushAudioExternalStreamData:data:` + +`pushAudioExternalStreamData(streamId, data)` + +设置音量 + +`setAudioExternalStreamVolume(streamId, type, vol)` + +`setAudioExternalStreamVolume:type:volume:` + +`setAudioExternalStreamVolume(streamId, type, vol)` + +获取音量 + +`getAudioExternalStreamVolume(streamId, type)` + +`getAudioExternalStreamVolume:type:` + +`getAudioExternalStreamVolume(streamId, type)` + +清空缓存 + +`clearAudioExternalStreamBuffer(streamId, fadeoutMs)` + +`clearAudioExternalStreamBuffer:fadeoutMs:` + +`clearAudioExternalStreamBuffer(streamId, fadeoutMs)` + +移除流 + +`removeAudioExternalStream(streamId)` + +`removeAudioExternalStream:` + +`removeAudioExternalStream(streamId)` + +### Push 数据最佳实践 + +- 需要循环调用 `pushAudioExternalStreamData`，保证数据 push 成功 + +- 返回错误码 110（缓冲区满）时短暂 Sleep 30ms 后重试，不要丢弃数据 + +- 引擎退出前先停止推送循环，再调用 `removeAudioExternalStream` + +- 实时采集每帧 10ms 长，有数据就调用 push；从文件解析每帧 40ms 长，间隔 30ms 调用 push 一次 + + +## **音频帧数据回调** + +音频帧回调允许开发者在音频管线的不同位置获取原始 PCM 数据，用于音频分析、自定义处理、录制等场景。 + +### 支持的数据源位置 + +**数据源** + +**枚举值** + +**说明** + +Captured + +0 + +采集后的原始音频数据（未经 3A 处理） + +ProcessCaptured + +1 + +经过 3A 处理后的音频数据，需要 Connect 成功后才回调数据 + +Publish + +2 + +即将推流的音频数据（需要 Connect 成功） + +Playback + +3 + +即将播放的音频数据（远端下行） + +### 回调配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +sampleRate + +int + +48000 + +回调音频的采样率 + +channels + +int + +1 + +回调音频的声道数，支持 1/2 + +mode + +AoqAudioObserverMode + +ReadOnly + +只读(0)/读写(1) 模式 + +### 使用步骤 + +1. **注册观察者**：调用 `setAudioFrameObserver` 设置音频帧回调监听器 + +2. **启用数据源**：调用 `enableAudioFrameObserver` 选择需要监听的数据源位置，开启回调 + +3. **处理回调数据**：在回调函数中获取 PCM 数据 + + +### API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +注册观察者 + +`setAudioFrameObserver(listener)` + +`setAudioFrameObserver:` + +`setAudioFrameObserver(observer)` + +启用回调 + +`enableAudioFrameObserver(enabled, source, config)` + +`enableAudioFrameObserver:audioSource:config:` + +`enableAudioFrameObserver(enabled, source, config)` + +### 回调方法 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +采集数据 + +`onCapturedAudioFrame(frame)` + +`onCapturedAudioFrame:` + +`onCapturedAudioFrame(frame)` + +3A 后数据 + +`onProcessCapturedAudioFrame(frame)` + +`onProcessCapturedAudioFrame:` + +`onProcessCapturedAudioFrame(frame)` + +推流数据 + +`onPublishAudioFrame(trackType, frame)` + +`onPublishAudioFrame:frame:` + +`onPublishAudioFrame(trackType, frame)` + +播放数据 + +`onPlaybackAudioFrame(frame)` + +`onPlaybackAudioFrame:` + +`onPlaybackAudioFrame(frame)` + +## **音频状态与路由** + +SDK 自动监测音频设备的状态变化和路由切换，并通过回调通知应用层。 + +### 设备状态码 + +**状态码** + +**值** + +**说明** + +AoqAudioDeviceNone + +0 + +初始状态 + +RecordStarting + +1 + +采集启动中 + +RecordStarted + +2 + +采集已启动 + +RecordStopping + +3 + +采集停止中 + +RecordStopped + +4 + +采集已停止 + +RecordFail + +5 + +采集失败 + +PlayStarting + +6 + +播放启动中 + +PlayStarted + +7 + +播放已启动 + +PlayStopping + +8 + +播放停止中 + +PlayStopped + +9 + +播放已停止 + +PlayFail + +10 + +播放失败 + +### 设备路由类型 + +**路由** + +**值** + +**说明** + +Default + +0 + +默认 + +Headset + +1 + +有线耳机 + +Earpiece + +2 + +听筒 + +HeadsetNoMic + +3 + +无麦克风耳机 + +SpeakerPhone + +4 + +扬声器 + +Usb + +5 + +USB 设备 + +Bluetooth + +6 + +蓝牙 SCO + +BluetoothA2dp + +7 + +蓝牙 A2DP + +### 回调对照 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +设备状态变化 + +`onAudioDeviceStateChanged(state)` + +`onAudioDeviceStateChanged:` + +`onAudioDeviceStateChanged(state, reason)` + +路由变化 + +`onAudioDeviceRouteChanged(routeType)` + +`onAudioDeviceRouteChanged:` + +`onAudioDeviceRouteChanged(routeType)` + +设备中断 + +`onAudioDeviceInterrupted(interrupt)` + +`onAudioDeviceInterrupted:` + +`onAudioDeviceInterrupted(interrupt)` + +文件状态 + +`onAudioFileState(state)` + +`onAudioFileState:` + +`onAudioFileState(fileId, stateCode, errorCode)` + +## **音频错误码与警告码** + +### 音频错误码 + +**错误码** + +**值** + +**说明** + +AoqErrorCodeAudio + +100 + +通用音频错误 + +AudioExternalBufferFull + +110 + +外部缓冲区已满 + +AudioDevice + +120 + +设备通用错误 + +RecordingAuthFailed + +121 + +麦克风权限失败 + +RecordingOccupied + +122 + +麦克风被占用 + +RecordingBackgroundStart + +123 + +后台启动录音 + +RecordingStartFail + +124 + +录音启动失败 + +PlayoutOccupied + +125 + +播放设备被占用 + +PlayoutBackgroundStart + +126 + +后台启动播放 + +PlayoutStartFail + +127 + +播放启动失败 + +EarpieceRequiresVoipMode + +128 + +听筒需要启用 VoIP 模式 + +### 音频警告码 + +**警告码** + +**值** + +**说明** + +AoqWCAudio + +100 + +通用音频警告 + +AudioHowling + +101 + +啸叫检测 + +AudioDevice + +120 + +设备通用警告 + +MicEnumerateError + +121 + +麦克风枚举错误 + +MicStartTimeout + +122 + +麦克风启动超时 + +RecordingError + +123 + +录音错误 + +SpeakerEnumerateError + +124 + +扬声器枚举错误 + +SpeakerStartTimeout + +125 + +扬声器启动超时 + +PlayoutError + +126 + +播放错误 + +## **iOS 专有：AVAudioSession 控制** + +iOS 平台提供了 `setAudioSessionRestriction` 接口，可精细控制 SDK 对系统 AVAudioSession 的管理权限。 + +**控制项** + +**说明** + +SetCategory + +SDK 是否有权设置 Session 类别 + +ConfigureSession + +SDK 是否有权配置 Session 参数 + +DeactivateSession + +SDK 是否有权停用 Session + +ActivateSession + +SDK 是否有权激活 Session + +通过按位组合传入 restriction 值，可限制 SDK 对 AVAudioSession 的控制范围，避免与应用层其他音频组件冲突。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md new file mode 100644 index 00000000..433f6552 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md @@ -0,0 +1,171 @@ +# 连接状态管理 + +本章节介绍 AOQ Client SDK 的连接状态机及其对应的 API 调用。 + +## **连接状态图** + +下图描述了 AOQ Client SDK 的连接状态迁移关系： + +AOQ Client SDK 连接状态迁移图 + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3811364871/p1088081.svg) + +## **状态说明** + +**状态** + +**枚举值** + +**说明** + +链接中（Connecting） + +1 + +调用 `connect` 后进入，正在与 AI Service 建立连接 + +已连接（Connected） + +2 + +连接建立成功，可正常收发音视频和数据消息 + +失败（Failed） + +3 + +连接异常（鉴权失败/超时/服务端拒绝等），SDK 内部会自动迁移到已断开 + +已断开（Disconnected） + +0 + +初始状态/主动断开/异常断开后的终态 + +## **状态迁移规则** + +1. **App 调用** `connect` → 进入 `Connecting` 状态。 + +2. **链接成功** → 从 `Connecting` 迁移到 `Connected`。 + +3. **链接异常** → 从 `Connecting` 迁移到 `Failed`，随后 SDK 自动迁移到 `Disconnected`。 + +4. **链接异常** → 从 `Connected` 迁移到 `Failed`，随后 SDK 自动迁移到 `Disconnected`。 + +5. **App 调用** `disconnect` → 从 `Connected` 迁移到 `Disconnected`。 + + +**说明**：`Failed` 是瞬态，SDK 触发 `onConnectionStatusChange(Failed)` 后会自动迁移到 `Disconnected`，业务层无需手动调用 disconnect。 + +## **connect API** + +调用 `connect` 发起与 AI Service 的连接，传入由 AppServer allocate 接口返回的鉴权凭证。 + +### **方法签名** + +**Android：** + +``` +public abstract int connect(@NonNull AoqConnectConfig config); +``` + +**iOS：** + +``` +- (int)connect:(AoqConnectConfig * _Nonnull)config; +``` + +**Ohos：** + +``` +connect(config: AoqConnectConfig): number; +``` + +**返回值：**`0` 表示调用成功（异步建连）；`< 0` 表示失败。 + +### **行为说明** + +- 调用后触发 `onConnectionStatusChange(connecting)` 回调。 + +- 如果链接成功时触发 `onConnectionStatusChange(connected)` 回调。 + +- 如果链接失败时触发 `onConnectionStatusChange(failed)` 回调。 + + +## **disconnect API** + +调用 `disconnect` 主动断开与 AI Service 的连接。 + +### **方法签名** + +**Android：** + +``` +public abstract int disconnect(); +``` + +**iOS：** + +``` +- (int)disconnect; +``` + +**Ohos：** + +``` +disconnect(): number; +``` + +**返回值：**`0` 表示成功；`< 0` 表示失败。 + +### **行为说明** + +- 调用后触发 `onConnectionStatusChange(Disconnected)` 回调。 + +- 引擎不会自动释放，可重新调用 `connect` 进行重连。 + +- 未连接状态下调用 `disconnect` 是安全的，返回 0。 + + +## **onConnectionStatusChange 回调** + +连接状态变化时，SDK 通过此回调通知业务层。 + +**Android：** + +``` +public void onConnectionStatusChange( + @NonNull AoqClientEngine.AoqConnectionStatus status) {} +``` + +**iOS：** + +``` +- (void)onConnectionStatusChange:(AoqConnectionStatus)status; +``` + +**Ohos：** + +``` +onConnectionStatusChange?: (status: AoqConnectionStatus) => void; +``` + +### **示例代码** + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + switch status { + case .connecting: + print("正在连接...") + case .connected: + print("连接成功") + // 连接成功后可发送 session.update + case .failed: + print("连接失败") + // SDK 会自动迁移到 disconnected，无需手动 disconnect + case .disconnected: + print("已断开") + // 可根据业务决定是否重连 + } +} +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md new file mode 100644 index 00000000..670fc6f0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md @@ -0,0 +1,253 @@ +# 自定义音频采集 + +介绍如何使用 AOQ Client SDK 实现自定义音频采集功能，包括外部音频流的添加、PCM 数据推送和管理。 + +## **功能介绍** + +AOQ Client SDK 内部音频模块可满足应用中对基本音频功能的需求，但在特定场景中，SDK 内部的音频采集模块可能无法满足开发需求，需要实现自定义音频采集功能，例如： + +- 解决音频采集设备被占用问题。 + +- 需要从定制的采集系统、音频文件中获取音频数据后交给 SDK 传输。 + +- 需要将 AI TTS 生成的音频数据通过 SDK 推流传输。 + + +AOQ Client SDK 支持灵活的自定义采集功能，允许用户根据业务场景自行管理音频设备与音频源。外部音频流的数据会与内部采集的音频数据混音后一起推流发送。 + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +### **1\. 打开或关闭音频采集** + +需要先开启音频采集，外部音频流输入的数据会与内部采集数据混音后一起推流。如果不需要内部麦克风采集，可以设置 `isExternal=true` 关闭内部采集设备。 + +``` +// 方式一：开启内部采集，外部音频流数据会与麦克风数据混音推流 +AoqClientEngine.AoqAudioCaptureConfig config = new AoqClientEngine.AoqAudioCaptureConfig(); +config.isExternal = false; // 使用内部麦克风采集 +config.isVoipMode = false; +engine.startAudioCapture(config); + +// 方式二：关闭内部采集，仅推送外部音频流数据 +AoqClientEngine.AoqAudioCaptureConfig config = new AoqClientEngine.AoqAudioCaptureConfig(); +config.isExternal = true; // 不打开麦克风，由外部音频流提供数据 +engine.startAudioCapture(config); +``` + +### **2\. 连接成功后，添加外部音频流** + +在 `onConnectionStatusChange` 回调状态变为 `AoqConnectionStatusConnected` 后，调用 `addAudioExternalStream` 添加外部音频流。需要指定一个唯一的 `streamId` 用于后续推送数据和管理。 + +如果需要音频 3A 处理（回声消除、噪声抑制、自动增益），请配置 `AoqAudioExternalStreamConfig` 中的 `enable3A` 参数。 + +``` +// 在 onConnectionStatusChange 回调中确认连接成功后添加 +@Override +public void onConnectionStatusChange(AoqClientEngine.AoqConnectionStatus status) { + if (status == AoqClientEngine.AoqConnectionStatus.AoqConnectionStatusConnected) { + addExternalAudioStream(); + } +} + +private void addExternalAudioStream() { + AoqClientEngine.AoqAudioExternalStreamConfig config = new AoqClientEngine.AoqAudioExternalStreamConfig(); + config.sampleRate = 48000; // 采样率，需与实际音频数据一致 + config.channels = 1; // 声道数 + config.publishVolume = 100; // 推流音量 [0-100] + config.playoutVolume = 0; // 本地播放音量 [0-100]，0 表示不本地播放 + config.maxBufferDuration = 1000; // 最大缓冲时长（毫秒） + config.enable3A = true; // 是否对输入 PCM 进行 3A 处理 + + String streamId = "external_audio_1"; + int ret = engine.addAudioExternalStream(streamId, config); + if (ret == 0) { + mExternalStreamId = streamId; + } +} +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +AoqTrackTypeAudio + +音频轨道类型 + +codecType + +AoqEncoderType + +AoqEncoderTypeAudioPCM + +音频流格式 + +channels + +int + +1 + +声道数 + +sampleRate + +int + +48000 + +采样率（Hz） + +playoutVolume + +int + +100 + +播放音量 \[0-100\] + +publishVolume + +int + +100 + +推流音量 \[0-100\] + +maxBufferDuration + +int + +1000 + +最大缓冲时长（毫秒） + +enable3A + +boolean + +false + +是否对输入 PCM 进行 3A 处理 + +### **3\. 实现自采集模块或从文件获取 PCM 数据** + +自定义采集功能需要根据业务场景自行采集并处理音频数据，之后将数据传入 SDK 进行传输。常见的数据来源： + +- **麦克风采集**：通过 Android AudioRecord 采集 PCM 数据。 + +- **文件读取**：从本地 PCM/WAV 音频文件中解析获取 PCM 数据。 + +- **AI TTS**：从语音合成引擎获取 PCM 数据。 + +- **网络流**：从网络音频流中解码获取 PCM 数据。 + + +音频数据需要为 PCM 格式，并记录对应的采样率、声道数等参数，用于构造 `AoqAudioFrameData` 对象。 + +### **4\. 通过外部音频流 ID 推送音频数据到 SDK** + +调用 `pushAudioExternalStreamData` 接口，将采集到的 PCM 音频数据传入 SDK。 + +- 从硬件采集：建议采集 10ms 为一帧数据，采集到数据就 push 给 SDK。 + +- 从文件解析：建议 40ms 为一帧数据，每 push 一帧 Sleep 30ms 后 push 下一帧。 + +- 需要维护一个 `running` 标记，当引擎退出或 stream ID 被删除时退出推送循环。 + + +``` +// 成员变量：控制推送循环的运行标记 +private volatile boolean mPushRunning = false; + +// 推送单帧音频数据 + +private void pushAudioData(byte[] audioData, int bytesRead) { + + if (engine == null || mExternalStreamId == null || bytesRead <= 0) { + return; + } + + int channels = 1; + int bytesPerSample = 2; // 16bit PCM + int sampleRate = 48000; + + // 构造音频帧数据 + AoqClientEngine.AoqAudioFrameData frameData = new AoqClientEngine.AoqAudioFrameData(); + frameData.dataPtr = audioData; + frameData.dataSize = bytesRead; + frameData.numOfSamples = bytesRead / (channels * bytesPerSample); + frameData.bytesPerSample = bytesPerSample; + frameData.numOfChannels = channels; + frameData.samplesPerSec = sampleRate; + + // 推送数据，处理缓冲区满的情况 + int ret; + final int WAIT_MS = 30; + + do { + // 检查运行标记和 stream ID 是否仍有效 + if (!mPushRunning || mExternalStreamId == null) { + break; + } + ret = engine.pushAudioExternalStreamData(mExternalStreamId, frameData); + if (ret == 110) { // AoqErrorCodeAudioExternalBufferFull + try { + Thread.sleep(WAIT_MS); + } catch (InterruptedException e) { + break; + } + } else { + break; + } + } while (true); +} +``` + +**注意事项：** + +- 需要在连接成功且添加外部音频流之后再开始推送数据。 + +- 需要按照数据的实际长度设置 `AoqAudioFrameData` 的 `numOfSamples`。 + +- 调用 `pushAudioExternalStreamData` 时，可能出现内部缓冲区满（错误码 110）而导致失败，需要等待重试。 + +- 实时采集建议 10ms 一帧数据 push，有数据就调用 push，注意处理内部缓冲区满（错误码 110）。 + +- 从文件解析建议 40ms 一帧数据，间隔 30ms 调用 push，注意处理内部缓冲区满（错误码 110）。 + +- 引擎退出（`destroy`）或 stream ID 被移除前，必须先设置 `mPushRunning = false` 停止推送循环，避免在已释放的资源上操作。 + + +### **5\. 移除外部音频流** + +当不再需要发布自定义采集的音频时，先停止推送循环，再调用 `removeAudioExternalStream` 接口移除外部音频流。 + +``` +// 先停止推送 +stopPushAudio(); +// 再移除外部音频流 +engine.removeAudioExternalStream(mExternalStreamId); +mExternalStreamId = null; +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md new file mode 100644 index 00000000..1d94d8e7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md @@ -0,0 +1,235 @@ +# 自定义音频播放 + +AOQ Client SDK 支持自定义音频播放功能，通过音频帧回调机制将解码后的 PCM 数据回调给应用层，由开发者自行实现音频渲染播放。 + +## **功能介绍** + +AOQ Client SDK 内部音频模块默认会将接收到的远端音频数据通过系统扬声器/听筒播放，但在特定场景中，SDK 内部的音频播放模块可能无法满足开发需求，需要实现自定义音频播放功能，例如： + +- 需要将接收到的音频数据输出到自定义的播放设备或音频处理管线。 + +- 需要对接收到的音频数据进行二次处理（如 AI 语音识别、音效处理等）。 + +- 解决音频播放设备被占用的问题。 + + +AOQ Client SDK 支持灵活的自定义播放功能，通过音频帧回调机制，将解码后的 PCM 数据回调给应用层，由开发者自行实现音频渲染播放。 + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +### **1\. 开启音频播放（外部模式）** + +调用 `startAudioPlayer` 时设置 `isExternal=true`，关闭 SDK 内部的音频渲染设备，由应用层自行处理音频播放。 + +``` +AoqClientEngine.AoqAudioPlaybackConfig config = new AoqClientEngine.AoqAudioPlaybackConfig(); +config.isExternal = true; // 关闭 SDK 内部播放，由应用层自行渲染 +config.channel = 1; // 声道数 +engine.startAudioPlayer(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +isVoipMode + +boolean + +false + +是否启用 VoIP 模式（硬件AEC），移动端有效 + +isDefaultSpeaker + +boolean + +true + +是否默认扬声器，移动端有效 + +isExternal + +boolean + +false + +是否外部播放模式，true 时 SDK 不打开播放设备 + +channel + +int + +1 + +声道数 + +### **2\. 设置音频帧回调监听** + +调用 `setAudioFrameObserver` 设置音频帧数据回调监听器，实现 `onPlaybackAudioFrame` 回调方法接收播放 PCM 数据。 + +``` +engine.setAudioFrameObserver(new AoqClientListener.AoqAudioFrameListener() { + @Override + public void onPlaybackAudioFrame(@NonNull AoqClientEngine.AoqAudioFrameData frame) { + // 在此处理接收到的播放音频数据 + // frame.dataPtr: PCM 数据 + // frame.numOfSamples: 采样点数 + // frame.numOfChannels: 声道数 + // frame.samplesPerSec: 采样率 + // frame.bytesPerSample: 每采样点字节数 + playPcmData(frame); + } +}); +``` + +### **3\. 开启播放数据回调** + +调用 `enableAudioFrameObserver` 开启播放位置的音频帧回调，指定数据源为 `AoqAudioSourcePlayback`。 + +``` +AoqClientEngine.AoqAudioObserverConfig observerConfig = new AoqClientEngine.AoqAudioObserverConfig(); +observerConfig.sampleRate = 48000; // 回调音频采样率 +observerConfig.channels = 1; // 回调音频声道数 +observerConfig.mode = AoqClientEngine.AoqAudioObserverMode.AoqAudioObserverModeReadOnly; // 只读模式 + +engine.enableAudioFrameObserver( + true, // 开启回调 + AoqClientEngine.AoqAudioSource.AoqAudioSourcePlayback, // 播放数据源 + observerConfig +); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +sampleRate + +int + +48000 + +回调音频采样率（Hz） + +channels + +int + +1 + +回调音频声道数 + +mode + +AoqAudioObserverMode + +AoqAudioObserverModeReadOnly + +读写模式 + +### **4\. 实现自定义音频渲染** + +在 `onPlaybackAudioFrame` 回调中接收到 PCM 数据后，由应用层自行实现音频渲染播放。常见的实现方式： + +- **Android AudioTrack**：通过 AudioTrack 将 PCM 数据写入系统音频设备播放。 + +- **AI 语音识别**：将 PCM 数据传入 ASR 引擎进行语音识别。 + +- **音效处理**：对 PCM 数据进行音效处理后再播放。 + +- **文件存储**：将接收到的音频数据保存到本地文件。 + + +``` +// 示例：使用 Android AudioTrack 播放 +private AudioTrack mAudioTrack; +private volatile boolean mPlayRunning = false; + +private void initAudioTrack(int sampleRate, int channels) { + int channelConfig = (channels == 2) + ? AudioFormat.CHANNEL_OUT_STEREO + : AudioFormat.CHANNEL_OUT_MONO; + int bufferSize = AudioTrack.getMinBufferSize( + sampleRate, channelConfig, AudioFormat.ENCODING_PCM_16BIT); + + mAudioTrack = new AudioTrack( + AudioManager.STREAM_VOICE_CALL, + sampleRate, + channelConfig, + AudioFormat.ENCODING_PCM_16BIT, + bufferSize, + AudioTrack.MODE_STREAM); + mAudioTrack.play(); + mPlayRunning = true; +} + +private void playPcmData(AoqClientEngine.AoqAudioFrameData frame) { + if (!mPlayRunning || mAudioTrack == null) { + return; + } + if (frame.dataPtr != null && frame.dataSize > 0) { + mAudioTrack.write(frame.dataPtr, 0, frame.dataSize); + } +} +``` + +**注意事项：** + +- `onPlaybackAudioFrame` 回调在 SDK 内部线程触发，回调中的 `frame.dataPtr` 仅在回调期间有效，异步使用需自行拷贝。 + +- AudioTrack.write 是阻塞操作，在回调中直接写入即可，SDK 内部会按节奏回调。 + +- 需要维护 `mPlayRunning` 标记，当引擎退出或停止播放时退出处理逻辑。 + + +### **5\. 停止自定义播放** + +当不再需要自定义播放时，先关闭音频帧回调，再停止播放设备，释放 AudioTrack 资源。 + +``` +// 1. 关闭播放位置的音频帧回调 +AoqClientEngine.AoqAudioObserverConfig observerConfig = new AoqClientEngine.AoqAudioObserverConfig(); +engine.enableAudioFrameObserver( + false, // 关闭回调 + AoqClientEngine.AoqAudioSource.AoqAudioSourcePlayback, + observerConfig +); + +// 2. 移除音频帧回调监听 +engine.setAudioFrameObserver(null); + +// 3. 停止 SDK 音频播放 +engine.stopAudioPlayer(); + +// 4. 释放 AudioTrack 资源 +mPlayRunning = false; +if (mAudioTrack != null) { + mAudioTrack.stop(); + mAudioTrack.release(); + mAudioTrack = null; +} +``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md new file mode 100644 index 00000000..358f530f --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md @@ -0,0 +1,647 @@ +# 自定义视频输入 + +介绍 AOQ Client SDK 自定义视频输入的两种模式：原始帧模式和编码帧模式，以及各模式的配置方法和示例代码。 + +## **功能介绍** + +AOQ Client SDK 内部视频模块可满足应用中对基本视频功能的需求，但在特定场景中，SDK 内部的视频采集模块可能无法满足开发需求，需要实现自定义视频采集功能，例如： + +- 解决摄像头设备被占用或不兼容问题。 + +- 需要从定制的采集系统、视频文件中获取视频数据后交给 SDK 传输。 + +- 需要将 AI 生成的画面、屏幕录制、虚拟摄像头等内容通过 SDK 推流传输。 + + +AOQ Client SDK 支持两种自定义视频采集模式： + +- **原始帧模式**：自行采集原始视频帧（BGRA、I420、NV12、NV21 等格式），通过 `pushExternalVideoCapturedFrame` 推送给 SDK 进行编码和传输。SDK 内部完成编码、传输等完整流程。 + +- **编码帧模式**：自行完成视频编码（目前支持 JPEG），通过 `pushExternalVideoEncodedFrame` 直推已编码数据给 SDK，跳过 SDK 内部编码器，直接打包发送。 + + +## **示例代码** + +暂无 + +## **前提条件** + +- 已创建引擎实例（调用 `createEngine`）。 + +- 已成功连接服务器（`onConnectionStatusChange` 回调状态为 `AoqConnectionStatusConnected`）。 + + +## **功能实现** + +根据业务场景选择以下两种模式之一。两种模式不可混用：同一时间只能使用其中一种推送接口。 + +## **模式一：原始帧模式** + +自行采集原始视频帧（BGRA、I420、NV12、NV21 等格式），推送给 SDK 进行编码和传输。SDK 内部完成编码、传输等完整流程。 + +### **1\. 配置视频编码参数** + +SDK 内部编码器会对推送的原始帧进行编码，可根据业务需要调整编码参数。 + +``` +AoqClientEngine.AoqVideoCodecConfig config = new AoqClientEngine.AoqVideoCodecConfig(); +config.width = 1280; +config.height = 720; +config.fps = 2; +config.bitrate = 500000; // 起始码率 500kbps +config.minBitrate = 128000; // 最小码率 128kbps +config.keyframeInterval = 2; +// isExternal 保持默认 false，SDK 内部编码 + +engine.setVideoEncoderConfig(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +AoqTrackTypeVideo + +视频轨道类型 + +codecType + +AoqEncoderType + +AoqEncoderTypeVideoH264 + +编码器类型 + +width + +int + +720 + +编码宽度（像素） + +height + +int + +1280 + +编码高度（像素） + +fps + +int + +5 + +帧率 + +bitrate + +int + +500000 + +起始码率（bps） + +minBitrate + +int + +128000 + +最小码率（bps） + +keyframeInterval + +int + +2 + +关键帧间隔（秒） + +isExternal + +boolean + +false + +原始帧模式保持 false + +mirrorMode + +AoqMirrorMode + +AoqMirrorModeDisabled + +镜像模式 + +orientationMode + +AoqOrientationMode + +AoqOrientationModeAuto + +画面方向模式 + +### **2\. 以外部采集模式启动视频采集** + +调用 `startVideoCapture` 并设置 `isExternal=true`，告知 SDK 不打开摄像头，由外部源提供视频帧。这是原始帧模式的前置条件，未调用则 SDK 不会消费推送的帧数据。 + +``` +AoqClientEngine.AoqVideoCaptureConfig config = new AoqClientEngine.AoqVideoCaptureConfig(); +config.isExternal = true; // 不打开摄像头，由外部源推送视频帧 +// isExternal=true 时 width/height/fps 无效，实际分辨率和帧率由推送数据决定 +int ret = engine.startVideoCapture(config); +``` + +**参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +width + +int + +1280 + +采集宽度（`isExternal=true` 时无效） + +height + +int + +720 + +采集高度（`isExternal=true` 时无效） + +fps + +int + +15 + +采集帧率（`isExternal=true` 时无效） + +isExternal + +boolean + +false + +true：不打开摄像头，由外部源推送帧数据 + +cameraDirection + +AoqCameraDirection + +AoqCameraDirectionFront + +摄像头方向（`isExternal=true` 时无效） + +### **3\. 推送原始视频帧** + +调用 `pushExternalVideoCapturedFrame` 接口，将采集到的原始视频帧传入 SDK。SDK 内部完成编码和传输。 + +支持的视频帧格式：BGRA、I420、NV12、NV21、RGBA。Apple 平台额外支持 CVPixelBuffer 零拷贝格式。 + +#### **3.1 BGRA 格式** + +BGRA 为打包格式，每个像素 4 字节（Blue、Green、Red、Alpha），一帧数据量 = width x height x 4。 + +``` +// 构造 BGRA 视频帧 +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatBGRA; +frame.width = 1280; +frame.height = 720; + +frame.data = bgraBytes; // byte[]，长度 = width * height * 4 + +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.2 I420 格式** + +I420 为三平面格式（Y、U、V 分离），Y 平面大小 = width x height，U/V 平面各为 (width/2) x (height/2)。 + +``` +// 构造 I420 视频帧 +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatI420; +frame.width = 1280; +frame.height = 720; + +frame.dataY = yPlane; // byte[]，长度 = width * height + +frame.dataU = uPlane; // byte[]，长度 = (width/2) * (height/2) + +frame.dataV = vPlane; // byte[]，长度 = (width/2) * (height/2) + +frame.strideY = 1280; // Y 平面行字节数 +frame.strideU = 640; // U 平面行字节数 +frame.strideV = 640; // V 平面行字节数 +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.3 NV12 / NV21 格式** + +NV12 和 NV21 为半平面格式，Y 平面 + UV 交错平面。NV12 为 UV 交替排列，NV21 为 VU 交替排列。数据量 = width x height x 3 / 2，打包在 `data` 字段中。 + +``` +// 构造 NV12 视频帧（NV21 同理，修改 format 即可） +AoqClientEngine.AoqVideoFrame frame = new AoqClientEngine.AoqVideoFrame(); +frame.format = AoqClientEngine.AoqVideoPixelFormat.AoqVideoPixelFormatNV12; +frame.width = 1280; +frame.height = 720; + +frame.data = nv12Bytes; // byte[]，长度 = width * height * 3 / 2 + +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoCapturedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +#### **3.4 CVPixelBuffer 格式（Apple 平台）** + +iOS / macOS 平台支持直接传递 `CVPixelBufferRef`，实现零拷贝传输，避免内存拷贝带来的性能开销。 + +``` +// iOS / macOS 平台 +let frame = AoqVideoFrame() +frame.format = .cvPixelBuffer +frame.width = 1280 +frame.height = 720 +frame.pixelBuffer = pixelBuffer // CVPixelBufferRef +frame.timeStamp = Int64(Date().timeIntervalSince1970 * 1000) + +// SDK 内部异步持有 pixelBuffer，需要额外 +1 引用计数 +// SDK 消费完毕后会自行释放 +let _ = Unmanaged.passRetained(pixelBuffer) + +engine.pushExternalVideoCapturedFrame(.video, frame: frame) +``` + +### **4\. 停止原始帧采集** + +当不再需要推送视频帧时，先停止推帧定时器，再调用 `stopVideoCapture` 关闭视频采集。 + +``` +// 1. 停止推帧定时器 +stopExternalFramePush(); +// 2. 停止视频采集 +engine.stopVideoCapture(); +``` + +## **模式二：编码帧模式** + +自行完成视频编码（目前支持 JPEG），直推已编码数据给 SDK，跳过 SDK 内部编码器，直接打包发送。此模式**不需要调用** `startVideoCapture` 等采集相关接口。 + +### **1\. 配置视频编码参数并启用外部编码** + +调用 `setVideoEncoderConfig` 并设置 `isExternal=true`，告知 SDK 跳过内部编码器，由外部提供已编码数据。 + +``` +AoqClientEngine.AoqVideoCodecConfig config = new AoqClientEngine.AoqVideoCodecConfig(); +config.width = 1280; +config.height = 720; +config.fps = 2; +config.isExternal = true; // 跳过内部编码，由外部推送已编码帧 + +engine.setVideoEncoderConfig(config); +``` + +设置完成后即可直接推送编码帧，无需调用 `startVideoCapture`。 + +### **2\. 推送编码视频帧** + +调用 `pushExternalVideoEncodedFrame` 接口，将已编码的视频数据直传给 SDK。目前仅支持 JPEG 编码格式。 + +``` +// 从 Bitmap 生成 JPEG 数据 +android.graphics.Bitmap bmp = android.graphics.Bitmap.createBitmap( + width, height, android.graphics.Bitmap.Config.ARGB_8888); +// ... 填充 Bitmap 内容 ... + +java.io.ByteArrayOutputStream baos = new java.io.ByteArrayOutputStream(); +bmp.compress(android.graphics.Bitmap.CompressFormat.JPEG, 85, baos); +bmp.recycle(); + +// 构造编码帧并推送 +AoqClientEngine.AoqVideoEncodedFrame frame = new AoqClientEngine.AoqVideoEncodedFrame(); +frame.codec = AoqClientEngine.AoqVideoCodecType.AoqVideoCodecTypeJPEG; +frame.data = baos.toByteArray(); +frame.width = width; +frame.height = height; +frame.timeStamp = System.currentTimeMillis(); + +int ret = engine.pushExternalVideoEncodedFrame( + AoqClientEngine.AoqTrackType.AoqTrackTypeVideo, frame); +``` + +**AoqVideoEncodedFrame 参数说明：** + +**参数** + +**类型** + +**默认值** + +**说明** + +codec + +AoqVideoCodecType + +AoqVideoCodecTypeJPEG + +编码格式，目前仅支持 JPEG + +data + +byte\[\] + +null + +编码后的数据 + +width + +int + +0 + +画面宽度（像素） + +height + +int + +0 + +画面高度（像素） + +timeStamp + +long + +0 + +时间戳（毫秒），为 0 时 SDK 使用本地时钟补充 + +### **3\. 停止编码帧推送** + +编码帧模式无需管理采集设备，停止推帧定时器即可。 + +``` +stopExternalFramePush(); +``` + +## **视频帧格式参考** + +### **AoqVideoFrame（原始帧模式使用）** + +**字段** + +**类型** + +**说明** + +format + +AoqVideoPixelFormat + +像素格式 + +width + +int + +画面宽度（像素） + +height + +int + +画面高度（像素） + +data + +byte\[\] + +打包格式数据（NV12/NV21/BGRA/RGBA） + +dataY + +byte\[\] + +I420 Y 平面数据 + +dataU + +byte\[\] + +I420 U 平面数据 + +dataV + +byte\[\] + +I420 V 平面数据 + +strideY + +int + +I420 Y 平面行字节数 + +strideU + +int + +I420 U 平面行字节数 + +strideV + +int + +I420 V 平面行字节数 + +textureId + +int + +Android 纹理 ID（TextureOES/Texture2D） + +transformMatrix + +float\[\] + +纹理变换矩阵（4x4 行优先） + +eglContext + +EGLContext + +Android 共享 EGL context（纹理模式使用） + +pixelBuffer + +CVPixelBufferRef + +Apple 零拷贝 CVPixelBuffer（仅 iOS/macOS） + +timeStamp + +long + +时间戳（毫秒），为 0 时 SDK 用本地时钟补充 + +### **AoqVideoPixelFormat 枚举值** + +**枚举值** + +**数值** + +**说明** + +AoqVideoPixelFormatUnknown + +0 + +未知格式 + +AoqVideoPixelFormatI420 + +1 + +I420 三平面格式 + +AoqVideoPixelFormatNV12 + +2 + +NV12 半平面格式（UV 交替） + +AoqVideoPixelFormatNV21 + +3 + +NV21 半平面格式（VU 交替） + +AoqVideoPixelFormatBGRA + +4 + +BGRA 打包格式 + +AoqVideoPixelFormatRGBA + +5 + +RGBA 打包格式 + +AoqVideoPixelFormatCVPixelBuffer + +6 + +Apple CVPixelBuffer（仅 iOS/macOS） + +AoqVideoPixelFormatTextureOES + +7 + +Android OES 外部纹理 + +AoqVideoPixelFormatTexture2D + +8 + +Android 2D 纹理 + +### **AoqVideoEncodedFrame（编码帧模式使用）** + +**字段** + +**类型** + +**说明** + +codec + +AoqVideoCodecType + +编码格式 + +data + +byte\[\] + +编码后的数据 + +width + +int + +画面宽度（像素） + +height + +int + +画面高度（像素） + +timeStamp + +long + +时间戳（毫秒），为 0 时 SDK 用本地时钟补充 + +### **AoqVideoCodecType 枚举值** + +**枚举值** + +**数值** + +**说明** + +AoqVideoCodecTypeJPEG + +0 + +JPEG 编码格式 + +## **注意事项** + +- 原始帧模式：必须先调用 `startVideoCapture(isExternal=true)` 再推送帧，否则 SDK 返回参数错误。 + +- 编码帧模式：只需调用 `setVideoEncoderConfig(isExternal=true)` 即可推送，**不需要**调用 `startVideoCapture`。 + +- 原始帧模式与编码帧模式不可混用：同一时间只能使用其中一种推送接口。 + +- 编码帧模式目前仅支持 JPEG 格式。 + +- 视频帧数据在推送后由 SDK 内部管理生命周期，调用方无需在推送后继续持有数据引用。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md new file mode 100644 index 00000000..69b625b7 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md @@ -0,0 +1,115 @@ +# 媒体流发送管理 + +`enableSendMediaStream` 用于控制客户端是否向 AI 服务发送音频/视频媒体流，在 AOQ 协议接入场景中精确控制发送时机。 + +## **概述** + +`enableSendMediaStream` 用于控制客户端是否向 AI 服务发送音频/视频媒体流。在 AOQ 协议接入场景中，部分模型要求在收到 `session.updated` 确认后才能接收媒体数据，因此需要通过此接口精确控制发送时机。 + +## **API 定义** + +**iOS / Mac** + +``` +// iOS / Mac +func enableSendMediaStream(_ trackType: AoqTrackType, enable: Bool) +``` + +**Android** + +``` +// Android +void enableSendMediaStream(AoqTrackType trackType, boolean enable) +``` + +**OHOS (ArkTS)** + +``` +// OHOS (ArkTS) +enableSendMediaStream(trackType: AoqTrackType, enable: boolean): void +``` + +**参数说明：** + +**参数** + +**类型** + +**说明** + +trackType + +AoqTrackType + +媒体轨道类型：`.audio` 或 `.video` + +enable + +Bool + +`true` = 开启发送，`false` = 暂停发送 + +## **控制流程** + +典型的媒体流控制流程如下： + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0575914871/p1088082.svg) + +## **使用示例（iOS Swift）** + +### **connect 前禁用发送** + +``` +// 连接前关闭音视频发送，避免模型未就绪时收到数据 +engine.enableSendMediaStream(.audio, enable: false) +engine.enableSendMediaStream(.video, enable: false) + +// 发起连接 +engine.connect(config) +``` + +### **收到 session.updated 后开启** + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + + if type == "session.updated" { + // AI 侧已确认会话配置，开启媒体发送 + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +## **注意事项** + +- **调用时机**：必须在 `createEngine` 之后调用，引擎未创建时调用无效。 + +- **默认行为**：如果不调用此接口，connect 成功后 SDK 会立即开始发送媒体流。 + +- **模型兼容性**：部分模型要求先收到 `session.updated` 再接收媒体数据，建议统一采用"先禁用、后开启"模式。 + +- **独立控制**：音频和视频可独立控制，例如仅发送音频不发送视频。 + + +## **常见场景** + +**场景** + +**操作** + +**说明** + +连接模型前 + +`enable(.audio, false)` + +等待 session.updated 再发送 + +收到 session.updated + +`enable(.audio, true)` + +AI 已就绪，开始发送 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md new file mode 100644 index 00000000..f71bf047 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md @@ -0,0 +1,1186 @@ +# 视频常用功能介绍 + +AOQ Client SDK 提供了完整的视频能力，覆盖视频采集、渲染显示、编码配置、帧数据回调、外部视频输入等核心场景。本文档基于 Android（Java）、iOS（Objective-C）、Ohos（ArkTS）三个平台的公开 API，对视频常用功能进行统一介绍。 + +## **1\. 视频采集** + +### 1.1 功能说明 + +视频采集用于打开设备摄像头，将实时视频帧数据送入 SDK 编码推流管线。SDK 支持两种采集模式： + +- **内部采集（默认）**：SDK 自动管理摄像头设备的打开、帧采集和关闭，支持前后置摄像头切换。 + +- **外部采集**：由应用自行管理摄像头或其他视频源，采集到的帧数据通过 `pushExternalVideoCapturedFrame` 接口输入 SDK。 + + +### 1.2 采集配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +width + +int + +1280 + +采集宽度（像素），外部采集时无效 + +height + +int + +720 + +采集高度（像素），外部采集时无效 + +fps + +int + +15 + +采集帧率，外部采集时由送帧节奏决定 + +isExternal + +bool + +false + +是否使用外部采集模式 + +cameraDirection + +AoqCameraDirection + +Front(0) + +摄像头方向，外部采集时无效 + +### 1.3 摄像头方向枚举 + +**枚举值** + +**数值** + +**说明** + +AoqCameraDirectionFront + +0 + +前置摄像头 + +AoqCameraDirectionBack + +1 + +后置摄像头 + +### 1.4 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +开启采集 + +startVideoCapture(config) + +startVideoCapture:config: + +startVideoCapture(config) + +关闭采集 + +stopVideoCapture() + +stopVideoCapture + +stopVideoCapture() + +切换摄像头 + +switchCamera(direction) + +switchCamera: + +switchCamera(direction) + +### 1.5 使用示例 + +**Android** + +``` +AoqVideoCaptureConfig config = new AoqVideoCaptureConfig(); +config.width = 1280; +config.height = 720; +config.fps = 15; +config.cameraDirection = AoqCameraDirection.AoqCameraDirectionFront; +engine.startVideoCapture(config); +``` + +**iOS** + +``` +AoqVideoCaptureConfig *config = [[AoqVideoCaptureConfig alloc] init]; +config.width = 1280; +config.height = 720; +config.fps = 15; +config.cameraDirection = AoqCameraDirectionFront; +[engine startVideoCapture:config]; +``` + +**Ohos** + +``` +let config: AoqVideoCaptureConfig = { + width: 1280, + height: 720, + fps: 15, + cameraDirection: AoqCameraDirection.AoqCameraDirectionFront +}; +engine.startVideoCapture(config); +``` + +## **2\. 视频渲染** + +### 2.1 功能说明 + +视频渲染用于将本地采集或远端接收的视频帧数据显示到屏幕上。SDK 支持设置本地预览窗口和远端渲染窗口，通过 `trackType` 区分视频流（Video）和屏幕共享流（Screen）。 + +### 2.2 渲染模式 + +**枚举值** + +**数值** + +**说明** + +AoqRenderModeAuto + +0 + +自动模式 + +AoqRenderModeStretch + +1 + +拉伸平铺，画面可能变形 + +AoqRenderModeFill + +2 + +填充黑边，画面完整显示 + +AoqRenderModeCrop + +3 + +裁剪模式，画面内容可能丢失 + +### 2.3 画布配置 + +**参数** + +**类型** + +**默认值** + +**说明** + +view + +平台视图 + +null + +渲染视图（Android: SurfaceView/TextureView, iOS: UIView, Ohos: XComponent） + +renderMode + +AoqRenderMode + +Auto(0) + +渲染显示模式 + +### 2.4 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置本地预览 + +setLocalView(trackType, canvas) + +setLocalView:trackType:canvas: + +setLocalView(trackType, canvas) + +设置远端渲染 + +setRemoteView(trackType, canvas) + +setRemoteView:trackType:canvas: + +setRemoteView(trackType, canvas) + +**说明** + +**平台差异**：Android 使用 SurfaceView 或 TextureView 作为渲染容器；iOS 使用 UIView（内部通过 AoqRenderView 封装，支持 Metal 加速）；Ohos 使用 XComponent（通过 AoqXComponentController 管理 native 渲染视图）。 + +## **3\. 视频编码配置** + +### 3.1 功能说明 + +设置视频编码参数，包括编码格式、分辨率、帧率、码率、关键帧间隔、镜像和方向等。通过 `trackType` 区分视频轨道和屏幕共享轨道的编码配置。 + +### 3.2 编码配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +trackType + +AoqTrackType + +Video(1) + +轨道类型：Video + +codecType + +AoqEncoderType + +VideoH264(3) + +编码格式 + +width + +int + +720 + +编码宽度 + +height + +int + +1280 + +编码高度 + +fps + +int + +5 + +编码帧率 + +bitrate + +int + +500000 + +目标码率（bps） + +minBitrate + +int + +128000 + +最小码率（bps） + +keyframeInterval + +int + +2 + +关键帧间隔（秒） + +mirrorMode + +AoqMirrorMode + +Disabled(0) + +镜像模式 + +orientationMode + +AoqOrientationMode + +Auto(0) + +方向模式 + +isExternal + +bool + +false + +外部编码模式（true 时由应用推送已编码帧） + +### 3.3 编码格式枚举 + +**枚举值** + +**数值** + +**说明** + +AoqEncoderTypeVideoH264 + +3 + +H.264 编码 + +AoqEncoderTypeVideoJpeg + +4 + +JPEG 编码（用于外部编码帧） + +### 3.4 镜像模式 + +**枚举值** + +**数值** + +**说明** + +AoqMirrorModeDisabled + +0 + +禁用镜像 + +AoqMirrorModeEnabled + +1 + +启用镜像 + +### 3.5 方向模式 + +**枚举值** + +**数值** + +**说明** + +AoqOrientationModeAuto + +0 + +自动方向 + +AoqOrientationModePortrait + +1 + +竖屏方向 + +AoqOrientationModeLandscape + +2 + +横屏方向 + +### 3.6 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +设置编码参数 + +setVideoEncoderConfig(config) + +setVideoEncoderConfig: + +setVideoEncoderConfig(config) + +## **4\. 外部视频帧输入** + +### 4.1 功能说明 + +外部视频帧输入允许应用将自定义的视频帧数据推送到 SDK，用于外部采集或外部编码场景。支持两种推送方式： + +- **推送原始帧**：将未编码的像素数据（I420/NV12/NV21/BGRA/RGBA 等格式）推送给 SDK，由 SDK 进行编码。 + +- **推送已编码帧**：将已编码的数据（如 JPEG）直推给 SDK，SDK 不做二次编码，直接打包发送。 + + +通过 `trackType` 路由，`AoqTrackTypeVideo` 对应视频采集的外部帧，`AoqTrackTypeScreen` 对应屏幕共享的外部帧。 + +### 4.2 像素格式枚举 + +**枚举值** + +**数值** + +**说明** + +**平台支持** + +AoqVideoPixelFormatI420 + +1 + +I420 三平面 + +全平台 + +AoqVideoPixelFormatNV12 + +2 + +NV12 双平面 + +全平台 + +AoqVideoPixelFormatNV21 + +3 + +NV21 双平面 + +全平台 + +AoqVideoPixelFormatBGRA + +4 + +BGRA 打包 + +全平台 + +AoqVideoPixelFormatRGBA + +5 + +RGBA 打包 + +全平台 + +AoqVideoPixelFormatCVPixelBuffer + +6 + +Apple 零拷贝 + +仅 iOS + +AoqVideoPixelFormatTextureOES + +7 + +OES 纹理 + +仅 Android + +AoqVideoPixelFormatTexture2D + +8 + +2D 纹理 + +仅 Android + +### 4.3 原始视频帧数据结构 (AoqVideoFrame) + +**字段** + +**类型** + +**说明** + +format + +AoqVideoPixelFormat + +像素格式 + +width + +int + +宽度（像素） + +height + +int + +高度（像素） + +data + +byte\[\] / ArrayBuffer + +打包格式数据（NV12/NV21/BGRA/RGBA） + +dataY / dataU / dataV + +byte\[\] / ArrayBuffer + +I420 三平面数据 + +strideY / strideU / strideV + +int + +I420 三平面步长 + +textureId + +int + +纹理 ID（Android TextureOES/Texture2D 时有效） + +transformMatrix + +float\[16\] + +4x4 纹理变换矩阵（Android） + +eglContext + +EGLContext + +共享 EGL 上下文（Android） + +pixelBuffer + +CVPixelBufferRef + +Apple 零拷贝（iOS） + +timeStamp + +long + +时间戳（ms），0 时 SDK 用本地时钟补 + +### 4.4 已编码视频帧数据结构 (AoqVideoEncodedFrame) + +**字段** + +**类型** + +**默认值** + +**说明** + +codec + +AoqVideoCodecType + +JPEG(0) + +编码格式 + +data + +byte\[\] / ArrayBuffer + +\- + +编码后数据 + +width + +int + +\- + +宽度（像素） + +height + +int + +\- + +高度（像素） + +timeStamp + +long + +0 + +时间戳（ms） + +### 4.5 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +推送原始帧 + +pushExternalVideoCapturedFrame(trackType, frame) + +pushExternalVideoCapturedFrame:frame: + +pushExternalVideoCapturedFrame(trackType, frame) + +推送已编码帧 + +pushExternalVideoEncodedFrame(trackType, frame) + +pushExternalVideoEncodedFrame:frame: + +pushExternalVideoEncodedFrame(trackType, frame) + +## **5\. 视频帧数据回调** + +### 5.1 功能说明 + +视频帧回调允许开发者在视频管线的不同位置获取原始帧数据，用于视频分析、自定义处理、录制等场景。支持只读和读写两种模式，读写模式下可修改帧数据并写回 SDK。 + +### 5.2 支持的数据源位置 + +**数据源** + +**枚举值** + +**说明** + +Captured + +0 + +采集后的视频数据（前处理前） + +PreEncode + +1 + +编码前的视频数据（前处理后） + +Remote + +2 + +远端解码后、渲染前的视频数据 + +### 5.3 回调配置参数 + +**参数** + +**类型** + +**默认值** + +**说明** + +format + +AoqVideoPixelFormat + +I420(1) + +期望回调的像素格式 + +alignment + +AoqVideoObserverAlignment + +Default(0) + +宽度对齐策略 + +mode + +AoqVideoObserverMode + +ReadOnly(0) + +只读(0)/读写(1) 模式 + +mirrorApplied + +bool + +false + +是否对回调数据应用镜像 + +### 5.4 宽度对齐枚举 + +**枚举值** + +**数值** + +**说明** + +AoqVideoObserverAlignmentDefault + +0 + +默认对齐 + +AoqVideoObserverAlignmentEven + +1 + +2 字节对齐 + +AoqVideoObserverAlignment4 + +2 + +4 字节对齐 + +AoqVideoObserverAlignment8 + +3 + +8 字节对齐 + +AoqVideoObserverAlignment16 + +4 + +16 字节对齐 + +### 5.5 使用步骤 + +1. **注册观察者**：调用 `setVideoFrameObserver` 设置视频帧回调监听器 + +2. **启用数据源**：调用 `enableVideoFrameObserver` 选择需要监听的数据源位置，开启回调 + +3. **处理回调数据**：在回调函数中获取帧数据（仅回调期间有效，异步使用需自行拷贝） + + +### 5.6 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +注册观察者 + +setVideoFrameObserver(listener) + +setVideoFrameObserver: + +setVideoFrameObserver(observer) + +启用回调 + +enableVideoFrameObserver(enabled, source, config) + +enableVideoFrameObserver:videoSource:config: + +enableVideoFrameObserver(enabled, source, config) + +### 5.7 回调方法 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +采集后数据 + +onCapturedVideoFrame(frame) + +onCapturedVideoFrame: + +onCapturedVideoFrame(frame) + +编码前数据 + +onPreEncodeVideoFrame(trackType, frame) + +onPreEncodeVideoFrame:frame: + +onPreEncodeVideoFrame(trackType, frame) + +远端数据 + +onRemoteVideoFrame(trackType, frame) + +onRemoteVideoFrame:frame: + +onRemoteVideoFrame(trackType, frame) + +**说明** + +回调方法返回 `true`/`YES` 表示数据已修改、需写回 SDK（仅 ReadWrite 模式且 I420 格式时生效）。 + +## **7\. 媒体流发送控制** + +### 7.1 功能说明 + +控制本地媒体流的发送开关，通过 `trackType` 路由到不同轨道（Audio/Video/Screen）。停用发送后，采集和编码继续运行，但数据不会发送到远端。 + +### 7.2 API 对照 + +**功能** + +**Android** + +**iOS** + +**Ohos** + +控制流发送 + +enableSendMediaStream(trackType, enable) + +enableSendMediaStream:enable: + +enableSendMediaStream(trackType, enable) + +### 7.3 轨道类型枚举 + +**枚举值** + +**数值** + +**说明** + +AoqTrackTypeAudio + +0 + +音频轨道 + +AoqTrackTypeVideo + +1 + +视频轨道 + +AoqTrackTypeData + +2 + +数据轨道 + +## **8\. 视频设备状态监控** + +### 8.1 功能说明 + +SDK 自动监测视频采集设备（摄像头）的状态变化，并通过 `onVideoDeviceStateChanged` 回调通知应用层。 + +### 8.2 设备状态码 + +**状态码** + +**值** + +**说明** + +AoqVideoDeviceNone + +0 + +初始状态 + +AoqVideoDeviceCaptureStarting + +1 + +采集启动中 + +AoqVideoDeviceCaptureStarted + +2 + +采集已启动 + +AoqVideoDeviceCaptureStopping + +3 + +采集停止中 + +AoqVideoDeviceCaptureStopped + +4 + +采集已停止 + +AoqVideoDeviceCaptureFail + +5 + +采集失败 + +### 8.3 回调对照 + +**回调** + +**Android** + +**iOS** + +**Ohos** + +设备状态变化 + +onVideoDeviceStateChanged(state) + +onVideoDeviceStateChanged: + +onVideoDeviceStateChanged(state) + +## **9\. 视频错误码与警告码** + +### 9.1 视频错误码 + +**错误码** + +**值** + +**说明** + +AoqErrorCodeVideo + +200 + +通用视频错误 + +VideoExternalBufferFull + +210 + +视频外部缓冲区已满 + +VideoDevice + +220 + +视频设备通用错误 + +CameraOpenFail + +221 + +摄像头打开失败 + +CameraAuthFailed + +222 + +摄像头权限被拒绝 + +CameraOccupied + +223 + +摄像头被占用 + +CameraRunningError + +224 + +摄像头运行错误 + +VideoCodec + +230 + +视频编解码通用错误 + +EncoderInitFail + +231 + +编码器初始化失败 + +VideoRender + +240 + +视频渲染通用错误 + +RenderCreateFail + +241 + +渲染器创建失败 + +RenderDrawError + +242 + +渲染绘制错误 + +Screen + +300 + +屏幕共享通用错误 + +**说明** + +Android 额外错误码：ScreenPermissionDenied(310) 屏幕共享权限被拒绝、ScreenForegroundServiceFailed(311) 前台服务启动失败。 + +### 9.2 视频警告码 + +**警告码** + +**值** + +**说明** + +AoqWCVideo + +200 + +通用视频警告 + +CameraEnumerateError + +201 + +摄像头枚举错误 + +EncoderSwitched + +202 + +编码器切换警告 + +RenderDowngrade + +203 + +渲染降级警告 + +## **附录：完整视频 API 方法列表** + +**分类** + +**方法名** + +**说明** + +采集控制 + +startVideoCapture + +打开视频采集设备 + +采集控制 + +stopVideoCapture + +关闭视频采集设备 + +采集控制 + +switchCamera + +切换前后置摄像头 + +渲染控制 + +setLocalView + +设置本地预览窗口 + +渲染控制 + +setRemoteView + +设置远端渲染窗口 + +编解码 + +setVideoEncoderConfig + +设置视频编码参数 + +外部输入 + +pushExternalVideoCapturedFrame + +推送原始视频帧 + +外部输入 + +pushExternalVideoEncodedFrame + +推送已编码视频帧 + +屏幕共享 + +startScreenCapture + +启动屏幕采集 + +屏幕共享 + +stopScreenCapture + +停止屏幕采集 + +流控制 + +enableSendMediaStream + +控制媒体流发送 + +帧回调 + +setVideoFrameObserver + +注册视频帧观察者 + +帧回调 + +enableVideoFrameObserver + +启用/禁用视频帧回调 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md new file mode 100644 index 00000000..a291cc23 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md @@ -0,0 +1,520 @@ +# 通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话 + +本文档说明如何在 Android、iOS、HarmonyOS 平台接入 AOQ Client SDK，实现 AOQ+qwen3.5-omni-plus-realtime 音视频通话功能。 + +## **SDK 获取** + +AOQ Client SDK 及音频 Opus 插件请参见[SDK下载](https://help.aliyun.com/zh/model-studio/realtime-sdk-download)。Opus 编码以独立插件形式提供，请根据您的场景按需引入。 + +## **SDK 导入** + +请根据不同平台将核心 SDK 产物导入工程依赖目录，并在工程配置中声明相关权限。 + +### **Android** + +将 `AoqClientSdk-release.aar` 放入工程 `app/libs/` 目录，将 `libPluginOpus.so` 按 ABI 放入 `app/libs/armeabi-v7a/` 和 `app/libs/arm64-v8a/`，并在 `app/build.gradle` 中： + +``` +android { + defaultConfig { + minSdk 21 + ndk { abiFilters 'armeabi-v7a', 'arm64-v8a' } + } + sourceSets { main { jniLibs.srcDirs = ['libs'] } } + packagingOptions { + // 避免与宿主工程的同名 so 冲突 + pickFirsts += ['lib/*/*.so'] + } +} + +dependencies { + implementation fileTree(dir: 'libs', include: ['*.aar']) +} +``` + +在 `AndroidManifest.xml` 声明权限： + +``` + + + + + +``` + +其中 `RECORD_AUDIO` 和 `CAMERA` 为运行时权限，应用需在运行时调用 Android `ActivityCompat.requestPermissions()` 方法，主动向 Android 系统申请用户授权。 + +### **iOS（framework）** + +1. 将 `AoqClientSdk.framework` 与 `PluginOpus.framework` 拖入 Xcode 工程，在 Target > General > Frameworks, Libraries, and Embedded Content 中选择 **Embed & Sign**。 + +2. 权限声明：在 Xcode 中选中您的 Target > Info > Custom iOS Target Properties，添加以下两项权限用途描述： + + **Key** + + **Value** + + `NSMicrophoneUsageDescription` + + 用于实时语音通话 + + `NSCameraUsageDescription` + + 用于实时视频通话 + +3. Swift 工程：`import AoqClientSdk`；Objective-C 工程：`#import `。 + + +### **HarmonyOS（har）** + +1. 将 `aoq-client-sdk.har` 放入工程 `libs/` 目录，将 `libPluginOpus.so` 按 ABI 放入 `entry/libs/armeabi-v7a/` 和 `entry/libs/arm64-v8a/`；并在 `entry/oh-package.json5` 中声明。 + +2. 在 `entry/src/main/module.json5` 添加权限： + + ``` + "requestPermissions": [ + { "name": "ohos.permission.INTERNET" }, + { "name": "ohos.permission.MICROPHONE", + "reason": "$string:perm_mic_reason", + "usedScene": { "abilities": ["EntryAbility"], "when": "inuse" } }, + { "name": "ohos.permission.CAMERA", + "reason": "$string:perm_camera_reason", + "usedScene": { "abilities": ["EntryAbility"], "when": "inuse" } } + ] + ``` + +3. 在 `EntryAbility` 中通过 `abilityAccessCtrl.createAtManager().requestPermissionsFromUser` 触发运行时授权。 + + +## **AppServer获取Token** + +请按照[Token鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication)的 AOQ 章节搭建获取 Token 的 AppServer。每次通话前，客户端需要向业务侧 AppServer 请求一次 Token。 + +## **实现 AI 音视频通话** + +![111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0075914871/p1088080.svg) + +### **创建引擎并设置回调** + +调用 `createEngine` 接口创建 `AoqClientEngine` 实例。 + +**iOS：** + +``` +let config = AoqCreateConfig() +config.workDir = workDir +engine = AoqClientEngine.createEngine(config, delegate: self) +``` + +实现 `AoqEngineDelegate` 协议监听 `onConnectionStatusChange`、`onDataMsg`、`onError` 等回调。 + +**Android：** + +``` +AoqCreateConfig config = new AoqCreateConfig(); +config.workDir = appCtx.getFilesDir().getAbsolutePath(); +engine = AoqClientEngine.createEngine(appCtx, config, this); +``` + +**HarmonyOS：** + +``` +const config: AoqCreateConfig = { workDir: context.filesDir, extras: '' }; +engine = AoqClientEngine.createEngine(config, this, context); +``` + +### **启动音视频采集与播放** + +调用 `startAudioCapture` 与 `startAudioPlayer` 启动本地音频采集与播放；调用 `startVideoCapture` 启动摄像头，并通过 `setLocalView` 将 SDK 渲染目标绑定到业务侧的预览控件。 + +**iOS：** + +``` +// 音频采集 +let capCfg = AoqAudioCaptureConfig() +capCfg.channel = 1; capCfg.isExternal = false +engine.startAudioCapture(capCfg) + +// 音频播放 +let playCfg = AoqAudioPlaybackConfig() +playCfg.channel = 1; playCfg.isExternal = false +engine.startAudioPlayer(playCfg) + +// 视频采集 +let vidCfg = AoqVideoCaptureConfig() +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15 +engine.startVideoCapture(vidCfg) + +// 为本地预览画面设置渲染视图 +let canvas = AoqVideoCanvas() +canvas.view = localPreview +canvas.renderMode = .crop +engine.setLocalView(.video, canvas: canvas) +``` + +**Android：** + +``` +// 音频采集 +AoqAudioCaptureConfig capCfg = new AoqAudioCaptureConfig(); +capCfg.channel = 1; capCfg.isExternal = false; +engine.startAudioCapture(capCfg); + +// 音频播放 +AoqAudioPlaybackConfig playCfg = new AoqAudioPlaybackConfig(); +playCfg.channel = 1; playCfg.isExternal = false; +engine.startAudioPlayer(playCfg); + +// 视频采集 +AoqVideoCaptureConfig vidCfg = new AoqVideoCaptureConfig(); +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15; +engine.startVideoCapture(vidCfg); + +// 为本地预览画面设置渲染视图 +AoqVideoCanvas canvas = new AoqVideoCanvas(); +canvas.view = localPreview; +canvas.renderMode = AoqRenderMode.AoqRenderModeCrop; +engine.setLocalView(AoqTrackType.AoqTrackTypeVideo, canvas); +``` + +**HarmonyOS：** + +``` +// 音频采集 +const capCfg: AoqAudioCaptureConfig = { channel: 1, isExternal: false }; +engine.startAudioCapture(capCfg); + +// 音频播放 +const playCfg: AoqAudioPlaybackConfig = { channel: 1, isExternal: false }; +engine.startAudioPlayer(playCfg); + +// 视频采集 +const vidCfg: AoqVideoCaptureConfig = { width: 720, height: 1280, fps: 15, isExternal: false }; +engine.startVideoCapture(vidCfg); + +// 为本地预览画面设置渲染视图 +const canvas: AoqVideoCanvas = { view: localCtrl, renderMode: AoqRenderMode.AoqRenderModeCrop }; +engine.setLocalView(AoqTrackType.AoqTrackTypeVideo, canvas); +``` + +### **获取连接凭证** + +由业务 AppServer 代理百炼请求，参见[Token鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication)。 + +### **设置编解码及建立连接** + +设置编解码参数后调用 `connect`。 + +注意：qwen3.5-omni-plus-realtime 要求客户端在收到服务端的 `session.updated` 之后才能开始发送媒体数据。为避免 `connect` 建联成功到 `session.updated` 到达之间的空档期误推媒体，在 `connect` 之前对上行音频与视频轨道分别调用 `enableSendMediaStream(trackType, false)`，将上行推流暂时关闭。WebSocket事件说明详见[客户端事件](https://help.aliyun.com/zh/model-studio/client-events)。 + +**iOS：** + +``` +// 音频编解码配置 +let encCfg = AoqAudioCodecConfig() +encCfg.codecType = .audioPCM; encCfg.sampleRate = 16000; encCfg.channel = 1 +engine.setAudioEncoderConfig(encCfg) +engine.setAudioDecoderConfig(encCfg) + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(.audio, enable: false) +engine.enableSendMediaStream(.video, enable: false) + +// 建立连接 +let conn = AoqConnectConfig() +conn.token = token; conn.sid = sid; conn.certFingerprint = cert +conn.relayEndpoints = endpoints; conn.workspaceIdHash = workspaceIdHash + +let aTrack = AoqTrackParam(); aTrack.trackType = .audio +let vTrack = AoqTrackParam(); vTrack.trackType = .video +let dTrack = AoqTrackParam(); dTrack.trackType = .data +conn.publishTracks = [aTrack, vTrack, dTrack] +conn.subscribeTracks = [aTrack, dTrack] +engine.connect(conn) +``` + +**Android：** + +``` +// 音频编解码配置 +AoqAudioCodecConfig encCfg = new AoqAudioCodecConfig(); +encCfg.codecType = AoqEncoderType.AoqEncoderTypeAudioPCM; +encCfg.sampleRate = 16000; encCfg.channel = 1; +engine.setAudioEncoderConfig(encCfg); +engine.setAudioDecoderConfig(encCfg); + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, false); +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, false); + +// 建立连接 +AoqConnectConfig conn = new AoqConnectConfig(); +conn.token = token; conn.sid = sid; conn.certFingerprint = cert; +conn.relayEndpoints.addAll(endpoints); conn.workspaceIdHash = workspaceIdHash; + +AoqTrackParam aTrack = new AoqTrackParam(); aTrack.trackType = AoqTrackType.AoqTrackTypeAudio; +AoqTrackParam vTrack = new AoqTrackParam(); vTrack.trackType = AoqTrackType.AoqTrackTypeVideo; +AoqTrackParam dTrack = new AoqTrackParam(); dTrack.trackType = AoqTrackType.AoqTrackTypeData; +conn.publishTracks.add(aTrack); +conn.publishTracks.add(vTrack); +conn.publishTracks.add(dTrack); +conn.subscribeTracks.add(aTrack); +conn.subscribeTracks.add(dTrack); +engine.connect(conn); +``` + +**HarmonyOS：** + +``` +// 音频编解码配置 +const encCfg: AoqAudioCodecConfig = { + codecType: AoqEncoderType.AoqEncoderTypeAudioPCM, + sampleRate: 16000, channel: 1 +}; +engine.setAudioEncoderConfig(encCfg); +engine.setAudioDecoderConfig(encCfg); + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, false); +engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, false); + +// 建立连接 +const conn: AoqConnectConfig = { + token, sid, certFingerprint: cert, + relayEndpoints: endpoints, + workspaceIdHash, + publishTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeVideo }, + { trackType: AoqTrackType.AoqTrackTypeData } + ], + subscribeTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ] +}; +engine.connect(conn); +``` + +**重要**：AOQ SDK 在建联后会默认发送媒体数据，此示例演示了连接模型时关闭媒体发送的能力。 + +### **配置 AI 会话** + +在 `onConnectionStatusChange(Connected)` 回调中通过 `sendDataMsg` 发送 `session.update` 消息（业务自定义 JSON，包含 modalities、voice、instructions、turn\_detection 等会话参数），完成会话握手，WebSocket事件说明详见[客户端事件](https://help.aliyun.com/zh/model-studio/client-events)。 + +**iOS：** + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + if status == .connected { sendSessionUpdate() } +} + +private func sendSessionUpdate() { + let json = """ + { + // 该事件的id，由客户端生成 + "event_id": "event_ToPZqeobitzUJnt3QqtWg", + // 事件类型，固定为session.update + "type": "session.update", + // 会话配置 + "session": { + // 输出模态，支持设置为["text"]（仅输出文本）或["text","audio"]（输出文本与音频）。 + "modalities": [ + "text", + "audio" + ], + // 输出音频的音色 + "voice": "Ethan", + // 输入音频格式，当前仅支持设置为pcm。输入音频为16 kHz采样率的PCM音频流。 + "input_audio_format": "pcm", + // 输出音频格式，当前仅支持设置为pcm。输出音频为24 kHz采样率的PCM音频流。 + "output_audio_format": "pcm", + // 系统消息，用于设定模型的目标或角色。 + "instructions": "你是某五星级酒店的AI客服专员，请准确且友好地解答客户关于房型、设施、价格、预订政策的咨询。请始终以专业和乐于助人的态度回应，杜绝提供未经证实或超出酒店服务范围的信息。", + // 是否开启语音活动检测。若需启用，需传入一个配置对象，服务端将据此自动检测语音起止。 + // 设置为null表示由客户端决定何时发起模型响应。 + "turn_detection": { + // VAD类型，取值为server_vad或semantic_vad。使用qwen3.5-omni-realtime模型时推荐设为semantic_vad。 + "type": "semantic_vad", + // VAD检测阈值。建议在嘈杂的环境中增加，在安静的环境中降低。 + "threshold": 0.5, + // 检测语音停止的静音持续时间，超过此值后会触发模型响应 + "silence_duration_ms": 800 + } + } + } + """ + let msg = AoqDataMsg() + msg.data = json.data(using: .utf8)! + engine.send(msg) +} +``` + +**Android：** + +``` +@Override +public void onConnectionStatusChange(AoqConnectionStatus status) { + if (status == AoqConnectionStatus.AoqConnectionStatusConnected) { + sendSessionUpdate(); + } +} + +private void sendSessionUpdate() { + String sessionUpdateJson = /* 与 Swift 示例中相同的 session.update JSON */; + AoqDataMsg msg = new AoqDataMsg(); + msg.data = sessionUpdateJson.getBytes(StandardCharsets.UTF_8); + engine.sendDataMsg(msg); +} +``` + +**HarmonyOS：** + +``` +onConnectionStatusChange(status: AoqConnectionStatus): void { + if (status === AoqConnectionStatus.AoqConnectionStatusConnected) { + this.sendSessionUpdate(); + } +} + +private sendSessionUpdate(): void { + const sessionUpdateJson = /* 与 Swift 示例中相同的 session.update JSON */; + const msg: AoqDataMsg = { data: new TextEncoder().encode(sessionUpdateJson).buffer }; + this.engine.sendDataMsg(msg); +} +``` + +### **收到 session.updated 后开启媒体发送** + +在 `onDataMsg` 回调中解析下行消息，收到模型回复 `session.updated` 的时候，对上一步禁推的每个轨道类型调用 `enableSendMediaStream(trackType, true)` 放开推流。下面为代码示例，WebSocket事件说明详见[服务端事件](https://help.aliyun.com/zh/model-studio/server-events)。 + +**iOS：** + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + if type == "session.updated" { + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +**Android：** + +``` +@Override +public void onDataMsg(AoqDataMsg msg) { + if (msg == null || msg.data == null) return; + try { + JSONObject obj = new JSONObject(new String(msg.data, StandardCharsets.UTF_8)); + if ("session.updated".equals(obj.optString("type"))) { + engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, true); + engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, true); + } + } catch (JSONException ignored) {} +} +``` + +**HarmonyOS：** + +``` +onDataMsg(msg: AoqDataMsg): void { + if (!msg?.data) return; + try { + const text = new TextDecoder('utf-8').decode(new Uint8Array(msg.data)); + const obj = JSON.parse(text) as { type?: string }; + if (obj.type === 'session.updated') { + this.engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeAudio, true); + this.engine.enableSendMediaStream(AoqTrackType.AoqTrackTypeVideo, true); + } + } catch (_) { /* 非 JSON，忽略 */ } +} +``` + +**重要** + +1. 模型必须在收到 `session.updated` 后才开启媒体流发送，否则 AI 侧可能还未准备好接收数据。 + +2. 建连时添加的音频轨道和视频轨道（即 AOQ 媒体通道）会自动将数据传输到服务端。 + + 1. 音频：通过音频轨道直接传输，无需发送 `input_audio_buffer.append` 事件。 + + 2. 视频：通过视频轨道发送画面帧，无需发送 `input_image_buffer.append` 事件。 + + +### **断开连接与销毁引擎** + +``` +engine.disconnect() +AoqClientEngine.destroy() +``` + +## **典型场景** + +### **打断（Barge-in）** + +- SDK 与百炼深度融合，支持百炼模型的打断消息会在新一轮对话开始时打断上一轮次。 + +- SDK 提供**本地播放器打断**接口 `interruptAudioPlayer`，当用户主动需要停止时可以调用打断 API 实现此功能。 + + +``` +// iOS +engine.interruptAudioPlayer(.audio, fadeMs: 100) +``` + +### **静音 / 取消静音** + +静音后 SDK 仍在采集音频，但只推送静音帧，`session` 不会中断。 + +``` +engine.muteAudioCapture(true); // 静音麦克风（采集仍在跑，但只送静音帧） +engine.muteAudioCapture(false); // 恢复 +``` + +### **切换前后摄像头** + +``` +// 传入期望切换到的方向枚举即可 +engine.switchCamera(AoqCameraDirection.AoqCameraDirectionFront); +engine.switchCamera(AoqCameraDirection.AoqCameraDirectionBack); +``` + +### **通话字幕与ASR结果显示** + +服务端通过下行数据消息推送 ASR 结果与 AI 文本回复。业务侧在 `onDataMsg` 回调中根据 `type` 字段分流即可。WebSocket事件说明详见[服务端事件](https://help.aliyun.com/zh/model-studio/server-events)。 + +## **注意事项** + +1. **单例语义**：`createEngine` 是单例，重复调用返回同一实例；`destroy` 后才能重新创建。多页面共用建议在 Application/Ability 级管理引擎生命周期。 + +2. **本地预览 View 类型**： + + - Android：`SurfaceView` 或 `TextureView`；其它类型不支持。 + + - iOS：任意 `UIView` 子类。 + + - HarmonyOS：请参考 SDK 文档。 + +3. **音频路由变化**：耳机插拔、蓝牙连接等会触发 `onAudioDeviceRouteChanged`，业务侧通常无需处理；如果 UI 上显示"扬声器/听筒"开关，需要根据该回调同步状态。 + +4. **后台续传**：如需通话切到后台后继续传音频，`Info.plist` 必须开启 `UIBackgroundModes = audio`，并在前台时正确激活 `AVAudioSession`（SDK 会处理大部分情况，业务侧用 `setAudioSessionRestriction:` 可精细控制是否让 SDK 接管）。 + + +## **Demo 示例下载** + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5762814871/p1087917.png) + +示例源码下载： + +**iOS：**[aoqdemo.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/grpnos/aoqdemo.zip) + +## **相关文档** + +- AOQ Client SDK 详细 API：[AOQ SDK简介](https://help.aliyun.com/zh/model-studio/realtime-api-aoq-sdk-desc/) + +- qwen3.5-omni-plus-realtime 模型客户端事件：[客户端事件](https://help.aliyun.com/zh/model-studio/client-events) + +- qwen3.5-omni-plus-realtime 模型服务端事件：[服务端事件](https://help.aliyun.com/zh/model-studio/server-events) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md new file mode 100644 index 00000000..e743e6b6 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md @@ -0,0 +1,512 @@ +# 通过WebRTC使用多模态交互套件实现实时通话 + +本文档说明如何在浏览器端通过 WebRTC + JavaScript 接入通义多模态交互套件（multimodal-dialog），实现与多模态 AI 应用的实时音视频交互。 + +**说明** + +多模态交互套件面向 **AI/AR 眼镜、学习机、智能机器人**等硬件场景，提供可视化应用配置、预置 Agent/插件、音色管理等完整业务能力。WebRTC 模式下音频通过 UDP 直接传输，内置回声消除和降噪，适合浏览器端低延迟交互场景。 + +## **前提条件及注意事项** + +1. 已在百炼控制台完成以下准备： + + - 已[配置 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并将其[设置到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + + - 创建多模态交互应用，获取 **Workspace ID** 和 **App ID**（详情请参见[应用创建](https://help.aliyun.com/zh/model-studio/multimodal-app-creation)）。 + + - 在应用中完成模型、音色、提示词、Agent/插件等配置（详情请参见[应用配置](https://help.aliyun.com/zh/model-studio/multimodal-app-configuration)）。 + +2. 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +3. 浏览器需要麦克风权限；如需视频交互，还需摄像头权限。 + +4. 浏览器无法直接向服务端发起 SDP 交换请求（受 CORS 限制），Demo 中需通过终端执行 curl 命令完成连接建立；正式产品中由业务后端代理时不存在此限制。 + + +## **实现实时通话** + +以下时序图展示了整个 WebRTC 实时通话的完整流程： + +WebRTC 多模态交互套件实时通话流程时序图 + +![1111](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6265914871/p1088078.svg) + +### **创建 RTCPeerConnection** + +调用浏览器原生 `RTCPeerConnection` 构造函数创建连接实例。服务端采用 ICE-lite 模式直连，无需配置 ICE 服务器。 + +``` +pc = new RTCPeerConnection(); +``` + +同时注册关键回调： + +``` +pc.onconnectionstatechange = () => { + if (pc.connectionState === 'connected') { + // 连接成功 + } else if (['failed', 'closed', 'disconnected'].includes(pc.connectionState)) { + // 连接断开，清理资源 + endSession(); + } +}; + +pc.ontrack = (e) => { + // 将远端音频流绑定到 audio 元素播放 + const remoteAudio = document.createElement('audio'); + remoteAudio.autoplay = true; + remoteAudio.srcObject = e.streams[0]; + document.body.appendChild(remoteAudio); +}; +``` + +### **获取本地媒体流** + +通过 `navigator.mediaDevices.getUserMedia` 获取麦克风权限（必须），以及摄像头权限（可选）。 + +**纯音频模式：** + +``` +const localStream = await navigator.mediaDevices.getUserMedia({ audio: true }); +``` + +**音视频模式（需要 AI 视觉理解时）：** + +``` +const localStream = await navigator.mediaDevices.getUserMedia({ + audio: true, + video: { + facingMode: { ideal: 'environment' }, // 后置摄像头 + frameRate: { ideal: 30, max: 30 }, + width: { ideal: 640 }, + height: { ideal: 480 } + } +}); +``` + +**说明** + +视频帧率根据场景调整。需要 AI 实时理解画面的场景（如物体识别、场景描述）建议 15-30 fps。 + +### **添加媒体轨道到 PeerConnection** + +将本地音频轨道添加到 PeerConnection。如果开启了视频，视频轨道一并添加。 + +``` +localStream.getTracks().forEach(track => pc.addTrack(track, localStream)); +``` + +**维持视频发送质量（可选）：** + +在弱网环境下，浏览器可能自动降低视频分辨率。可通过设置 sender 参数尽量维持： + +``` +const sender = pc.getSenders().find(s => s.track && s.track.kind === 'video'); +if (sender) { + const params = sender.getParameters(); + if (!params.encodings || params.encodings.length === 0) params.encodings = [{}]; + params.encodings[0].scaleResolutionDownBy = 1.0; + params.encodings[0].maxBitrate = 2500000; // 2.5 Mbps + params.encodings[0].maxFramerate = 30; + params.degradationPreference = 'maintain-resolution'; + await sender.setParameters(params); +} +``` + +### **创建 DataChannel** + +创建名为 `oai-events` 的 DataChannel，用于与服务端交换控制消息（run-task、事件通知等）。 + +``` +const dc = pc.createDataChannel('oai-events'); + +dc.onopen = () => { + console.log('DataChannel open'); + // DataChannel 就绪后发送 run-task + sendStartMessage(dc); +}; + +dc.onmessage = (e) => { + const evt = JSON.parse(e.data); + handleServerEvent(evt, dc); +}; +``` + +同时监听 `pc.ondatachannel` 以处理服务端主动创建的 DataChannel： + +``` +pc.ondatachannel = (event) => { + const ch = event.channel; + if (ch.label === 'txt' || ch.label === 'oai-events') { + ch.onopen = () => sendStartMessage(ch); + ch.onmessage = (e) => handleServerEvent(JSON.parse(e.data), ch); + } +}; +``` + +### **生成 Offer SDP** + +调用 `createOffer` 并 `setLocalDescription`，浏览器生成包含本地媒体能力描述的 Offer SDP。服务端采用 ICE-lite 模式，客户端无需等待 ICE candidates 收集完成，`setLocalDescription` 后即可发送。 + +``` +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); + +// offer.sdp 即为待发送的 Offer SDP 字符串 +``` + +### **交换 SDP（HTTP POST）** + +将 Offer SDP 通过 HTTP POST 发送到服务端 WebRTC 端点，服务端返回 Answer SDP。 + +**Endpoint 格式：**`{workspace_id}.{region}.maas.aliyuncs.com`，其中 `workspace_id` 为百炼工作空间 ID（如 `llm-xxxxxxxxxx`），`region` 为部署区域（如 `cn-beijing`）。创建工作空间后即可在百炼控制台获取。 + +**请求配置：** + +**配置项** + +**说明** + +请求地址 + +`POST https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog` + +Content-Type + +`application/sdp` + +请求头 + +`Authorization: Bearer {DASHSCOPE_API_KEY}` + +请求体 + +客户端生成的 Offer SDP 字符串 + +响应 + +成功：HTTP 200，返回服务端 Answer SDP 字符串 + +``` +const API_KEY = 'your-api-key'; // 百炼控制台获取 +const WORKSPACE_ID = '{workspace-id}'; // 百炼控制台获取 +const REGION = 'cn-beijing'; +const SIGNALING_URL = `https://${WORKSPACE_ID}.${REGION}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog`; + +const resp = await fetch(SIGNALING_URL, { + method: 'POST', + headers: { + 'Content-Type': 'application/sdp', + 'Authorization': `Bearer ${API_KEY}`, + }, + body: offer.sdp, +}); + +if (!resp.ok) throw new Error('SDP 交换失败: ' + resp.status); +const answerSdp = await resp.text(); +``` + +**说明** + +由于浏览器端受 CORS 限制无法直接请求服务端，Demo 中需用户手动在终端执行 curl 命令。正式产品中应通过业务后端代理此请求。 + +**curl 命令格式：** + +``` +curl -X POST 'https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog' \ + -H 'Content-Type: application/sdp' \ + -H 'Authorization: Bearer $DASHSCOPE_API_KEY' \ + --data-binary '' +``` + +### **设置 Answer SDP 建立连接** + +将从服务端获取的 Answer SDP 设置为远端描述，WebRTC 连接随即建立。 + +``` +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +// 连接建立完成，pc.connectionState 将变为 'connected' +``` + +### **发送 run-task 启动会话** + +WebRTC 连接建立、DataChannel 就绪后，客户端发送 **run-task** 消息启动多模态会话。 + +``` +let currentTaskId = null; + +function sendStartMessage(channel) { + currentTaskId = generateTaskId(); + const msg = { + payload: { + input: { + workspace_id: '{workspace-id}', + app_id: '{app-id}', + directive: 'Start' + }, + task_group: 'aigc', + task: 'multimodal-generation', + function: 'generation', + model: 'multimodal-dialog', + parameters: { + client_info: { + user_id: '{user-id}', + device: { uuid: '{device-uuid}' }, + network: { ip: '{client-ip}' } + }, + upstream: { + mode: 'duplex', + sample_rate: '16000', + type: 'AudioAndVideo' // 'Audio' 或 'AudioAndVideo' + }, + dialog_attributes: { + vocabulary_id: '{vocabulary-id}' // 可选 + }, + downstream: { + voice: 'longanhuan', + sample_rate: 24000, + audio_format: 'pcm' + } + } + }, + header: { + streaming: 'duplex', + action: 'run-task', + task_id: currentTaskId + } + }; + channel.send(JSON.stringify(msg)); +} + +function generateTaskId() { + if (window.crypto && typeof window.crypto.randomUUID === 'function') { + return window.crypto.randomUUID().replace(/-/g, ''); + } + const r = () => Math.random().toString(16).slice(2); + return (Date.now().toString(16) + r() + r()).slice(0, 32); +} +``` + +**参数说明：** + +**参数路径** + +**类型** + +**说明** + +`payload.input.workspace_id` + +string + +百炼工作空间 ID，在控制台「应用管理」中获取 + +`payload.input.app_id` + +string + +多模态交互应用 ID，在控制台「应用管理」中获取 + +`payload.input.directive` + +string + +固定值 `Start` + +`payload.parameters.client_info.user_id` + +string + +业务系统中的用户标识，用于日志追踪 + +`payload.parameters.client_info.device.uuid` + +string + +设备唯一标识，用于设备维度的数据分析 + +`payload.parameters.upstream.mode` + +string + +交互模式：`duplex`（全双工）、`push2talk`（按住说话）、`tap2talk`（点击说话） + +`payload.parameters.upstream.type` + +string + +上行媒体类型：`Audio`（纯语音）或 `AudioAndVideo`（音视频） + +`payload.parameters.upstream.sample_rate` + +string + +上行音频采样率，通常 `16000` + +`payload.parameters.downstream.voice` + +string + +下行音色，可在百炼控制台音色列表中选取 + +`payload.parameters.downstream.sample_rate` + +number + +下行音频采样率，通常 `24000` + +`payload.parameters.dialog_attributes.vocabulary_id` + +string + +可选，热词表 ID，提升专有名词识别准确率 + +**说明** + +应用中的模型选择、提示词、Agent/插件、知识库等配置均在百炼控制台可视化完成，无需通过代码传入。run-task 只需指定 `workspace_id` 和 `app_id`，服务端会自动加载对应配置。 + +### **实时对话** + +run-task 发送成功后，进入实时对话状态： + +- **上行**：浏览器采集的音频/视频通过 RTP 协议自动发送到服务端 + +- **下行音频**：AI 语音回复通过 `ontrack` 回调接收并播放 + +- **下行事件**：通过 DataChannel 接收业务事件 + + +``` +function handleServerEvent(evt, channel) { + const type = evt.type || evt.header?.action; + + switch (type) { + case 'open_videochat': + // 服务端请求开启视频通道 + // 延迟数秒响应，确保视频处理通道就绪 + setTimeout(() => { + channel.send(JSON.stringify({ + payload: { + input: { text: '', type: 'prompt', directive: 'RequestToRespond' }, + parameters: { + biz_params: { + videos: [{ action: 'connect', type: 'voicechat_video_channel' }] + } + } + }, + header: { + streaming: 'duplex', + action: 'continue-task', + task_id: currentTaskId + } + })); + }, 3000); + break; + + default: + console.log('[服务端事件]', type, evt); + break; + } +} +``` + +**说明** + +**open\_videochat 机制**：当 `upstream.type` 设为 `AudioAndVideo` 时，客户端已通过 WebRTC 发送视频轨道。服务端在需要时（如 AI Agent 判断需要"看"画面）会推送 `open_videochat` 事件，客户端需回复 `continue-task` 确认视频通道建立。该机制允许按需开启视频处理，节省服务端资源。 + +**静音 / 取消静音：** + +``` +// 静音 +localStream.getAudioTracks().forEach(t => { t.enabled = false; }); +// 取消静音 +localStream.getAudioTracks().forEach(t => { t.enabled = true; }); +``` + +**开启/关闭视频：** + +``` +// 关闭视频 +localStream.getVideoTracks().forEach(t => { t.enabled = false; }); +// 开启视频 +localStream.getVideoTracks().forEach(t => { t.enabled = true; }); +``` + +### **结束会话与资源清理** + +通话结束后需要正确释放所有资源，避免内存泄漏和设备占用。 + +``` +function endSession() { + // 1. 关闭 DataChannel + if (dataChannel) { + dataChannel.close(); + dataChannel = null; + } + + // 2. 停止本地媒体流（释放麦克风/摄像头） + if (localStream) { + localStream.getTracks().forEach(t => t.stop()); + localStream = null; + } + + // 3. 关闭 PeerConnection + if (pc) { + pc.close(); + pc = null; + } + + // 4. 重置状态 + currentTaskId = null; +} +``` + +## **注意事项** + +1. **API Key 安全**：切勿将 API Key 硬编码在前端代码中。生产环境应通过后端服务代理 SDP 交换请求，API Key 仅存放在服务端。 + +2. **CORS 限制**：浏览器端无法直接调用百炼 API 进行 SDP 交换，正式产品中需要通过后端代理转发请求。 + +3. **HTTPS 要求**：`getUserMedia` 在非 localhost 环境下要求页面必须通过 HTTPS 提供服务。 + +4. **交互模式选择**：多模态套件支持三种交互模式——`duplex`（全双工，用户可随时打断）、`push2talk`（按住说话）、`tap2talk`（点击说话）。根据硬件形态选择合适的模式。 + +5. **视频帧率**：视频理解场景建议 15-30 fps；如果仅需偶尔拍照识别，可降低帧率节省带宽。 + +6. **浏览器兼容性**：推荐使用 Chrome 90+、Edge 90+、Firefox 85+、Safari 15+。 + +7. **单实例限制**：同一页面同时只应维护一个 `RTCPeerConnection` 实例，创建新会话前需先关闭旧连接。 + + +## **完整 Demo 示例下载** + +以下是一个完整的 HTML 页面 Demo，可直接在浏览器中运行体验多模态交互。由于浏览器 CORS 限制，SDP 交换通过 curl 命令手动完成。 + +[webrtc\_multimodel\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/ifirko/webrtc_multimodel_demo.html) + +**使用步骤：** + +1. 在浏览器中打开该文件。 + +2. 填写连接配置：Endpoint（格式为 `{workspace_id}.{region}.maas.aliyuncs.com`）、API Key、Workspace ID 和 App ID。 + +3. 如需视频交互，勾选“开启视频”。 + +4. 点击**开始会话**，允许浏览器访问麦克风（及摄像头）。 + +5. 如果浏览器能直接发起请求（无 CORS 限制），将自动完成连接；否则页面会显示 curl 命令，复制到终端执行后将返回的 Answer SDP 粘贴回页面即可。 + +6. 连接建立后，对着麦克风说话即可与多模态 AI 实时对话。 + +7. 通话结束后可点击“下载远端音频”保存 AI 回复的录音。 + + +## **相关文档** + +- [通义多模态交互开发套件产品概述](https://help.aliyun.com/zh/model-studio/multimodal-products-overview) + +- [多模态交互套件使用指南](https://help.aliyun.com/zh/model-studio/multimodal-guidelines/) + +- [多模态交互 SDK（Python/Java）GitHub 示例代码](https://github.com/aliyun/alibabacloud-bailian-speech-demo/tree/master/samples/conversation/multimodal_dialog) + +- [WebRTC API (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/WebRTC_API) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md new file mode 100644 index 00000000..b5d496b5 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md @@ -0,0 +1,388 @@ +# 通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话 + +本文档说明如何在浏览器端通过 WebRTC + JavaScript 接入百炼 Realtime API，实现与 qwen3.5-omni-plus-realtime 模型的实时音视频通话。 + +**说明** + +WebRTC 适合浏览器端、低延迟语音场景，音频通过 UDP 直接传输，内置回声消除和降噪。WebRTC 仅支持服务端 VAD 模式（`server_vad` 或 `semantic_vad`），不支持手动模式。 + +## **前提条件及注意事项** + +1. 已[配置 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并将其[设置到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +2. 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +3. 浏览器需要麦克风权限；如需视频通话，还需摄像头权限。 + +4. 浏览器无法直接向服务端发起 SDP 交换请求（受 CORS 限制），Demo 中通过终端执行 curl 命令完成连接建立；正式使用时由业务 AppServer 代理完成，无此限制。 + + +## **实现 AI 音视频通话** + +以下时序图展示了整个 WebRTC 音视频通话的完整流程： + +WebRTC 音视频通话流程时序图 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5665914871/p1088079.png) + +### **创建 RTCPeerConnection** + +调用浏览器原生 `RTCPeerConnection` 创建连接实例，无需配置 ICE 服务器（服务端会处理 NAT 穿透）。 + +``` +pc = new RTCPeerConnection({ iceServers: [ ] }); +``` + +注册关键回调： + +``` +// 连接状态监听 +pc.onconnectionstatechange = () => { + if (!pc) return; + if (pc.connectionState === 'connected') { + setStatus('已连接，请说话', 'connected'); + } else if (["failed", "closed", "disconnected"].includes(pc.connectionState)) { + endSession(true); + } +}; + +// 接收远端音频流并播放 + 启动录制 +pc.ontrack = async (e) => { + const stream = e.streams[0]; + ensureHiddenAudioEl(); + hiddenRemoteAudioEl.srcObject = stream; + try { await hiddenRemoteAudioEl.play(); } catch {} + startRecordingRemoteStream(stream); +}; +``` + +### **获取本地媒体流** + +通过一次 `getUserMedia` 调用获取所需的音频（必须）和视频（可选）。是否开启视频由用户勾选"开启视频"复选框决定。 + +``` +const wantVideo = !!sendVideoCheckbox.checked; + +const constraints = wantVideo + ? { + audio: true, + video: { + facingMode: { ideal: "user" }, + frameRate: { ideal: 30, max: 30 }, + width: { ideal: 640 }, + height: { ideal: 480 }, + } + } + : { audio: true }; + +localStream = await navigator.mediaDevices.getUserMedia(constraints); +``` + +**说明** + +音频和视频通过**同一次** `getUserMedia` 调用获取，而非分开请求。视频预览帧率为 30fps（本地流畅预览），发送帧率会通过 Canvas 降至 2fps。 + +### **添加媒体轨道到 PeerConnection** + +**添加音频轨道：** + +``` +localStream.getAudioTracks().forEach(t => { + pc.addTrack(t, localStream); + gatedAudioTracks.push(t); +}); +``` + +**添加视频轨道（可选，通过 Canvas 降帧至 2fps）：** + +Canvas 尺寸从摄像头实际分辨率动态获取，而非硬编码： + +``` +const sendFps = 2; +const settings = localStream.getVideoTracks()[0].getSettings(); +sendCanvas = document.createElement("canvas"); +sendCanvas.width = settings.width || 640; // 动态获取实际宽度 +sendCanvas.height = settings.height || 480; // 动态获取实际高度 +sendCanvasCtx = sendCanvas.getContext("2d", { alpha: false }); + +sendCanvasStream = sendCanvas.captureStream(sendFps); // 2fps +const lowFpsTrack = sendCanvasStream.getVideoTracks()[0]; +pc.addTrack(lowFpsTrack, sendCanvasStream); +gatedVideoTracks.push(lowFpsTrack); + +// requestAnimationFrame 循环：将摄像头画面绘制到 Canvas +const pump = () => { + if (!sendCanvasCtx || !sendCanvas) return; + try { sendCanvasCtx.drawImage(localVideo, 0, 0, sendCanvas.width, sendCanvas.height); } catch {} + sendRafId = requestAnimationFrame(pump); +}; +sendRafId = requestAnimationFrame(pump); +``` + +**媒体门控（关键）：** + +添加轨道后立即禁止发送，确保在收到 `session.created` 之前不推送媒体数据： + +``` +// 1. 禁用所有轨道的 enabled +gateMedia(false); // track.enabled = false + +// 2. 将 sender 的 track 替换为 null，彻底阻止发送 +audioSender = pc.getSenders().find(s => s.track?.kind === 'audio'); +videoSender = pc.getSenders().find(s => s.track?.kind === 'video'); +audioTrack = audioSender?.track; +videoTrack = videoSender?.track; +await audioSender?.replaceTrack(null); +await videoSender?.replaceTrack(videoTrack ? null : undefined); +``` + +**说明** + +等价于其他 SDK 中的 `enableSendMediaStream(false)`，必须在收到 `session.created` 后才恢复发送。 + +### **创建 DataChannel** + +创建名为 `oai-events` 的 DataChannel，用于与 AI 服务端交换会话控制事件。 + +``` +const dc = pc.createDataChannel('oai-events'); + +dc.onopen = () => console.log("DC open"); +dc.onmessage = (e) => { + handleDcMessage(e.data, dc); +}; + +// 同时监听服务端主动创建的 DataChannel +pc.ondatachannel = (event) => { + const ch = event.channel; + ch.onmessage = (e) => { + handleDcMessage(e.data, ch); + }; +}; +``` + +### **生成 Offer SDP** + +调用 `createOffer()` 并设置本地描述，等待 ICE 候选收集完成后获取完整的 Offer SDP。 + +``` +pc.onicegatheringstatechange = () => { + if (!pc) return; + if (pc.iceGatheringState === "complete" && pc.localDescription?.sdp) { + const sdp = pc.localDescription.sdp; + // ICE 收集完成，Offer SDP 可用 + // 自动生成 curl 命令供用户使用 + } +}; + +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); +``` + +**说明** + +必须等待 `iceGatheringState === "complete"` 后再使用 SDP，此时 SDP 中包含所有 ICE 候选信息。 + +### **交换 SDP（通过 curl 命令或业务 AppServer）** + +将 Offer SDP 发送到百炼服务端，获取 Answer SDP。Demo 中通过 curl 命令完成： + +``` +curl -X POST 'https://{endpoint}/api/v1/webrtc/realtime?model=qwen3.5-omni-plus-realtime' \ + -H 'Content-Type: application/sdp' \ + -H 'Authorization: Bearer $DASHSCOPE_API_KEY' \ + --data-binary '' +``` + +**说明** + +生产环境中，此步骤应由业务 AppServer 代理完成，避免前端暴露 API Key。`{endpoint}` 为 Realtime API 接入地址。 + +### **设置 Answer SDP 建立连接** + +将服务端返回的 Answer SDP 设置为远端描述，WebRTC 连接即开始建立。注意 SDP 格式需要规范化处理： + +``` +function normalizeSdpForSetRemote(sdp) { + sdp = String(sdp).trim().replace(/\r?\n/g, "\r\n"); + if (!sdp.endsWith("\r\n")) sdp += "\r\n"; + return sdp; +} + +const answerSdp = normalizeSdpForSetRemote(txt); +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +``` + +**说明** + +SDP 规范要求行尾为 `\r\n`，`normalizeSdpForSetRemote` 负责处理不同来源的换行符兼容问题。 + +### **配置 AI 会话（session.update）** + +连接建立后，服务端通过 DataChannel 发送 `session.created` 事件。收到后需： + +1. 解除媒体门控，恢复音视频发送 + +2. 发送 `session.update` 配置会话参数 + + +**解除门控并恢复媒体：** + +``` +function handleDcMessage(data, channel) { + let obj; + try { obj = JSON.parse(data); } catch (err) { return; } + + if (obj?.type === "session.created") { + // 解除门控：恢复 track.enabled + gateMedia(true); + // 恢复 sender 的实际 track + if (audioSender) audioSender.replaceTrack(audioTrack); + if (videoSender && videoTrack) videoSender.replaceTrack(videoTrack); + // 发送会话配置 + sendUpdate(channel); + } +} +``` + +**session.update 消息体：** + +``` +const update = { + event_id: `event_${Date.now()}`, + type: "session.update", + session: { + input_audio_format: "pcm", + input_audio_transcription: { model: "qwen3-asr-flash-realtime" }, + instructions: "You are a helpful assistant.", + modalities: ["text", "audio"], + output_audio_format: "pcm", + smooth_output: false, + turn_detection: { + prefix_padding_ms: 500, + silence_duration_ms: 800, + threshold: 0.5, + type: "server_vad", + }, + }, +}; +if (channel && channel.readyState === "open") channel.send(JSON.stringify(update)); +``` + +**说明** + +`turn_detection.type` 可设为 `server_vad`（基于音量检测）或 `semantic_vad`（基于语义检测）。WebRTC 模式不支持手动 VAD。 + +### **实时对话** + +连接建立后，音视频通过 RTP 实时传输。远端 AI 语音通过 `ontrack` 回调接收并播放，同时使用 MediaRecorder 录制以便下载。 + +**接收远端音频并录制：** + +``` +pc.ontrack = async (e) => { + const stream = e.streams[0]; + ensureHiddenAudioEl(); + hiddenRemoteAudioEl.srcObject = stream; + try { await hiddenRemoteAudioEl.play(); } catch {} + startRecordingRemoteStream(stream); // 启动录制 +}; + +function startRecordingRemoteStream(remoteStream) { + const audioTracks = remoteStream.getAudioTracks(); + if (!audioTracks.length) return; + const audioStream = new MediaStream(audioTracks); + + recordedChunks = [ ]; + + mediaRecorder = new MediaRecorder(audioStream, { mimeType: 'audio/webm' }); + mediaRecorder.ondataavailable = (e) => { + if (e.data && e.data.size > 0) recordedChunks.push(e.data); + }; + mediaRecorder.onstop = () => { + audioBlob = new Blob(recordedChunks, { type: 'audio/webm' }); + // 录制结束后可下载 + }; + mediaRecorder.start(); +} +``` + +**DataChannel 事件统一展示：** + +所有通过 DataChannel 收发的事件（包括 `session.created`、`response.audio_transcript.done` 等）统一通过事件面板展示，支持展开查看完整 JSON： + +``` +function pushEventFromDataChannel(eventObj) { + const ts = eventObj.timestamp || nowTs(); + events.unshift({ event: eventObj, timestamp: ts }); + renderEvents(); +} +``` + +### **结束会话与资源清理** + +结束通话时需依次清理所有资源，顺序很重要： + +``` +function endSession(silent = false) { + // 1. 停止 Canvas 降帧循环 + if (sendRafId) cancelAnimationFrame(sendRafId); + sendRafId = 0; + if (sendCanvasStream) sendCanvasStream.getTracks().forEach(t => t.stop()); + sendCanvasStream = null; sendCanvasCtx = null; sendCanvas = null; + + // 2. 停止录制 + try { if (mediaRecorder && mediaRecorder.state !== "inactive") mediaRecorder.stop(); } catch {} + mediaRecorder = null; + + // 3. 停止本地媒体流 + if (localStream) { + localStream.getTracks().forEach(t => t.stop()); + localStream = null; + } + + // 4. 关闭 PeerConnection + if (pc) { try { pc.close(); } catch {} pc = null; } + + // 5. 清理远端音频元素 + if (hiddenRemoteAudioEl) { + try { hiddenRemoteAudioEl.pause(); } catch {} + hiddenRemoteAudioEl.srcObject = null; + hiddenRemoteAudioEl.remove(); + hiddenRemoteAudioEl = null; + } +} +``` + +**说明** + +结束后可通过"下载远端音频"按钮下载 AI 回复的录音（WebM 格式）。 + +## **注意事项** + +1. **媒体门控必须在 session.created 后解除**：在服务端发送 `session.created` 之前推送媒体数据会被丢弃，必须通过 `replaceTrack(null)` 彻底阻断发送。 + +2. **视频降帧通过 Canvas 实现**：本地预览 30fps，发送至服务端仅 2fps，通过 `captureStream(2)` 控制，节省带宽。 + +3. **SDP 格式规范化**：设置 Answer SDP 前必须确保行尾为 `\r\n`，否则 `setRemoteDescription` 可能失败。 + +4. **视频为可选功能**：用户未勾选视频时，仅请求音频权限，不会触发摄像头授权弹窗。 + +5. **远端音频自动录制**：通过 MediaRecorder 录制 AI 回复的音频流，会话结束后可下载 WebM 格式文件。 + +6. **WebRTC 仅支持服务端 VAD**：不支持 `manual` 模式，可选 `server_vad`（音量检测）或 `semantic_vad`（语义检测）。 + + +## **完整 demo 下载** + +完整示例代码请下载：[webrtc\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/ychtmj/webrtc_demo.html)。 + +## **相关文档** + +- [WebRTC API (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/WebRTC_API) + +- [RTCPeerConnection (MDN)](https://developer.mozilla.org/zh-CN/docs/Web/API/RTCPeerConnection) + +- qwen3.5-omni-plus-realtime 模型客户端事件：[客户端事件](https://help.aliyun.com/zh/model-studio/client-events) + +- qwen3.5-omni-plus-realtime 模型服务端事件：[服务端事件](https://help.aliyun.com/zh/model-studio/server-events) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md new file mode 100644 index 00000000..217e42ed --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md @@ -0,0 +1,180 @@ +# Realtime API简介 + +Realtime API 是一系列针对性能、延迟、抗弱网、对接成本、适配性提供多种对接方式的方法，供客户灵活选择。 + +## **概述** + +Realtime API 支持 **WebSocket**、**WebRTC** 和 **AOQ（AI over QUIC）**三种传输协议，开发者可以根据业务场景灵活选择。 + +**维度** + +**WebSocket** + +**WebRTC** + +**AOQ** + +适用场景 + +服务端集成、快速原型验证 + +浏览器端互动、传统音视频通话 + +AI 多模态实时交互、弱网场景、混合数据传输 + +浏览器兼容性 + +原生支持 + +原生支持 + +不支持 + +接入难度 + +极低 + +中等 + +低 + +弱网对抗 + +差 + +良好 + +极致 + +数据类型 + +文本/音频/图像 + +音视频 + 文本 + +音视频 + 文本 + +建连速度 + +慢 + +慢 + +快 + +回声消除/降噪 + +无，需客户端自行处理 + +内置 + +内置 + +AI 场景适配 + +基础，适合纯文本或低实时性场景 + +传统设计，AI 场景需额外适配 + +原生为 AI 多模态数据特征深度定制 + +端侧平台支持 + +全平台（任何支持 WebSocket 的环境） + +浏览器、移动端 + +Android / iOS / HarmonyOS + +开发者可根据实际需求选择协议方案： + +- **WebSocket 方案**：适合服务端集成、快速原型验证、对接入门槛要求极低的场景。通过 DashScope SDK 可快速实现实时语音对话。 + +- **WebRTC 方案**：适合需要浏览器原生支持、已有 WebRTC 基础设施的传统音视频通话场景，内置回声消除和降噪能力。 + +- **AOQ 方案**：适合对延迟、弱网对抗、多模态数据传输有极致要求的 AI 实时交互场景，同时内置回声消除和降噪能力，尤其是移动端原生应用。 + + +## **模型/应用支持力度** + +不同协议对模型和应用的支持情况如下： + +**模型/应用类型** + +**模型** + +**AOQ** + +**WebRTC** + +**WebSocket** + +实时全模态 + +qwen3.5-omni-plus-realtime + +支持 + +支持 + +支持 + +qwen3.5-omni-flash-realtime + +支持 + +支持 + +支持 + +qwen3.5-livetranslate-flash-realtime + +支持 + +支持 + +支持 + +多模态开发套件 + +multimodal-dialog + +不支持 + +支持 + +支持 + +实时语音识别 + +Fun-ASR系列模型 + +不支持 + +不支持 + +支持 + +实时语音合成 + +CosyVoice系列模型 + +不支持 + +不支持 + +支持 + +实时语音对话 + +qwen-audio-3.0-realtime-plus、qwen-audio-3.0-realtime-flash + +不支持 + +不支持 + +支持 + +**说明** + +模型的名称、上下文、价格、快照版本等信息请参见[阿里云百炼控制台](https://bailian.console.aliyun.com/cn-beijing#/home)；并发限流条件请参考[限流](https://help.aliyun.com/zh/model-studio/rate-limit)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md new file mode 100644 index 00000000..147179b9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md @@ -0,0 +1,291 @@ +# 实现接通模型/应用 + +介绍如何通过 AOQ、WebRTC、WebSocket 三种协议接入 Realtime API 模型或应用，包含各协议的连接流程、时序图和代码示例。 + +## **AOQ 接入** + +AOQ 基于 QUIC 协议深度定制，适合移动端原生应用，支持音频/视频/数据混合传输，内置极致抗弱网能力。以下以 iOS Demo 为例。 + +### **整体流程时序图** + +![AOQ中文1](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5755914871/p1088073.jpg) + +### **创建引擎并设置回调** + +``` +let config = AoqCreateConfig() +config.workDir = workDir +config.enableDumpAudio = false +engine = AoqClientEngine.createEngine(config, delegate: self) +``` + +实现 `AoqEngineDelegate` 协议监听 `onConnectionStatusChange`、`onDataMsg`、`onError` 等回调。 + +### **启动音频采集与播放** + +``` +// 音频采集 +let capCfg = AoqAudioCaptureConfig() +capCfg.channel = 1; capCfg.isExternal = false +engine.startAudioCapture(capCfg) + +// 音频播放 +let playCfg = AoqAudioPlaybackConfig() +playCfg.channel = 1; playCfg.isExternal = false +engine.startAudioPlayer(playCfg) + +// 视频采集（可选） +let vidCfg = AoqVideoCaptureConfig() +vidCfg.width = 720; vidCfg.height = 1280; vidCfg.fps = 15 +engine.startVideoCapture(vidCfg) +``` + +### **获取连接凭证** + +由业务 AppServer 代理百炼请求，参考 [Token 鉴权](https://help.aliyun.com/zh/model-studio/realtime-token-authentication) 章节。 + +### **设置编解码及建立连接** + +设置编解码参数后调用 `connect`： + +``` +// 音频编解码配置 +let encCfg = AoqAudioCodecConfig() +encCfg.codecType = .audioPCM; encCfg.sampleRate = 16000; encCfg.channel = 1 +engine.setAudioEncoderConfig(encCfg) +engine.setAudioDecoderConfig(encCfg) + +// connect 前关闭媒体发送，待 session.updated 后再开启 +engine.enableSendMediaStream(.audio, enable: false) + +let config = AoqConnectConfig() +config.token = token +config.sid = sid +config.certFingerprint = certificate +config.relayEndpoints = relayEndpoints +config.workspaceIdHash = workspaceIdHash +config.publishTracks = [audioTrack, dataTrack] +config.subscribeTracks = [audioTrack, dataTrack] +engine.connect(config) +``` + +**重要** + +**重要**：AOQ SDK 在建联后会默认发送媒体数据，此示例演示了连接模型时关闭媒体发送的能力。 + +### **配置 AI 会话** + +连接成功后发送 `session.update` 的示例，详见[模型客户端事件参考](https://help.aliyun.com/zh/model-studio/client-events)： + +``` +func onConnectionStatusChange(_ status: AoqConnectionStatus) { + if status == .connected { sendSessionUpdate() } +} + +private func sendSessionUpdate() { + let json = """ + { + // 该事件的id，由客户端生成 + "event_id": "event_ToPZqeobitzUJnt3QqtWg", + // 事件类型，固定为session.update + "type": "session.update", + // 会话配置 + "session": { + // 输出模态，支持设置为["text"]（仅输出文本）或["text","audio"]（输出文本与音频）。 + "modalities": [ + "text", + "audio" + ], + // 输出音频的音色 + "voice": "Ethan", + // 输入音频格式，当前仅支持设置为pcm。输入音频为16 kHz采样率的PCM音频流。 + "input_audio_format": "pcm", + // 输出音频格式，当前仅支持设置为pcm。输出音频为24 kHz采样率的PCM音频流。 + "output_audio_format": "pcm", + // 系统消息，用于设定模型的目标或角色。 + "instructions": "你是某五星级酒店的AI客服专员，请准确且友好地解答客户关于房型、设施、价格、预订政策的咨询。请始终以专业和乐于助人的态度回应，杜绝提供未经证实或超出酒店服务范围的信息。", + // 是否开启语音活动检测。若需启用，需传入一个配置对象，服务端将据此自动检测语音起止。 + // 设置为null表示由客户端决定何时发起模型响应。 + "turn_detection": { + // VAD类型，取值为server_vad或semantic_vad。使用qwen3.5-omni-realtime模型时推荐设为semantic_vad。 + "type": "semantic_vad", + // VAD检测阈值。建议在嘈杂的环境中增加，在安静的环境中降低。 + "threshold": 0.5, + // 检测语音停止的静音持续时间，超过此值后会触发模型响应 + "silence_duration_ms": 800 + } + } + } + """ + let msg = AoqDataMsg() + msg.data = json.data(using: .utf8)! + engine.send(msg) +} +``` + +### **收到 session.updated 后开启媒体发送** + +收到模型回复 `session.updated` 的示例，详见[模型服务器事件参考](https://help.aliyun.com/zh/model-studio/server-events)： + +``` +func onDataMsg(_ msg: AoqDataMsg) { + guard let obj = try? JSONSerialization.jsonObject(with: msg.data) as? [String: Any], + let type = obj["type"] as? String else { return } + if type == "session.updated" { + engine.enableSendMediaStream(.audio, enable: true) + engine.enableSendMediaStream(.video, enable: true) + } +} +``` + +**重要** + +**重要**： + +1. 模型必须在收到 `session.updated` 后才开启媒体流发送，否则 AI 侧可能还未准备好接收数据。 + +2. 建连时添加的音频轨道和视频轨道（即 AOQ 媒体通道）会自动将数据传输到服务端。 + + 1. 音频：通过音频轨道直接传输，无需发送 `input_audio_buffer.append` 事件。 + + 2. 视频：通过视频轨道发送画面帧，无需发送 `input_image_buffer.append` 事件。 + + +### **断开连接与销毁引擎** + +``` +engine.disconnect() +AoqClientEngine.destroy() +``` + +## **WebRTC 接入** + +WebRTC 协议不提供 SDK，Web 端可以通过 JavaScript，其他端可以通过开源项目或者第三方支持标准 WebRTC 协议的 RTC 服务商进行接入。以下文档以 Web 端 JavaScript 为例进行介绍。 + +### **整体流程图** + +![AOQ中文2](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5755914871/p1088074.jpg) + +### **建立连接** + +``` +# pip install aiortc aiohttp certifi +import asyncio, aiohttp, ssl, certifi +from aiortc import RTCPeerConnection, RTCConfiguration, RTCSessionDescription +from aiortc.mediastreams import AudioStreamTrack + +API_KEY = "your-api-key" +MODEL = "目标模型" +SIGNALING_URL = f"https://{{endpoint}}/api/v1/webrtc/realtime?model={MODEL}" + +async def connect(): + pc = RTCPeerConnection(RTCConfiguration(iceServers=[])) + + # 添加音频轨道，确保 Offer SDP 包含 m=audio（服务端必需） + pc.addTrack(AudioStreamTrack()) + + # 创建 DataChannel 以触发 SDP 协商（名称可自定义，服务端会通过名为 "txt" 的通道推送事件） + pc.createDataChannel("oai-events") + + # SDP 交换：创建 Offer 并发送到服务端 + offer = await pc.createOffer() + await pc.setLocalDescription(offer) + + async with aiohttp.ClientSession() as session: + async with session.post( + SIGNALING_URL, + ssl=ssl.create_default_context(cafile=certifi.where()), + data=offer.sdp.encode("utf-8"), + headers={ + "Content-Type": "application/sdp", + "Authorization": f"Bearer {API_KEY}", + }, + ) as resp: + if not resp.ok: + raise Exception(f"SDP 交换失败: {resp.status} {await resp.text()}") + answer_sdp = await resp.text() + + print("=== Offer SDP ===") + print(offer.sdp) + print("=== Answer SDP ===") + print(answer_sdp) + + # ICE 建连自动完成 + await pc.setRemoteDescription(RTCSessionDescription(sdp=answer_sdp, type="answer")) + print("WebRTC 连接已建立") + return pc +``` + +### **配置目标模型参数** + +监听模型返回的 DataChannel 消息保证交互时序： + +``` +pc.ondatachannel = (event) => { + const ch = event.channel; + ch.onmessage = (e) => { + let obj; + try { obj = JSON.parse(e.data); } + catch (err) { + return; + } + if (obj?.type === "session.created") { + sendUpdate(event.channel); + //开始推送音视频 + audioSender?.replaceTrack(audioTrack); + videoSender?.replaceTrack(videoTrack); + } + }; +}; +``` + +### **收发媒体数据** + +建连时添加的音频轨道和视频轨道（即 RTP 媒体通道）会自动将数据传输到服务端。 + +- 音频：通过音频轨道（RTP）直接传输，无需发送 `input_audio_buffer.append` 事件。 + +- 图片：通过视频轨道（RTP）发送画面帧，不支持 `input_image_buffer.append` 事件。 + + +**说明** + +WebRTC 仅支持服务端 VAD 模式（`server_vad` 或 `semantic_vad`），不支持手动模式。 + +### **Demo 源码** + +#### **前提条件** + +- 使用支持 WebRTC 的现代浏览器（Chrome、Edge、Firefox、Safari 等）。 + +- 浏览器需要麦克风权限。 + +- 浏览器无法直接向服务端发起建立连接的请求（受浏览器跨域安全策略限制），因此需要通过终端执行 curl 命令来完成连接建立。 + + +#### **运行示例** + +新建一个 HTML 文件，命名为 `webrtc_demo.html`，并将以下代码复制到文件中： + +[webrtc\_demo.html](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/crtwmi/webrtc_demo.html)。 + +在浏览器中打开此文件，按以下步骤操作： + +1. 点击开始会话，页面会自动生成 Offer SDP 和对应的 curl 命令。 + +2. 点击复制 curl 命令，在终端中执行。命令返回的内容即为 Answer SDP。 + +3. 将 Answer SDP 粘贴到页面的 Answer SDP 文本框中，点击设置 Answer 即可建立连接并开始语音对话。 + + +## **WebSocket 接入** + +可以通过 DashScope SDK 或者模型的 API 进行接入，详见： + +- [实时全模态](https://help.aliyun.com/zh/model-studio/realtime#bdaa43cdd7hsd) + +- [多模态开发套件](https://help.aliyun.com/zh/model-studio/multimodal-interaction-protocol/) + +- [实时语音识别](https://help.aliyun.com/zh/model-studio/fun-asr-realtime-websocket-api) + +- [实时语音合成](https://help.aliyun.com/zh/model-studio/cosyvoice-websocket-api) diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md new file mode 100644 index 00000000..69982edf --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md @@ -0,0 +1,60 @@ +# SDK下载 + +本文提供AOQ SDK下载链接，介绍如何集成SDK，以及SDK相关信息。 + +## **AOQ SDK下载** + +**版本** + +**平台** + +**下载** + +**更新日期** + +**更新说明** + +v1.0.1 + +Android + +[AoqClientSdk-v1.0.1.aar](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/sqsjum/AoqClientSdk-v1.0.1.aar) + +[libPluginOpus.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/oxzenn/libPluginOpus.zip) + +2026-07-09 + +1. 支持AOQ协议接入模型/应用 + +2. 支持音视频编解码参数 + +3. 支持音视频设备采集及播放 + +4. 支持自定义外部输入音频采集 + +5. 支持自定义外部输入视频采集 + +6. 支持自定义外部输入视频编码 + +7. 支持控制媒体流的发送 + + +iOS + +[AoqClientSdk-v1.0.1.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/vlyumo/AoqClientSdk-v1.0.1.zip) + +[PluginOpus.framework.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/rbvmfq/PluginOpus.framework.zip) + +Harmony + +[AoqClientSdk-v1.0.1.har](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/zwlikp/AoqClientSdk-v1.0.1.har) + +[libPluginOpus.zip](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260715/wvtyrh/libPluginOpus.zip) + +**重要** + +AOQ使用音频插件的方式加载Opus编解码器，如果需要Opus编解码时需要下载opus插件并加载到工程内。 + +## **WebSocket SDK下载** + +参见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md new file mode 100644 index 00000000..6830d3a4 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md @@ -0,0 +1,358 @@ +# Token鉴权 + +介绍 Realtime API 的 Token 鉴权机制，包括 API Key 的获取方式以及 WebSocket、WebRTC、AOQ 三种协议的建连鉴权方法。 + +## **概述** + +Realtime API 使用 **API Key** 进行身份认证。无论您选择 WebSocket、WebRTC 还是 AOQ 协议接入，均通过 HTTP 请求头中的 `Authorization` 字段携带 Bearer Token 完成身份验证。 + +鉴权发生在**建连阶段**，连接建立后的音视频/数据传输无需重复鉴权。 + +三种协议的鉴权差异： + +**协议** + +**鉴权时机** + +**鉴权方式** + +**说明** + +WebSocket + +WebSocket 连接握手时 + +HTTP Header `Authorization: Bearer ` + +客户端或服务端直接携带 API Key 建连 + +WebRTC + +SDP 交换 HTTP 请求时 + +HTTP Header `Authorization: Bearer ` + +客户端或服务端携带 API Key 发起 SDP 交换 + +AOQ + +业务 AppServer 请求网关时 + +HTTP Header `Authorization: Bearer ` + +API Key 仅在服务端使用，客户端使用网关返回的 Token + +## **获取 API Key** + +### **步骤 1：开通百炼服务** + +1. 访问[阿里云百炼控制台](https://bailian.console.aliyun.com/cn-beijing#/home)并登录您的阿里云账号。 + +2. 如果是首次使用，按照页面提示完成服务开通。 + + +### **步骤 2：创建 API Key** + +1. 在控制台左侧导航栏中，选择 **API Key 管理**。 + +2. 点击 **创建 API Key**，选择关联的业务空间。 + +3. 创建完成后，请**立即复制并妥善保存** API Key。 + + +**重要** + +**安全提示**：API Key 是您访问服务的唯一凭证，请勿将其硬编码到客户端代码中或提交到代码仓库。建议通过环境变量或后端服务下发的方式管理。 + +## **建连鉴权详解** + +### **AOQ 协议鉴权** + +AOQ 采用**服务端代理鉴权**模式：API Key 仅在业务 AppServer 侧使用，客户端使用网关返回的临时 Token 建连，避免 API Key 暴露在客户端。 + +![Token鉴权](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3935914871/p1088069.jpg) + +#### **百炼网关请求 curl 示例** + +``` +curl -X POST \ + "https://{endpoint}/api/v1/webrtc/realtime?model=qwen3.5-omni-plus-realtime" \ + -H "Content-Type: application/json" \ + -H "Authorization: Bearer ${DASHSCOPE_API_KEY}" \ + -H "x-dashscope-rtc-transport: moq" \ + -d '{"clientIp": ${客户端真实IP}}' +``` + +#### **请求字段说明** + +**配置项** + +**值** + +**说明** + +endpoint + +根据业务情况选择接入域名 + +指定对应的接入域名，详情请参见[选择地域、服务部署范围和接入域名](https://help.aliyun.com/zh/model-studio/regions/) + +Content-Type + +`application/json` + +\- + +Authorization + +`Bearer ` + +必填 + +x-dashscope-rtc-transport + +`moq` + +**指定使用 AOQ 协议** + +clientIp + +选填。客户端真实公网 IP + +不填写时，使用请求百炼网关的 IP 作为客户端 IP；若填写，则以 clientIp 作为客户端 IP。Realtime API 会参考客户端 IP 提供最佳的 Relay 接入点信息 + +#### **响应示例** + +``` +{ + "sid": "1d06b55683db49bba67a407902f62d02:1782706970:69aecdc5...", + "aoqTokenForClient": "ecc1a46015d5496ca4ff7a48281eb739", + "clientRelayEndpoints": [{"endpoint": "121.199.XX.XX", "port": 8443}], + "clientRelayCertFingerprint": "sha256/99843495...", + "sidExpiresInSecs": 7200, + "extraInfo": {"workspaceIdHash": "2021b6f98cea4cff"} +} +``` + +#### **响应字段说明** + +**字段** + +**说明** + +sid + +会话唯一标识 + +aoqTokenForClient + +客户端连接令牌，传给 SDK 的 token 字段 + +clientRelayEndpoints + +Relay 接入点数组（endpoint + port） + +clientRelayCertFingerprint + +Relay TLS 证书指纹 + +sidExpiresInSecs + +会话过期时间（秒） + +extraInfo.workspaceIdHash + +工作区 ID 哈希 + +#### **AOQ Client SDK 连接示例** + +## **iOS (Swift)** + +``` +let resp = try JSONDecoder().decode(AllocateResponse.self, from: responseData) + +let config = AoqConnectConfig() +config.token = resp.aoqTokenForClient +config.sid = resp.sid +config.certFingerprint = resp.clientRelayCertFingerprint +config.relayEndpoints = resp.clientRelayEndpoints.map { item in + let ep = AoqRelayEndpoint() + ep.endpoint = item.endpoint + ep.port = item.port + return ep +} +config.workspaceIdHash = resp.extraInfo?.workspaceIdHash ?? "" + +let audioTrack = AoqTrackParam() +audioTrack.trackType = .audio +let dataTrack = AoqTrackParam() +dataTrack.trackType = .data +config.publishTracks = [audioTrack, dataTrack] +config.subscribeTracks = [audioTrack, dataTrack] + +engine.connect(config) +``` + +## **Android (Java)** + +``` +JSONObject obj = new JSONObject(responseText); +AoqClientEngine.AoqConnectConfig cfg = new AoqClientEngine.AoqConnectConfig(); +cfg.token = obj.optString("aoqTokenForClient", ""); +cfg.sid = obj.optString("sid", ""); +cfg.certFingerprint = obj.optString("clientRelayCertFingerprint", ""); + +JSONArray arr = obj.optJSONArray("clientRelayEndpoints"); +if (arr != null) { + for (int i = 0; i < arr.length(); i++) { + JSONObject o = arr.optJSONObject(i); + AoqClientEngine.AoqRelayEndpoint ep = new AoqClientEngine.AoqRelayEndpoint(); + ep.endpoint = o.optString("endpoint", ""); + ep.port = o.optInt("port", 0); + cfg.relayEndpoints.add(ep); + } +} + +JSONObject ext = obj.optJSONObject("extraInfo"); +cfg.workspaceIdHash = ext != null ? ext.optString("workspaceIdHash", "") : ""; + +AoqClientEngine.AoqTrackParam audio = new AoqClientEngine.AoqTrackParam(); +audio.trackType = AoqClientEngine.AoqTrackType.AoqTrackTypeAudio; +AoqClientEngine.AoqTrackParam data = new AoqClientEngine.AoqTrackParam(); +data.trackType = AoqClientEngine.AoqTrackType.AoqTrackTypeData; +cfg.publishTracks.add(audio); +cfg.publishTracks.add(data); +cfg.subscribeTracks.add(audio); +cfg.subscribeTracks.add(data); + +engine.connect(cfg); +``` + +## **OHOS (ArkTS)** + +``` +const obj = JSON.parse(responseText) as Record; +const cfg: AoqConnectConfig = { + token: String(obj['aoqTokenForClient'] ?? ''), + sid: String(obj['sid'] ?? ''), + certFingerprint: String(obj['clientRelayCertFingerprint'] ?? ''), + relayEndpoints: (obj['clientRelayEndpoints'] as Array).map(item => ({ + endpoint: String(item['endpoint'] ?? ''), + port: Number(item['port'] ?? 0) + })), + workspaceIdHash: String((obj['extraInfo'] as any)?.['workspaceIdHash'] ?? ''), + publishTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ], + subscribeTracks: [ + { trackType: AoqTrackType.AoqTrackTypeAudio }, + { trackType: AoqTrackType.AoqTrackTypeData } + ] +}; +engine.connect(cfg); +``` + +**说明** + +`clientIp` 为请求体中的非必填字段。不填写时，使用请求百炼网关的 IP 作为客户端 IP；若填写，则以 clientIp 作为客户端 IP。建议由业务 AppServer 在服务端获取客户端真实 IP 后填入，以获得最佳的 Relay 接入点。 + +## **WebRTC 协议鉴权** + +WebRTC 通过 HTTP POST 请求完成 SDP 交换，鉴权在此阶段完成。客户端将 Offer SDP 发送给服务端，服务端返回 Answer SDP。 + +**配置项** + +**值** + +**说明** + +请求方法 + +POST + +\- + +请求地址 + +`https://{endpoint}/api/v1/webrtc/realtime?model={model_name}` + +替换 endpoint 和 model\_name + +Content-Type + +`application/sdp` + +请求体为 SDP 字符串 + +Authorization + +`Bearer ` + +必填 + +响应 + +HTTP 200，返回 Answer SDP + +失败返回 4xx + +**说明** + +WebRTC 功能目前为白名单开放，请联系商务经理获取 Endpoint。 + +``` +const pc = new RTCPeerConnection(); +const stream = await navigator.mediaDevices.getUserMedia({ audio: true }); +stream.getAudioTracks().forEach(t => pc.addTrack(t, stream)); +pc.createDataChannel('oai-events'); + +const offer = await pc.createOffer(); +await pc.setLocalDescription(offer); + +// 等待 ICE 收集完成后发送 +const resp = await fetch(API_URL, { + method: 'POST', + headers: { + 'Content-Type': 'application/sdp', + 'Authorization': `Bearer ${API_KEY}`, + }, + body: pc.localDescription.sdp, +}); +const answerSdp = await resp.text(); +await pc.setRemoteDescription({ type: 'answer', sdp: answerSdp }); +``` + +## **WebSocket 协议鉴权** + +WebSocket 鉴权最为简单，客户端在建立 WebSocket 连接时直接通过 HTTP Header 携带 API Key。 + +**配置项** + +**值** + +**说明** + +连接地址 + +`wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model={model_name}` + +华北2（北京） + +Authorization + +`Bearer ` + +必填 + +``` +import websocket, os +API_KEY = os.getenv("DASHSCOPE_API_KEY") +URL = "wss://dashscope.aliyuncs.com/api-ws/v1/realtime?model=qwen3.5-omni-plus-realtime" +ws = websocket.WebSocketApp(URL, header=["Authorization: Bearer " + API_KEY]) +ws.run_forever() +``` + +**说明** + +您也可以使用 [DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk) 方式接入。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md index 53fb9ef1..455f7550 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md @@ -19,7 +19,39 @@ ## 支持的模型 -`qwen3-max`、`qwen3-max-2026-01-23`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-02-15`、`qwen3.5-plus-2026-04-20`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash`、`qwen3-coder-next`。 +### 华北2（北京） + +**中国内地部署范围** + +`qwen3.8-max-preview`（[Token Plan](https://help.aliyun.com/zh/model-studio/token-plan-overview)）、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3-max`、`qwen3-max-2026-01-23`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-04-20`、`qwen3.5-plus-2026-02-15`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash`、`qwen3.5-ocr`、`qwen-plus-character`、`qwen-flash-character` + +### 新加坡 + +**国际部署范围** + +`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3-max`、`qwen3-max-2026-01-23`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-04-20`、`qwen3.5-plus-2026-02-15`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash`、`qwen-plus-character`、`qwen-flash-character` + +### 美国（弗吉尼亚） + +**全球部署范围** + +`` `qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.5-plus`、`qwen3.5-plus-2026-02-15`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16`、`qwen3.5-flash`、`qwen3.5-flash-2026-02-23`、`qwen3.6-35b-a3b`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-27b`、`qwen3.5-35b-a3b` `` + +### 德国（法兰克福） + +**全球部署范围** + +`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.5-397b-a17b`、`qwen3.5-122b-a10b`、`qwen3.5-35b-a3b`、`qwen3.5-27b` + +### 日本（东京） + +**日本部署范围** + +`qwen3.7-plus`、`qwen3.7-plus-2026-05-26` + +**全球部署范围** + +`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.6-plus`、`qwen3.6-plus-2026-04-02`、`qwen3.6-flash`、`qwen3.6-flash-2026-04-16` ## 服务地址 @@ -771,7 +803,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/ **使用方式**：在请求 Header 中添加 `x-dashscope-session-cache: enable` 开启，或设置为 `disable` 关闭。默认值为 `disable`。 -**支持的模型：**`qwen3-max`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.5-plus`、`qwen3.6-flash`、`qwen3.5-flash`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash` +**支持的模型：**`qwen3.8-max-preview`、`qwen3.7-max`、`qwen3.7-max-2026-05-20`、`qwen3.7-max-2026-06-08`、`qwen3.7-plus`、`qwen3.7-plus-2026-05-26`、`qwen3.6-plus`、`qwen3.5-plus`、`qwen3.6-flash`、`qwen3.5-flash`、`qwen3-max`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus`、`qwen3-coder-flash` > Session 缓存 最小可缓存提示词长度为 1024 Token，缓存有效期为 5 分钟。相关约束限制与[显式缓存](https://help.aliyun.com/zh/model-studio/context-cache)一致。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md index b7487808..37eae351 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md @@ -16,14 +16,14 @@ **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 **items** `_array_`（可选） @@ -948,3 +948,5 @@ console.log("第二轮响应:", response2.output_text); - 创建会话或添加消息项时，`items` 最多包含20条。 - `metadata` 最多16对键值对，key最大长度64字符，value最大长度512字符。 + +- 会话信息最多保存7天。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md index d94c7a68..480aa12c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md @@ -64,9 +64,7 @@ text-embedding-async-v1 通用文本向量批处理接口API支持通过HTTP和DashScope SDK进行调用。 -在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 +在调用前，先[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 ## HTTP调用 @@ -79,14 +77,16 @@ HTTP调用仅支持异步模式，需通过两步完成： **通过HTTP调用时需配置的endpoint：** -`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ### 创建任务 ##### **请求参数** ``` -curl -X POST 'https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ +curl -X POST 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -H 'X-DashScope-Async: enable' \ @@ -232,7 +232,7 @@ url `_string_` **（必选）** ### **根据任务ID查询结果** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ##### **请求参数** @@ -240,10 +240,10 @@ url `_string_` **（必选）** 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://dashscope-intl.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -398,6 +398,9 @@ SDK与HTTP接口的参数名基本一致，参数结构根据不同语言的SDK ``` from dashscope import BatchTextEmbedding +import dashscope +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" result = BatchTextEmbedding.call(BatchTextEmbedding.Models.text_embedding_async_v1, url="https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241016/nigwvr/text_embedding_file.txt", @@ -416,8 +419,11 @@ import com.alibaba.dashscope.exception.NoApiKeyException; import com.alibaba.dashscope.task.AsyncTaskListParam; import com.alibaba.dashscope.task.AsyncTaskListResult; import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; public static void basicCall() throws ApiException, NoApiKeyException { BatchTextEmbeddingParam param = BatchTextEmbeddingParam.builder() .model(BatchTextEmbedding.Models.TEXT_EMBEDDING_ASYNC_V1) @@ -430,6 +436,8 @@ public class Main { } public static void main(String[] args) { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; try { basicCall(); } catch (ApiException | NoApiKeyException e) { @@ -446,7 +454,10 @@ public class Main { ``` from dashscope import BatchTextEmbedding +import dashscope from http import HTTPStatus +# 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 +dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" # 创建异步任务 def create_async_task(): @@ -508,8 +519,11 @@ import com.alibaba.dashscope.exception.NoApiKeyException; import com.alibaba.dashscope.task.AsyncTaskListParam; import com.alibaba.dashscope.task.AsyncTaskListResult; import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; public class Main { + // 以下为华北2（北京）地域的配置，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的配置不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; /**创建批处理任务*/ public static BatchTextEmbeddingResult createTask() throws ApiException, NoApiKeyException { @@ -738,7 +752,7 @@ url `_string_` **（必选）** 任务状态 -- SUCCESSED: 任务执行成功 +- SUCCEEDED: 任务执行成功 - FAILED: 任务执行失败 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md index 79c7fb66..c8e22c77 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md @@ -4,13 +4,13 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 ## 适用范围 -为确保调用成功，请务必保证模型、endpoint URL 和 API Key 均属于**同一地域**。跨地域调用将会失败。 +为确保调用成功，请务必保证模型、endpoint URL和API Key 均属于**同一地域**。跨地域调用将会失败。 - [**选择模型**](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL。 -- **配置 API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +- **配置API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 **说明** @@ -19,14 +19,14 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,7 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +64,7 @@ HappyHorse图生视频模型，以首帧图片为基础，支持通过文本描 ## 图生视频-基于首帧 ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -169,7 +166,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 1. 公网URL： - - 支持 HTTP 或 HTTPS 协议。 + - 支持HTTP或HTTPS协议。 - 示例值：https://xxx/xxx.png。 @@ -326,14 +323,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -342,8 +335,6 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - **说明** - **轮询建议**：视频生成过程约需数分钟，建议采用**轮询**机制，并设置合理的查询间隔（如 15 秒）来获取结果。 @@ -363,7 +354,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md index b93ad55b..97c7bc86 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md @@ -19,14 +19,14 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,7 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +64,7 @@ HappyHorse-参考生视频模型支持传入**多张参考图像**，通过**文 ## 参考生视频（多图像） ``` +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -368,14 +365,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -384,8 +377,6 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - **说明** - **轮询建议**：视频生成过程约需数分钟，建议采用**轮询**机制，并设置合理的查询间隔（如 15 秒）来获取结果。 @@ -403,7 +394,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md index 385ff8f8..2db67804 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md @@ -19,14 +19,14 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,7 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +64,7 @@ HappyHorse文生视频模型，输入文本提示词生成物理真实、运动 ## 文生视频 ``` +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -266,14 +263,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -282,8 +275,6 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - **说明** - **轮询建议**：视频生成过程约需数分钟，建议采用**轮询**机制，并设置合理的查询间隔（如 15 秒）来获取结果。 @@ -303,7 +294,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md index edfc94c8..b5a84b4f 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md @@ -19,14 +19,14 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 **重要** -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -38,14 +38,10 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -54,7 +50,7 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -68,6 +64,7 @@ HappyHorse 视频编辑模型支持输入视频与参考图，结合文本指令 ## 视频编辑（指令+参考图） ``` +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ @@ -369,14 +366,10 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **美国（弗吉尼亚）** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -385,8 +378,6 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - **说明** - **轮询建议**：视频编辑过程约需数分钟，建议采用**轮询**机制，并设置合理的查询间隔（如 15 秒）来获取结果。 @@ -404,7 +395,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md index 4a7b656c..f2b1777b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md @@ -27,7 +27,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -43,7 +45,8 @@ 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -67,7 +70,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -107,7 +111,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -136,7 +141,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`、`kling/kling-v3-video-generation` 。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -169,7 +175,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：`kling/kling-v3-omni-video-generation`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -640,7 +647,7 @@ audio直接影响费用，请前往[百炼控制台](https://bailian.console.ali ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -659,7 +666,7 @@ audio直接影响费用，请前往[百炼控制台](https://bailian.console.ali ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md index 7fea4312..7e98f28e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md @@ -17,20 +17,26 @@ 为确保调用成功，请务必保证模型、endpoint URL 和 API Key 均属于**同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#183f1b6fa0lox)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 - **配置 API Key**：获取该地域的[API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 图生视频任务耗时较长（通常为1-5分钟），API采用异步调用的方式。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +52,8 @@ 支持模型：pixverse/pixverse-c1-it2v、pixverse/pixverse-v6-it2v、pixverse/pixverse-v5.6-it2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +86,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景即可，不支持设置 `shot_type`参数。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -110,7 +118,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景，并设置 `shot_type` 为`multi`， 即可生成有声多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -155,7 +164,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#183f1b6fa0lox)。 +模型名称。 可选值： @@ -383,7 +392,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -402,7 +411,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md index 99e60eb8..557607a9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md @@ -17,20 +17,26 @@ 为确保调用成功，请务必保证**模型、Endpoint URL 和 API Key 均属于同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#f32f686472enw)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 - **配置 API Key**：选择地域并[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 由于首尾帧生视频任务耗时较长（通常为1-5分钟），API采用异步调用。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +52,8 @@ 支持模型：pixverse/pixverse-c1-kf2v、pixverse/pixverse-v6-kf2v、pixverse/pixverse-v5.6-kf2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -95,7 +102,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#f32f686472enw)。 +模型名称。 可选值： @@ -310,7 +317,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -329,7 +336,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md index 7a9a606c..61775d7e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入视频和音频文件，生成对口型视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +82,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 传入视频和 TTS 文本，由模型合成语音并生成对口型视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -273,7 +277,7 @@ auto ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md index f6766e71..44892823 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入角色图片和动作视频，生成角色模仿动作的视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -162,7 +165,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md index c90d87c5..1e14eac3 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md @@ -30,7 +30,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -46,7 +48,8 @@ 通过 `media` 传入参考图像，设置 `size` 和 `duration` 控制视频分辨率和时长。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -83,7 +86,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在 `media` 传入参考图像，并传入`ref_name`用于设置参考图像的主体名称。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -759,7 +763,7 @@ audio直接影响费用，请在调用前确认[爱诗-参考生视频](https:// ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -778,7 +782,7 @@ audio直接影响费用，请在调用前确认[爱诗-参考生视频](https:// ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md index 4775d1a3..320e8241 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md @@ -17,7 +17,7 @@ 为确保调用成功，请务必保证**模型、Endpoint URL 和 API Key 均属于同一地域**。跨地域调用将会失败。 -- [**选择模型**](https://help.aliyun.com/zh/model-studio/use-video-generation#3ad2d09509ldb)：确认模型所属的地域。 +- [**选择模型**](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)：确认模型所属的地域。 - **选择 URL**：选择对应的地域 Endpoint URL，支持HTTP URL或 DashScope SDK URL。 @@ -26,13 +26,19 @@ - **安装 SDK**：如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 +**说明** + +本文的示例代码适用于**华北2（北京）地域**。 + ## HTTP调用 由于文生视频任务耗时较长（通常为1-5分钟），API采用异步调用。整个流程包含 **"创建任务 -> 轮询获取"** 两个核心步骤，具体如下： ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -48,7 +54,8 @@ 支持模型：pixverse/pixverse-c1-t2v、pixverse/pixverse-v6-t2v、pixverse/pixverse-v5.6-t2v。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -74,7 +81,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景即可，不支持设置 `shot_type`参数。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -99,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 在`prompt`中描述多镜头场景，并设置 `shot_type` 为`multi`， 即可生成有声多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -140,7 +149,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener **model** `_string_` **（必选）** -模型名称。模型输出规格请参见[模型列表](https://help.aliyun.com/zh/model-studio/use-video-generation#3ad2d09509ldb)。 +模型名称。 可选值： @@ -574,7 +583,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -593,7 +602,7 @@ audio直接影响费用，请在调用前确认[模型价格](https://help.aliyu ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ @@ -798,7 +807,9 @@ SDK 的参数命名与[HTTP接口](#pv101a0h2http)基本一致，参数结构根 若版本过低，可能会触发 "url error, please check url!" 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -**北京地域**：`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +**北京地域**：`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -811,7 +822,7 @@ import dashscope import os # 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -881,7 +892,7 @@ import dashscope import os # 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -984,13 +995,16 @@ if __name__ == '__main__': 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -**北京地域**：`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +**北京地域**：`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 ##### 请求示例 ``` +// 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 // Copyright (c) Alibaba, Inc. and its affiliates. import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; @@ -1006,7 +1020,7 @@ public class Text2Video { static { // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -1080,6 +1094,7 @@ public class Text2Video { ##### 请求示例 ``` +// 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 // Copyright (c) Alibaba, Inc. and its affiliates. import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; @@ -1097,7 +1112,7 @@ public class Text2Video { static { // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md index 43f148d0..ac899560 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md @@ -34,7 +34,9 @@ ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -50,7 +52,8 @@ 传入视频文件，输出超分辨率视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 调用时请将 {WorkspaceId} 替换为真实的业务空间ID。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -159,7 +162,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md index c12815fa..fcb9c26c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-图生视频模型根据**输入图像**和**文本提示词**，生成一 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -40,8 +42,10 @@ Vidu-图生视频模型根据**输入图像**和**文本提示词**，生成一 ## 图生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -299,7 +303,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -318,7 +322,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md index 5547a29f..425a4bc4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-首尾帧生视频模型基于**首帧图像**、**尾帧图像和文本提 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -40,8 +42,10 @@ Vidu-首尾帧生视频模型基于**首帧图像**、**尾帧图像和文本提 ## 首尾帧生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -292,7 +296,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -311,7 +315,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md index 833a58c2..84700aa6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md @@ -27,7 +27,9 @@ Vidu-参考生视频模型支持传入**参考图片**和**文本提示词**， ### **步骤 1：创建任务获取任务 ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -42,8 +44,10 @@ Vidu-参考生视频模型支持传入**参考图片**和**文本提示词**， 支持模型：vidu/viduq3-ad\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,8 +83,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：vidu/viduq3-drama\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -118,8 +124,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 参考生视频（仅参考图像） +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -155,8 +163,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持模型：vidu/viduq2-pro\_reference2video。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -595,7 +605,7 @@ vidu/viduq3-drama\_reference2video 不支持该参数，默认输出有声视频 ### **步骤 2：根据任务 ID 查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -614,7 +624,7 @@ vidu/viduq3-drama\_reference2video 不支持该参数，默认输出有声视频 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md index bde2d5df..1254f4e7 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md @@ -29,7 +29,9 @@ Vidu-文生视频模型基于**文本提示词**，生成一段流畅的视频 ### **步骤1：创建任务获取任务ID** -**北京地域**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +**北京地域**：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -42,8 +44,10 @@ Vidu-文生视频模型基于**文本提示词**，生成一段流畅的视频 ## 文生视频 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -322,7 +326,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ### **步骤2：根据任务ID查询结果** -**北京地域**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +**北京地域**：`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` **说明** @@ -341,7 +345,7 @@ duration直接影响费用，按秒计费，时间越长费用越高，请在调 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ @@ -546,7 +550,7 @@ SDK 的参数命名与[HTTP接口](#vd101a0h2http)基本一致，参数结构根 请确保 DashScope Python SDK 版本**不低于** `**1.25.8**`，再运行以下代码。 -**北京地域**：`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +**北京地域**：`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## 同步调用 @@ -558,8 +562,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") @@ -628,8 +632,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" api_key = os.getenv("DASHSCOPE_API_KEY") @@ -728,7 +732,7 @@ if __name__ == '__main__': 请确保 DashScope Java SDK 版本**不低于** `**2.22.6**`，再运行以下代码。 -**北京地域**：`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +**北京地域**：`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## 同步调用 @@ -749,8 +753,8 @@ import com.alibaba.dashscope.utils.Constants; public class Text2Video { static { - // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -836,8 +840,8 @@ import com.alibaba.dashscope.utils.Constants; public class Text2Video { static { - // 以下为北京地域url - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为北京地域URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md index 08695f0a..083f04cc 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md @@ -29,9 +29,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -45,13 +50,13 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -67,7 +72,8 @@ 基于首帧图像和音频生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -101,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 传入首帧和尾帧生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -134,7 +141,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 基于首段视频片段，让模型生成后续内容。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -526,13 +534,13 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -553,10 +561,10 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -754,13 +762,13 @@ SDK 的参数命名与[HTTP接口](#9c71bffa84zm6)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -773,8 +781,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -854,8 +862,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-general-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -973,13 +981,13 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -1005,29 +1013,12 @@ import java.util.List; public class Image2Video { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.InputRequiredException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.util.ArrayList; -import java.util.List; - -public class Image2Video { - - static { - // 以下为新加坡地域URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/en/model-studio/get-api-key + // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key static String apiKey = System.getenv("DASHSCOPE_API_KEY"); public static void syncCall() { @@ -1093,29 +1084,12 @@ import java.util.List; public class Image2Video { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 - -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesis; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisParam; -import com.alibaba.dashscope.aigc.videosynthesis.VideoSynthesisResult; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.InputRequiredException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.util.ArrayList; -import java.util.List; - -public class Image2Video { - - static { - // 以下为新加坡地域URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/en/model-studio/get-api-key + // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key static String apiKey = System.getenv("DASHSCOPE_API_KEY"); public static void asyncCall() { diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md index bf2dd8df..352341c9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md @@ -29,9 +29,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -41,14 +46,12 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` @@ -57,7 +60,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -75,7 +78,9 @@ 可通过设置`"prompt_extend": true`和`"shot_type":"multi"`启用。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -102,7 +107,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 若不提供 `input.audio_url` ，模型将根据视频内容自动生成匹配的背景音乐或音效。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -127,7 +134,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 如需为视频指定背景音乐或配音，可通过 `input.audio_url` 参数传入自定义音频的 URL。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -156,7 +165,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -182,7 +193,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 示例：下载[img\_base64](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20250722/pmcjis/img_base64.txt)文件，并将完整内容粘贴至`img_url`参数中。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -203,11 +216,13 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener - prompt 字段将被忽略，建议留空。 -- 特效的可用性与模型相关。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +- 特效的可用性与模型相关。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -228,7 +243,9 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过 negative\_prompt 指定生成的视频避免出现“花朵”元素。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -322,7 +339,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 支持输入的格式： -1. 公网URL: +1. 公网URL： - 支持 HTTP 或 HTTPS 协议。 @@ -381,7 +398,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 视频特效模板的名称。若未填写，表示不使用任何视频特效。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 示例值：flying，表示使用“魔法悬浮”特效。 @@ -596,14 +613,12 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `GET https://dashscope-us.aliyuncs.com/api/v1/tasks/{task_id}` @@ -612,7 +627,7 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -633,10 +648,10 @@ audio直接影响费用，有声视频与无声视频价格不同，请前往百 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -881,18 +896,16 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -根据模型所在地域设置 `**base_http_api_url**`: +根据模型所在地域设置 `**base_http_api_url**`： ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `dashscope.base_http_api_url = 'https://dashscope-us.aliyuncs.com/api/v1'` @@ -901,7 +914,7 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -919,8 +932,9 @@ from dashscope import VideoSynthesis import mimetypes import dashscope -# 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +# 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -1032,8 +1046,9 @@ from http import HTTPStatus from dashscope import VideoSynthesis import dashscope -# 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +# 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -1145,18 +1160,16 @@ if __name__ == '__main__': 若版本过低，可能会触发 “url error, please check url!” 等错误。请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)进行更新。 -根据模型所在地域设置 `**baseHttpApiUrl**`: +根据模型所在地域设置 `**baseHttpApiUrl**`： ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - ## **弗吉尼亚** `Constants.baseHttpApiUrl = "https://dashscope-us.aliyuncs.com/api/v1";` @@ -1165,7 +1178,7 @@ if __name__ == '__main__': `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -1198,8 +1211,9 @@ import java.util.Map; public class Image2Video { static { - // 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + // 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -1351,8 +1365,9 @@ import java.util.Map; public class Image2Video { static { - // 以下为北京地域url，获取url：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + // 获取URL：https://help.aliyun.com/zh/model-studio/image-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md index 7dc1d88e..dd4f30e4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md @@ -23,9 +23,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -35,13 +40,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -57,7 +64,8 @@ 根据首帧、尾帧和prompt生成视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -82,7 +90,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 格式参见[如何输入图像](https://help.aliyun.com/zh/model-studio/image-to-video-first-and-last-frames-guide#69308edfbebnx)。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -104,10 +113,11 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 必须传入`first_frame_url`和`template`，无需传入prompt和last\_frame\_url。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -129,7 +139,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 通过 negative\_prompt 指定生成的视频避免出现“人物”元素。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -274,7 +285,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/image2video 视频特效模板的名称。使用此参数时，仅需传入 `first_frame_url`。 -不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 +不同模型支持不同的特效模板。调用前请查阅[万相-图生视频-视频特效列表](https://help.aliyun.com/zh/model-studio/wanx-video-effects)，以免调用失败。 示例值：hufu-1，表示使用“唐韵翩然”特效。 @@ -412,13 +423,15 @@ duration直接影响费用，按秒计费，调用前请确认百炼控制台。 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -441,10 +454,10 @@ duration直接影响费用，按秒计费，调用前请确认百炼控制台。 请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 -> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中{WorkspaceId}需替换为真实的业务空间ID。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -641,13 +654,15 @@ SDK 的参数命名与[HTTP接口](https://help.aliyun.com/zh/model-studio/text- ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### **示例代码** @@ -666,7 +681,7 @@ import mimetypes import dashscope # 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -768,7 +783,7 @@ from http import HTTPStatus from dashscope import VideoSynthesis import dashscope -# 以下为华北2（北京）地域的URL，各地域的URL不同。 +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ### Java SDK调用 @@ -816,7 +831,7 @@ import java.util.Map; public class Kf2vSync { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ## 异步调用 @@ -851,7 +866,7 @@ import java.util.Map; public class Kf2vAsync { static { - // 以下为华北2（北京）地域的URL，各地域的URL不同。 + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 ``` ## **使用限制** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md index 8477b7b4..b5562bbf 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -37,13 +42,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -53,7 +60,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -69,7 +76,8 @@ 通过`reference_urls`传入图像和视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -99,7 +107,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过`reference_urls`传入多个视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -125,7 +134,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过`reference_urls`传入单个视频URL。同时设置`shot_type`为`multi`，生成多镜头视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -150,7 +160,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 当生成无声视频时，**必须显式设置** `parameters.audio = false`。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -505,13 +516,15 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -521,7 +534,7 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -542,10 +555,10 @@ audio直接影响费用，有声视频与无声视频价格不同，请在调用 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -735,13 +748,15 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 美国 @@ -751,7 +766,7 @@ SDK 的参数命名与[HTTP接口](#42703589880ts)基本一致，参数结构根 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -766,7 +781,7 @@ import dashscope import os # 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -814,7 +829,7 @@ from dashscope import VideoSynthesis import dashscope # 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -877,19 +892,21 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **法兰克福** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -916,7 +933,7 @@ public class Ref2Video26 { static { // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -986,7 +1003,7 @@ public class Ref2Video26Async { static { // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md index 03862462..b5ad73b5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md @@ -23,9 +23,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -39,13 +44,15 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -55,7 +62,7 @@ `POST https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -74,7 +81,8 @@ ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -100,7 +108,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 可通过 `input.audio_url` 参数传入自定义音频的 URL。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -125,7 +134,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 若不提供 `input.audio_url` ，模型将根据视频内容自动生成匹配的背景音乐或音效。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -149,7 +159,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener > wan2.6 及wan2.5系列模型默认生成有声视频。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -172,7 +183,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener 通过 negative\_prompt 排除“花朵”元素，避免其出现在视频画面中。 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -477,13 +489,15 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -493,7 +507,7 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 `GET https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -514,10 +528,10 @@ duration直接影响费用。费用 = 单价（基于分辨率）× 时长（秒 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -743,13 +757,15 @@ SDK 的参数命名与HTTP接口基本一致，参数结构根据语言特性进 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -759,7 +775,7 @@ SDK 的参数命名与HTTP接口基本一致，参数结构根据语言特性进 `dashscope.base_http_api_url = 'https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -771,8 +787,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -844,8 +860,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同，获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -955,13 +971,15 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **弗吉尼亚** @@ -971,7 +989,7 @@ if __name__ == '__main__': `Constants.baseHttpApiUrl = "https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -994,8 +1012,8 @@ import java.util.Map; public class Text2Video { static { - // 以下为北京地域url，各地域的url不同，获取url：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" @@ -1093,8 +1111,8 @@ public class Text2Video { static { - // 以下为北京地域url，各地域的url不同，获取url：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。获取URL：https://help.aliyun.com/zh/model-studio/text-to-video-api-reference + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md index 0fea1012..eb4e31d9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -33,20 +38,23 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 请求参数 ## **多图参考** ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -71,7 +79,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频重绘 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -92,8 +101,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **局部编辑** ``` -# 如果使用华北2（北京）地域的模型，需要将url替换为：https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -117,7 +126,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频延展 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -137,7 +147,8 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 视频画面扩展 ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ --header 'X-DashScope-Async: enable' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ @@ -1168,22 +1179,24 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 请求参数 ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md index 9fa72fb6..4c0b78dd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md @@ -465,7 +465,7 @@ duration直接影响费用，按秒计费，请在调用前确认[模型价格]( ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md index a7b65d67..e6707b3c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md @@ -53,6 +53,8 @@ **新加坡地域**：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis` +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + **说明** - 创建成功后，使用接口返回的 `task_id` 查询结果，task\_id 有效期为 24 小时。**请勿重复创建任务**，轮询获取即可。 @@ -64,7 +66,7 @@ ## 视频换人 -以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 ``` curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2video/video-synthesis' \ @@ -256,13 +258,13 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 新加坡 `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -285,7 +287,7 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi 请将`0385dc79-5ff8-4d82-bcb6-xxxxxx`替换为真实的task\_id。 -> 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 +> 以下为华北2（北京）地域的URL。请将 {WorkspaceId} 替换为您的百炼业务空间ID，各地域的URL不同。 ``` curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/0385dc79-5ff8-4d82-bcb6-xxxxxx \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md index dbdb443a..3808f5dd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md @@ -253,13 +253,13 @@ curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/servi `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 #### 新加坡 `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md index 24b5e582..d6728f38 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md @@ -21,9 +21,14 @@ **重要** -百炼为新加坡地域推出了业务空间专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope-intl.aliyuncs.com` 迁移至新域名。 +阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 +- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` + +- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` + + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 ## HTTP调用 @@ -33,13 +38,13 @@ ## **北京** -`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` ## **新加坡** `POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -53,7 +58,7 @@ ## 纯指令编辑（修改视频风格） ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -79,7 +84,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **指令+参考图编辑（局部替换）** ``` -curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis' \ -H 'X-DashScope-Async: enable' \ -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ @@ -499,13 +504,13 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## **北京** -`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` ## **新加坡** `GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 **说明** @@ -526,10 +531,10 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/services/aigc/video-gener ## 查询任务结果 -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 +将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时，并请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ``` -curl -X GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} \ +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" ``` @@ -729,13 +734,13 @@ SDK 的参数命名与[HTTP接口](#e9e21dd3a6945)基本一致，参数结构根 ## **北京** -`dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1'` +`dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1'` ## **新加坡** `dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1'` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -749,8 +754,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -855,8 +860,8 @@ from dashscope import VideoSynthesis import dashscope import os -# 以下为北京地域URL，各地域的URL不同 -dashscope.base_http_api_url = 'https://dashscope.aliyuncs.com/api/v1' +# 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key @@ -997,13 +1002,13 @@ if __name__ == '__main__': ## **北京** -`Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1";` +`Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";` ## **新加坡** `Constants.baseHttpApiUrl = "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1";` -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 +调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 ## 同步调用 @@ -1034,8 +1039,8 @@ import java.util.Base64; public class VideoEdit { static { - // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" @@ -1184,8 +1189,8 @@ import java.util.Base64; public class VideoEdit { static { - // 以下为北京地域url，各地域的url不同 - Constants.baseHttpApiUrl = "https://dashscope.aliyuncs.com/api/v1"; + // 以下为华北2（北京）地域的URL，调用时请将 {WorkspaceId} 替换为真实的业务空间ID，各地域的URL不同。 + Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; } // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md index 732930c9..a9d58fee 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md @@ -4,7 +4,7 @@ ## **模型调优流程** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/8010204871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3396534871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) ## **步骤一：选择调优方式** @@ -776,3 +776,21 @@ Checkpoint 有保存时长限制，超过保存时长后将被自动清理，届 **说明** 您可以在模型调优任务列表中，点击训练失败任务右侧的**日志**，查看具体的训练失败原因。 + +### **微调后模型体验仍回答基座模型身份怎么办？** + +**原因说明** + +控制台**模型体验**页面不支持设置 system prompt，导致微调后的身份设定无法生效。 + +**解决方案** + +可以通过以下方式验证微调模型的身份设定效果： + +- **方案一：通过 API 调用传入 system prompt** + + 调用 API 时，在请求的 `system` 字段中填写身份设定内容，即可让模型按照微调后的设定进行回答。 + +- **方案二：在百炼控制台模型调试页面测试** + + 前往百炼控制台**模型调试**页面，在系统提示词输入框中填写身份设定内容，再进行测试。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md index ae2396d5..27d187cf 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md @@ -25,7 +25,7 @@ ### **模型调优流程** -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2700204871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7786534871/CAEQZhiBgMDg9PGS2hkiIDNlZDFiMGRlMTJhOTQ1YzJhMmNjNDM3NzQ1ZjNiOGZk4608430_20240830103738.564.svg) 详情参见： diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md index e688c82e..5c1b36a4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md @@ -115,6 +115,144 @@ curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/files' \ 请将`<替换为训练数据集的文件id>`完整替换为上一步获取的`file_id`。完整参数说明与格式约束请参见[超参数](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#5f391e4b3cezf)。 +**超参数** + +**字段** + +**类型** + +**必选** + +**描述** + +**推荐值** + +max\_steps + +int + +是 + +**训练总步数**。控制训练时长的核心参数。max\_steps 决定训练迭代次数，max\_token\_length 决定每步处理的数据量。建议不少于 500 步以确保模型充分收敛；大数据集可适当增加步数。 + +800 + +eval\_steps + +int + +是 + +**验证间隔**。取值需≥0。训练期间每隔多少个 steps 进行一次验证评估，用于阶段性评估模型训练效果。同时保存当前 step 的模型文件。 + +200 + +learning\_rate + +float + +是 + +**学习率**。控制模型权重更新的幅度。过高可能导致模型变差，过低则变化不明显。推荐使用默认值。 + +3e-5 + +generation\_type + +string + +是 + +**生成模式**。`"t2i"`：文生图模式；`"i2i"`：图生图模式。决定训练数据格式和推理方式。 + +t2i + +max\_pixels + +string + +是 + +**训练图片的最大分辨率**。例如 "1k"、"2k"（1K 即 1024×1024，2K 即 2048×2048）。设置训练集中图片分辨率的像素总数（宽×高）上限，系统仅对超过该值的图片进行缩放处理，未超限的图片保持原样。建议三个分辨率参数（max\_pixels、max\_token\_length、val\_img\_size）保持一致。 + +文生图："2k" +图生图："1k" + + +val\_img\_size + +string + +是 + +**验证图生成分辨率**。例如 "1k"、"2k"（1K 即 1024×1024，2K 即 2048×2048）。训练过程中验证评估时生成图片的目标分辨率。 + +文生图："2k" +图生图："1k" + + +max\_token\_length + +string + +是 + +**每步训练的最大 Token 长度**。例如 "1k"、"2k"。与 max\_steps 共同控制训练过程：max\_steps 决定迭代次数，max\_token\_length 决定每步处理的数据量。 + +文生图："2k" +图生图："1k" + + +gradient\_clip + +float + +是 + +**梯度裁剪**。对所有可训练参数做全局梯度范数裁剪的阈值，防止梯度爆炸。设为 -1 表示不裁剪。 + +0.5 + +weight\_decay + +float + +是 + +**权重衰减**。AdamW 解耦式权重衰减系数，对所有可训练参数生效，用于正则化防止过拟合。 + +0.02 + +lora\_rank + +int + +是 + +**LoRA 低秩矩阵的维数**。该值决定了微调参数量的大小。数值越大，模型拟合能力越强，但训练速度会变慢。取值必须为 2n（如 16、32、64）。 + +32 + +save\_total\_limit + +int + +否 + +**Checkpoint 保存数量上限**。限制最多保存的模型数量。系统将始终只保存训练生成的最后 N 个 Checkpoint（N 为该参数值）。 + +10 + +split + +float + +否 + +**训练集划分比例**。取值范围为 (0, 1)。仅在未指定 `validation_datasets` 时生效。此参数用于从训练集中自动按比例拆分出验证集。例如，0.9 表示 90% 训练集，10% 验证集。 + +0.9 + ``` curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md index 34eb3507..955a59f1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md @@ -109,10 +109,6 @@ curl --location --request POST 'https://dashscope.aliyuncs.com/api/v1/files' \ 使用步骤1中的文件ID启动训练任务。 -**说明** - -不同模型的微调参数的值有所差异，超参数设置请参见[超参数](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#5f391e4b3cezf)，更多调用示例请参见[请求示例](https://help.aliyun.com/zh/model-studio/wan-generation-finetune-api-reference#1a9196bd16o9h)。 - **请求示例** 请将`<替换为训练数据集的文件id>`完整替换为上一步获取的`file_id`。 @@ -130,12 +126,12 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ ], "training_type": "efficient_sft", "hyper_parameters": { - "n_epochs": 400, + "n_epochs": 50, "batch_size": 1, "learning_rate": 2e-5, "split": 0.9, "max_split_val_dataset_sample": 5, - "eval_epochs": 50, + "eval_epochs": 20, "max_pixels": 102400, "save_total_limit": 10, "lora_rank": 32, @@ -157,12 +153,12 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ ], "training_type": "efficient_sft", "hyper_parameters": { - "n_epochs": 400, + "n_epochs": 50, "batch_size": 4, "learning_rate": 2e-5, "split": 0.9, "max_split_val_dataset_sample": 5, - "eval_epochs": 50, + "eval_epochs": 20, "max_pixels": 262144, "save_total_limit": 10, "lora_rank": 32, @@ -171,6 +167,140 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes' \ }' ``` +**超参数（hyper\_parameters）** + +**字段** + +**类型** + +**必选** + +**描述** + +**推荐值** + +batch\_size + +int + +是 + +**批次大小**。一次性送入模型进行训练的数据条数。 + +- wan2.7-i2v：推荐为 1。 + +- wan2.5-i2v-preview：推荐为 4。 + +- wan2.2-i2v-flash：推荐为 4。 + +- wan2.2-kf2v-flash：推荐为 4。 + + +以模型为准 + +n\_epochs + +int + +是 + +**训练循环次数**。steps = n\_epochs × ⌈数据集大小 / batch\_size⌉。建议总步数 ≥ 800。 + +> 例如：数据集 5 条，batch\_size=2，每轮步数=⌈5/2⌉=3，最小 n\_epochs = 800/3 ≈ 267。 + +> 推荐训练轮数会根据数据量自动调整。数据越少，需要更多轮数来充分学习；数据越多，每轮包含的样本越多，因此所需轮数会减少。50 epochs 主要适用于 2 条左右的小数据集；当数据量达到 50-60 条视频时，通常建议训练约 3000-5000 steps 即可。 + +50 + +learning\_rate + +float + +是 + +**学习率**。控制模型权重更新幅度。过高可能导致模型变差，过低则变化不明显。 + +2e-5 + +eval\_epochs + +int + +是 + +**验证间隔**。取值需 ≥ `n_epochs/10`。每隔多少个 epoch 进行一次验证评估并保存 Checkpoint。 + +20 + +max\_pixels + +int + +是 + +**训练视频的最大分辨率**（像素总数 = 宽×高）。系统仅对超过该值的视频进行缩放处理。 + +- wan2.7-i2v：推荐 102400。范围 36864～123904。 + +- wan2.5-i2v-preview：推荐 36864。范围 16384～36864。 + +- wan2.2-i2v-flash：推荐 262144。范围 65536～262144。 + +- wan2.2-kf2v-flash：推荐 262144。范围 65536～262144。 + + +以模型为准 + +split + +float + +否 + +**训练集划分比例**。取值 (0,1)，仅在未指定 validation\_datasets 时生效。 + +0.9 + +max\_split\_val\_dataset\_sample + +int + +否 + +**自动划分验证集的最大样本数**。验证集数量 = min(总数×(1−split), 此值)。 + +5 + +save\_total\_limit + +int + +否 + +**Checkpoint 保存数量上限**。系统只保存最后 N 个 Checkpoint。 + +10 + +lora\_rank + +int + +否 + +**LoRA 低秩矩阵维数**。取值须为 2n（16/32/64）。 + +32 + +lora\_alpha + +int + +否 + +**LoRA 权重缩放系数**。取值须为 2n（16/32/64）。 + +32 + **响应示例** 关注 `output` 中的三个关键参数： @@ -1276,7 +1406,7 @@ curl --location 'https://dashscope.aliyuncs.com/api/v1/fine-tunes/<替换为微 - **n\_epochs (训练轮数)** - - 默认值：**400**，推荐使用默认值。若需调整，请遵循 **“总训练步数 (Steps) ≥ 800”** 的原则。 + - 默认值：**50**，推荐使用默认值。若需调整，请遵循 **”总训练步数 (Steps) ≥ 800”** 的原则。 - 总步数计算公式： `steps = n_epochs × 向上取整（训练集大小 / batch_size）。` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md index bb37de23..895966ad 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/models.md @@ -12,6 +12,16 @@ [ +qwen3.8-max-preview + +](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +仅 Token Plan 可用 + +qwen3.8-max-preview 目前仅面向 Token Plan 订阅用户提供，[前往开通 Token Plan →](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +[ + qwen3.7-max ](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max) @@ -558,92 +568,20 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/a [ -kimi-k2.7-code +kimi/kimi-k3 -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi-k2.7-code) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi%2Fkimi-k3?serviceSite=asia-pacific-china) -华北2（北京）新加坡德国（法兰克福）美国（弗吉尼亚） +华北2（北京） -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 -模型 ID`kimi-k2.7-code` +模型 ID`kimi/kimi-k3` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - [ glm-5.2 @@ -888,7 +826,7 @@ qwen3.5-omni-plus 华北2（北京）新加坡 -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 模型 ID`qwen3.5-omni-plus` @@ -896,19 +834,7 @@ Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?t API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 模型 ID`qwen3.5-omni-plus` @@ -916,106 +842,22 @@ Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-south API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`qwen3.5-omni-plus` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - [ -kimi-k2.7-code +kimi/kimi-k3 -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi-k2.7-code) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/kimi%2Fkimi-k3?serviceSite=asia-pacific-china) -华北2（北京）新加坡德国（法兰克福）美国（弗吉尼亚） +华北2（北京） -OpenAI 兼容Anthropic 兼容DashScope +OpenAI 兼容 -模型 ID`kimi-k2.7-code` +模型 ID`kimi/kimi-k3` Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/compatible-mode/v1` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/eu-central-1?tab=globalset#/efm/business_management).eu-central-1.maas.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/eu-central-1?tab=model#/api-key) - -OpenAI 兼容Anthropic 兼容DashScope - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/apps/anthropic` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - -模型 ID`kimi-k2.7-code` - -Base URL`https://dashscope-us.aliyuncs.com/api/v1` - -API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/api-key) - ### 生成 通过文本或图片生成图像与视频，支持编辑、参考与高分辨率输出 @@ -1024,41 +866,41 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/us-east-1?tab=model#/a [ -wan2.7-image-pro +qwen-image-3.0-pro -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/wan2.7-image-pro) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-image-3.0-pro) 华北2（北京）新加坡 -模型 ID`wan2.7-image-pro` +模型 ID`qwen-image-3.0-pro` -Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` +Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`wan2.7-image-pro` +模型 ID`qwen-image-3.0-pro` -Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` +Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) [ -qwen-image-2.0-pro +wan2.7-image-pro -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-image-2.0-pro) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/wan2.7-image-pro) 华北2（北京）新加坡 -模型 ID`qwen-image-2.0-pro` +模型 ID`wan2.7-image-pro` -Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` +Request URL`https://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen-image-2.0-pro` +模型 ID`wan2.7-image-pro` -Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` +Request URL`https://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) @@ -1258,18 +1100,24 @@ API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api- [ -cosyvoice-v3.5-plus +qwen-audio-3.0-tts-plus -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/cosyvoice-v3.5-plus) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-audio-3.0-tts-plus) -华北2（北京） +华北2（北京）新加坡 -模型 ID`cosyvoice-v3.5-plus` +模型 ID`qwen-audio-3.0-tts-plus` Request URL`wss://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api-ws/v1/inference` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) +模型 ID`qwen-audio-3.0-tts-plus` + +Request URL`wss://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api-ws/v1/inference` + +API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) + [ MiniMax/speech-2.8-hd @@ -1426,24 +1274,18 @@ API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=mod [ -qwen3.5-omni-plus-realtime +qwen-audio-3.0-realtime-plus -](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.5-omni-plus-realtime) +](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen-audio-3.0-realtime-plus) -华北2（北京）新加坡 +华北2（北京） -模型 ID`qwen3.5-omni-plus-realtime` +模型 ID`qwen-audio-3.0-realtime-plus` Request URL`wss://[{WorkspaceId}](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management).cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime` API Key[获取↗](https://bailian.console.aliyun.com/cn-beijing?tab=model#/api-key) -模型 ID`qwen3.5-omni-plus-realtime` - -Request URL`wss://[{WorkspaceId}](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=globalset#/efm/business_management).ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime` - -API Key[获取↗](https://modelstudio.console.aliyun.com/ap-southeast-1?tab=model#/api-key) - [ qwen3.5-omni-plus diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md index 8e375ee1..fbd2a16b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/rate-limit.md @@ -665,6 +665,14 @@ qwen3.6-flash-2026-04-16 1,000,000 +qwen3.6-flash-us + +美国 + +15,000 + +5,000,000 + qwen3.5-plus 全球 @@ -2161,17 +2169,17 @@ qwen-vl-ocr-latest 中国内地 -1,200 +6,000 -6,000,000 +30,000,000 qwen-vl-ocr-2025-11-20 中国内地 -1,200 +6,000 -6,000,000 +30,000,000 qwen-vl-ocr-2025-08-28 @@ -4695,17 +4703,21 @@ kimi-k2.7-code > **含输入与输出Token** -kimi/kimi-k2.7-code-highspeed +kimi/kimi-k3 中国内地 500 -> 同一个阿里云百炼API Key 下，在 4 个模型中共享 500 RPM 限流配额。即这 4 个模型的每分钟请求总数加起来不能超过 500。 +> 同一个阿里云百炼API Key 下，在 5 个模型中共享 500 RPM 限流配额。即这 5 个模型的每分钟请求总数加起来不能超过 500。 3,000,000 -> 同一个阿里云百炼API Key 下，在 4 个模型中共享 3000000 TPM 限流配额。即这 4 个模型的每分钟 Token 消耗总数加起来不能超过 3000000。 +> 同一个阿里云百炼API Key 下，在 5 个模型中共享 3000000 TPM 限流配额。即这 5 个模型的每分钟 Token 消耗总数加起来不能超过 3000000。 + +kimi/kimi-k2.7-code-highspeed + +中国内地 kimi/kimi-k2.7-code @@ -4907,7 +4919,15 @@ glm-5.1 glm-5.2 -全球 +国际 + +500 + +1,000,000 + +glm-5.1 + +国际 500 @@ -4945,7 +4965,7 @@ ZHIPU/GLM-5.1 200 -10,000,000 +3,000,000 ZHIPU/GLM-5 @@ -5107,6 +5127,14 @@ stepfun/step-3.7-flash **同时处理中任务数量（并发数）** +qwen-image-3.0-pro + +中国内地 + +1 次/分钟 + +同步接口无限制 + qwen-image-2.0-pro 中国内地 @@ -5263,6 +5291,14 @@ qwen-mt-image **同时处理中任务数量（并发数）** +qwen-image-3.0-pro + +国际 + +1 次/分钟 + +同步接口无限制 + qwen-image-2.0-pro 国际 @@ -5273,7 +5309,7 @@ qwen-image-2.0-pro qwen-image-2.0-pro-2026-06-22 -中国内地 +国际 2 次/分钟 @@ -7223,7 +7259,7 @@ happyhorse-1.1-t2v 中国内地 -10 +5 5 @@ -7231,7 +7267,7 @@ happyhorse-1.1-i2v 中国内地 -10 +5 5 @@ -7239,7 +7275,7 @@ happyhorse-1.1-r2v 中国内地 -10 +5 5 @@ -7247,7 +7283,7 @@ happyhorse-1.0-t2v 中国内地 -10 +5 5 @@ -7255,7 +7291,7 @@ happyhorse-1.0-i2v 中国内地 -10 +5 5 @@ -7263,7 +7299,7 @@ happyhorse-1.0-r2v 中国内地 -10 +5 5 @@ -7271,7 +7307,7 @@ happyhorse-1.0-video-edit 中国内地 -10 +5 5 @@ -7291,7 +7327,7 @@ happyhorse-1.1-t2v 全球 -10 +5 5 @@ -7299,7 +7335,7 @@ happyhorse-1.1-i2v 全球 -10 +5 5 @@ -7307,7 +7343,7 @@ happyhorse-1.1-r2v 全球 -10 +5 5 @@ -7315,7 +7351,7 @@ happyhorse-1.0-t2v 全球 -10 +5 5 @@ -7323,7 +7359,7 @@ happyhorse-1.0-i2v 全球 -10 +5 5 @@ -7331,7 +7367,7 @@ happyhorse-1.0-r2v 全球 -10 +5 5 @@ -7339,7 +7375,7 @@ happyhorse-1.0-video-edit 全球 -10 +5 5 @@ -7359,7 +7395,7 @@ happyhorse-1.1-t2v 国际 -10 +5 5 @@ -7367,7 +7403,7 @@ happyhorse-1.1-i2v 国际 -10 +5 5 @@ -7375,7 +7411,7 @@ happyhorse-1.1-r2v 国际 -10 +5 5 @@ -7383,7 +7419,7 @@ happyhorse-1.0-t2v 国际 -10 +5 5 @@ -7391,7 +7427,7 @@ happyhorse-1.0-i2v 国际 -10 +5 5 @@ -7399,7 +7435,7 @@ happyhorse-1.0-r2v 国际 -10 +5 5 @@ -7407,7 +7443,7 @@ happyhorse-1.0-video-edit 国际 -10 +5 5 @@ -7427,7 +7463,7 @@ happyhorse-1.1-t2v 全球 -10 +5 5 @@ -7435,7 +7471,7 @@ happyhorse-1.1-i2v 全球 -10 +5 5 @@ -7443,7 +7479,7 @@ happyhorse-1.1-r2v 全球 -10 +5 5 @@ -7451,7 +7487,7 @@ happyhorse-1.0-t2v 全球 -10 +5 5 @@ -7459,7 +7495,7 @@ happyhorse-1.0-i2v 全球 -10 +5 5 @@ -7467,7 +7503,7 @@ happyhorse-1.0-r2v 全球 -10 +5 5 @@ -7475,7 +7511,7 @@ happyhorse-1.0-video-edit 全球 -10 +5 5 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md index d9c3ff39..775296d9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/get-started-with-models/regions.md @@ -18,7 +18,7 @@ 3. 推理结果回到接入地域存储，再响应给应用（用户静态数据始终存于所选地域）。 -![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/7478182871/CAEQchiBgIDPq_PR9xkiIDJhZDdiNzAxMGFiODRhNmRiMDYxYjNjNGU2NTJkMDYw7466796_20260515102254.505.svg) +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5407814871/CAEQchiBgIDPq_PR9xkiIDJhZDdiNzAxMGFiODRhNmRiMDYxYjNjNGU2NTJkMDYw7466796_20260515102254.505.svg) ## 选择地域和服务部署范围 @@ -104,7 +104,7 @@ 推荐在生产环境中使用，具备更高并发承载能力与网络隔离性，保障大流量场景下的稳定、低延迟访问体验。 -存量业务兼容，建议[迁移至业务空间专属域名](#section-migrate-domain)。 +存量业务兼容，建议迁移至[业务空间专属域名](#section-migrate-domain)。 快速体验、功能验证，不建议用于生产环境。 @@ -247,13 +247,9 @@ HTTP、SSE 从 Dashscope 域名或试用域名迁移到业务空间专属域名只需两步，无需修改业务逻辑代码： -1. **获取业务空间专属域名**： +1. **获取业务空间专属域名**：在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)页面，复制 **API Host** 列的内容。 - - 方式一：在[API Key 创建](https://bailian.console.aliyun.com/cn-beijing#/api-key)后的弹窗中，复制 **API Host** 。 - - - 方式二：在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)页面，复制 **API Host** 列的内容。 - -2. **替换 Base URL 中的域名**：将原域名替换为业务空间专属域名。以华北2（北京）地域为例，`llm-xxx` 为业务空间 ID： +2. **替换请求地址中的域名**：将复制的 API Host（如 `llm-xxx.cn-beijing.maas.aliyuncs.com`）替换代码中原有的域名部分，以华北2（北京）地域为例，`llm-xxx` 为业务空间 ID： - OpenAI 兼容接口：从 `https://dashscope.aliyuncs.com/compatible-mode/v1` 替换为 `https://llm-xxx.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md new file mode 100644 index 00000000..6d7dae69 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-data-overview/model-log-backflow.md @@ -0,0 +1,330 @@ +# 日志回流 + +日志回流将 SLS（日志服务）推理日志转化为可用于模型微调或评测的结构化数据集。 + +## 功能概述 + +日志回流功能将 SLS（日志服务）中的推理日志数据回流到百炼平台，经过格式化处理后生成结构化数据集（JSONL 格式），可用于模型微调或模型评测。回流产出的是结构化数据，而非原始日志的直接副本。 + +### 支持范围 + +日志回流支持创建以下两类数据集： + +- **训练集**：训练场景为文本生成，训练方式支持 SFT（监督微调）、DPO（直接偏好优化）和 CPT（持续预训练）。 + +- **评测集**：支持文本生成场景。 + + +日志回流目前仅在**华北2（北京）**和**新加坡** Region 可用，其他 Region 不显示日志回流入口。单次回流上限为 10 万条，可多次回流到同一数据集的不同版本以积累更多数据。 + +日志回流支持**平台存储**（默认）和对象存储 **OSS 挂载**（需额外授权）两种存储方式。存储方式的差异和选择指引，请参见[创建日志回流数据集](#sec-create)。 + +回流生成的数据集可直接用于下游任务：训练集可用于[模型调优](https://help.aliyun.com/zh/model-studio/model-training-on-console#topic-2529531)，评测集用于模型评测。数据也支持后续进行[数据清洗](https://help.aliyun.com/zh/model-studio/data-processing#topic-2761062)。 + +## 开通并授权相关服务 + +使用日志回流前，请确认当前 Region 为华北2（北京）或新加坡（其他 Region 不显示日志回流入口），并在[模型监控](https://bailian.console.aliyun.com/#/model-telemetry)页面完成以下服务开通和权限授权。全部完成后授权配置抽屉自动关闭，进入日志回流表单。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086527.png) + +### 开启审计日志和推理日志 + +审计日志和推理日志各需完成三个步骤，共六个条件全部满足后才能使用日志回流。审计日志是推理日志的前置依赖，必须先完成审计日志的全部步骤。 + +**审计日志** + +1. 授权 SLS 服务关联角色：单击**立即授权**，授权 AliyunServiceRoleForSFMAccessSLS 角色。未授权时显示红色**未授权**标签。 + +2. 开通 SLS 日志服务：未开通时显示**未开通**状态和跳转链接，单击后前往 SLS 控制台完成开通。 + +3. 开启审计日志：单击**创建并开启审计日志**，系统创建 LogStore（日志库）实例并轮询等待就绪（最多 60 秒）。 + + +**推理日志** + +1. 授权推理日志的 SLS 服务关联角色。 + +2. 确认 SLS 日志服务已开通。 + +3. 开启推理日志：审计日志未开启时该按钮置灰，需先完成审计日志开启。 + + +**重要** + +开启必须按序：先审计日志，再推理日志。关闭必须反序：先关闭推理日志，再关闭审计日志。关闭日志后已有数据不可复原，操作前请确认不再需要日志数据。推理日志开启后 SLS 持续产生存储和读写费用，不再需要时应及时关闭。 + +### OSS 多角色授权（仅 OSS 挂载模式） + +选择 OSS 挂载存储方式时，需额外完成以下授权： + +1. 在弹出的授权弹窗中，勾选数据访问授权协议。 + +2. 单击**一键授权**，系统自动授权以下两个服务关联角色： + + - AliyunServiceRoleForAccessCusOss（OSS 写入角色） + + - AliyunServiceRoleForSFMDataHubOSSImport（DataHub 导入角色） + + +下表汇总了日志回流涉及的服务关联角色。 + +**角色名称** + +**用途** + +**授权时机** + +AliyunServiceRoleForSFMAccessSLS + +百炼访问 SLS 日志数据 + +审计日志和推理日志各授权一次 + +AliyunServiceRoleForAccessCusOss + +OSS 数据写入 + +选择 OSS 挂载存储时 + +AliyunServiceRoleForSFMDataHubOSSImport + +DataHub 数据导入 + +选择 OSS 挂载存储时 + +## 创建日志回流数据集 + +日志回流提供三个入口，均可进入配置表单创建数据集： + +### 模型监控列表页 + +在[**模型监控**列表页](https://bailian.console.aliyun.com/#/model-telemetry)顶部，单击**日志回流**。首次使用时先完成授权配置，授权通过后自动展示日志回流表单。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086529.png) + +### 模型监控详情页 + +在模型监控详情页的时间选择器区域，单击**日志回流**。从此入口进入时，表单自动预填当前页面的时间范围、API Key 和模型，且模型不可修改。 + +### 数据管理页 + +在**数据管理**页面新建数据集时，导入方式选择**日志回流**。该选项仅在训练集（文本生成 + SFT）或评测集（文本生成）时可见。 + +### 配置回流参数 + +进入日志回流表单后，按从上到下的顺序配置以下参数。部分参数有前置依赖：API Key 过滤需先选时间范围，模型选择需先选时间范围和 API Key。修改时间范围、数据类型、训练场景或训练方式会联动重置其他参数，建议严格按顺序填写。各参数取值说明见下表。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1086595.png) + +**预估日志回流数据**：系统根据筛选条件显示预估的回流数据条数。超过 10 万条时红色警告，超出部分不会被回流。查询结果过多时，**确定**按钮禁用，需缩小筛选范围。 + +**说明** + +表单联动重置规则：修改时间范围会重置 API Key（回到**全部**）和模型选择（清空）；修改数据类型、训练场景或训练方式会重置存储位置和导入方式。 + +**重要** + +存储方式、数据类型和训练方式在创建后均不可更改，选择前请仔细确认。 + +**参数** + +**说明** + +**是否必填** + +**取值说明** + +回流位置 + +数据集的存储方式 + +是 + +平台存储（默认）或 OSS 挂载。评测集时 OSS 挂载禁用。创建后不可更改 + +数据集名称 + +数据集在列表中的显示名称 + +是 + +中文、英文、数字、下划线、斜杠、连字符，最大 50 字符。建议采用 功能场景\_模型名\_时间 格式命名。创建后不可修改 + +数据集描述 + +数据集用途补充说明 + +否 + +最大 200 字符 + +类型与格式 + +数据集用途类型 + +是 + +训练集或评测集。选择评测集时训练场景和训练方式隐藏。创建后不可更改 + +训练场景 + +训练场景类型（仅训练集显示） + +是 + +当前仅支持文本生成 + +训练方式 + +微调方法（仅训练集显示） + +是 + +SFT、DPO、CPT，选项由系统动态展示。创建后锁定 + +时间范围 + +回流日志的时间段 + +是 + +最近 30 天（含当天），精确到时分秒。修改会重置 API Key 和模型选择 + +API Key 过滤 + +按 API Key 筛选日志数据 + +是 + +全部（不过滤）、其他（排除已列出 Key）、或选择具体 Key（多选） + +模型选择 + +回流目标模型 + +是 + +最多 10 个，按能力类型过滤不匹配模型置灰 + +OSS 文件夹路径 + +数据存储的目标目录（仅 OSS 挂载模式显示） + +是 + +Bucket 需在同一 Region + +## 查看回流结果 + +提交日志回流任务后，在[**数据管理**列表页](https://bailian.console.aliyun.com/#/efm/data_ass)查看数据集和导入进度。 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3463064871/p1088579.png) + +### 列表页展示 + +日志回流创建的数据集在列表页中的导入方式显示为**日志回流**，存储位置根据创建时的选择显示为**平台存储**或**OSS 挂载**。 + +可在列表页查看任务的导入状态。任务失败时，可查看系统返回的具体失败原因。 + +平台存储模式下，导入完成后系统自动发布数据集版本，无需手动操作。 + +### 详情页信息 + +数据集详情页展示的信息根据存储方式有所不同： + +- **OSS 挂载**：展示发布状态、数据量、创建时间、FileID、数据类型、导入状态和 OSS 挂载地址。 + +- **平台存储（OSS 导入）**：展示发布状态、数据量、创建时间、FileID、数据类型、导入状态和 OSS 导入地址。 + +- 其他情况：展示发布状态、数据量、创建时间、FileID、数据类型和导入状态。 + + +## 追加日志回流数据 + +在[数据管理](https://bailian.console.aliyun.com/#/efm/data_ass)页面，向已有数据集追加新一批日志回流数据，有以下两种方式： + +### 导入数据页 + +进入已有数据集的**导入数据**页面，选择**日志回流**作为导入方式。表单参数与创建流程一致，详见[创建日志回流数据集](#sec-create)。此方式额外支持按工作空间过滤。 + +此方式适用于所有存储类型的数据集，包括 OSS 挂载数据集。 + +### 新增版本弹窗 + +在数据集详情页，单击**新增版本**，在弹窗中选择日志回流导入方式。仅平台存储数据集可用此方式。 + +**说明** + +OSS 挂载数据集不支持**新增版本**操作，按钮置灰。需通过**导入数据**页追加。 + +### 增量回流最佳实践 + +推荐分批次回流，针对不同时段或不同模型分别执行回流任务，逐步积累高质量训练集。每批次可精准选择表现良好的模型和业务高峰时段的数据，确保数据质量优于一次性大量回流。 + +## 常见问题 + +**API Key 过滤的「其他」包含哪些请求？** + +**问题**：API Key 过滤中**其他**包含哪些请求？ + +**其他**过滤掉已列出 Key 的请求，包含已删除的 Key 或其他工作空间产生的日志。**全部**则不按 Key 过滤。三种模式（全部/其他/逐个选择）的完整说明见[创建日志回流数据集](#sec-create)的 API Key 过滤参数。 + +**10 万条上限是数据集总量限制吗？** + +**问题**：单次回流上限 10 万条，是否意味着数据集总量也被限制在 10 万条？ + +不是。10 万条是单次回流的上限，并非数据集的总量上限。可以多次回流到同一数据集的不同版本，累计数据量不受此限制。例如分批次针对不同时段回流，每次 10 万条以内，最终数据集可以积累远超 10 万条的数据。 + +**平台存储和 OSS 挂载有什么区别？** + +**问题**：两种存储方式除了数据存储位置不同，还有哪些差异？ + +两种存储方式的核心差异对比如下。 + +**维度** + +**平台存储** + +**OSS 挂载** + +额外授权 + +无需额外授权 + +需授权两个服务关联角色并勾选数据访问协议 + +新增版本 + +支持 + +不支持（按钮置灰），需通过导入数据页追加 + +数据访问 + +通过控制台访问 + +可在 OSS Bucket 直接查看和管理 JSONL 文件 + +评测集 + +支持 + +不支持（选项禁用） + +自动发布 + +导入完成后自动发布版本 + +不自动发布 + +**审计日志和推理日志是同一个操作吗？** + +**问题**：开启日志需要分别操作审计日志和推理日志，两者有什么关系？ + +两者需分别开启，且必须先完成审计日志才能开启推理日志。操作步骤详见[开通并授权相关服务](#sec-authorize)。 + +**预估数据量为什么和实际回流结果不一致？** + +**问题**：提交前显示的预估数据量与回流完成后的实际条数有差异。 + +预估数据量是基于筛选条件的近似估算值，实际回流完成后的数据条数可能略有差异，这是正常现象。预估值用于帮助判断是否需要调整筛选条件（如超过 10 万条时提示缩小范围），不代表精确计数。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md index fbf3abc6..5b3f64e9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/model-deployment-introduction.md @@ -108,7 +108,7 @@ 1. 预付费按天计费。无法提前退费 -2. 如果单位时间内使用超出购买的吞吐量，将自动切换成百炼提供的[模型调用](https://help.aliyun.com/zh/model-studio/model-pricing)服务。 +2. 如果单位时间内使用超出购买的吞吐量，按创建时选择的溢出策略处理：自动溢出则切换为该模型的[模型调用](https://help.aliyun.com/zh/model-studio/model-pricing)按量付费，仅使用 PTU 容量则返回 429。 预付费购买后，若在首月内提前退订，日单价（≈ 月单价 / 30）将按 **1.2** 倍计费 @@ -137,9 +137,9 @@ - 后付费时，如果账户欠费，部署的资源将继续保留并计费 24 小时，在这 24 小时内服务仍可正常使用。超过 24 小时后系统停止计费，模型部署进入欠费状态，底层资源将被删除，但模型部署任务仍会保留。补足欠费后，系统将重新分配资源并恢复使用（恢复后继续产生费用）。如果您不希望继续产生费用，可删除模型部署任务，删除成功后将不再计费。 -当模型输入超过最长输入 Token 或 超出购买的 TPM 量时，相关调用将自动切换为当前模型的按量付费模式。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 +当模型输入超过最长输入 Token 时，相关调用将自动切换为当前模型的按量付费模式；超出购买的 TPM 量时，按创建时选择的溢出策略处理（「自动溢出」切换为按量付费，「仅使用 PTU 容量」返回 429）。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 -- 此时，调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 +- 此时（仅「自动溢出」策略下），调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 - TPM 统计请前往：[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)。 @@ -483,15 +483,33 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-35B-A3B qwen3.6-35b-a3b -MU8 x 1 +MU1 x 8 -¥47 +¥432 -¥22,400 +¥208,944 + +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 MU9 x 1 @@ -509,6 +527,12 @@ MU1 x 2 ¥52,236 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-Plus-2026-04-02 qwen3.6-plus-2026-04-02 @@ -529,12 +553,6 @@ PD分离模式：¥417,888 qwen3.5-397b-a17b -MU2 x 8 - -¥504 - -¥240,288 - MU3 x 8 MU3 x 16（PD分离模式） @@ -547,6 +565,12 @@ PD分离模式：¥2,192 PD分离模式：¥1,055,504 +MU6 x 16 + +¥400 + +¥193,424 + 千问3.5-122B-A10B qwen3.5-122b-a10b @@ -557,11 +581,17 @@ MU1 x 4 ¥104,472 -MU2 x 8 +MU3 x 8 -¥504 +¥1,096 -¥240,288 +¥527,752 + +MU6 x 16 + +¥400 + +¥193,424 千问3.5-35B-A3B @@ -579,10 +609,40 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + +MU9 x 1 + +¥51 + +¥24,600 + 千问3.5-27B qwen3.5-27b +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 + +MU8 x 1 + +¥47 + +¥22,400 + MU9 x 1 ¥51 @@ -599,6 +659,12 @@ MU1 x 2 ¥52,236 +MU2 x 2 + +¥126 + +¥60,072 + MU8 x 1 ¥47 @@ -625,12 +691,24 @@ MU1 x 2 qwen3.5-plus-2026-02-15 +MU1 x 8 + MU1 x 16（PD分离模式） +¥432 + PD分离模式：¥864 +¥208,944 + PD分离模式：¥417,888 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 8 MU3 x 16（PD分离模式） @@ -659,57 +737,25 @@ MU2 x 8 ¥240,288 -千问3-Next-80B-A3B-Instruct - -qwen3-next-80b-a3b-instruct - -MU1 x 2 - -¥108 - -¥52,236 - 千问3-32B qwen3-32b -MU1 x 4 - -¥216 - -¥104,472 - -MU6 x 4 - -¥100 - -¥48,356 - -千问3-30B-A3B - -qwen3-30b-a3b - -MU9 x 2 - -¥102 - -¥49,200 - -千问3-30B-A3B-Instruct-2507 +MU6 x 16 -qwen3-30b-a3b-instruct-2507 +¥400 -MU1 x 4 +¥193,424 -¥216 +千问3-30B-A3B-Thinking-2507 -¥104,472 +qwen3-30b-a3b-thinking-2507 -MU2 x 8 +MU1 x 2 -¥504 +¥108 -¥240,288 +¥52,236 千问3-8B @@ -759,12 +805,6 @@ MU1 x 2 ¥52,236 -MU5 x 1 - -¥21 - -¥10,139 - 千问3-Embedding-0.6B qwen3-embedding-0.6b @@ -837,11 +877,11 @@ MU5 x 1 qwen2.5-72b-instruct -MU1 x 4 +MU1 x 8 -¥216 +¥432 -¥104,472 +¥208,944 千问2.5-开源版-32B @@ -879,26 +919,6 @@ MU5 x 1 ¥10,139 -千问2.5-开源版-3B - -qwen2.5-3b-instruct - -MU5 x 1 - -¥21 - -¥10,139 - -千问-Flash-2025-07-28 - -qwen-flash-2025-07-28 - -MU1 x 4 - -¥216 - -¥104,472 - 千问-Plus-2025-07-28 qwen-plus-2025-07-28 @@ -955,6 +975,12 @@ GLM-5.1 glm-5.1 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 16（PD分离模式） PD分离模式：¥2,192 @@ -987,6 +1013,16 @@ PD分离模式：¥800 PD分离模式：¥386,848 +GLM-4.7-Flash + +glm-4.7-flash + +MU3 x 16（PD分离模式） + +PD分离模式：¥2,192 + +PD分离模式：¥1,055,504 + #### DeepSeek **模型名称** @@ -1007,11 +1043,11 @@ DeepSeek-v4-Flash deepseek-v4-flash -MU1 x 8 +MU3 x 8 -¥432 +¥1,096 -¥208,944 +¥527,752 DeepSeek-v3.2 @@ -1039,16 +1075,6 @@ PD分离模式：¥480,576 **最小计费：天** -MiniMax-M2.5 - -MiniMax-M2.5 - -MU1 x 16（PD分离模式） - -PD分离模式：¥864 - -PD分离模式：¥417,888 - Kimi-K2.5 kimi-k2.5 @@ -1091,15 +1117,21 @@ MU2 x 8 **最小计费：天** -千问3-VL-235B-A22B-Instruct +千问3-VL-32B-Instruct -qwen3-vl-235b-a22b-instruct +qwen3-vl-32b-instruct -MU1 x 4 +MU2 x 8 -¥216 +¥504 -¥104,472 +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 千问3-VL-8B-Instruct @@ -1111,6 +1143,12 @@ MU1 x 2 ¥52,236 +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-4B-Instruct qwen3-vl-4b-instruct @@ -1131,6 +1169,16 @@ MU5 x 1 ¥10,139 +千问3-VL-Embedding-2B + +qwen3-vl-embedding-2b + +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-Flash-2025-10-15 qwen3-vl-flash-2025-10-15 @@ -1161,16 +1209,6 @@ MU6 x 4 ¥48,356 -千问VL-OCR-2025-11-20 - -qwen-vl-ocr-2025-11-20 - -MU6 x 4 - -¥100 - -¥48,356 - #### 千问 Omni **模型名称** @@ -1251,6 +1289,14 @@ MU5 **元/千Token** +千问3.5-27B（邀测中） + +qwen3.5-27b + +¥0.0018 + +¥0.0048 + 千问3-32B qwen3-32b @@ -1281,6 +1327,16 @@ qwen3-8b 思考模式：¥0.005 +千问3-4B-Instruct-2507 + +qwen3-4b-instruct-2507 + +¥0.0003 + +非思考模式：¥0.0012 + +思考模式：¥0.003 + 千问2.5-开源版-72B qwen2.5-72b-instruct @@ -1313,6 +1369,14 @@ qwen2.5-7b-instruct ¥0.001 +千问2-开源版-7B + +qwen2-7b-instruct + +¥0.001 + +¥0.002 + #### 千问VL **基础模型** @@ -1359,6 +1423,14 @@ qwen2.5-vl-7b-instruct ¥0.005 +千问2.5-VL-3B-Instruct + +qwen2.5-vl-3b-instruct + +¥0.0012 + +¥0.0036 + 如果需要部署更多模型，请参考此[解决方案](https://www.aliyun.com/solution/tech-solution/deepseek-r1-for-platforms)并结合具体业务需求选择最适合的部署方案。 ## 部署方法 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md index 37f2cbeb..230abbfc 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md @@ -1,6 +1,6 @@ # 预置吞吐长输入与缓存 -本文介绍 PTU（预置吞吐）部署的长输入和前缀缓存能力，包括额度消耗规则、容量计算器使用方法和 API 响应字段说明。 +本文介绍 PTU（预置吞吐）部署的长输入和前缀缓存能力，包括额度消耗规则、**预置吞吐额度计算器**使用方法和 API 响应字段说明。 ## 功能概述 @@ -12,12 +12,12 @@ PTU 部署支持长输入请求（部分模型最高 200K token）和前缀缓 - 前缀缓存优惠：部分模型支持前缀缓存，命中缓存的输入 token 按折扣系数消耗额度（具体折扣率因模型而异），可降低多轮对话和重复前缀场景的额度消耗。 -- 自动转按量计费：超出 PTU 额度或输入超过模型上限（千问 128K / DeepSeek 64K）时，请求自动转为按量计费，无需修改调用代码。 +- 溢出策略：创建 PTU 时可选——自动溢出至按量计费（默认，业务不中断）或仅使用 PTU 容量（超出返回 429、不产生额外费用）。输入超过模型上限（千问 128K / DeepSeek 64K）仍自动转为按量计费。 **重要** -自动转按量计费后，费用按对应模型的按量付费单价计算。建议通过容量计算器合理规划 PTU 额度，避免意外费用。 +自动溢出策略下转为按量计费后，费用按对应模型的按量付费单价计算（仅使用 PTU 容量策略下超出返回 429、不产生额外费用）。建议通过**预置吞吐额度计算器**合理规划 PTU 额度，避免意外费用。 常见于长文档分析（合同、研报摘要）和多轮对话（客服、编程助手）等输入超 32K token 的场景。 @@ -44,6 +44,8 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 + + deepseek-v4-pro @@ -87,17 +89,17 @@ qwen3.7-plus-2026-05-26 输入合计 = 31.940 KTPM（比无缓存节省 43%） ``` -## 使用容量计算器估算额度 +## 使用**预置吞吐额度计算器**估算额度 **说明** 建议在创建或扩容前使用计算器评估长输入场景的额度需求，避免额度不足导致请求转为按量计费。购买上限以控制台实际展示为准。 -前提条件：已开通百炼服务并具备 PTU 部署权限。登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_deploy/create)，在**模型部署** > **创建部署**页面（或在已有部署详情页单击**扩容**），选择可部署的PTU（预置吞吐）模型后，展开**容量计算器**。 +前提条件：已开通百炼服务并具备 PTU 部署权限。登录[百炼控制台](https://bailian.console.aliyun.com/#/efm/model_deploy/create)，在**模型部署** > **创建部署**页面（或在已有部署详情页单击**扩容**），选择可部署的PTU（预置吞吐）模型后，展开**预置吞吐额度计算器**。 ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4645961871/p1082157.png) -容量计算器根据业务负载自动推荐 TPM 额度。填写以下参数后，计算器输出推荐的输入 TPM 和输出 TPM。 +**预置吞吐额度计算器**根据业务负载自动推荐额度。填写以下参数后，计算器输出建议购买的输入 KTPM 和输出 KTPM。 **参数** @@ -109,25 +111,25 @@ qwen3.7-plus-2026-05-26 业务高峰期每分钟的请求数。 -RPM 越大，建议购买的输入和输出 TPM 同比增大。 +RPM 越大，建议购买输入 KTPM 和输出 KTPM 同比增大。 平均输入长度（token） 每条请求的平均输入 token 数。 -输入越长，所处阶梯越高，系数越大，建议购买的输入 TPM 越高。不同模型的阶梯边界不同，以控制台实际展示为准。 +输入越长，所处阶梯越高，系数越大，建议购买输入 KTPM 越高。不同模型的阶梯边界不同，以控制台实际展示为准。 平均输出长度（token） 每条请求的平均输出 token 数。 -输出越长，系数可能越大，建议购买的输出 TPM 越高。 +输出越长，系数可能越大，建议购买输出 KTPM 越高。 -预估缓存命中率（%） +缓存命中率（%） 请求中重复前缀被缓存命中的比例。实际命中率取决于请求内容的重复程度，以运行结果为准。 -命中率越高，输入容量消耗越慢，建议购买的输入 TPM 越低。仅影响输入 TPM，不影响输出 TPM。 +命中率越高，输入容量消耗越慢，建议购买输入 KTPM 越低。仅影响输入 KTPM，不影响输出 KTPM。 ## API 响应字段说明 @@ -271,7 +273,7 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 - Token 用量与缓存命中：包含 `cached_tokens` 数据系列，可查看缓存命中量占总输入的比例。 -- 配额内/外调用次数：了解超出 PTU 额度后转为按量计费的请求占比。 +- 配额内/外调用次数：了解超出 PTU 额度后的请求占比（自动溢出策略下转为按量计费，仅使用 PTU 容量策略下返回 429）。 更多监控指标和操作方式，请参见[模型监控](https://help.aliyun.com/zh/model-studio/model-telemetry)。 @@ -280,7 +282,7 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 **Q: 超出 PTU 额度时会怎样？** -请求自动转为按量计费。API 响应中 `service_tier` 字段不返回或返回 `default`，同时响应头包含 `x-dashscope-ptu-overflow:true`。业务不会中断。 +取决于创建时选择的溢出策略：「自动溢出」策略下，请求自动转为按量计费，API 响应中 `service_tier` 字段不返回或返回 `default`，同时响应头包含 `x-dashscope-ptu-overflow:true`，业务不会中断；「仅使用 PTU 容量」策略下，超出请求返回 429 错误，不产生额外费用。 **Q: 单次输入超过模型上限时会怎样？** @@ -296,4 +298,4 @@ PTU 部署的运行监控通过百炼平台的模型监控功能实现，支持 **Q: 利用率为什么超过 100%？** -部分模型（如 glm-5.1）的长输入阶梯系数使实际额度消耗高于原始 token 数。利用率 = 折算后消耗 ÷ 购买额度。超过 100% 表示消耗速度超过购买额度，超出部分自动转为按量计费，不影响服务可用性。 +部分模型（如 glm-5.1）的长输入阶梯系数使实际额度消耗高于原始 token 数。利用率 = 折算后消耗 ÷ 购买额度。超过 100% 表示消耗速度超过购买额度，超出部分按溢出策略处理（自动溢出则转为按量计费、不影响服务可用性；仅使用 PTU 容量则返回 429）。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md index 5e614b57..eee5b087 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/image-model.md @@ -43,20 +43,22 @@ FLUX.2 - 写实人像和产品照片 -### 何时使用qwen-image-2.0-pro +### 何时使用Qwen Image - 需要使用负向提示词排除输出中的特定元素 - 需要每次调用生成最多6张图片变体（Wan标准模式最多支持4张） +- 需要复杂版面生成、小字精准渲染或多语言字体支持 → 选择 qwen-image-3.0-pro（邀测中） + ## 图片编辑 推荐使用`wan2.7-image-pro`，它支持多图参考（最多9张输入图片）、边界框交互式编辑以及角色一致性多图生成。详细使用方法请参见[图像编辑-千问](https://help.aliyun.com/zh/model-studio/qwen-image-edit-guide)和[图像编辑-万相2.7/2.6/2.5](https://help.aliyun.com/zh/model-studio/wan-image-edit)。 -### 何时使用qwen-image-2.0-pro +### 何时使用Qwen Image -如果编辑时需要使用负向提示词，请使用`qwen-image-2.0-pro`（生成和编辑使用同一个模型ID）。 +如果编辑时需要使用负向提示词，请使用`qwen-image-3.0-pro`（邀测中）或`qwen-image-2.0-pro`（生成和编辑使用同一个模型ID）。 ## 推荐模型 @@ -108,9 +110,9 @@ FLUX.2 2048x2048 -`qwen-image-2.0-pro` +`qwen-image-3.0-pro` -负向提示词、最多6张图片变体 +复杂版面生成、小字渲染、多语言字体 支持 @@ -272,6 +274,16 @@ qwen-image-2.0-pro的快速版本 **最大分辨率** +`qwen-image-3.0-pro` + +支持 + +支持 + +6 + +2048x2048 + `qwen-image-2.0-pro` 支持 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md index 9a3c0b1a..dc841b66 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/s2s-model.md @@ -14,13 +14,13 @@ OpenAI GPT Realtime、Gemini 3.1 Live -`qwen3.5-omni-plus-realtime` +`qwen-audio-3.0-realtime-plus` 成本敏感对话 OpenAI gpt-4o-mini Realtime -`qwen3.5-omni-flash-realtime` +`qwen-audio-3.0-realtime-flash` 实时翻译 / 同传 @@ -74,7 +74,7 @@ Gemini 3.1 Live ## 实时还是文件模式？ -- **实时（WebSocket）**：适用于语音助手、呼叫中心、同声传译等实时语音交互场景。音频流式输入，语音流式输出。模型名称中包含`-realtime`。 +- **实时（WebSocket）**：适用于语音助手、呼叫中心、同声传译等实时语音交互场景。音频流式输入，语音流式输出。 - **文件模式（HTTP）**：可以用延迟换取更好的效果，适用于视频配音、播客翻译、离线内容处理等场景。文件模式下还支持 Function Calling、联网搜索、思考模式、视频上下文等附带能力（详见下方“S2S 单模型的附带能力”）。 @@ -91,13 +91,13 @@ Gemini 3.1 Live 语音助手 / 客服对话 -`qwen3.5-omni-plus-realtime` +`qwen-audio-3.0-realtime-plus` WebSocket 成本敏感的对话 -`qwen3.5-omni-flash-realtime` +`qwen-audio-3.0-realtime-flash` WebSocket @@ -127,27 +127,27 @@ WebSocket ## S2S 单模型的附带能力 -以下能力由 Qwen3.5-Omni / Qwen3-Omni 模型在 S2S 单模型路线下直接提供。Pipeline 路线中，对应能力需要由其中的 LLM 等组件分别支持。 +以下能力由 Qwen3.5-Omni / Qwen3-Omni 模型在 S2S 单模型路线下直接提供；其中 Function Calling 也可由 Qwen-Audio Realtime 提供。Qwen-Audio Realtime 不支持联网搜索和思考模式。Pipeline 路线中，对应能力需要由其中的 LLM 等组件分别支持。 ### Function Calling -让模型根据听到和看到的内容执行操作 -- 查询知识库、查询日程、触发工作流。使用Qwen3.5 Omni（WebSocket与HTTP模式） 或 Qwen3 Omni（HTTP模式）。 +让模型根据听到和看到的内容执行操作 -- 查询知识库、查询日程、触发工作流。使用 Qwen3.5 Omni（WebSocket 与 HTTP 模式）、Qwen3 Omni（HTTP 模式）或 Qwen-Audio Realtime（WebSocket 模式）。 **说明** -Qwen3.5-Omni/Qwen3-Omni实时（WebSocket）模式和Livetranslate模型不支持此功能。Qwen-Audio Realtime（WebSocket）支持Function Calling。 +Qwen3.5-Omni / Qwen3-Omni 实时（WebSocket）模式和 Livetranslate 模型不支持此功能。 ### 联网搜索 -让模型检索实时信息，回答关于时事、股价、天气等问题。使用Qwen3.5 Omni（HTTP和WebSocket），包括Plus和Flash系列。模型自主决定是否搜索。 +让模型检索实时信息，回答关于时事、股价、天气等问题。使用 Qwen3.5 Omni（HTTP 和 WebSocket），包括 Plus 和 Flash 系列。模型自主决定是否搜索。Qwen-Audio Realtime 不支持此功能。 **说明** -Qwen3-Omni-Flash和Livetranslate模型不支持此功能。 +Qwen3-Omni-Flash 和 Livetranslate 模型不支持此功能。 ### 思考模式 -当回答质量比延迟更重要时，使用Qwen3 Omni（HTTP模式）。模型在回复前会逐步推理，适用于视频分析、批量打标等场景。 +当回答质量比延迟更重要时，使用 Qwen3 Omni（HTTP 模式）。模型在回复前会逐步推理，适用于视频分析、批量打标等场景。Qwen-Audio Realtime 不支持此功能。 **说明** @@ -157,7 +157,7 @@ Qwen3-Omni-Flash和Livetranslate模型不支持此功能。 以下模型系列均支持语音翻译： -- **Qwen3.5-Livetranslate**：支持 60 种语言互译，其中 29 种支持音频+文本输出、31 种仅支持文本输出，覆盖中文、英语、法语、德语、俄语、日语、韩语、西班牙语、葡萄牙语、阿拉伯语等主流语种。。 +- **Qwen3.5-Livetranslate**：支持 60 种语言互译，其中 29 种支持音频+文本输出、31 种仅支持文本输出，覆盖中文、英语、法语、德语、俄语、日语、韩语、西班牙语、葡萄牙语、阿拉伯语等主流语种。 - **Qwen3-Livetranslate**：支持18种语言 + 5种中文方言，约3秒延迟，开箱即用。文件模式支持输入视频以获得上下文感知的翻译精度。其中7种语言仅输出文本（不输出语音）。 @@ -992,6 +992,50 @@ HTTP ## 所有模型 +### Qwen-Audio + +**模型** + +**API** + +**输入** + +**Function Calling** + +**联网搜索** + +**思考模式** + +**翻译** + +`qwen-audio-3.0-realtime-plus` + +WebSocket + +音频、文本 + +支持 + +\-- + +\-- + +\-- + +`qwen-audio-3.0-realtime-flash` + +WebSocket + +音频、文本 + +支持 + +\-- + +\-- + +\-- + ### Qwen3.5-Omni **模型** @@ -1256,50 +1300,6 @@ HTTP 18 -### Qwen-Audio - -**模型** - -**API** - -**输入** - -**Function Calling** - -**联网搜索** - -**思考模式** - -**翻译** - -`qwen-audio-3.0-realtime-plus` - -WebSocket - -音频、文本 - -支持 - -\-- - -\-- - -\-- - -`qwen-audio-3.0-realtime-flash` - -WebSocket - -音频、文本 - -支持 - -\-- - -\-- - -\-- - ### 旧版模型 以下模型不再更新，新项目建议使用Qwen3.5-Omni。 @@ -1356,6 +1356,8 @@ WebSocket 选定模型后，参考对应的调用文档： +- Qwen-Audio Realtime（WebSocket，实时语音对话）→ [实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides) + - Qwen3.5-Omni / Qwen3-Omni（WebSocket，实时）→ [实时（Qwen-Omni-Realtime）](https://help.aliyun.com/zh/model-studio/realtime) - Qwen3.5-Omni / Qwen3-Omni（HTTP，文件）→ [非实时（Qwen-Omni）](https://help.aliyun.com/zh/model-studio/qwen-omni) @@ -1363,5 +1365,3 @@ WebSocket - Qwen3.5-Livetranslate（WebSocket，实时）→ [实时语音/音视频翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-5-livetranslate-flash-realtime) - Qwen3-Livetranslate（HTTP，文件）→ [音视频文件翻译-千问](https://help.aliyun.com/zh/model-studio/qwen3-livetranslate-flash) - -- Qwen-Audio Realtime（WebSocket，实时语音对话）→ [实时语音对话（Qwen-Audio-Realtime）](https://help.aliyun.com/zh/model-studio/qwen-audio-realtime-user-guides) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md index 560fb0a8..6f2eda70 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/text-generation-model.md @@ -4,7 +4,7 @@ ## 做 AI 编程或 Agent 开发（OpenClaw、Claude Code、Hermes 等）该选哪个模型？ -推荐 `qwen3.7-plus`——能力与成本均衡，完整工具调用支持，1M 上下文适合大型代码库。如需最强推理能力，可选择 `qwen3.7-max`。 +推荐 `qwen3.7-plus`——能力与成本均衡，完整工具调用支持，1M 上下文适合大型代码库。如需最强推理能力，可选择`qwen3.8-max-preview`（ Token Plan 可用），或 `qwen3.7-max`。 ## 从闭源模型迁移到百炼? @@ -18,7 +18,7 @@ GPT-5.5、Claude Opus 4.7、Gemini 3.1 Pro -`qwen3.7-max` +`qwen3.7-max`、`qwen3.8-max-preview`（仅 Token Plan 可用） 平衡 @@ -34,7 +34,7 @@ GPT-5.4-mini、Claude Haiku 4.5、Gemini 3.1 Flash ## 应用场景 -聊天机器人、内容生成、摘要总结、文档处理等场景，推荐使用 `qwen3.7-plus`，能力与成本均衡，拥有100万上下文窗口和完整的内置工具。确认效果满足需求后，可以尝试 `qwen3.6-flash` 来降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。如需最强推理能力，可选择 `qwen3.7-max`（百万 token 上下文），但成本较高。 +聊天机器人、内容生成、摘要总结、文档处理等场景，推荐使用 `qwen3.7-plus`，能力与成本均衡，拥有100万上下文窗口和完整的内置工具。确认效果满足需求后，可以尝试 `qwen3.6-flash` 来降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。如需最强推理能力，可选择 `qwen3.7-max`（百万 token 上下文）；也可选择 `qwen3.8-max-preview`（ Token Plan 可用）。 ### 办公场景（非编程） @@ -42,7 +42,7 @@ GPT-5.4-mini、Claude Haiku 4.5、Gemini 3.1 Flash 确认效果满足需求后，可尝试 `qwen3.6-flash` 降低成本，效果接近旗舰模型，且拥有相同的上下文长度和功能支持。 -如需最强推理能力（如复杂数据分析、多步逻辑推演），可选择 `qwen3.7-max`，但成本较高。 +如需最强推理能力（如复杂数据分析、多步逻辑推演），可选择 `qwen3.7-max`，但成本较高；也可选择 `qwen3.8-max-preview`（仅 Token Plan 可用）。 处理超长文档（如同时审阅多份合同、大规模文献梳理）时，推荐 `qwen-long`——上下文窗口达 1000 万 Token，可完整处理大体量文档。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md index 65c53e25..e784e1ba 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md @@ -54,7 +54,7 @@ ElevenLabs Multilingual v3 `qwen-audio-3.0-tts-plus`、`MiniMax/speech-2.8-hd` -`qwen-audio-3.0-tts-flash`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） +`qwen-audio-3.0-tts-plus`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） - **使用标准语音合成**：当内置音色库能满足需求，希望快速上手、无需额外配置时。 @@ -89,7 +89,7 @@ ElevenLabs Multilingual v3 推荐模型 -`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` +`qwen-audio-3.0-tts-plus`、`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` `cosyvoice-v3.5-plus`、`cosyvoice-v3.5-flash` @@ -145,7 +145,7 @@ Qwen-Audio-TTS WebSocket / HTTP -不支持 +支持 不支持 @@ -205,7 +205,7 @@ HTTP WebSocket / HTTP -不支持 +支持 不支持 @@ -225,7 +225,7 @@ WebSocket / HTTP - 系统音色（因音色而异）：中文（普通话）、英文 -- 声音复刻音色（方言通过指令控制功能进行设置）：中文（普通话、广东话、重庆话、东北话、甘肃话、贵州话、浙江话、河北话、河南话、湖北话、湖南话、江西话、宁波话、宁夏话、青岛话、陕西话、山西话、山东话、上海话、四川话、云南话）、英文、日语、韩语、德语、法语、意大利语、俄语、葡萄牙语、泰语、印尼语、马来语、越南语 +- 声音复刻音色（方言通过指令控制功能进行设置）：中文（普通话、广东话、重庆话、东北话、甘肃话、贵州话、浙江话、河北话、河南话、湖北话、湖南话、江西话、宁波话、宁夏话、青岛话、陕西话、山西话、山东话、上海话、四川话、云南话）、英语、日语、韩语、俄语、法语、德语、葡萄牙语、泰语、印尼语、越南语、西班牙语、意大利语、马来西亚语、菲律宾语、阿拉伯语 ### CosyVoice diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md index 8e525acd..ecfee151 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md @@ -10,7 +10,7 @@ TPM 预留为指定模型锁定专属推理容量，确保业务高峰期不受 - 专属模型 code：创建 TPM 预留后，系统自动生成专属模型 code，您需要将 API 请求中的 `model` 参数替换为该 code。 -- 超额不中断：超出预留容量的请求自动降级为按量计费处理，无需修改代码。 +- 溢出策略：创建时可选超额处理方式——自动溢出至按量计费（默认，业务不中断）或仅使用预留容量（超出返回 429，不产生额外费用）。 ## **方案对比与选型** @@ -61,7 +61,7 @@ TPM 预留 **流量可预估、不能接受限流** -超出自动降级公共池按量，不中断 +可选：自动溢出按量（默认）/仅预留容量返回429 替换 model 参数即可 @@ -73,7 +73,7 @@ PTU（[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-in **高吞吐高性能** -超出转按量 +可选：自动溢出按量（默认）/仅PTU容量返回429 替换 model 参数即可 @@ -199,7 +199,7 @@ DeepSeek-v4-Pro `退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)` -- 超出保障额度后自动降级为标准按量计费，服务不中断。可在详情页**超额降级统计**中查看降级次数。 +- 溢出策略为「自动溢出」时：超出保障额度自动降级为标准按量计费，服务不中断，可在详情页**超额降级统计**查看降级次数；为「仅使用预留容量」时：超出返回 429，不产生额外费用。 - 服务到期后 2 小时内：实例仍为运行中，可继续调用，可续费；到期后 2~14 小时：实例已停止，不可调用，仍可续费；到期 14 小时后：实例已删除，不可恢复。 @@ -232,6 +232,7 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 + deepseek-v4-pro @@ -357,6 +358,14 @@ Qwen 系列 在输入框中输入天数，取值范围与购买时长一致。 + 溢出策略 + + 预留容量耗尽时，超出部分请求的处理方式。 + + 是 + + 自动溢出至按 token 付费（默认，超出转按量、业务不中断）/ 仅使用预留容量（超出返回 429、不产生额外费用） + 2. 确认参数后单击**立即购买**，在费用确认弹窗中核对费用，单击**确认支付**。 3. 在 TPM 预留详情页的**概览** Tab，找到**专属模型 code**，单击复制。 @@ -451,7 +460,7 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 - 使用率趋势：可切换输入/输出方向，展示预留容量线和实际用量。 -- 超额降级统计：展示超出预留容量后被降级处理的次数。 +- 超额降级统计：展示超出预留容量后被降级处理的次数（仅「自动溢出」策略下产生降级）。 #### 监控 @@ -476,6 +485,10 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 单击**扩缩容**，在弹窗中调整输入 TPM 和输出 TPM。 +**说明** + +输入 TPM 和输出 TPM 支持调整为 0：归 0 后不再产生容量费用，且专属模型 code 继续保留，避免因到期或退订导致 code 失效。但归 0 属于减配，已使用部分按 1.5 倍系数结算违约金（详见上方计费与使用说明）。 + #### 续费 单击**续订**，选择续费时长并完成支付。如已开启**到期自动续费**，系统在到期前一天 08:00 自动扣款续费。 @@ -524,7 +537,7 @@ RPM 越大，建议购买的输入和输出 TPM 同比增大。 **Q: 超出预留容量时会怎样？** -超出预留容量的请求自动降级为按量计费处理，服务不中断。可在详情页概览 Tab 的**超额降级统计**中查看降级次数和时间。频繁降级时建议扩容。 +取决于创建时选择的溢出策略：「自动溢出」策略下，超出预留容量的请求自动降级为按量计费，服务不中断，可在详情页概览 Tab 的**超额降级统计**查看降级次数和时间，频繁降级时建议扩容；「仅使用预留容量」策略下，超出请求返回 429 错误，不产生额外费用，频繁 429 时建议扩容。 **Q: 专属模型 code 怎么获取？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md index 27e89c0c..a229ac38 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-monitoring/model-telemetry.md @@ -11,9 +11,9 @@ ## 支持的模型 -- **监控：普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的所有模型，包括基于它们调优后的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)；**高级监控**支持北京、新加坡、弗吉尼亚地域下的所有模型**。** +- **监控：普通监控**支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的所有模型，包括基于它们调优后的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)；**高级监控**支持北京、上海、新加坡、弗吉尼亚地域下的所有模型**。** -- **告警功能：**支持北京、新加坡地域下的所有模型。 +- **告警功能：**支持北京、新加坡、弗吉尼亚地域下的所有模型。 ## **监控模型运行** @@ -22,9 +22,9 @@ > 列表记录按“模型 + 业务空间”维度生成。新模型在首次数据同步完成后自动加入列表（普通监控的延迟通常为小时级，请耐心等待；如需分钟级的数据洞察，请使用[高级监控](#2e5f2f0dffijg)）。 -列表顶部「监控数据看板」以卡片形式汇总**模型总量**、**总调用次数**、**总失败次数**、**平均调用时长**、**平均首包时长**。 +列表顶部「**监控数据**」以卡片形式汇总**模型总量**、**总调用次数**、**总失败次数**、**平均调用时长**、**平均首包时长**。 -「模型监控」表格列出各模型的**模型 Code**、**业务空间**、**调用总量**、**调用失败量**、**失败率**、**平均调用时长**、**平均首包时长**（除模型 Code、业务空间外均可排序），操作列提供**监控**、**日志**入口。 +「模型监控」表格列出各模型的**模型Code**、**业务空间**、**调用总量**、**调用失败量**、**失败率**、**平均调用时长**、**平均首Token延时**（除模型Code、业务空间外均可排序），操作列提供**监控**、**日志**入口。列表工具栏还提供**日志回流**入口，可将推理日志回流为训练数据集。 > 默认业务空间成员可查看所有业务空间的模型调用情况；子业务空间成员仅能查看当前空间的数据，无法切换查看其他业务空间数据。 @@ -80,7 +80,11 @@ ### **查看某次调用的 Token 消耗** -> 该功能目前仅适用于**华北2（北京）**地域的部分模型。 +> 该功能目前适用于**华北2（北京）**、新加坡地域的部分模型，弗吉尼亚地域同样支持。 + +**说明** + +**数据说明：**推理日志（高级监控）从调用发生到可查询存在分钟级延迟，请耐心等待；普通监控的用量汇总（如调用次数、Token 总量）延迟为小时级，高峰期可能达 1-2 小时。如遇无数据或查不到记录的情况，请先确认已等待足够的数据同步时间。仅记录**开启推理日志后**的调用数据，开通前的历史调用无法追溯。 1. 使用主账号（[或拥有足够权限的子账号](#f9d06146c0xe0)）登录，在目标业务空间的[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)页面，点击右上角的**模型监控配置**，按照指引依次开通审计日志和推理日志。 @@ -88,7 +92,7 @@ 2. 在模型监控列表中找到目标模型，点击其右侧**操作**列的**日志**。 -3. **日志**页签展示该模型的[实时推理](#f131611173sdx)调用记录，**用量**字段即为本次调用的Token消耗。 +3. **日志**页签以表格形式展示该模型的[实时推理](#f131611173sdx)调用记录，表格包含**Request ID/调用时间**、**调用时长**、**状态码**（支持筛选）、**错误码**、**用量**、**请求和响应**、**操作**等列。其中**用量**字段即为本次调用的Token消耗。 ### **创建异常消耗告警** @@ -100,10 +104,14 @@ **重要** -该功能目前仅适用于**华北2（北京）**地域的部分模型。 +该功能目前适用于**华北2（北京）**、新加坡地域的部分模型，弗吉尼亚地域同样支持。 模型监控支持查看模型的每一次对话，包括输入、输出及耗时，是故障排查和内容审计的关键工具。 +**说明** + +**数据说明：**日志从调用发生到可查询存在分钟级延迟，请耐心等待。如遇实时更新延迟或查不到记录的情况，请先确认已等待足够时间（普通监控用量汇总为小时级延迟，高级监控/推理日志为分钟级）。仅记录**开启推理日志后**的调用数据，开通前的历史调用无法追溯。 + ### **步骤一：开通日志** 使用主账号（[或拥有足够权限的子账号](#f9d06146c0xe0)）登录，在目标业务空间的[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)页面，点击右上角的**模型监控配置**，按照指引依次开通审计日志和推理日志。 @@ -116,7 +124,7 @@ 1. 在模型监控列表中找到目标模型，点击其右侧**操作**列的**日志**。 -2. **日志**页签展示该模型的[实时推理](#f131611173sdx)调用记录，**请求和响应**字段分别对应本次调用的输入与输出。 +2. **日志**页签以表格形式展示该模型的[实时推理](#f131611173sdx)调用记录，表格包含**Request ID/调用时间**、**调用时长**、**状态码**（支持筛选）、**错误码**、**用量**、**请求和响应**、**操作**等列。其中**请求和响应**字段分别对应本次调用的输入与输出。 **支持请求和响应的模型** @@ -152,14 +160,20 @@ - 三方模型：deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp +并非所有模型都支持推理日志（请求/响应内容记录）。是否支持由模型本身决定，与模型是否为多模态无关。当所选模型不支持时，界面会显示**当前模型暂不支持日志**。 + +请注意：请求和响应内容**仅在开启推理日志后**才会被采集，开通前的历史调用不会补录。若某次调用的输出内容缺失或存在日志缺失，请先确认该模型是否支持推理日志，以及推理日志是否已在调用发生前完成开通。 + ## **建立主动告警** **重要** -该功能目前仅适用于新加坡和华北2（北京）地域。 +该功能目前适用于新加坡、北京和弗吉尼亚地域。 模型的静默失败（如超时、Token消耗突增），传统应用日志难以发现。模型监控支持对监控指标（如成本、失败率、响应延迟）设置告警。一旦指标出现异常，系统立即告警。 +模型告警页面包含**告警规则**和**告警历史**两个页签。**告警历史**页签可按告警时间、告警规则、告警等级、状态筛选查看历史告警记录，点击详情可查看告警详情。 + ### **步骤一：开启高级监控** 1. 使用主账号（[或拥有足够权限的子账号](#54ea9ba526ovz)）登录，在目标业务空间的模型监控（[北京](https://bailian.console.aliyun.com/?tab=model#/model-telemetry) 或 [新加坡](https://modelstudio.console.aliyun.com/?tab=dashboard#/model-telemetry)）页面，点击右上角的**模型监控配置**。 @@ -171,7 +185,7 @@ 1. 在模型告警（[北京](https://bailian.console.aliyun.com/?tab=model#/model-alert) 或[新加坡](https://modelstudio.console.aliyun.com/?tab=dashboard#/model-alert)）页面，点击右上角的**创建告警规则**。 -2. 在对话框中，选择要监控的模型和监控模板，确认无误后点击**创建**。当指定的监控指标（如调用统计或性能指标）出现异常时，系统将通知您的团队。 +2. 在对话框中，选择要监控的模型和监控模板，确认无误后点击**确定**。当指定的监控指标（如调用统计或性能指标）出现异常时，系统将通知您的团队。 - **通知方式：**支持短信、电子邮件、电话、钉钉群机器人、企业微信机器人及Webhook。 @@ -289,12 +303,20 @@ 模型非首包时长p99 + **TPS** + + model\_tps\_per\_request + + 单次请求输出 Token 速度（TPS），每秒生成 Token 数，衡量模型生成速度（仅高级监控支持） + **用量** model\_usage 模型用量总和 + **关于 TPS 指标：**`model_tps_per_request` 仅在**高级监控**中展示，高级监控为收费功能。TPS（每秒生成 Token 数）与非首包时长（每 Token 的平均生成耗时）呈倒数关系（TPS ≈ 1 ÷ 非首包时长均值）。排查响应慢的问题时，建议结合首 Token 延时（TTFT）、非首 Token 延时及输入 Token 量综合分析，单次调用总耗时还受输入长度、网络等因素影响，不能仅凭 TPS 判断。TPS 触发的是按请求维度的限流，区别于 TPM（每分钟 Token 数）的按账号维度限流。 + - **HTTP API：**`{HTTP API}`需替换为前面[步骤一](#title-tkb-ds1-4p5)获取的HTTP API地址。 - **Authorization：**需将阿里云账号的 `AccessKey:AccessKeySecret` 拼接后进行Base64编码，并以 `Basic 编码后字符串` 的形式提供。 @@ -530,6 +552,12 @@ 模型非首包时长p99 + **TPS** + + model\_tps\_per\_request + + 单次请求输出 Token 速度（TPS），每秒生成 Token 数，衡量模型生成速度（仅高级监控支持） + **用量** model\_usage @@ -715,6 +743,12 @@ } ``` +- **API Key 用量限额：**当前不支持为单个 API Key 设置月度或每日 Token 消耗上限并自动停服（即不支持硬性阻断，无法实现额度用尽后自动禁用 API 调用）。如需防止意外欠费，可采用以下替代方案： + + - **告警通知 + 手动禁用 Key：**开启**高级监控**（收费功能）后，可对 Token 消耗配置告警阈值，当消耗超出阈值时接收消费预警通知，再人工介入处理（如手动禁用对应 API Key）。普通监控仅支持基础用量查看，不支持配置告警。 + + - **免费额度用完即停：**适用于仅使用免费配额的场景，开启后免费额度耗尽时自动停止调用。 + ## **计费说明** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md index a108688f..7c8723b1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/model-release-notes.md @@ -33,6 +33,30 @@ **功能说明** +7月21日 + +平台功能 + +记忆库商业化通知 + +记忆库商业化通知[了解详情](https://www.aliyun.com/notice/118464) + +7月16日 + +平台功能 + +企业知识库（旧）下线通知 + +企业知识库（旧）下线通知[了解详情](https://www.aliyun.com/notice/118448) + +7月16日 + +平台功能 + +Managed Agent商业化通知 + +Managed Agent商业化通知[了解详情](https://www.aliyun.com/notice/118456) + 7月14日 平台功能 @@ -177,7 +201,7 @@ Coding Plan Coding Plan 联网搜索 MCP 升级 -Coding Plan 联网搜索 MCP 升级 Streamable HTTP 协议，前 2000 次免费，[了解详情](https://help.aliyun.com/zh/model-studio/web-search-for-coding-plan) +Coding Plan 联网搜索 MCP 升级 Streamable HTTP 协议，前 2000 次免费，[了解详情](https://help.aliyun.com/zh/model-studio/web-search-mcp) 6月10日 @@ -339,7 +363,7 @@ Token Plan Token Plan 团队版团队管理上线 -Token Plan 团队版新增团队管理：支持 SSO/钉钉登录、席位分配、Credits 用量监控，[了解详情](https://help.aliyun.com/zh/model-studio/token-plan-team#tp05-h-enter) +Token Plan 团队版新增团队管理：支持 SSO/钉钉登录、席位分配、Credits 用量监控，[了解详情](https://help.aliyun.com/zh/model-studio/token-plan-team-management#tp05-h-enter) 5月4日 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md index 1fffce5d..7ac5da7c 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/release-notes/newly-released-models.md @@ -14,6 +14,26 @@ **功能说明** +图像生成 + +2026-07-21 + +中国内地 + +qwen-image-3.0-pro + +Qwen-Image-3.0-Pro 系列模型邀测上线，支持长文本输入与图中图密集排版，能够一次性精准生成报纸、分镜、菜单及试卷等复杂版面；具备10像素小字精准渲染能力，生动还原微表情、毛孔与发丝等摄影级细节，并支持 12 国语言、多种字体及主流网页、游戏界面的高保真仿真。[千问-图像生成与编辑3.0](https://help.aliyun.com/zh/model-studio/qwen-image-generation-and-editing-api-reference) + +文生文与视觉理解 + +2026-07-17 + +中国内地 + +kimi/kimi-k3 + +Kimi K3 是 Kimi 迄今能力最强的旗舰模型，拥有 2.8 万亿参数，原生支持视觉理解，并拥有 100 万 token 上下文窗口，面向长程编程、知识工作和推理等前沿智能场景而设计。[Kimi-月之暗面](https://help.aliyun.com/zh/model-studio/kimi-api-by-moonshot-ai) + 视频对口型 2026-07-15 @@ -62,7 +82,7 @@ Qwen-Audio端到端实时语音大模型兼顾语音推理能力与双工对话 qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash -Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景，首包延时控制在 200ms 以内。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) 文生图/参考生图 @@ -2928,6 +2948,16 @@ qwen1.5-110b-chat **功能说明** +图像生成 + +2026-07-21 + +国际 + +qwen-image-3.0-pro + +Qwen-Image-3.0-Pro 系列模型邀测上线，支持长文本输入与图中图密集排版，能够一次性精准生成报纸、分镜、菜单及试卷等复杂版面；具备10像素小字精准渲染能力，生动还原微表情、毛孔与发丝等摄影级细节，并支持 12 国语言、多种字体及主流网页、游戏界面的高保真仿真。[千问-图像生成与编辑3.0](https://help.aliyun.com/zh/model-studio/qwen-image-generation-and-editing-api-reference) + 语音合成 2026-07-14 @@ -2936,7 +2966,7 @@ qwen1.5-110b-chat qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash -Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景，首包延时控制在 200ms 以内。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) +Qwen-Audio-TTS语音合成模型上线，新增更多小语种和中文方言支持，增强了指令遵循与细粒度标签控制能力，音质和表现力全面提升。其中 Plus 版本面向高品质专业场景，Flash 版本面向低延迟实时交互场景。[实时语音合成](https://help.aliyun.com/zh/model-studio/realtime-tts-user-guide) 参考生视频 @@ -2956,7 +2986,7 @@ wan2.7-r2v-2026-06-12 kimi-k2.7-code -Kimi K2.7 Code 模型新加坡地域上线。以编码为中心的智能体模型，专为长程软件工程任务优化，仅支持思考模式。[](#) +Kimi K2.7 Code 模型新加坡地域上线。以编码为中心的智能体模型，专为长程软件工程任务优化，仅支持思考模式。[Kimi-阿里云](https://help.aliyun.com/zh/model-studio/kimi-api) 图像生成 @@ -3852,6 +3882,16 @@ qwen3-asr-flash-realtime、qwen3-asr-flash-realtime-2025-10-27 **功能说明** +推理模型 + +2026-07-17 + +美国 + +qwen3.6-flash-us + +Qwen3.6 原生视觉语言 Flash 系列模型，在整体性能上较 Qwen3.5-Flash 显著提升。重点增强了智能体编程能力（在多项代码智能体基准上大幅超越前代）、数学推理和代码推理能力；在视觉能力方面，空间智能显著增强，其中物体定位和目标检测表现尤为突出。 + 文生文 2026-07-14 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md new file mode 100644 index 00000000..54844525 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/after-sales-service-scope.md @@ -0,0 +1,57 @@ +# 阿里云百炼平台售后服务范围说明 + +## **阿里云百炼平台售后服务范围说明** + +欢迎您使用阿里云百炼。本《**阿里云百炼平台售后服务范围说明》是对您使用阿里云百炼相关产品和服务时适用的售后服务范围的说明。** + +1\. 在您购买的服务期限内，我们将为您提供如下售后基础服务，即通过官网、电话及阿里云APP提供7×24的电话咨询（95187、400电话）、智能在线和标准工单支持。支持范围包括： + +（1）关于阿里云百炼模型服务与产品功能、架构的咨询； + +（2）使用、配置阿里云百炼模型服务的最佳实践； + +（3）阿里云百炼模型服务的使用咨询、技术问题及故障诊断； + +（4）阿里云百炼API及阿里云百炼官方SDK问题的故障诊断； + +（5）与阿里云百炼管理控制台相关的问题； + +（6）与阿里云相关的账号问题咨询支持； + +（7）与阿里云相关的财务、合同及计费问题的咨询支持。 + +2\. 阿里云百炼将以阿里云官网页面公布的[客户服务权益](https://www.aliyun.com/service/customer-service-benefits?spm=5176.support-home.J_3451238410.1.12d1156fPBBxO0)标准向您提供相应的售后服务支持。 + +3\. 阿里云百炼同时提供付费版的售后增值服务（包括支持计划等），该等服务需在阿里云官网订购后生效使用。您还可通过阿里云获得其他付费的售后服务，具体详见阿里云的网站相关页面的收费售后服务内容。如您的项目需要更深度的技术支持（如业务代码编写指导、定制化集成方案等），建议联系阿里云商务经理沟通定制化服务方案。 + +4\. 为了方便您的生产或使用，如您选择将阿里云百炼服务与外部（非阿里云百炼平台上的）第三方工具或产品进行对接，阿里云将尽商业上合理的努力为您提供第三方工具在接入阿里云百炼模型推理服务过程中的方向性建议，但针对非阿里云百炼服务相关的问题，我们无法提供专业意见。 + +（1）我们提供建议的范围包括： + +（i）确认阿里云百炼API接口及服务端的可用状态； + +（ii）阿里云百炼官方API调用示例及SDK使用说明参考； + +（iii）协助核查阿里云百炼服务端调用明细和计费记录； + +（iv）基本的连通性测试建议（如通过curl等标准工具测试阿里云百炼服务地址的可达性）。 + +**（2）我们提供建议的范围不包括：** + +（i）第三方工具（如Cursor、Windsurf、Cline、OpenClaw等）的安装、部署、配置、升级及日常使用指导； + +（ii）第三方工具的产品功能、交互逻辑及内部实现问题的排查； + +（iii）其他云厂商、企业或社区提供的产品或服务的配置与运维； + +（iv）用户业务代码的编写、调试与实现； + +（v）用户本地环境（含内网、代理、VPN、防火墙、操作系统等）导致的连通性或兼容性问题的排查； + +（vi）第三方工具内部显示的Token数量、费用预估值或调用统计与阿里云计费数据之间的差异解释； + +（vii）所有第三方工具的安装、补丁更新、测试、故障诊断、优化等日常运维服务； + +（viii）基于阿里云百炼模型服务原生能力之上的第三方自建业务相关支持。 + +5\. 但请您注意，除我们另有书面说明外，第三方工具不构成我们的代理、受托或联合服务主体，**我们不对外部第三方工具的任何陈述、承诺或行为承担责任。**您知悉并确认，阿里云仅负责阿里云百炼平台自身的运营维护，即百炼服务端的技术架构、API接口、计量计费系统、控制台功能等；**第三方工具的运行维护（如AI编程工具的安装配置、开源代理框架的部署调优等）由您及相应工具提供方负责。**当您的问题出现在百炼模型服务的使用过程中，但其原因、责任范围或依赖关系已超出阿里云百炼平台本身可直接提供支持和保障的范围时——通常涉及您侧系统、外部第三方服务、网络环境、账号权限、业务流程或非标集成，阿里云将协助进行初步排查。若经排查确认问题来源于非阿里云侧，阿里云将给予方向性建议并引导您联系相应的服务主体。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md index fff36d4d..53d67f33 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md @@ -208,6 +208,10 @@ 等待时间取决于您的具体限流值（RPS/RPM）。例如，如果您的限流是120 RPM（每分钟查询数），即每秒2次请求。如果您在0.2秒内连续提交了2次请求，第3次请求就会被限流，您需要等待大约0.8秒后才能再次成功提交。 +17. **qwen-plus-latest 这个模型具体对应哪个系列？是 Qwen3.7 还是 Qwen3.5？** + + qwen-plus-latest 是 qwen-plus 的最新版本，属于 Qwen3 系列，而非 Qwen3.5 或 Qwen3.7 系列。另外 Qwen3.5、Qwen3.7 等是独立的模型系列，与 Qwen3 系列并列，并非 Qwen3 的子版本。 + ## **模型幻觉问题** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/related-agreements.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/related-agreements.md index 7b67c5ee..c071bfc9 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/related-agreements.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/support/related-agreements.md @@ -4,7 +4,7 @@ - [阿里云百炼模型推理服务等级协议（SLA）](https://terms.alicdn.com/legal-agreement/terms/b_end_product_protocol/20250923215800868/20250923215800868.html) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) +- [阿里云百炼体验功能特别说明](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20260716114753386/20260716114753386.html) - [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/bill-query-and-cost-management.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/bill-query-and-cost-management.md index 76084a81..2e671bba 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/bill-query-and-cost-management.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/bill-query-and-cost-management.md @@ -8,7 +8,7 @@ ### **费用概览** -登录[百炼控制台](https://bailian.console.aliyun.com/?tab=model)，单击顶部**模型**标签页，在左侧菜单选择**用量 & 费用** > [**费用概览**](https://bailian.console.aliyun.com/?tab=model#/costing-balance/overview)，选择**账期月份**： +登录[百炼控制台](https://bailian.console.aliyun.com/?tab=model)，单击顶部**模型**标签页，选择[**费用概览**](https://bailian.console.aliyun.com/?tab=model#/costing-balance)，选择**账期月份**： > 该页面仅展示**大模型推理**相关费用。**模型训练**和**知识库**等费用请通过[账单详情](#29f8b9b9a4lmc)查看。 @@ -31,14 +31,14 @@ 2. 选择**产品名称**为**大模型服务平台百炼**，单击**搜索**。 -3. 单击账单列表右上角的导出图标，将账单下载到本地。 +3. 单击页面顶部的**导出明细**，将账单下载到本地。 4. 打开文件，找到 实例 ID（出账粒度）列，根据下文规则进行核对。 #### **2\. 解读关键字段** -**“实例 ID（出账粒度）”字段**以英文分号 `;` 分隔，完整格式为`ApiKeyID;业务空间 ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`。 +**“实例 ID（出账粒度）”字段**以英文分号 `;` 分隔，完整格式为`ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`。 - 格式 A：标准调用（包含ApiKeyID） @@ -68,7 +68,7 @@ - 查询 API Key：复制账单中的 `ApiKeyID`，前往[百炼API Key管理](https://bailian.console.aliyun.com/?tab=model#/api-key)页面查找对应的 Key 名称。 -- 查询业务空间：复制账单中的 `业务空间ID`，前往[百炼控制台](https://bailian.console.aliyun.com/?tab=model#/api-key)，点击左侧菜单底部的**默认业务空间**，点击**业务空间详情**，确认具体空间ID。您也可以切换到其他业务空间。 +- 查询业务空间：复制账单中的 `业务空间ID`，前往[业务空间管理](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)页面，在列表的**业务空间ID**列确认具体空间。当前所处的业务空间显示在控制台顶部右上角的**默认业务空间**切换器中，悬停可查看其地域、业务空间ID与创建时间等详情。 - 调用渠道说明： @@ -83,13 +83,13 @@ 给**业务空间**绑定**标签**，可按部门或项目归集费用。 -1. **获取业务空间信息**：在[**业务空间管理**](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)确定标签绑定的业务空间**Workspace ID**（示例：llm-xxx），并在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)确定业务空间的**地域**信息。 +1. **获取业务空间信息**：在[**业务空间管理**](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management)确定标签绑定的**业务空间ID**（示例：llm-xxx），并在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)确定业务空间的**地域**信息。 2. **绑定标签**： 1. 在[**标签管理**](https://resourcemanager.console.aliyun.com/tags#/)页面选择**资源绑定标签。** - 2. 资源选择方式选择“**输入多个资源ID**”，在产品选项卡搜索并选择“**大模型服务平台百炼:业务空间**”并选择业务空间对应地域，资源ID输入框中填写**Workspace ID**，完成后点击绑定标签按钮执行操作。 + 2. 资源选择方式选择“**输入多个资源ID**”，在产品选项卡搜索并选择“**大模型服务平台百炼:业务空间**”并选择业务空间对应地域，资源ID输入框中填写**业务空间ID**，完成后点击绑定标签按钮执行操作。 3. 在绑定标签页面中，创建标签键值或使用已创建的预置标签与业务空间绑定。当完成键值输入或选择好预置标签后，点击**确定**完成业务空间标签的绑定。 @@ -102,16 +102,14 @@ **账户可用额度 < 0** 视为欠费，可能导致模型调用等服务暂停。在[费用与成本首页](https://billing-cost.console.aliyun.com/home)悬停**账户可用额度**区域可查看，公式为：可用额度 =（现金余额 + 信控额度）-（当月未结清 + 历史未结清）。 -- **欠费影响**：按账单**商品名称**维度判定。 +- **欠费影响**：账户欠费将导致**按量付费（后付费）**模型调用等服务暂停，能否继续使用取决于计费方式。 - - 仍有**免费额度**：可继续使用，用完后停用。 - - - 仍有**节省计划**或**资源包**额度：可继续使用。 + - **免费额度、节省计划、资源包**：三者均用于抵扣按量付费费用，**欠费期间即使仍有剩余额度，也无法调用模型**，需结清欠费后恢复。 - 已购 **Coding Plan** 或 **Token Plan**：套餐额度独立于账户余额，欠费期间可继续使用，但会导致自动续费失败，到期后无法续用。 - - 以上额度均无：该商品下服务将**暂停**，需结清欠费后恢复。 - +- **代金券与余额说明**：代金券（含学生权益优惠券）不计入账户可用额度中的现金余额，后付费场景下系统会冻结当月消费金额，账户实际余额须大于当月冻结金额方可正常调用模型，详情参见[代金券说明文档](https://help.aliyun.com/zh/user-center/voucher-management)。 + - **结清欠费**：在[费用与成本](https://usercenter2.aliyun.com/home)页面单击**充值汇款**，输入金额并完成支付。 - **预防欠费**：在[高额消费预警](https://usercenter2.aliyun.com/home/alarm-threshold)页面设置消费阈值，达阈值即提醒。 @@ -121,21 +119,21 @@ 不再使用百炼时，按以下方式关停对应服务即可停止计费。 -- **停止模型推理**：停止代码中的 API 调用、关闭控制台的模型体验，即不再产生费用。为防止意外调用，可在[**API-KEY**](https://bailian.console.aliyun.com/?apiKey=1&tab=globalset#/efm/api_key)页面删除已创建的 Key。 +- **停止模型推理**：停止代码中的 API 调用、关闭控制台的模型体验，即不再产生费用。为防止意外调用，可在[**API-KEY**](https://bailian.console.aliyun.com/?tab=model#/api-key)页面删除已创建的 Key。 - **停止模型训练**：没有正在进行的训练任务时即不产生费用。 -- **取消 Coding Plan 订阅**：Coding Plan 为包月订阅产品，到期自动停止，中途不支持取消和退款。如已开启自动续费，请在[Coding Plan](https://bailian.console.aliyun.com/?tab=model#/efm/coding_plan) 页面关闭自动续费。 +- **停止 Coding Plan 计费**：Coding Plan 为包月订阅产品，到期自动停止，中途不支持取消和退款。如已开启自动续费，请在[Coding Plan](https://bailian.console.aliyun.com/?tab=model#/efm/subscription/coding-plan) 页面关闭自动续费。 - **退订 Token Plan 团队版**：在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面按席位退订，未产生用量消耗的席位可退订，退款原路退回支付账户。如不再续费，请关闭自动续费。 - **停止模型部署**：根据部署时的计费方式操作不同： - - 按模型调用量计费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，或删除[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)防止意外调用。 + - **按 Token 调用计费（后付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，或删除[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)防止意外调用。 - - 按算力使用时长计费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型。 + - **按算力单元或模型单元计费（后付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型。 - - 包月预付费：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，然后在[退订管理](https://usercenter2.aliyun.com/refund/refund)页面退订实例。退订时按已消费金额扣减，退回剩余金额（详见[退订说明](https://help.aliyun.com/zh/user-center/user-guide/refund-management/)）。 + - **按预置吞吐单元计费（预付费）**：[下线](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)已部署的模型，然后在[退订管理](https://usercenter2.aliyun.com/refund/refund)页面退订实例。退订时已使用部分按 1.5 倍系数结算（详见退订说明），退回剩余金额。 ## **常见问题** @@ -161,11 +159,11 @@ **原因：**账单的“计费项”统一显示为“大模型文本消耗量”，未直接展示模型名称。 -**解决方案：**查看[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页的**实例 ID（出账粒度）**列。字段以英文分号分隔，紧跟业务空间 ID（如 llm-xxx）之后的字段即为模型名称。例：`12xxx;llm-xxx;**qwen3.6-plus**;context_0-128k_input_token;bmp;0`表示 qwen3.6-plus 模型。 +**解决方案：**查看[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)页的**实例 ID（出账粒度）**列。字段以英文分号分隔，紧跟业务空间ID（如 llm-xxx）之后的字段即为模型名称。例：`12xxx;llm-xxx;**qwen3.6-plus**;context_0-128k_input_token;bmp;0`表示 qwen3.6-plus 模型。 **在哪里查看模型调用次数和统计？** -进入[阿里云百炼控制台](https://bailian.console.aliyun.com/?tab=model)，右上角选择目标地域，单击顶部**模型**标签页，在左侧菜单选择**用量 & 费用** > [模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)。 +进入[阿里云百炼控制台](https://bailian.console.aliyun.com/?tab=model)，右上角选择目标地域，单击顶部**模型**标签页，选择[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)。 **按量付费是实时扣款吗？** @@ -190,3 +188,27 @@ 2. 检查应用代码或百炼应用配置中是否开启了 `enable_search`，如不再需要联网搜索，将该参数设为 `false` 或移除。 3. 如已停止所有调用但仍有扣费，检查是否有其他 API Key 或应用仍在运行，可在[API Key 管理](https://bailian.console.aliyun.com/?tab=model#/api-key)页面逐一排查或删除不再使用的 Key。 + + +**为什么没有主动调用 API 也会产生费用？** + +**原因：**百炼的模型部署按使用时长计费，模型完成部署即状态为**运行中**时开始收费，不依赖 API 调用。即使未主动通过 API 调用该模型，只要部署状态为**运行中**就会持续产生费用。 + +**解决方案：** + +- 前往[**模型部署**](https://bailian.console.aliyun.com/?tab=model#/efm/model_deploy)页面，下线不再使用的已部署模型，停止按时长计费。 + +- 如需防止意外调用产生推理费用，可在[**API-KEY**](https://bailian.console.aliyun.com/?tab=model#/api-key)页面删除不再使用的 Key（注意：删除后无法恢复，请谨慎操作）。 + + +**如何判断账户是否被盗用？** + +如果怀疑账户被他人盗用产生非预期费用，按以下步骤排查： + +1. 在[账单详情](https://usercenter2.aliyun.com/finance/expense-report/expense-detail)中筛选**大模型服务平台百炼**，查看**实例 ID（出账粒度）**列中的 `ApiKeyID`，确认产生费用的 API Key。 + +2. 前往[API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key)，核对每个 Key 的**创建时间**，确认是否为本人创建。API Key 管理页面仅显示创建时间，不显示调用时间。 + +3. 查看调用时段分布，判断是否存在非本人操作的异常调用模式：进入[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)页面，按**模型**或 **API Key ID** 筛选，切换至**列表**视图查看调用时间分布。 + +4. 如发现未授权调用，立即在[API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key)删除对应 API Key 并重新生成。更新所有合法调用方使用新 Key。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md index 7137579d..6acc767a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md @@ -48,7 +48,7 @@ **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.7-max @@ -278,7 +278,7 @@ qwen3-max-preview **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-max @@ -902,7 +902,7 @@ qwen3.7-max-2026-05-20 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **非思考模式** @@ -1296,7 +1296,7 @@ qwen-plus-2025-04-28 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-plus-2025-01-25 @@ -2486,7 +2486,7 @@ qwen3.6-plus-2026-04-02 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.6-flash @@ -2706,6 +2706,24 @@ qwen3.6-flash-2026-04-16 28.8元 +qwen3.6-flash-us + +美国 + +非思考和思考模式 + +0 当前能力等同于qwen3.5-flash-2026-02-23 @@ -3070,6 +3088,8 @@ qwen3.5-flash 非思考和思考模式 +0 [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 + +中国内地 + +20元 + +100元 + +无 + kimi/kimi-k2.7-code-highspeed > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 @@ -9256,8 +9290,6 @@ kimi/kimi-k2.7-code-highspeed 54元 -无 - kimi/kimi-k2.7-code > [上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)享有折扣 @@ -9314,7 +9346,7 @@ kimi/kimi-k2.5 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） glm-5.2 @@ -9540,7 +9572,7 @@ glm-5.1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） glm-5.2 @@ -9742,7 +9774,7 @@ ZHIPU/GLM-5 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） MiniMax-M2.5 @@ -9940,7 +9972,13 @@ stepfun/step-3.7-flash **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） + +qwen-image-3.0-pro + +中国内地 + +限时免费 qwen-image-2.0-pro @@ -10046,6 +10084,12 @@ qwen-image **输出单价** +qwen-image-3.0-pro + +国际 + +限时免费 + qwen-image-2.0-pro > 当前能力等同于qwen-image-2.0-pro-2026-04-22 @@ -10138,7 +10182,13 @@ qwen-image **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） + +qwen-image-3.0-pro + +中国内地 + +限时免费 qwen-image-2.0-pro @@ -10252,6 +10302,12 @@ qwen-image-edit **输出单价** +qwen-image-3.0-pro + +国际 + +限时免费 + qwen-image-2.0-pro > 当前能力等同于qwen-image-2.0-pro-2026-04-22 @@ -10346,7 +10402,7 @@ qwen-image-edit **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-mt-image @@ -10374,7 +10430,7 @@ qwen-mt-image **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） z-image-turbo @@ -10420,7 +10476,7 @@ z-image-turbo **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.6-t2i @@ -10576,7 +10632,7 @@ wan2.6-t2i **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-image-pro @@ -10674,7 +10730,7 @@ wan2.6-image **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.5-i2i-preview @@ -10720,7 +10776,7 @@ wan2.5-i2i-preview **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-sketch-to-image-lite @@ -10744,7 +10800,7 @@ wanx-sketch-to-image-lite **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-x-painting @@ -10770,7 +10826,7 @@ wanx-x-painting **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-style-repaint-v1 @@ -10794,7 +10850,7 @@ wanx-style-repaint-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-background-generation-v2 @@ -10818,7 +10874,7 @@ wanx-background-generation-v2 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-out-painting @@ -10842,7 +10898,7 @@ image-out-painting **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-instance-segmentation @@ -10868,7 +10924,7 @@ image-instance-segmentation **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） image-erase-completion @@ -10894,7 +10950,7 @@ image-erase-completion **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-virtualmodel @@ -10924,7 +10980,7 @@ virtualmodel-v2 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） shoemodel-v1 @@ -10950,7 +11006,7 @@ shoemodel-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx-poster-generation-v1 @@ -10997,7 +11053,7 @@ facechain-finetune 50次 -有效期：申请通过后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） facechain-generation @@ -11007,7 +11063,7 @@ facechain-generation 500张 -有效期：申请通过后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） ### **创意文字生成-WordArt锦书** @@ -11023,7 +11079,7 @@ facechain-generation **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wordart-texture @@ -11058,7 +11114,7 @@ wordart-semantic **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） aitryon @@ -11308,7 +11364,7 @@ vidu/viduq2-fast\_reference2image **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-music-preview @@ -11344,7 +11400,7 @@ fun-music-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-audio-3.0-tts-plus @@ -11404,7 +11460,7 @@ qwen-audio-3.0-tts-flash **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-instruct-flash @@ -11442,7 +11498,7 @@ qwen3-tts-instruct-flash-2026-01-26 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vd-2026-01-26 @@ -11468,7 +11524,7 @@ qwen3-tts-vd-2026-01-26 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vc-2026-01-22 @@ -11494,7 +11550,7 @@ qwen3-tts-vc-2026-01-22 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-flash @@ -11542,7 +11598,7 @@ qwen3-tts-flash-2025-09-18 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-tts-flash @@ -11694,7 +11750,7 @@ qwen3-tts-flash-2025-09-18 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-instruct-flash-realtime @@ -11732,7 +11788,7 @@ qwen3-tts-instruct-flash-realtime-2026-01-22 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vd-realtime-2026-01-15 @@ -11768,7 +11824,7 @@ qwen3-tts-vd-realtime-2025-12-16 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-vc-realtime-2026-01-15 @@ -11800,7 +11856,7 @@ qwen3-tts-vc-realtime-2025-11-27 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-tts-flash-realtime @@ -11846,7 +11902,7 @@ qwen3-tts-flash-realtime-2025-09-18 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-tts-realtime @@ -11994,7 +12050,7 @@ qwen3-tts-flash-realtime-2025-09-18 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-voice-enrollment @@ -12036,7 +12092,7 @@ qwen-voice-enrollment **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-voice-design @@ -12078,7 +12134,7 @@ qwen-voice-design **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） cosyvoice-v3.5-plus @@ -12240,7 +12296,7 @@ MiniMax/speech-02-turbo **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **输入：音频** @@ -12396,7 +12452,7 @@ qwen3-livetranslate-flash-realtime-2025-09-22 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **输入：音频** @@ -12496,7 +12552,7 @@ qwen3-livetranslate-flash-2025-12-01 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-asr-flash-filetrans @@ -12624,7 +12680,7 @@ qwen3-asr-flash-2025-09-08 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-asr-flash-realtime @@ -12692,7 +12748,7 @@ qwen3-asr-flash-realtime-2025-10-27 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-asr @@ -12788,7 +12844,7 @@ fun-asr-flash-2026-06-15 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） fun-asr-realtime @@ -12979,7 +13035,7 @@ paraformer-realtime-8k-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -13056,7 +13112,7 @@ qwen-audio-3.0-realtime-flash **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-t2v @@ -13208,7 +13264,7 @@ happyhorse-1.0-t2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-i2v @@ -13360,7 +13416,7 @@ happyhorse-1.0-i2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.1-r2v @@ -13512,7 +13568,7 @@ happyhorse-1.0-r2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） happyhorse-1.0-video-edit @@ -13620,7 +13676,7 @@ happyhorse-1.0-video-edit **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-t2v-2026-06-12 @@ -13918,7 +13974,7 @@ wan2.6-t2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-i2v-2026-04-25 @@ -14014,7 +14070,7 @@ wan2.7-i2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.6-i2v-flash @@ -14356,7 +14412,7 @@ wan2.6-i2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-kf2v-flash @@ -14429,7 +14485,7 @@ wan2.1-kf2v-plus **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-r2v-2026-06-12 @@ -14663,7 +14719,7 @@ wan2.6-r2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.7-videoedit @@ -14691,7 +14747,7 @@ wan2.7-videoedit **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wanx2.1-vace-plus @@ -14762,7 +14818,7 @@ wan2.1-vace-plus **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-s2v-detect @@ -14805,7 +14861,7 @@ wan2.2-s2v **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-animate-move @@ -14817,7 +14873,7 @@ wan2.2-animate-move 50秒 -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） 专业模式`wan-pro` @@ -14865,7 +14921,7 @@ wan2.2-animate-move **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） wan2.2-animate-mix @@ -14877,7 +14933,7 @@ wan2.2-animate-mix 50秒 -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） 专业模式`wan-pro` @@ -14924,7 +14980,7 @@ wan2.2-animate-mix **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） animate-anyone-detect-gen2 @@ -14967,7 +15023,7 @@ animate-anyone-gen2 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） emo-detect-v1 @@ -15007,7 +15063,7 @@ emo-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） liveportrait-detect @@ -15042,7 +15098,7 @@ liveportrait **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） emoji-detect-v1 @@ -15074,7 +15130,7 @@ emoji-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） videoretalk @@ -15100,7 +15156,7 @@ videoretalk **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） video-style-transform @@ -16422,7 +16478,7 @@ Tripo/Tripo-P1.0 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3.7-text-embedding @@ -16522,7 +16578,7 @@ text-embedding-v3 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） **文本** @@ -16594,7 +16650,7 @@ multimodal-embedding-v1 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen3-vl-rerank @@ -16680,7 +16736,7 @@ farui-plus **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） tongyi-intent-detect-v3 @@ -16712,7 +16768,7 @@ tongyi-intent-detect-v3 **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） qwen-plus-character @@ -16858,7 +16914,7 @@ qwen-plus-character **免费额度**[（注）](https://help.aliyun.com/zh/model-studio/new-free-quota#591f3dfedfyzj) -有效期：阿里云百炼开通后90天内 +有效期：自开通百炼/模型发布/申请通过之日起90天内（以较晚者为准） gui-plus diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md index 8bc376aa..85ba833e 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-training-and-deployment-billing.md @@ -474,9 +474,9 @@ wan2.2-kf2v-flash - 后付费时，如果账户欠费，部署的资源将继续保留并计费 24 小时，在这 24 小时内服务仍可正常使用。超过 24 小时后系统停止计费，模型部署进入欠费状态，底层资源将被删除，但模型部署任务仍会保留。补足欠费后，系统将重新分配资源并恢复使用（恢复后继续产生费用）。如果您不希望继续产生费用，可删除模型部署任务，删除成功后将不再计费。 -当模型输入超过最长输入 Token 或 超出购买的 TPM 量时，相关调用将自动切换为当前模型的按量付费模式。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 +当模型输入超过最长输入 Token 时，相关调用将自动切换为当前模型的按量付费模式；超出购买的 TPM 量时，按创建时选择的溢出策略处理（「自动溢出」切换为按量付费，「仅使用 PTU 容量」返回 429）。此时，推理性能可能下降，将受业务空间中当前快照模型的公共流量的管控，[费用](https://help.aliyun.com/zh/model-studio/model-pricing)按模型调用（按量付费）标准计收。 -- 此时，调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 +- 此时（仅「自动溢出」策略下），调用 API 返回 Header 将包含：`x-dashscope-ptu-overflow:true`。 - TPM 统计请前往：[模型监控（北京）](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)。 @@ -820,15 +820,33 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-35B-A3B qwen3.6-35b-a3b -MU8 x 1 +MU1 x 8 -¥47 +¥432 -¥22,400 +¥208,944 + +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 MU9 x 1 @@ -846,6 +864,12 @@ MU1 x 2 ¥52,236 +MU3 x 8 + +¥1,096 + +¥527,752 + 千问3.6-Plus-2026-04-02 qwen3.6-plus-2026-04-02 @@ -866,12 +890,6 @@ PD分离模式：¥417,888 qwen3.5-397b-a17b -MU2 x 8 - -¥504 - -¥240,288 - MU3 x 8 MU3 x 16（PD分离模式） @@ -884,6 +902,12 @@ PD分离模式：¥2,192 PD分离模式：¥1,055,504 +MU6 x 16 + +¥400 + +¥193,424 + 千问3.5-122B-A10B qwen3.5-122b-a10b @@ -894,11 +918,17 @@ MU1 x 4 ¥104,472 -MU2 x 8 +MU3 x 8 -¥504 +¥1,096 -¥240,288 +¥527,752 + +MU6 x 16 + +¥400 + +¥193,424 千问3.5-35B-A3B @@ -916,10 +946,40 @@ MU2 x 8 ¥240,288 +MU3 x 8 + +¥1,096 + +¥527,752 + +MU9 x 1 + +¥51 + +¥24,600 + 千问3.5-27B qwen3.5-27b +MU2 x 8 + +¥504 + +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 + +MU8 x 1 + +¥47 + +¥22,400 + MU9 x 1 ¥51 @@ -936,6 +996,12 @@ MU1 x 2 ¥52,236 +MU2 x 2 + +¥126 + +¥60,072 + MU8 x 1 ¥47 @@ -962,12 +1028,24 @@ MU1 x 2 qwen3.5-plus-2026-02-15 +MU1 x 8 + MU1 x 16（PD分离模式） +¥432 + PD分离模式：¥864 +¥208,944 + PD分离模式：¥417,888 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 8 MU3 x 16（PD分离模式） @@ -996,41 +1074,25 @@ MU2 x 8 ¥240,288 -千问3-Next-80B-A3B-Instruct - -qwen3-next-80b-a3b-instruct - -MU1 x 2 - -¥108 - -¥52,236 - 千问3-32B qwen3-32b -MU1 x 4 - -¥216 - -¥104,472 - -MU6 x 4 +MU6 x 16 -¥100 +¥400 -¥48,356 +¥193,424 -千问3-30B-A3B +千问3-30B-A3B-Thinking-2507 -qwen3-30b-a3b +qwen3-30b-a3b-thinking-2507 -MU9 x 2 +MU1 x 2 -¥102 +¥108 -¥49,200 +¥52,236 千问3-8B @@ -1080,12 +1142,6 @@ MU1 x 2 ¥52,236 -MU5 x 1 - -¥21 - -¥10,139 - 千问3-Embedding-0.6B qwen3-embedding-0.6b @@ -1154,6 +1210,16 @@ MU5 x 1 ¥10,139 +千问2.5-开源版-72B + +qwen2.5-72b-instruct + +MU1 x 8 + +¥432 + +¥208,944 + 千问2.5-开源版-32B qwen2.5-32b-instruct @@ -1190,26 +1256,6 @@ MU5 x 1 ¥10,139 -千问2.5-开源版-3B - -qwen2.5-3b-instruct - -MU5 x 1 - -¥21 - -¥10,139 - -千问-Flash-2025-07-28 - -qwen-flash-2025-07-28 - -MU1 x 4 - -¥216 - -¥104,472 - 千问-Plus-2025-07-28 qwen-plus-2025-07-28 @@ -1266,6 +1312,12 @@ GLM-5.1 glm-5.1 +MU2 x 8 + +¥504 + +¥240,288 + MU3 x 16（PD分离模式） PD分离模式：¥2,192 @@ -1298,6 +1350,16 @@ PD分离模式：¥800 PD分离模式：¥386,848 +GLM-4.7-Flash + +glm-4.7-flash + +MU3 x 16（PD分离模式） + +PD分离模式：¥2,192 + +PD分离模式：¥1,055,504 + ###### DeepSeek **模型名称** @@ -1318,11 +1380,11 @@ DeepSeek-v4-Flash deepseek-v4-flash -MU1 x 8 +MU3 x 8 -¥432 +¥1,096 -¥208,944 +¥527,752 DeepSeek-v3.2 @@ -1350,16 +1412,6 @@ PD分离模式：¥480,576 **最小计费：天** -MiniMax-M2.5 - -MiniMax-M2.5 - -MU1 x 16（PD分离模式） - -PD分离模式：¥864 - -PD分离模式：¥417,888 - Kimi-K2.5 kimi-k2.5 @@ -1402,15 +1454,21 @@ MU2 x 8 **最小计费：天** -千问3-VL-235B-A22B-Instruct +千问3-VL-32B-Instruct -qwen3-vl-235b-a22b-instruct +qwen3-vl-32b-instruct -MU1 x 4 +MU2 x 8 -¥216 +¥504 -¥104,472 +¥240,288 + +MU3 x 8 + +¥1,096 + +¥527,752 千问3-VL-8B-Instruct @@ -1422,6 +1480,12 @@ MU1 x 2 ¥52,236 +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-4B-Instruct qwen3-vl-4b-instruct @@ -1442,6 +1506,16 @@ MU5 x 1 ¥10,139 +千问3-VL-Embedding-2B + +qwen3-vl-embedding-2b + +MU5 x 1 + +¥21 + +¥10,139 + 千问3-VL-Flash-2025-10-15 qwen3-vl-flash-2025-10-15 @@ -1472,16 +1546,6 @@ MU6 x 4 ¥48,356 -千问VL-OCR-2025-11-20 - -qwen-vl-ocr-2025-11-20 - -MU6 x 4 - -¥100 - -¥48,356 - ###### 千问 Omni **模型名称** @@ -1562,6 +1626,14 @@ MU5 **元/千Token** +千问3.5-27B（邀测中） + +qwen3.5-27b + +¥0.0018 + +¥0.0048 + 千问3-32B qwen3-32b @@ -1592,6 +1664,16 @@ qwen3-8b 思考模式：¥0.005 +千问3-4B-Instruct-2507 + +qwen3-4b-instruct-2507 + +¥0.0003 + +非思考模式：¥0.0012 + +思考模式：¥0.003 + 千问2.5-开源版-72B qwen2.5-72b-instruct @@ -1624,6 +1706,14 @@ qwen2.5-7b-instruct ¥0.001 +千问2-开源版-7B + +qwen2-7b-instruct + +¥0.001 + +¥0.002 + ##### 千问VL **基础模型** @@ -1670,6 +1760,14 @@ qwen2.5-vl-7b-instruct ¥0.005 +千问2.5-VL-3B-Instruct + +qwen2.5-vl-3b-instruct + +¥0.0012 + +¥0.0036 + ### **图像生成模型-万相** 经过SFT-LoRA高效微调的万相图像生成模型，部署免费，调用按微调的基础模型的标准调用价格计费。模型训练和部署流程请参见[图像生成模型调优](https://help.aliyun.com/zh/model-studio/wan-image-generation-finetune-guide)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md index 6e121bcc..f8407dc6 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/new-free-quota.md @@ -10,11 +10,11 @@ ### 有效期 -免费额度的有效期为 30～90 天，从开通阿里云百炼或模型申请通过之日起计算。额度到期或耗尽后，继续调用模型推理服务将[产生计费](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 +免费额度的有效期为 90 天，从开通阿里云百炼、模型发布或模型申请通过之日起计算（以较晚者为准）。额度到期或耗尽后，继续调用模型推理服务将[产生计费](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 **重要** -自**2025年9月8日11点**起，首次开通阿里云百炼的用户，获赠的新人免费额度有效期调整为 90 天，在此之前已开通的用户不受影响，详情参考[阿里云百炼新人免费额度有效期调整通知](https://help.aliyun.com/zh/model-studio/new-free-quota-validity-adjustment)。 +**2025年9月8日11点**前已开通阿里云百炼的用户，免费额度有效期可能不足90天；在此之后开通的用户有效期为90天。详情参考[阿里云百炼新人免费额度有效期调整通知](https://help.aliyun.com/zh/model-studio/new-free-quota-validity-adjustment)。 免费额度过期后自动失效，不支持补发、延期或重置： @@ -212,9 +212,12 @@ 免费额度列显示**无免费额度**或**免费额度**区域不显示，可能由以下原因之一导致： -- **免费额度已到期或耗尽**：免费额度的有效期为 30～90 天，从开通阿里云百炼或模型申请通过之日起计算，到期或耗尽后将不再显示，继续调用模型将产生计费。 +- **免费额度已到期或耗尽**：免费额度的有效期为 90 天，从开通阿里云百炼、模型发布或模型申请通过之日起计算（以较晚者为准），到期或耗尽后将不再显示，继续调用模型将产生计费。 + - **该模型所在地域或服务部署范围不享有免费额度**：仅华北2（北京）地域且服务部署范围为中国内地的模型、以及仅新加坡地域且服务部署范围为国际的模型享有免费额度，其他地域和部署范围无免费额度。 + - **该模型本身不提供免费额度**：部分模型不参与新人免费额度发放。 + ### 使用哪种 API Key 才能消耗免费额度？ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md index 5ce9dfe5..a1867096 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/savings-plan-and-resource-package.md @@ -80,7 +80,7 @@ AI 通用型节省计划是针对大模型按量付费使用场景设计的折 - C 类：qwen3.6-max-preview、DeepSeek、Kimi、GLM、MiniMax、HappyHorse - > 三方直供模型不支持抵扣，详情参见[三方直供模型支持抵扣 AI 通用型节省计划吗？](#85a29cab67489) + > 三方直供模型不支持抵扣，其中 DeepSeek、Kimi、GLM 已有阿里云直供版本可抵扣，MiniMax 暂无阿里云直供版本。详情参见[三方直供模型支持抵扣 AI 通用型节省计划吗？](#85a29cab67489) **每月承诺消费金额范围** @@ -802,7 +802,7 @@ ASR模型按秒计费，TTS模型按字符计费，请前往[百炼控制台](ht - 根据[退订规则](https://help.aliyun.com/zh/user-center/cancel-subscription/)，预付费商品未发生使用的部分，可按未使用额度费用[申请退款](https://billing-cost.console.aliyun.com/refund/refund?commodityType=RESOURCE_PLANS&refundType=NOREASON_REFUND)；已使用的部分则无法退款。 -**使用限制**：资源包按模型名称严格匹配，**跨版本或子型号不通用**，请以资源包购买页标注的适用模型为准。例如，qwen-plus 资源包不支持抵扣 qwen-max 或 qwen-turbo 的调用费用；若需要同时覆盖多个模型版本的调用费用，建议选择 AI 通用型节省计划。 +**使用限制**：资源包按模型名称严格匹配，**跨版本或子型号不通用**，请以资源包购买页标注的适用模型为准。例如，qwen-plus 资源包不支持抵扣 qwen-max 的调用费用；若需要同时覆盖多个模型版本的调用费用，建议选择 AI 通用型节省计划。 ### **大语言模型推理资源包** @@ -812,48 +812,36 @@ ASR模型按秒计费，TTS模型按字符计费，请前往[百炼控制台](ht [大语言模型推理资源包 qwen-max](https://common-buy.aliyun.com/?commodityCode=sfm_llminference2_dp_cn#/buy) -[大语言模型推理资源包 qwen-turbo](https://common-buy.aliyun.com/?commodityCode=sfm_llminference3_dp_cn#/buy) - **适用地域** 华北2（北京） 华北2（北京） -华北2（北京） - **适用模型** qwen-plus 及 qwen-plus-latest的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） qwen-max的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） -qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)） - **包含输入和输出总Tokens** 1,200万/1.1亿 1,800万/3,900万/3.9亿/11.7亿/19.5亿 -3,500万/3.5亿/17.5亿/35亿 - **价格（元）** 11.66/114.4 57.6/125/1250/3750/6250 -11.45/114.45/572.25/1144.5 - **有效期** 自购买日起生效，有效期可选 3 个月、6 个月或 1 年。 自购买之日起有效期为 1 年。 -自购买之日起有效期为 1 年。 - **使用限制** - **qwen-plus**、**qwen-plus-latest** @@ -866,7 +854,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m - [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)、[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)、[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)产生的费用。 -- **qwen-max**、**qwen-turbo** +- **qwen-max** - 仅支持抵扣实时推理产生的费用（[非思考模式](https://help.aliyun.com/zh/model-studio/deep-thinking)，包含输入和输出），不支持抵扣[Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)、[上下文缓存](https://help.aliyun.com/zh/model-studio/context-cache)、[模型调优](https://help.aliyun.com/zh/model-studio/model-training-overview)、[模型部署](https://help.aliyun.com/zh/model-studio/model-deployment-introduction)产生的费用。 @@ -895,7 +883,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m **图像编辑**：qwen-image-edit-plus -**资源包容量 (生成图片张数)** +**资源包容量（生成图片张数）** 80/400 @@ -962,7 +950,7 @@ qwen-turbo的实时推理服务（[非思考模式](https://help.aliyun.com/zh/m ### **三方直供模型支持抵扣 AI 通用型节省计划吗？** -[C 类模型](#ho1f5x10wuun0)中，阿里直供的模型支持抵扣，三方直供的模型不支持抵扣。可以在[百炼模型广场](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)中通过模型卡片右上角标识（如"阿里直供"或"三方直供"标签）判断。 +[C 类模型](#ho1f5x10wuun0)中，阿里直供的模型支持抵扣，三方直供的模型不支持抵扣。目前，DeepSeek、Kimi、GLM 有阿里云直供版本，可通过 AI 通用型节省计划抵扣；MiniMax 暂无阿里云直供版本，暂不支持通过节省计划抵扣。可以在[百炼模型广场](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)中通过模型卡片右上角标识（如"阿里直供"或"三方直供"标签）查看最新的直供模型列表。 ### **购买节省计划后如何使用？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md index 1ab71393..57403f95 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md @@ -50,11 +50,11 @@ Coding Plan 整合了千问、GLM、Kimi 、MiniMax顶级模型，并兼容主 - 每月**90,000** 次请求 -- **限时优惠：**活动已结束，当前价格以下单页为准。 +- **限时优惠**：活动已结束，当前价格以下单页为准。 - **限量抢购**：名额有限、先到先得。每日 09:30:00（UTC+08:00）补充，可前往[Coding Plan 页面](https://www.aliyun.com/benefit/scene/codingplan)抢购。 -- **额度消耗：**单次提问将按实际“模型调用次数”扣除额度。简单任务约消耗 5-10 次，复杂任务约 10-30+ 次，实际消耗受任务难度、上下文及工具使用影响。在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)可以查看用量。 +- **额度消耗：**单次提问将按实际“模型调用次数”扣除额度。简单任务约消耗 5-10 次，复杂任务约 10-30+ 次，实际消耗受任务难度、上下文及工具使用影响。在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)可以查看用量。 - **额度恢复**： @@ -84,7 +84,7 @@ Coding Plan 整合了千问、GLM、Kimi 、MiniMax顶级模型，并兼容主 您需要获取并配置套餐专属的 API Key 和 Base URL，才能正确使用并抵扣套餐额度。 -- **API Key**：在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，获取Coding Plan 专属 API Key（格式为`sk-sp-xxxxx`）。 +- **API Key**：在[Coding Plan 页面](https://bailian.console.aliyun.com/cn-beijing/?tab=plan#/efm/subscription/coding-plan)，获取Coding Plan 专属 API Key（格式为`sk-sp-xxxxx`）。 - **Base URL**：后续需在 AI 工具中配置以下其中一个Base URL（因工具而异），具体操作请参见对应的AI工具文档。 @@ -99,33 +99,33 @@ Coding Plan 专属的 API Key 和 Base URL 与百炼按量计费的 API Key（`s ### **步骤三：接入AI工具** - [**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) +[**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) - [**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) +[**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) - [**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) +[**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) - [**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) +[**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) - [**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) +[**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) - [**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) +[**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) - [**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) +[**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) - [**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) +[**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) - [**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) +[**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) - [**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) +[**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) - [**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) +[**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) - [**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) +[**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) - [**Lingma**阿里云推出的智能编码辅助工具](https://help.aliyun.com/zh/model-studio/lingma-agent) +[**Lingma**阿里云智能编码助手，提供独立 IDE](https://help.aliyun.com/zh/model-studio/lingma-agent) - [**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) +[**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) [··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md similarity index 77% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md index 23651b2d..dad44bd5 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md @@ -1,16 +1,16 @@ # 添加视觉理解能力 -百炼 Coding Plan 中的部分模型（qwen3.6-plus、qwen3.5-plus、kimi-k2.5）原生支持视觉理解，可直接处理图片输入。对于 glm-5、MiniMax-M2.5 等纯文本模型，可通过添加本地 Skill 使其获得视觉能力。 +Token Plan 支持的部分模型（qwen3.7-plus 等）原生支持视觉理解，可直接处理图片输入。对于 glm-5、MiniMax-M2.5 等纯文本模型，可通过添加本地 Skill 使其获得视觉能力。 **说明** -运行图片理解 Skill 会消耗 Coding Plan 额度，无其他收费项。 +运行图片理解 Skill 会消耗 Token Plan Credits，无其他收费项。 ## 前提条件 -1. 已订阅 [Coding Plan](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 +1. 已订阅 [Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription)。 -2. 已在 Coding Plan 工具中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 +2. 已在 AI 工具中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 ## 视觉支持情况 @@ -21,11 +21,13 @@ **说明** -- qwen3.6-plus +- qwen3.8-max-preview + +- qwen3.7-plus -- qwen3.5-plus +- qwen3.6-plus -- kimi-k2.5 +- kimi-k2.5 等 是 @@ -51,7 +53,7 @@ ## 方法 1：直接使用视觉模型（推荐） -qwen3.6-plus、qwen3.5-plus 和 kimi-k2.5 具备视觉理解能力。如果经常需要处理图片，直接切换到这些模型是最简单、推荐的做法。 +qwen3.7-plus 等模型具备视觉理解能力。如果经常需要处理图片，直接切换到这些模型是最简单、推荐的做法。 **工具** @@ -59,15 +61,15 @@ qwen3.6-plus、qwen3.5-plus 和 kimi-k2.5 具备视觉理解能力。如果经 Claude Code -`/model qwen3.6-plus`或`/model qwen3.5-plus`或 `/model kimi-k2.5` +`/model qwen3.7-plus`或`/model qwen3.6-plus`或`/model qwen3.5-plus`或 `/model kimi-k2.5` OpenCode -`/models`→ 搜索并选择`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` +`/models`→ 搜索并选择`qwen3.7-plus`或`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` Qwen Code -`/model`→ 选择`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` +`/model`→ 选择`qwen3.7-plus`或`qwen3.6-plus`或`qwen3.5-plus`或`kimi-k2.5` 更多编程工具中的模型切换方式请参考[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。切换后可直接在对话中引用图片路径，或拖拽/粘贴图片。 @@ -91,9 +93,9 @@ Qwen Code --- name: image-analyzer description: 帮助没有视觉能力的模型进行图像理解。当需要分析图像内容、提取图片中的信息、文字、界面元素，或理解截图、图表、架构图等任何视觉内容时，使用此技能，传入图片路径即可获得描述信息。 - model: qwen3.6-plus + model: qwen3.7-plus --- - qwen3.6-plus具有视觉理解能力，请直接使用qwen3.6-plus模型进行图片理解。 + qwen3.7-plus具有视觉理解能力，请直接使用qwen3.7-plus模型进行图片理解。 ``` 创建完成后的目录结构如下： @@ -111,7 +113,7 @@ Qwen Code 2. 下载[aliyun.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260225/hxwnny/aliyun.png)到项目目录下，并提问：`请加载image-analyzer skill，描述一下 aliyun.png banner位置是什么信息。`可收到如下回复： - ![image.png](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5928202771/p1054884.png) + aliyun.png 为阿里云官网首页截图，banner 区域标题为 **Coding Plan 已支持 Qwen3.5**，正文介绍阿里云百炼支持 Qwen3.5、Kimi-k2.5、GLM-4.7 等模型，新客首月仅 7.9 元，页面提供**立即订阅**和**在线咨询**入口。 ### OpenCode @@ -128,13 +130,13 @@ Qwen Code **说明** - model 字段必须使用 OpenCode 配置文件中定义的 provider 和模型名称。参考 [OpenCode](https://help.aliyun.com/zh/model-studio/opencode) 文档的配置示例，应为`bailian-coding-plan/qwen3.6-plus`。 + model 字段必须使用 OpenCode 配置文件中定义的 provider 和模型名称。参考 [OpenCode](https://help.aliyun.com/zh/model-studio/opencode) 文档的配置示例，应为`bailian-token-plan/qwen3.7-plus`。 ``` --- description: Analyzes images using a vision-capable model. Use this agent when the user needs to understand image content, extract information from screenshots, diagrams, UI mockups, or any visual content. Invoke with @image-analyzer followed by the image path and your question. mode: subagent - model: bailian-coding-plan/qwen3.6-plus + model: bailian-token-plan/qwen3.7-plus tools: write: false edit: false @@ -156,7 +158,18 @@ Qwen Code 2. 下载[aliyun.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260225/hxwnny/aliyun.png)到项目目录下，通过`@`唤起`image-analyzer`并提问：`@image-analyzer，描述一下 aliyun.png banner位置是什么信息。`可收到如下回复： - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6472262771/p1055847.png) + ``` + Banner位置包含： + 左侧： + - 汉堡菜单图标 + - 阿里云橙色logo和"阿里云"文字 + - 导航菜单：大模型、产品、解决方案、权益、定价、云市场、伙伴、服务、了解阿里云 + 右侧： + - 搜索框(显示"大模型") + - 图标：蓝色圆圈、地球、耳机 + - 链接：文档、备案、控制台 + Build · glm-5 · 37.0s + ``` ## **常见问题** @@ -167,17 +180,17 @@ Qwen Code **解决方案**：在 OpenCode 配置文件的模型定义中添加 `modalities` 字段，将 `input` 设为 `["text", "image"]`，如下所示： -> 将sk-sp-xxx替换为Coding Plan API Key。 +> 将sk-sp-xxx替换为Token Plan API Key。 ``` { "$schema": "https://opencode.ai/config.json", "provider": { - "bailian-coding-plan-test": { + "bailian-token-plan": { "npm": "@ai-sdk/anthropic", - "name": "Model Studio Coding Plan", + "name": "Model Studio Token Plan", "options": { - "baseURL": "https://coding.dashscope.aliyuncs.com/apps/anthropic/v1", + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "sk-sp-xxx" }, "models": { @@ -255,7 +268,7 @@ Qwen Code "mode": "merge", "providers": { "bailian": { - "baseUrl": "https://coding.dashscope.aliyuncs.com/v1", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", "apiKey": "YOUR_API_KEY", "api": "openai-completions", "models": [ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md new file mode 100644 index 00000000..88c71bd8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md @@ -0,0 +1,79 @@ +# 接入 Harness 工具 + +Token Plan 支持的部分 Qwen 模型内置 Harness 工具，可为 AI 编程工具扩展联网搜索、代码解释器、网页抓取等能力。 + +**说明** + +适用于 Token Plan，不适用于 Coding Plan。 + +## **工具概览** + +**工具** + +**说明** + +联网搜索 + +检索互联网信息，结合搜索结果生成回答 + +代码解释器 + +在沙箱环境中编写与运行 Python 代码，用于数学计算、数据分析等场景 + +网页抓取 + +访问指定 URL 并提取内容，为大模型提供所需信息 + +以图搜图 + +根据输入图片从互联网搜索视觉相似的图片，适用于以图找同款、视觉内容溯源等场景 + +文搜图 + +根据文本描述从互联网搜索相关图片，适用于可视化问答、配图推荐等场景 + +## **支持的模型和工具** + +### **个人版** + +**模型** + +**支持的工具** + +qwen3.8-max-preview + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +qwen3.7-max + +联网搜索、代码解释器、网页抓取 + +qwen3.7-plus + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +### **团队版** + +**模型** + +**支持的工具** + +qwen3.8-max-preview + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +qwen3.7-max + +联网搜索、代码解释器、网页抓取 + +qwen3.7-plus + +联网搜索、代码解释器、网页抓取、以图搜图、文搜图 + +## **费用说明** + +Harness 工具按成功调用次数计费，费用从套餐 Credits 中抵扣。 + +## **使用方式** + +将 AI 编程工具的模型切换为上述支持 Harness 的 Qwen 模型，在对话中直接提问即可。模型会根据问题自动调用相应的内置工具，无需额外配置。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md index 4daf85ea..55d418ac 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md @@ -1,14 +1,6 @@ # 接入多模态生成模型 -图像生成模型需通过工具的扩展机制（Skill、Slash Command 或 Agent）接入。 - -## **前提：获取套餐专属凭证** - -在控制台「我的订阅」打开 Token Plan 套餐详情页，接入信息卡片展示套餐专属 API Key（以 `sk-sp-` 为前缀，掩码显示），支持生成、重置与复制 API Key。 - -**说明** - -套餐详情页「可使用模型」以文本、编程模型为主；图像生成模型不在该列表展示，需通过 `multimodal-generation` API 调用。 +Token Plan 中的图像生成、视频生成模型需通过工具的扩展机制（Skill、Slash Command 或 Agent）接入。 ## **示例：在 Claude Code 中接入图像生成模型** @@ -27,7 +19,7 @@ ## 步骤 -1. 从用户需求中提取 prompt（图片描述）、model、size（默认 1024*1024）。若用户明确指定了模型（如“模型=wan2.7-image”或“用 wan2.7-image 画”），必须严格使用用户指定的模型名，不要回退到默认模型；仅当用户未指定模型时才使用默认 qwen-image-2.0。常用图像生成模型有 qwen-image-2.0、qwen-image-2.0-pro、wan2.7-image、wan2.7-image-pro、z-image-turbo 等，完整列表以百炼模型列表为准。 +1. 从用户需求中提取 prompt（图片描述）、model、size（默认 1024*1024）。若用户明确指定了模型（如“模型=wan2.7-image”或“用 wan2.7-image 画”），必须严格使用用户指定的模型名，不要回退到默认模型；仅当用户未指定模型时才使用默认 qwen-image-2.0。常用图像生成模型有 qwen-image-2.0、qwen-image-2.0-pro、wan2.7-image、wan2.7-image-pro 等，完整列表以百炼模型列表为准。 2. 调用 API 生成图片（使用 Bash 工具执行 curl）： @@ -55,9 +47,73 @@ curl -s -X POST "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/services 在 Claude Code 中输入 `/text-to-image 画一只猫`。如需使用默认模型以外的图像生成模型，在指令中写明模型名即可，例如 `/text-to-image 用 wan2.7-image 画一只猫`。 +## **示例：在 Claude Code 中接入视频生成模型** + +以 Claude Code 为例，通过 Slash Command 接入视频生成模型。视频生成为异步接口，流程为"提交任务 → 轮询状态 → 下载视频"。 + +### **步骤一：创建 Slash Command** + +将套餐专属 API Key（以 `sk-sp-` 为前缀）配置为环境变量 `$ANTHROPIC_AUTH_TOKEN`，供后续 curl 鉴权使用。 + +在项目根目录创建 `.claude/commands/text-to-video.md`，写入以下内容： + +``` +调用 Token Plan 文生视频 API，根据描述生成视频并自动下载到本地。 + +用户需求：$ARGUMENTS + +## 步骤 + +1. 从用户需求中提取 prompt（视频描述）、model（默认 happyhorse-1.1-t2v）、resolution（默认 720P）、ratio（默认 16:9）、duration（默认 5 秒）。若用户明确指定了模型（如"模型=happyhorse-1.0-t2v"），必须严格使用用户指定的模型名。 + +2. 使用 Bash 工具执行以下脚本，一次性完成提交任务、等待完成、下载视频： + +```bash +#!/bin/bash +set -e + +TASK_RESPONSE=$(curl -s -X POST "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis" \ + -H "X-DashScope-Async: enable" \ + -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "", + "input": {"prompt": ""}, + "parameters": {"resolution": "", "ratio": "", "duration": } + }') + +TASK_ID=$(echo "$TASK_RESPONSE" | grep -o '"task_id":"[^"]*"' | head -1 | cut -d'"' -f4) +if [ -z "$TASK_ID" ]; then echo "提交失败: $TASK_RESPONSE"; exit 1; fi +echo "任务已提交，ID: $TASK_ID，等待生成..." + +while true; do + sleep 15 + STATUS_RESPONSE=$(curl -s "https://token-plan.cn-beijing.maas.aliyuncs.com/api/v1/tasks/$TASK_ID" \ + -H "Authorization: Bearer $ANTHROPIC_AUTH_TOKEN") + STATUS=$(echo "$STATUS_RESPONSE" | grep -o '"task_status":"[^"]*"' | cut -d'"' -f4) + if [ "$STATUS" = "SUCCEEDED" ]; then + VIDEO_URL=$(echo "$STATUS_RESPONSE" | grep -o '"video_url":"[^"]*"' | cut -d'"' -f4) + OUTPUT="generated_$(date +%Y%m%d_%H%M%S).mp4" + curl -s -o "$OUTPUT" "$VIDEO_URL" + echo "视频已下载: $(pwd)/$OUTPUT" + exit 0 + elif [ "$STATUS" = "FAILED" ]; then + echo "生成失败: $STATUS_RESPONSE"; exit 1 + fi + echo "生成中..." +done +``` + +3. 向用户展示生成的视频文件路径。 +``` + +### **步骤二：生成视频** + +在 Claude Code 中输入 `/text-to-video 一只白色的猫在阳台上晒太阳`。如需使用其他视频生成模型，在指令中写明模型名即可，例如 `/text-to-video 用 happyhorse-1.1-r2v 生成一只猫跳跃的视频`。 + ## **其他工具** -控制台套餐详情页「快速接入 AI 编程工具」入口提供 Qwen Code、Qoder、OpenClaw、Claude Code、OpenCode 等工具的接入文档。不同工具的扩展机制和配置文件路径如下表所示，该表为支持扩展机制的主流 AI 编程工具示例。将上述 Claude Code 示例中的配置内容保存到对应路径即可。 +不同工具的扩展机制和配置文件路径如下表所示，该表为支持扩展机制的主流 AI 编程工具示例。将上述 Claude Code 示例中的配置内容保存到对应路径即可。 工具 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md deleted file mode 100644 index aaf9db77..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md +++ /dev/null @@ -1,253 +0,0 @@ -# 工具调用 - -Token Plan 团队版支持通过模型内置工具和 MCP 服务两种方式为 AI 编程工具扩展能力，如联网搜索、代码解释器、网页抓取等。 - -## **工具概览** - -Token Plan 团队版提供两种方式接入工具： - -- **模型内置工具**：qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 模型的 Responses API 内置了联网搜索、代码解释器、网页抓取、以图搜图、文搜图五种工具。启用后，模型会在需要时自动调用相应工具。 - -- **MCP 服务**：其他模型（如 deepseek-v3.2、glm-5 等）可通过百炼 MCP 广场的 MCP 服务获取工具能力。本文以联网搜索 MCP 为例说明接入方式，其他 MCP 服务的接入方式类似。 - - -## **费用说明** - -### **模型内置工具** - -qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 模型内置工具的费用可通过 Token Plan 团队版抵扣，内置工具不额外收费，产生的 token 消耗统一从套餐 Credits 中抵扣。具体价格以[控制台模型详情页](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/detail/qwen3.6-plus)为准。 - -### **MCP 服务** - -百炼 MCP 广场提供联网搜索、代码解释器、网页抓取等 MCP 服务。联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费；其他 MCP 服务部分**限时免费**，每月提供一定免费额度。具体价格以[MCP 广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)各服务详情页为准。 - -## **使用方式** - -### **使用 qwen3.7-max / qwen3.7-plus / qwen3.6-plus / qwen3.6-flash 模型内置工具** - -将 AI 工具的模型设置为 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-plus` 或 `qwen3.6-flash`，在对话中直接提问即可。模型会根据问题自动调用相应的内置工具： - -**工具** - -**说明** - -联网搜索 - -检索互联网信息，结合搜索结果生成回答 - -代码解释器 - -调用模型时启用内置的 Python 代码解释器，可使模型在沙箱环境里编写与运行 Python 代码，以解决数学计算、数据分析等复杂问题。 - -网页抓取 - -网页抓取工具可以访问指定 URL 并提取内容，为大模型提供所需信息。 - -以图搜图 - -图搜图工具使模型能够根据输入图片从互联网搜索视觉相似的图片，并基于搜索结果进行分析和推理，适用于以图找同款、视觉内容溯源等场景。 - -文搜图 - -文搜图工具使模型能够根据文本描述从互联网搜索相关图片，并基于图片内容进行描述和推理，适用于可视化问答、配图推荐等场景。 - -### **通过 MCP 服务接入工具** - -其他模型可通过百炼 MCP 广场的 MCP 服务获取工具能力。以下以联网搜索 MCP 为例说明接入方式。 - -#### **前提条件** - -已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Token Plan 团队版专属 API Key（格式为 sk-sp-xxx）不同。 - -#### **开通 MCP 服务** - -1. 进入百炼的[MCP 广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)，找到需要的 MCP 服务（如联网搜索）。 - -2. 点击**立即开通**，确认开通。 - -3. 开通成功后，获取以下配置信息： - - - **Streamable HTTP Endpoint**：MCP 服务的连接地址。 - - - **API Key**：即百炼 API Key，是 MCP 服务的鉴权密钥。 - - -#### **接入工具** - -将 MCP 服务添加到 AI 编程工具中。以下以联网搜索 MCP 为例，示例中的 `YOUR_API_KEY` 需替换为百炼 API Key。接入其他 MCP 服务时，将 Endpoint 地址替换为对应服务的地址即可。 - -## OpenClaw - -1. 在终端执行如下命令安装 MCPorter。 - - ``` - npm install -g mcporter - ``` - -2. 在终端执行如下命令启用 MCPorter。 - - ``` - openclaw config set skills.entries.mcporter.enabled true - ``` - -3. 在 `~/.openclaw/workspace` 目录下，执行如下命令添加联网搜索 MCP。 - - ``` - mcporter config add WebSearch https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp --transport http --header "Authorization=Bearer YOUR_API_KEY" - ``` - -4. 执行如下命令确认 MCP 已安装。 - - ``` - mcporter list - ``` - -5. 执行如下命令使配置生效。 - - ``` - openclaw gateway restart - ``` - -6. 发送提问 `用 mcporter 搜索阿里云的新闻` 即可看到搜索结果。 - - -## OpenCode - -1. 在配置文件 `~/.config/opencode/opencode.json` 中写入 MCP 配置信息。 - - ``` - { - "mcp": { - "WebSearch": { - "type": "remote", - "httpUrl": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 若 `opencode.json` 中已有其他配置（如 provider），将 mcp 字段合并到现有配置中即可。 - -2. 在终端执行以下命令进入 OpenCode。 - - ``` - opencode - ``` - -3. 在对话框执行 `/mcps` 确认 `websearch` 状态为 Enabled。确认后按 Esc 退出。 - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Claude Code - -1. 在终端执行以下命令添加联网搜索 MCP 服务。 - - ``` - claude mcp add WebSearch https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp -t http -H "Authorization: Bearer YOUR_API_KEY" - ``` - - 终端返回 `Added SSE MCP server xx` 即表示添加成功。 - -2. 执行以下命令进入 Claude Code。 - - ``` - claude - ``` - -3. 在对话框执行 `/mcp` 命令，确认 `websearch` 的状态为 connected。 - -4. 按 Esc 退出 MCP 列表后，发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Qwen Code - -1. 在终端执行以下命令添加联网搜索 MCP。 - - ``` - qwen mcp add WebSearch \ - -t http \ - "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp" \ - -H "Authorization: Bearer YOUR_API_KEY" - ``` - -2. 在终端执行以下命令进入 Qwen Code。 - - ``` - qwen - ``` - -3. 在对话框执行 `/mcp` 命令确认 MCP 连接状态。 - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Kilo CLI - -1. 在配置文件 `~/.config/kilo/opencode.json` 中写入 MCP 配置信息。 - - ``` - { - "mcp": { - "websearch": { - "type": "remote", - "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 若 `opencode.json` 中已有其他配置（如 provider），将 mcp 字段合并到现有配置中即可。 - -2. 在终端执行以下命令查看 MCP 状态。Connected 即表示连接成功。 - - ``` - kilocode mcp list - ``` - -3. 在终端执行以下命令进入 Kilo CLI。 - - ``` - kilo - ``` - -4. 发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - - -## Kilo Code IDE 插件 - -1. 打开 Kilo Code IDE 插件，配置联网搜索 MCP 信息。 - - ``` - { - "mcpServers": { - "websearch": { - "type": "streamable-http", - "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", - "headers": { - "Authorization": "Bearer YOUR_API_KEY" - } - } - } - } - ``` - - 当联网搜索的 MCP 状态显示为绿色时，表示添加成功。 - -2. 返回对话界面，发送提问 `用 websearch MCP 搜索阿里云的新闻` 即可看到搜索结果。 - - > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md similarity index 51% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md index 3208f230..5fd78221 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md @@ -1,6 +1,6 @@ # 联网搜索 -在 Coding Plan 支持的编程工具中添加联网搜索工具，使模型能够检索实时信息。 +在 Token Plan 支持的编程工具中添加联网搜索工具，使模型能够检索实时信息。 ## 适用范围 @@ -8,11 +8,11 @@ ## 前提条件 -1. 已订阅 [Coding Plan](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 +1. 已订阅 [Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription)，详情请参见[快速开始](https://help.aliyun.com/zh/model-studio/coding-plan-quickstart)。 -2. 已在 Coding Plan 工具（如 Claude Code、Qwen Code）中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 +2. 已在 Token Plan 支持的工具（如 Claude Code、Qwen Code）中完成接入配置，且能正常对话，详情请参见[接入客户端/开发工具](https://help.aliyun.com/zh/model-studio/use-chat-client-or-development-tool/)。 -3. 已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Coding Plan 专属 API Key（格式为 sk-sp-xxx）不同。 +3. 已获取[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。此处的 API Key 为百炼通用 API Key（格式为 sk-xxx），用于调用 MCP 服务，与 Token Plan 专属 API Key（格式为 sk-sp-xxx）不同。 ## 开通或升级联网搜索 MCP @@ -27,6 +27,8 @@ - 联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费。如果使用第三方 MCP 服务，该 MCP 服务可能收费，以 MCP 服务的介绍信息为准。 + - 部分MCP服务支持**个人FC资源部署**，按实际调用时长和次数计费，适用于需要专属资源、指定资源地域等场景。 + 3. 开通成功后，可以获取以下配置信息： 1. **Streamable HTTP Endpoint**：MCP 服务的连接地址。联网搜索 MCP 的连接地址为`https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 @@ -38,10 +40,12 @@ 1. 进入百炼的[MCP广场](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/mcp-market)，找到**联网搜索** MCP 服务。 -2. 单击右侧**取消开通**，再单击**立即开通**。 +2. 单击右侧**取消开通**，再单击**立即开通** > **确认开通**。 - 联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费。如果使用第三方 MCP 服务，该 MCP 服务可能收费，以 MCP 服务的介绍信息为准。 + - 部分MCP服务支持**个人FC资源部署**，按实际调用时长和次数计费，适用于需要专属资源、指定资源地域等场景。 + 3. 重新开通成功后，即完成协议升级，可以获取以下配置信息： 1. **Streamable HTTP Endpoint**：MCP 服务的连接地址。联网搜索 MCP 的连接地址为`https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 @@ -78,8 +82,11 @@ ``` mcporter list ``` - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9753812771/p1055340.png) + ``` + mcporter 0.7.3 — Listing 1 server(s) (per-server timeout: 30s) + - websearch (1 tool, 2.4s) + [ok] Listed 1 server (1 healthy). + ``` 5. 在终端执行如下命令使配置生效。 @@ -89,7 +96,7 @@ 6. 发送提问`用 mcporter 搜索阿里云的新闻`即可看到搜索结果。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9753812771/p1055129.png) + 系统返回包含 4 条阿里云相关新闻的卡片，每条新闻显示标题、来源及发布时间，底部提示是否需要抓取某篇新闻的详细内容。 ## OpenCode @@ -120,13 +127,27 @@ 3. 在对话框执行`/mcps`确认`websearch`状态是否为 Enabled。确认后按 Esc 退出。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4753812771/p1054875.png) - 4. 发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054876.png) + ``` + 已为您搜索到最新的阿里云相关新闻： + 主要新闻摘要 + 1. 阿里云2026年战略目标 + 阿里云目标在2026年拿下中国AI云市场增量的80%，重点发力MaaS（模型即服务）和GPU算力服务。 + 目前阿里云上调用大模型API的客户中，有七成同时使用其GPU算力服务。 + 2. 阿里云服务器2026年优惠活动 + - 最便宜服务器：38元/年（2核2G、200M峰值带宽） + - 99元/年 ECS服务器（新老用户同享，续费同价） + - 199元/年企业专享服务器（2核4G、5M带宽） + 3. AI自动化工具生态 + 阿里云推出OpenClaw（原Clawdbot）AI + 自动化代理工具部署方案，支持一键部署到轻量应用服务器，可与阿里云百炼大模型无缝联动。 + 4. 企业智能客服 + 瓴羊Quick Service智能客服平台基于通义千问大模型，提供大模型驱动的语义理解、多轮对话、 + 知识库智能生成等功能。 + ``` ## Claude Code @@ -155,17 +176,54 @@ 3. 在对话框执行`/mcp`命令，确认`websearch`的状态为 connected。首次添加可能需要等待状态从 connecting 变成 connected。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054987.png) - 若连接状态显示 failed，请选中该 MCP 并选择 Reconnect 重连。若重试 1-2 次仍失败，请核实配置。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0613342771/p1055684.png) + ``` + /mcp + WebSearch MCP Server + Status: [ok] connected + Auth: [ok] authenticated + URL: https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp + xxx + Capabilities: tools + Tools: 1 tools + 1. View tools + 2. Re-authenticate + 3. Clear authentication + > 4. Reconnect + 5. Disable + ``` 4. 按 Esc 退出 MCP 列表后，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054990.png) + ``` + > 用websearch MCP搜索阿里云的新闻 + * websearch – bailian_web_search (MCP)(query: "阿里云 新闻", count: 10) + └ { + "status": 0, + "pages": [ + … +51 lines (ctrl+o to expand) + ] + } + * 搜索结果已返回，以下是阿里云的最新新闻摘要： + 阿里云近期新闻 + 1. 编程模型订阅服务推出（2026 年 2 月 25 日） + 阿里云百炼推出 Coding Plan 编程模型订阅服务，包含四款开源模型 API： + - Qwen3.5、GLM-5、MiniMax M2.5、Kimi K2.5 + - 用户可在 Qwen Code、Claude Code、Cline 等 AI 工具上无缝切换使用 + - 降低开发者模型选型与接入成本 + 2. 千问 3.5 新模型开源（2026 年 2 月 25 日） + 阿里发布三款中型千问 3.5 新模型： + - Qwen3.5-35B-A3B、Qwen3.5-122B-A10B、Qwen3.5-27B + - 基于 Qwen3.5-35B-A3B 的托管模型 Qwen3.5-Flash 已上线阿里云百炼 + - 每百万 Token 输入低至 0.2 元 + 3. 市场表现亮眼 + - 千问（Qwen）大模型调用量占比跃升至 32.1%，排名中国企业级大模型第一 + - 2026 财年第二季度收入 398.24 亿元，同比增长 34% + - AI 相关产品收入连续第九个季度实现三位数增长 + ``` ## Qwen Code @@ -197,13 +255,80 @@ 3. 在`Qwen Code`对话框执行`/mcp`命令以确认 MCP 连接状态。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/4753812771/p1054878.png) + ``` + > /mcp + Configured MCP servers: + * WebSearch – Ready (1 tool) + Tools: + – mcp__WebSearch__bailian_web_search + [tip] Tips: + – Use /mcp desc to show server and tool descriptions + – Use /mcp schema to show tool parameter schemas + – Use /mcp nodesc to hide descriptions + – Use /mcp auth to authenticate with OAuth-enabled servers + – Press Ctrl+T to toggle tool descriptions on/off + ``` 4. 发送提问`用 websearch MCP 搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及`websearch MCP`。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054879.png) + ``` + * 搜索结果已返回，我需要整理并总结其中与阿里云相关的主要新闻信息，以清晰的方式呈现给用户。 + * 已为您搜索到阿里云相关新闻，以下是主要内容： + 阿里云最新动态 + 1. 阿里云千问大模型表现亮眼 + - 2025 年下半年，阿里云千问（Qwen）在中国企业级大模型市场占比跃升至32.1%，几乎翻倍（上半年为 + 17.7%），位居第一 + - 字节豆包占比 21.3% 排名第二，DeepSeek 占比 18.4% 排名第三 + 2. AI 战略投入 + - 阿里豪掷3800 亿元押注 AI 基础设施 + - 2026 年，阿里云将 AI 相关收入目标增速上调至200% + - 2026 年 2 月推出30 亿元"春节请客计划"，联合淘宝、飞猪、盒马等以免单形式发放福利 + 3. Coding Plan 服务更新 + - 阿里云 Coding Plan 订阅服务新增支持 Qwen 3.5-Plus、GLM-4.7、Kimi-K2.5 等编程模型 + - 新用户首月首购享两折优惠 + 4. 技术进展 + - 除夕夜开源的千问 Qwen3.5-Plus 采用混合注意力机制和稀疏 MoE 架构 + - 总参数 3970 亿，激活 170 亿，推理速度达主流模型的 2 倍 + 需要我为您搜索更具体的阿里云新闻主题吗？ + > 输入您的消息或 @ 文件路径 + 按 ? 查看快捷键 1.8% context used + ``` + + +## Cline + +1. 点击MCP Servers，进入Configure页面，点击Configure MCP Servers。 + + ![2026-02-26\_13-58-09](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054977.jpg) + +2. 在配置文件`cline_mcp_settings.json`中写入 MCP 配置信息，并将`YOUR_API_KEY`替换为上一步骤获取的 API Key。配置完成后，保存文件。 + + ``` + { + "mcpServers": { + "websearch": { + "type": "http", + "url": "https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp", + "headers": { + "Authorization": "Bearer YOUR_API_KEY" + }, + "disabled": false + } + } + } + ``` + +3. 配置完成后，在Configure页面下可以查看MCP的相关信息。 + + ![2026-02-26\_14-05-44](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054984.jpg) + +4. 新开对话，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 + + > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 + + ![2026-02-26\_14-10-05](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9276802771/p1054989.jpg) ## Kilo CLI @@ -231,8 +356,14 @@ ``` kilocode mcp list ``` - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0248612771/p1055115.png) + ``` + MCP Servers + │ + * [ok] websearch connected + │ https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp + │ + └ 1 server(s) + ``` 3. 在终端中执行以下命令进入 Kilo CLI。 @@ -244,14 +375,31 @@ > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9228202771/p1054891.png) + ``` + 以下是阿里云的最新新闻要点： + 阿里云千问大模型市场份额跃升第一 + - 市场份额翻倍：根据沙利文2026年2月报告，阿里云千问（Qwen）占比跃升至*32.1%*，相较2025年上半年的17.7%几乎翻倍，超越字节豆包（21.3%）和DeepSeek(18.4%)，成为企业级大模型市场第一 + - 日均调用量激增：2025年下半年中国企业级大模型日均调用量达37.0万亿tokens，较上半年增长263% + AI基础设施投资战略 + - 3800亿元投资：阿里宣布豪掷3800亿元押注AI基础设施 + - 收入目标：2026年将AI相关收入目标增速上调至200% + - 市场目标：阿里云目标拿下中国AI云市场增量的80% + 千问模型技术进展 + - 开源Qwen3.5-Plus：除夕夜开源，采用混合注意力机制和稀疏MoE架构 + - 高效能设计：总参数3970亿，仅激活170亿，推理速度达到主流模型的2倍 + - 春节免单计划：千问启动30亿元"春节请客计划"，联合淘宝、飞猪、盒马等生态业务发放福利 + 其他动态 + - Coding Plan上新：支持千问3.5、GLM-4.7、Kimi-K2.5等编程模型，新用户首月两折优惠 + - 云服务器优惠：轻量应用服务器38元/年起，200M带宽不限流量 + Code · kimi-k2.5 · 49.8s + ``` ## Kilo Code IDE 插件 1. 打开Kilo Code IDE插件配置联网搜索 MCP 信息，并将`YOUR_API_KEY`替换为上一步骤获取的 API Key。配置完成后，保存文件。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055026.png) + 依次单击左上角齿轮图标，在侧边栏选择 **Agent Behaviour**，单击 **MCP Servers** 页签，然后单击底部 **Edit Global MCP** 按钮，打开 `mcp_settings.json` 配置文件，将以下 JSON 内容粘贴并保存。 ``` { @@ -269,14 +417,10 @@ 当联网搜索的MCP状态显示为绿色时，表示添加成功。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055021.png) - 2. 返回对话界面，发送提问`用websearch MCP搜索阿里云的新闻`即可看到搜索结果。 > 为了避免和其他工具混淆，建议提问时明确提及 websearch MCP。 - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3100902771/p1055023.png) - ## 常见问题 @@ -292,7 +436,7 @@ - 如果使用的 URL 为 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/sse`，说明开通的是旧版 SSE 协议，请将协议[升级至Streamable HTTP](#da7cd47fa8pcd)。 -3. **API Key 错误**：请确认使用了有效的百炼通用 API Key（格式为 sk-xxx，非 Coding Plan 专属 API Key），并已在命令中正确替换 YOUR\_API\_KEY。 +3. **API Key 错误**：请确认使用了有效的百炼通用 API Key（格式为 sk-xxx，非 Token Plan 专属 API Key），并已在命令中正确替换 YOUR\_API\_KEY。 4. **免费额度用尽**：联网搜索 MCP 全部用户前 2000 次调用免费，免费额度用尽后按 29 元/千次计费，请确认账户余额充足。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md deleted file mode 100644 index 76b285c1..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-faq.md +++ /dev/null @@ -1,292 +0,0 @@ -# 常见问题 - -Token Plan 团队版常见问题汇总，涵盖购买、使用、计量和性能相关的问题解答。 - -## **Token Plan 团队版和 Coding Plan 有什么区别？** - -**Token Plan 团队版** - -**Coding Plan** - -适用场景 - -一人公司/团队/企业日常办公 - -个人开发场景 - -支持的模型 - -文本生成、图像生成模型 - -文本生成模型 - -计费方式 - -按 Token 消耗抵扣 Credits - -按模型调用次数 - -使用频次 - -无每 5 小时/每周限额 - -每 5 小时/每周限额 - -API Key 和 Base URL - -在[管理后台](https://tokenplan-enterprise.bailian.aliyunportal.com)生成专属 API Key，Base URL 详见[快速开始](https://help.aliyun.com/zh/model-studio/token-plan-quickstart) - -在[Coding Plan 页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/coding-plan)获取专属 API Key 和专属 Base URL - -高峰期性能 - -多租户隔离 - -高峰期间可能排队 - -数据安全 - -承诺不使用数据训练模型 - -用户数据授权 - -## **接入与调用** - -### **如何在编程工具中使用图像生成模型？** - -图像生成模型使用独立的接口，无法通过文本模型的 Base URL 直接调用。需要通过工具的 Skill 或扩展机制接入，具体配置方法请参见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 - -### **常见报错及解决方案** - -**报错信息** - -**可能原因** - -**解决方案** - -**401 InvalidApiKey: No API-key provided.** - -请求头中未携带 API Key（`Authorization: Bearer` 或 `x-api-key` 均未传）。 - -在管理后台生成 API Key，并在工具中完成配置。 - -**401 InvalidApiKey: Invalid API-key provided.** - -1. 误用了百炼通用 API Key（sk-xxx 格式）或 Coding Plan 的 API Key - -2. Token Plan 团队版订阅过期 - -3. API Key 复制不完整或包含空格 - - -1. 确认使用的是 Token Plan 专属 API Key，确保完整且无空格。 - -2. 确认订阅是否过期。 - -3. 如仍报错，重置 API Key，重置后使用新 Key 配置。 - - -**404 model 'xxx' not found or not supported** - -**400 Model not exist.** - -1. 模型名称拼写错误或大小写错误 - -2. 模型 ID 不在套餐支持列表中 - - -1. 确认模型名称区分大小写，与套餐支持的模型 ID 一致。 - -2. 检查所选套餐是否包含该模型。 - - -**401 invalid access token or token expired** - -误用了 Coding Plan 或其他套餐的 Base URL - -Anthropic 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` - -OpenAI 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -**401 Incorrect API key provided** - -误用了百炼通用 Base URL（dashscope.aliyuncs.com） - -Anthropic 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` - -OpenAI 兼容端点：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -**400 InvalidParameter: Range of input length should be \[1, xxx\]** - -输入内容（含对话历史、代码上下文等）超出模型的最大上下文长度 - -新建会话清空历史，或使用工具自带的上下文压缩命令（如 Claude Code 的 `/compact`、Qwen Code 的 `/clear`）。也可切换上下文窗口更大的模型。 - -**400 InvalidParameter: url error, please check url！** - -Base URL 路径与协议不匹配。例如把 OpenAI 兼容路径配在 Anthropic 端点上，或反之。 - -按工具实际使用的协议选择对应的端点： - -- Anthropic 兼容协议（Claude Code 等）：以 `/apps/anthropic` 结尾。 - -- OpenAI 兼容协议（Cursor、Qwen Code 等）：以 `/compatible-mode/v1` 结尾。 - - -**400 InvalidParameter: Range of max\_tokens should be \[1, xxxx\]** - -请求中的 `max_tokens`（或工具配置中的最大输出长度）超出当前模型支持的最大输出 Token 数。 - -将 `max_tokens` 调整为不超过报错信息中提示的上限值。 - -**400 invalid\_parameter\_error: The thinking\_budget parameter must be a positive integer and not greater than xxxxx** - -工具配置中的思维链长度（如 `thinking_budget`、`budgetTokens`）超过当前模型支持的上限。各模型上限不同，以报错中的数值为准。 - -将思维链长度调整为不超过报错提示的上限值，或在不支持思考模式的模型上移除该配置项。 - -**400 data\_inspection\_failed: Input text data may contain inappropriate content.** - -输入或输出命中平台内容安全策略。 - -修改输入内容后重新提交。如多次触发，调整提示词避免敏感话题。 - -**429 API-Key Requests rate limit exceeded, please try again later.** - -短时间内请求过于密集，触发模型调用限流。 - -等待一分钟后重试；如频繁触发请降低请求频率，并确认 API Key 未被他人共享使用。 - -**429 Throttling.AllocationQuota: Allocated quota exceeded, please increase your quota limit.** - -**insufficient\_quota: You exceeded your current quota, please check your plan and billing details.** - -该报错可能由以下两种原因触发： - -**套餐额度已用尽**：坐席额度和共享用量包均已耗尽。 - -**触发模型调用限流**：即使套餐额度充足，每秒或每分钟消耗的 Token 数（TPS/TPM）超过模型限流阈值也会触发。限流按主账号维度计算，账号下所有 RAM 子账号、业务空间和 API Key 的调用量合并计算；即使每分钟总调用量未超限，短时间内的请求激增也可能触发。 - -**额度已用尽**：可加购坐席（加购后需将新坐席分配给成员后再使用）、加购共享用量包，或等待下一计费周期额度自动重置。 - -**触发限流**：等待约一分钟后重试，并采用平滑请求策略（如匀速调度、指数退避或请求队列缓冲）避免瞬时高峰。 - -**Connection error** - -Base URL 域名拼写错误或网络连接异常 - -检查 Base URL 域名拼写及网络连接。 - -## **产品功能相关** - -### **Token Plan 团队版的 API Key 能与其他套餐或普通 API 混用吗？** - -不能。Token Plan 团队版、Coding Plan 和百炼按量计费三者的 API Key 和 Base URL 互不相通，请勿混用。误用其他 API Key 不会抵扣 Token Plan 团队版的套餐额度。 - -### **能在多个工具中使用同一订阅吗？** - -可以。同一 API Key 可在全部兼容的 AI 编程和智能体工具中使用，额度共享消耗。每个成员持有独立的 API Key，不可共享给其他成员。 - -### **有哪些使用限制？** - -仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 - -### **团队管理入口在哪里？** - -阿里云主账号或 RAM 用户登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)后，在左侧菜单进入**我的订阅**，通过订阅卡片的**设置**、**用量分析**、**分配座席**入口进入成员管理与设置面板；也可点击**进入管理平台**跳转独立管理平台。通过 SSO 或钉钉加入的成员，通过管理员分发的**管理平台地址**（形如 tokenplan-enterprise.bailian.aliyunportal.com）登录管理平台。详见[访问入口](https://help.aliyun.com/zh/model-studio/token-plan-team#tp05-enter)。 - -### **成员如何获取 API Key？** - -管理员在管理后台创建成员账号并分配席位后，为成员生成 API Key。成员无法自行生成，需联系管理员获取。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 - -### **回收席位、修改角色、移出组织有什么区别？** - -这三个操作都在成员管理页面执行，但作用范围不同： - -- **回收席位**：撤销成员的席位使用权并将席位释放回席位池。成员失去使用权，但仍留在组织中，可被重新分配席位。 - -- **修改角色**：变更成员的权限（如管理员/普通成员），不影响席位分配和组织归属。 - -- **移出组织**：将成员从团队完全移除，席位自动回收至席位池，成员从成员列表消失。 - - -### **为什么 API Key 只能查看一次，丢失后如何处理？** - -为避免团队间 API Key 混用导致计费混乱，API Key 仅在首次生成或重置时显示，后续无法再次查看或复制。若 API Key 丢失，在**成员管理**页面找到对应成员，点击**重置**生成新 Key，原 Key 立即失效，需在工具中重新配置。 - -## **购买相关** - -### **可以同时购买多个套餐吗？** - -每个阿里云账号限购一个订阅，同一订阅下每种坐席类型均可购买多个。共享用量包可叠加购买，单次最多 1000 个。 - -### **可以单独购买共享用量包吗？** - -不可以。共享用量包是 Token Plan 团队版的附加商品，需先订阅 Token Plan 团队版坐席套餐后，才能购买共享用量包。 - -### **套餐是否支持退订？** - -支持按席位退订。在控制台**Token Plan 订阅详情页**的订阅明细中，点击对应席位的**退订**，已有用量消耗的席位不可退订。也可勾选多个席位后点击**批量退订**。退款原路退回支付账户，预计 1-3 个工作日到账。详见[订阅管理](https://help.aliyun.com/zh/model-studio/token-plan-overview#tp01-sub-mgmt)。 - -### **阿里云账号欠费是否影响 Token Plan 团队版的使用？** - -Token Plan 团队版为预付费订阅产品，只要套餐额度未用尽且订阅仍在有效期内，阿里云账号欠费不影响 Token Plan 团队版的正常使用。 - -### **如何加购坐席？** - -在 Token Plan 订阅详情页的订阅明细中，点击**加购座席**，选择坐席类型和数量后提交订单。加购后需将新坐席分配给成员才能使用。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 - -### **如何关闭或开启自动续费？** - -在 Token Plan 订阅详情页的订阅明细中，点击**关闭自动续费**或**开启自动续费**。关闭后订阅到期不自动续费，需手动续费。 - -### **套餐变更与退订重购有哪些限制？** - -Token Plan 团队版与 Coding Plan 是两个独立的订阅计划，不支持相互转换： - -- **套餐互转限制**：不支持将已购买的 Token Plan 团队版更换为 Coding Plan，也不支持将 Coding Plan 直接转换为 Token Plan 团队版（即使补差价也不行）。可以同时订阅这两个计划，各自独立计费。 - -- **退订重购注意事项**：退款并重新购买后，API Key 和 Base URL 会发生变更，需在工具中重新配置新的专属 API Key 才能正常使用。 - - -## **计量相关** - -### **Credits 抵扣规则是什么？** - -Token Plan 团队版实际消耗取决于每次请求中输入 Token、缓存 Token 和输出 Token 的组合。优先从坐席额度抵扣，坐席额度用尽后从共享用量包抵扣，全部用尽后服务暂停至下一计费周期或购买共享用量包补充额度。 - -### **如何查看用量？** - -在[Token Plan 订阅详情页](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)可查看套餐和共享用量包的用量详情。管理员还可在[管理后台](https://tokenplan-enterprise.bailian.aliyunportal.com)的用量分析页面查看全部成员的消耗明细。 - -### **用量如何重置？** - -坐席额度在每个订阅月到期时重置，未用完的额度不累积到下月。共享用量包额度购买后有效期为 1 个月，到期后需重新购买，不随座席额度按月重置。 - -### **超出限额之后怎么办？** - -坐席额度用尽后自动从共享用量包抵扣；全部额度用尽后服务暂停。可通过以下方式恢复： - -- 购买共享用量包补充额度。 - -- 等待下一计费周期额度自动重置。 - - -### **续费后为什么 Credits 没有增加？** - -Token Plan 团队版的坐席额度按订阅周期计算，每个订阅月到期时自动重置。续费（续订）仅延长订阅有效期或预定下一计费周期的额度，**不会叠加补充至当前计费周期**。 - -若当前周期额度已用尽且需立即恢复服务，续订下月额度无法即时补充，可通过以下方式恢复： - -- 购买共享用量包补充额度。 - -- 升级至更高规格的坐席。 - -- 加购坐席（加购后需将新坐席分配给成员方可使用）。 - - -## **数据安全** - -### **数据安全如何保障？** - -Token Plan 团队版承诺不使用对话数据训练模型，传输过程采用 HTTPS 加密，并基于多租户隔离架构保障企业级数据隔离。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md index 8abf5759..81a9bbc4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-overview.md @@ -1,299 +1,170 @@ -# Token Plan（团队版）概述 +# Token Plan 概述 -Token Plan 团队版是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持文本生成与图像生成模型，兼容主流 AI 编程与智能体工具，提供团队管理后台、数据安全保障，调用平稳运行。 +Token Plan 是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持多种 AI 编程和智能体工具。Token Plan 提供个人版和团队版两个版本，满足从个人开发者到企业团队的不同需求。 **说明** -Token Plan 团队版目前仅支持**华北2（北京）**地域。 +Token Plan 目前仅支持**华北2（北京）**地域，请在[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan)左上角将地域切换至**华北2（北京）**后购买并使用。 ## **产品简介** -Token Plan 团队版整合千问和三方模型，支持文本生成与图像生成。通过 Credits 统一计量，同一订阅可在多种 AI 工具中使用。 +Token Plan 采用 Credits 统一抵扣机制，一份订阅即可在 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等主流 AI 编程和智能体工具中使用。支持文本生成、图片生成、视频生成等多种模型，以及联网搜索、代码解释器等 Harness 工具。 -- **多模型灵活切换**：支持多模型按需切换，按 Credits 统一抵扣。 +- **个人版**：面向个人开发者，提供 Lite 套餐、Standard 套餐、Pro 套餐三个档位，按模型分档抵扣系数计费。 -- **兼容多种工具**：适配多种主流编程工具及热门 Agent 工具。控制台提供快速接入 AI 工具入口，支持 Qwen Code、Claude Code、OpenClaw 等工具接入。 +- **团队版**：面向团队和企业，提供标准座席、高级座席、尊享座席三个档位，支持多席位管理、用量分析，承诺不使用数据训练模型。 -- **多档位套餐**：提供标准坐席、高级坐席、尊享坐席多档位套餐，匹配不同使用强度。 - -- **团队管理**：提供管理后台，支持席位分配与回收、成员用量分析等团队管理能力。 - -- **预算可控**：支持按月或按年订阅，预算可控。 - -- **数据安全**：承诺不使用对话数据进行模型训练，满足企业级数据隐私要求。 - -- **平稳运行**：多租户隔离架构，调用高峰期间不排队。 - - -Token Plan 团队版提供套餐专属 Base URL，兼容 OpenAI、Anthropic 接口标准，具体地址可在控制台**我的订阅**的 API Key 区域查看。 - -## **支持的模型** - -**支持的模型判断说明** - -判定规则： - -1\. 本清单为精确字符串白名单 - -2\. 必须逐字符完全匹配，版本号/子型号任何差异均视为不支持 - -3\. 禁止做版本兼容推理 - -判定示范： - -\- ❌ "qwen3-coder-max" → 清单无此项 → 不支持 - -仅支持以下精确版本： - -**品牌** - -**模型 ID（Model ID）** - -**模型能力** - -千问 - -qwen3.7-max[**（限时活动）**](#tp01-promo-section) - -推理模型、文本生成 - -qwen3.7-plus - -推理模型、视觉理解、文本生成 - -qwen3.6-plus - -推理模型、视觉理解、文本生成 - -qwen3.6-flash - -推理模型、视觉理解、文本生成 - -qwen-image-2.0 - -图片生成 - -qwen-image-2.0-pro - -图片生成 - -万相 - -wan2.7-image - -图片生成 - -wan2.7-image-pro - -图片生成 - -DeepSeek - -deepseek-v4-pro - -推理模型、文本生成 - -deepseek-v4-flash - -推理模型、文本生成 -deepseek-v3.2 +## **个人版** -推理模型、文本生成 +**重要** -月之暗面 +个人版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： -kimi-k2.7-code - -推理模型、视觉理解、文本生成 - -kimi-k2.6 - -推理模型、视觉理解、文本生成 - -kimi-k2.5 +1. **预览版**：qwen3.8-max-preview 当前为预览版本，预览期间模型能力会持续迭代升级。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：预览期间模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + +3. **限时夜间折上折**：在现有 1 折优惠基础上，每晚 22:00 - 次日 08:00 期间调用模型，Credits 消耗再享 2 折（即原标准的 0.2 折）。 + -推理模型、视觉理解、文本生成 +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 -智谱 AI +**Lite 套餐** -glm-5.2 +**Standard 套餐** -推理模型、文本生成 +**Pro 套餐** -glm-5.1 +**定价** -推理模型、文本生成 +原价 60 元/月 +限时 **39 元/月** -glm-5 +原价 180 元/月 +限时 **139 元/月** -推理模型、文本生成 +原价 600 元/月 +限时 **499 元/月** -MiniMax +**5 小时限额** -MiniMax-M2.5 +700 Credits -推理模型、文本生成 +3,000 Credits -## **套餐与定价** +12,000 Credits -前往 [Token Plan 团队版购买页面](https://common-buy.aliyun.com/token-plan/)选择坐席类型、数量和订阅周期，完成订阅。主账号和 RAM 账号均可订阅。 +**每 7 天限额** -订阅周期支持按月购买、按年购买、连续包月包年。 +2,500 Credits -### **限时活动** +10,000 Credits -即日起至 2026 年 7 月 22 日 23:59（UTC+8），qwen3.7-max 模型 Credits 消耗减半，同时支持隐式缓存。 +40,000 Credits -### **Token Plan 团队版** +**并发 Agent** -提供标准坐席、高级坐席、尊享坐席三个档位，匹配不同使用强度。 +1-2 个 -席位（坐席）是 Token Plan 团队版的最小订阅单位，代表一个团队成员的使用名额。管理员在[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)中将席位分配给成员后，系统自动为该成员生成专属的 API Key。每个席位绑定一个成员、对应一个 API Key，不可共享。 +3-4 个 -**坐席类型** +6-8 个 -**价格** +**模型** -**额度** +支持文本生成、图像生成、视频生成等多种模型（[查看完整列表](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-models)） -**适用场景** +**Harness 工具** -标准坐席 +支持多种 Harness 工具，以 Credits 统一抵扣（[查看详情](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-harness)） -¥198/坐席/月 +## **团队版** -25,000 Credits/坐席/月 +**重要** -轻度使用 AI 辅助的团队成员 +团队版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： -高级坐席 +1. **预览版**：qwen3.8-max-preview 当前为预览版本，预览期间模型能力会持续迭代升级。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：预览期间模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + -¥698/坐席/月 +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 -100,000 Credits/坐席/月 +**标准座席 Standard** -日常高频使用 AI 编程或办公的团队成员 +**高级座席 Pro** -尊享坐席 +**尊享座席 Max** -¥1,398/坐席/月 +**共享用量包 Extra Bundle** -250,000 Credits/坐席/月 +**定价** -重度依赖 AI 的核心开发者或高强度使用者 +原价 198 元/座席/月 +限时 **150 元/座席/月** -### Token Plan 团队版 - 共享用量包 +原价 698 元/座席/月 +限时 **550 元/座席/月** -跨坐席共享的弹性用量包，当个别坐席用量超出套餐额度时，可从共享用量包中抵扣。每个共享用量包有效期为 1 个月，到期未使用的额度自动清零。持有多个共享用量包时，优先抵扣最近到期的用量包。 +**1,398 元/座席/月** -**档位** +5,000 元/个/月 -**价格** +**每月总额度** -**额度** +25,000 Credits/座席/月 -Token Plan 团队版 - 共享用量包 +100,000 Credits/座席/月 -¥5,000/个 +250,000 Credits/座席/月 625,000 Credits/个 -## **订阅管理** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面管理订阅： - -- **加购席位**：点击**加购座席**，新加席位与现有订阅统一到期，费用和 Credits 额度均按剩余时长折算。 - -- **升级席位**：在**订阅明细**中点击席位的**升级**，将低档位席位升级为高档位，按差价补缴费用。升级当月的 Credits 额度按**高低档位差值**、依本计费周期的剩余时长折算后补充到当月可用额度；自下一计费周期起按新档位的完整额度发放。实际到账额度以控制台**我的订阅**页面为准。 - -- **退订席位**：在**订阅明细**中点击席位的**退订**，按席位维度退订；已有用量消耗的席位不可退订。退款原路退回支付账户，预计 1-3 个工作日到账。 - -- **批量操作**：勾选多个席位后点击**批量升级**或**批量退订**。 - -- **续费**：点击**续费**按钮，续费周期与订阅时一致（按月订阅则按月续费，按年订阅则按年续费），到期前完成可避免服务中断。 - -- **自动续费**：点击**开启自动续费**，确认后次日生效，到期前 9 天系统按订阅周期自动扣款续费。如需关闭，点击**关闭自动续费**并确认，关闭后停止到期前自动扣款续费，订阅到期后自动停订。 - -- **加购共享用量包**：在**共享用量包**区域点击**前往购买**。 - - -## **Credits 计费机制** - -### **计费说明** +**5 小时限额** -单次请求消耗的 Credits **并非固定值**，由模型类型、Token 用量、思考模式及工具调用等动态决定。其中 Token 用量会随多轮对话累积的上下文（历史消息、代码、工具返回、检索内容等）持续增长，且部分模型按**上下文长度阶梯计费**（上下文越长，单价档位可能越高），因此同一模型在不同请求下的消耗可能相差较大。实际消耗以[控制台订阅页用量明细](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan)为准。 +无限制 -### **计算示例** +**7 天限额** -以 qwen3.6-plus 为例，预估单次请求消耗明细如下（不同模型的单价不同，实际以账单为准）： +无限制 -**Token 类型** +**模型** -**数量** +支持文本生成、图像生成、视频生成等多种模型（[查看完整列表](https://help.aliyun.com/zh/model-studio/token-plan-team-overview#tpt01-models)） -**消耗 Credits** +**Harness 工具** -输入 tokens +支持多种 Harness 工具，以 Credits 统一抵扣（[查看详情](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview#tpp01-harness)） -8,349 +**团队管理** -1.67 +支持多席位管理和用量分析（[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team-management)） -缓存 tokens +## **套餐限额** -40,794 +### **个人版** -0.82 +个人版采用 5 小时和 7 天两层固定窗口限额，限额单位为 Credits： -输出 tokens - -573 - -0.69 - -**合计** - -**约 3.18 Credits** - -**说明** - -上表**仅为单次请求的示例**，并不代表每次请求都固定消耗约 3 Credits。在 AI 编程、智能体等**多轮对话**场景中，每次请求都会携带累积的上下文（历史对话、代码、工具返回等），随着对话进行输入 Token 持续增多，单次消耗的 Credits 也会**相应上升**；若模型按上下文长度阶梯计费，长上下文可能进入更高价位档，消耗速度进一步加快。 - -如需控制消耗，建议：任务切换或话题变更时**及时开启新会话**、清理无关历史，以缩短上下文；对长文档、大代码库按需拆分输入；并在上述控制台订阅页用量明细中关注各模型的实时消耗趋势。 - -**说明** - -上表以 qwen3.6-plus 展示 Token 类型分布，其中**“缓存 tokens”一行为通用 Token 分类示意**。**Token Plan 团队版中，隐式缓存的命中取决于所选模型是否支持**——目前仅 qwen3.7-max 在限时活动期间支持隐式缓存（详见上文[限时活动](#tp01-promo-section)章节）；使用其他模型时，输入 Token 不会进入隐式缓存抵扣，此时“缓存 tokens”一行不适用。 - -### **抵扣顺序** - -1. 优先从坐席套餐的月度额度中抵扣。 +- **5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 -2. 坐席额度用尽后，从共享用量包中抵扣。持有多个共享用量包时，优先抵扣最近到期的用量包。 - -3. 全部额度用尽后，服务将暂停至下一计费周期或购买共享用量包补充额度。 +- **7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 -## **查看额度消耗情况** +任一层限额触顶即暂停服务，需等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 -**通过控制台**：登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)，在**我的订阅**页面查看总额度使用百分比、重置时间、团队席位分配情况，以及各席位与共享用量包的状态和到期时间。 +### **团队版** -**通过团队管理平台**：在**用量分析**页面，可查看近 1、7、30 天的 Credits 消耗趋势、各模型用量，以及每个成员的消耗明细。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 +团队版采用月度总额度制，无 5 小时和 7 天窗口限额。每个座席的月度额度在计费周期内可用，到期未使用的额度不结转。超出月度总额度后调用将被阻断，可购买共享用量包补充额度。 -## **使用细则** +## **常见问题** -1. **使用范围**：仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 - -2. **数据安全**：Token Plan 团队版不会使用对话数据训练模型。 - -3. **账号规范**：API Key 仅限已分配席位的成员本人使用，不可共享或公开泄露。 - -4. **退订与退款**：在控制台**我的订阅**页面按席位退订，已有用量消耗的席位不可退订。退款原路退回支付账户，预计 1-3 个工作日到账。 - -5. **服务地域**：Token Plan 团队版目前仅在特定地域提供服务，如需从海外调用，请确认符合当地法律法规要求。关于百炼支持的地域和服务部署范围，请参见[选择地域和服务部署范围](https://help.aliyun.com/zh/model-studio/regions/)。 - +### **个人版和团队版可以同时购买吗？** + +可以。同一阿里云账号可以同时持有个人版和团队版，各自独立计费。 -## 错误码 +### **关于 Coding Plan** -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 +Coding Plan 和 Token Plan 是两个独立的订阅产品，两者之间无法迁移或升级。Coding Plan Lite 已于 2026 年 3 月 20 日停止新购，2026 年 4 月 13 日停止续费和升级；Coding Plan Pro 为限量抢购，库存售罄后不再补充。推荐使用 **Token Plan**，支持更多模型和 Harness 工具。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md new file mode 100644 index 00000000..a8a11599 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md @@ -0,0 +1,172 @@ +# 常见问题 + +Token Plan 个人版的额度、购买、订阅和接入常见问题。 + +## **额度与限额** + +### 5 小时限额和 7 天限额是什么意思？ + +Token Plan 个人版采用 5 小时和每 7 天两层固定窗口限额，限额单位为 Credits。两层限额独立计算，任一层触顶即暂停服务，需等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 + +- **5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 + + 例如：Standard 套餐的 5 小时限额为 3,000 Credits。您在 7 月 20 日 10:00 首次调用，系统开启窗口（10:00 ~ 15:00）。10:00 消耗 2,000 Credits，11:00 消耗 1,000 Credits，累计 3,000 Credits 触顶暂停。需等到 15:00 窗口结束，额度重置为 3,000 Credits，服务恢复。 + +- **7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 + + 例如：Standard 套餐的 7 天限额为 10,000 Credits。您在 7 月 20 日首次调用，系统开启窗口（7 月 20 日 ~ 7 月 27 日）。7 月 20 日消耗 4,000 Credits，7 月 22 日消耗 6,000 Credits，累计 10,000 Credits 触顶暂停。需等到 7 月 27 日窗口结束，额度重置为 10,000 Credits，服务恢复。 + + +各档位限额如下： + +**档位** + +**5 小时限额** + +**7 天限额** + +Lite 套餐 + +700 Credits + +2,500 Credits + +Standard 套餐 + +3,000 Credits + +10,000 Credits + +Pro 套餐 + +12,000 Credits + +40,000 Credits + +### 5 小时限额到了但7 天限额还有余量，还能继续使用吗？ + +不能。任一层限额触顶即暂停服务，需等待对应窗口周期结束后额度重置。 + +### 7 天限额是固定日期重置吗？ + +不是。7 天限额采用固定窗口机制，自首次调用起计时 7 天，到期后额度重置。重置时间取决于您首次调用的时间，而非固定的日历日期（如每周一）。 + +### 额度用完了怎么办？ + +限额用完后调用会被阻断，不会按量计费。恢复方式： + +- 等待额度释放。 + +- 升级套餐。 + + +## **接入报错** + +### 常见报错及解决方案 + +**报错信息** + +**可能原因** + +**解决方案** + +401 InvalidApiKey: No API-key provided. + +请求头中未携带 API Key + +生成 API Key 并在工具中完成配置 + +401 InvalidApiKey: Invalid API-key provided. + +误用了按量计费的 API Key 或 Coding Plan 的 Key；订阅过期；Key 复制不完整 + +确认使用 Token Plan 个人版专属 API Key，确保完整且无空格 + +404 model 'xxx' not found or not supported + +模型名称拼写错误或不在支持列表 + +确认模型名称区分大小写，与套餐支持的模型 ID 一致。 + +401 invalid access token or token expired + +误用了 Coding Plan 或其他计费模式的 Base URL + +使用 Token Plan 个人版专属 Base URL + +401 Incorrect API key provided + +误用了百炼通用 Base URL（dashscope.aliyuncs.com） + +使用 Token Plan 个人版专属 Base URL + +429 Requests rate limit exceeded + +短时间内请求过于密集 + +等待一分钟后重试，降低请求频率 + +429 Allocated quota exceeded + +5 小时或7 天限额用尽 + +等待窗口释放额度 + +## **并发与性能** + +### 最多支持多少个 Agent 并发？ + +并发能力与套餐档位相关： + +**档位** + +**建议并发** + +Lite 套餐 + +可同时支持 1-2 个 Agent 并发运行 + +Standard 套餐 + +可同时支持 3-4 个 Agent 并发运行 + +Pro 套餐 + +可同时支持 6-8 个 Agent 并发运行 + +### 高峰期响应会变慢吗？ + +高峰期可能出现排队等待。如需更稳定的吞吐，可升级到更高档位或使用团队版。 + +## **使用规则** + +### "禁止 API 生产自动化调用"具体是什么意思？ + +Token Plan 个人版仅供个人通过官方指定工具（如 Cursor、Claude Code、Windsurf 等）进行交互式开发。不允许将 API Key 用于生产环境的自动化服务、批量脚本或后台定时任务等非交互场景。 + +### 多人共用一个账号可以吗？ + +不可以。Token Plan 个人版限单人使用，不允许多人共用同一账号或 API Key。如需多人协作，请使用 Token Plan 团队版。 + +## **购买与订阅** + +### RAM 用户可以使用 Token Plan 吗？ + +可以，需由主账号完成以下授权： + +1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + +2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +### 可以升配吗？升配后额度怎么算？ + +支持从低档位升级到更高档位。升级按剩余时长补缴差价，升级后每 5 小时限额和每 7 天限额立即提升至新档位对应额度。 + +### 可以降配吗？ + +不支持降配。如需更换为更低档位，可在订阅到期后重新购买。 + +### 自动续费怎么取消？ + +登录[百炼控制台 Token Plan](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan) 页面，在订阅管理中关闭自动续费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md new file mode 100644 index 00000000..09fb4b2a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md @@ -0,0 +1,234 @@ +# 概述 + +[Token Plan 个人版](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)是面向个人开发者的 AI 大模型订阅服务，以 Credits 统一计量，支持文本、多模态模型及 Harness 工具，适配主流 AI 编程和智能体工具。 + +**说明** + +Token Plan 个人版目前仅支持**华北2（北京）**地域。 + +## **核心特性** + +- **Credits 统一计量**：通过 Credits 统一抵扣不同模型和 Harness 工具的费用。 + +- **多模态模型支持**：覆盖文本生成、推理、视觉理解、图片生成、语音合成、语音识别、视频生成等能力。 + +- **Harness 工具集成**：支持联网搜索、文搜图、图搜图、网页抓取、代码解释器。 + +- **兼容多种工具**：适配 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等主流 AI 编程和智能体工具。 + + +## **套餐档位与定价** + +**Lite 套餐** + +**Standard 套餐** + +**Pro 套餐** + +**定价** + +原价 60 元/月 +限时 **39 元/月** + +原价 180 元/月 +限时 **139 元/月** + +原价 600 元/月 +限时 **499 元/月** + +**每 7 天限额** + +2,500 Credits + +10,000 Credits + +40,000 Credits + +**每 5 小时限额** + +700 Credits + +3,000 Credits + +12,000 Credits + +**并发 Agent** + +1-2 个 + +3-4 个 + +6-8 个 + +**权益** + +文本、视觉等多模态模型 + +联网搜索等 Harness 工具 + +适配主流工具并持续扩展 + +享受 Lite 套餐所有权益 + +4x Lite 套餐用量 + +享受 Standard 套餐所有权益 + +16x Lite 套餐用量 + +更高的并发上限 + +- **每 5 小时限额**：自首次调用起开启 5 小时计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 5 小时后额度重置。 + +- **每 7 天限额**：自首次调用起开启 7 天计时窗口，窗口期内累计消耗达到限额后暂停服务，需等待满 7 天后额度重置。 + + +任一层限额触顶即暂停服务，需等待对应窗口周期结束后额度重置。窗口期内未用完的额度不结转至下一周期。 + +## **Credits 计费机制** + +### **计费说明** + +单次消耗的 Credits 由模型类型、Token 用量、思考模式及工具调用等动态决定，实际消耗以[控制台订阅页](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)用量详情为准。 + +### **抵扣顺序** + +1. 每次调用消耗的 Credits 同时计入每 5 小时限额和每 7 天限额。 + +2. 任一层限额（5 小时或 7 天）触顶后，服务暂停，需等待对应窗口周期结束后额度重置，或升级套餐以获取更高额度上限。 + + +## **支持的模型** + +**重要** + +个人版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： + +1. **预览版**：qwen3.8-max-preview 当前为预览版本，**预览期间模型能力会持续迭代升级**。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：限时活动期间，模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + +3. **限时夜间折上折**：在现有 1 折优惠基础上，每晚 22:00 - 次日 08:00 期间调用模型，Credits 消耗再享 2 折（即原标准的 0.2 折）。 + + +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 + +**品牌** + +**模型 ID（Model ID）** + +**模型能力** + +千问 + +qwen3.8-max-preview + +推理模型、视觉理解、文本生成 + +qwen3.7-max + +推理模型、文本生成 + +qwen3.7-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-flash + +推理模型、视觉理解、文本生成 + +智谱 AI + +glm-5.2 + +推理模型、文本生成 + +DeepSeek + +deepseek-v4-pro + +推理模型、文本生成 + +万相 + +wan2.7-image + +图片生成 + +wan2.7-image-pro + +图片生成 + +HappyHorse + +happyhorse-1.1-i2v + +视频生成 + +happyhorse-1.1-t2v + +视频生成 + +happyhorse-1.1-r2v + +视频生成 + +## **支持的 Harness 工具** + +**工具能力** + +**工具名称** + +联网搜索 + +web\_search + +文搜图 + +t2i\_search + +图搜图 + +i2i\_search + +网页抓取 + +web\_extractor + +代码解释器 + +code\_interpreter + +## **订阅管理** + +### **升级** + +支持从低档位升级到更高档位。升级按剩余时长补缴差价，升级后每 5 小时限额和每 7 天限额立即提升至新档位对应额度。 + +### **续费** + +续费支持切换续费周期，也可一次续费多个周期。 + +- **手动续费**：支持选择不同续费时长，可续费多个周期。 + +- **自动续费**：开启后到期前系统自动扣款续费。 + + +续费仅延长订阅有效期，不会叠加补充至当前计费周期的额度。 + +### **其他** + +- 个人版暂不支持退订。 + +- 订阅到期后重新购买，API Key 会发生变更，需在工具中重新配置。新 API Key 可在控制台[**我的订阅**](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)页面的 API Key 区域获取。 + + +## **订阅前须知** + +1. **严禁 API 调用**：仅限在编程工具和智能体工具（如 Claude Code、Cursor、Qwen Code、Qoder、Qoder CN、OpenClaw 等）中使用，禁止以 API 调用的形式用于自动化脚本、自定义应用程序后端或任何非交互式批量调用场景。将套餐 API Key 用于允许范围之外的调用将被视为违规或滥用，可能会导致订阅被暂停或 API Key 被封禁。 + +2. **数据使用授权**：使用 Token Plan 个人版期间，模型输入以及模型生成的内容将用于服务改进与模型优化。停止使用 Token Plan 个人版服务可终止后续数据授权，但终止授权的范围不涵盖已授权使用的数据。详细条款请参见[阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html)第 5.2 条。 + +3. **账号使用规范**：套餐为订阅人专享使用，禁止共享。账号共享可能导致订阅权益受限。 + +4. **购买限制**：同一实名认证主体限购一份，可同时购买个人版和团队版。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md new file mode 100644 index 00000000..10116c46 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md @@ -0,0 +1,81 @@ +# 快速开始 + +三步完成 Token Plan 个人版订阅和接入：选择套餐、获取 API Key、配置 AI 工具。 + +## **步骤一：订阅 Token Plan 个人版** + +访问 [Token Plan 个人版购买页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)，选择套餐档位和订阅周期，完成订阅。 + +购买须知： + +- **RAM 用户授权**：RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + + 1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + + 2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +## **步骤二：获取 API Key 和 Base URL** + +- **API Key**：订阅完成后，在 Token Plan 控制台的**我的订阅**页面生成 API Key。API Key 仅在生成时完整显示一次，请立即复制并妥善保存。 + +- **Base URL**：根据 AI 工具支持的协议，选择对应的 Base URL。 + + +**协议** + +**Base URL** + +OpenAI 兼容 + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +Anthropic 兼容 + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + +**重要** + +Token Plan 的 API Key 以 `sk-sp-` 开头，与百炼通用 API Key（`sk-` 开头）格式不同，两者不可混用。Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离，必须配套使用。 + +## **步骤三：接入 AI 工具** + +将 API Key 和 Base URL 配置到 AI 工具中，即可开始使用。 + +[**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) + +[**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) + +[**Claude Code**AI 终端编码助手，支持自然语言编程](https://help.aliyun.com/zh/model-studio/claude-code) + +[**OpenCode**开源 AI 编程代理工具](https://help.aliyun.com/zh/model-studio/opencode) + +[**Cursor**AI 原生代码编辑器](https://help.aliyun.com/zh/model-studio/cursor) + +[**Codex**OpenAI 推出的命令行编程工具](https://help.aliyun.com/zh/model-studio/codex) + +[**Qwen Code**开源命令行 AI 编码工具](https://help.aliyun.com/zh/model-studio/qwen-code) + +[**QwenPaw**开源个人 AI 助手，支持本地与云端部署](https://help.aliyun.com/zh/model-studio/qwenpaw) + +[**Cherry Studio**多模型桌面客户端](https://help.aliyun.com/zh/model-studio/cherry-studio) + +[**Chatbox**跨平台 AI 桌面客户端](https://help.aliyun.com/zh/model-studio/chatbox) + +[**Cline**VS Code 扩展，智能代码补全和调试](https://help.aliyun.com/zh/model-studio/cline) + +[**Qoder**面向真实软件开发的 Agentic 编码平台](https://help.aliyun.com/zh/model-studio/qoder-agent) + +[**Lingma**阿里云智能编码助手，提供独立 IDE](https://help.aliyun.com/zh/model-studio/lingma-agent) + +[**Kilo CLI**轻量高性能命令行编程工具](https://help.aliyun.com/zh/model-studio/kilo-cli) + +[··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) + +## **可选：接入多模态生成模型** + +Token Plan 个人版支持多模态生成模型（wan2.7-image、happyhorse-1.1-t2v 等）。多模态生成模型使用独立的接口，需要通过 AI 工具的 Skill 或扩展机制接入，详见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 + +## **可选：接入 Harness 工具** + +通过 Harness 工具调用，模型可以在对话中调用联网搜索、文搜图、图搜图、网页抓取、代码解释器等扩展能力。当前仅 qwen3.7、qwen3.8 支持原生工具调用，通过 Responses API 直接调用，Harness 工具按抵扣系数消耗 Credits。详见[接入 Harness 工具](https://help.aliyun.com/zh/model-studio/token-plan-harness-tool)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md new file mode 100644 index 00000000..0b296d0c --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md @@ -0,0 +1,205 @@ +# 常见问题 + +Token Plan 团队版的常见问题解答，包括产品选择、Credits 计费与额度、模型与工具兼容性、使用限制、购买续费与退订等。 + +## **产品定位与套餐选择** + +### **个人版和团队版有什么区别？我该买哪个？** + +**对比项** + +**个人版** + +**团队版** + +适用场景 + +个人开发者 + +团队/企业 + +额度机制 + +5 小时 + 7 天固定窗口限额 + +固定月额度 + +团队管理 + +不支持 + +席位分配与回收、成员用量分析、SSO 接入 + +数据安全 + +数据使用遵循服务协议 + +承诺不使用对话数据训练模型 + +高峰期性能 + +高峰期可能出现等待 + +多租户隔离，高峰期不排队 + +个人开发者、日常使用 AI 编程工具，选择个人版即可；团队需要多人协作、统一管理席位和用量、对数据安全有更高要求，选择团队版。 + +### **个人版和团队版能同时买吗？额度是分开算的还是共享的？** + +可以同时购买。同一阿里云账号可以同时持有个人版和团队版，各自独立计费，额度不共享。 + +## **Credits 计费与额度** + +### **团队版的额度机制是怎样的？** + +团队版采用固定月额度，无 5 小时/7 天窗口限制。各坐席类型的月度额度： + +- **标准坐席**：25,000 Credits/坐席/月 + +- **高级坐席**：100,000 Credits/坐席/月 + +- **尊享坐席**：250,000 Credits/坐席/月 + + +坐席额度在每个订阅月到期时重置，未用完的额度不累积到下月。 + +### **坐席额度用完了怎么办？** + +坐席额度用尽后调用会被阻断，不会按量计费。恢复方式： + +- 等待下一个订阅月额度自动重置。 + +- 购买共享用量包（625,000 Credits/个，有效期 1 个月），团队内全部成员共享使用。 + + +## **模型与工具兼容性** + +### **Token Plan 团队版支持哪些模型？** + +团队版支持文本生成、推理、视觉理解、图片生成和语音模型。具体模型列表和抵扣系数请参见控制台的模型列表页面。 + +### **支持 Cursor / Claude Code / Cline 等第三方工具吗？** + +支持。Token Plan 兼容 OpenAI 和 Anthropic 协议，任何支持自定义 Base URL 和 API Key 的工具均可接入，包括 Cursor、Claude Code、Qwen Code、Qoder、Qoder CN、Cline、OpenClaw、Cherry Studio、Chatbox 等。具体配置方法请参见快速开始中的接入 AI 工具部分。 + +### **团队版的 API Key 能用在个人版上吗？** + +不能。个人版和团队版各自生成独立的 API Key，不可混用。系统会根据 API Key 自动识别对应的套餐。 + +### **Harness 工具是什么？** + +Harness 工具是模型内置的扩展能力，包括联网搜索、文搜图、图搜图、网页抓取、代码解释器等。团队版支持 Harness 工具，调用时按工具抵扣系数消耗 Credits。当前仅 qwen3.7 和 qwen3.8 系列模型支持原生 Harness 工具调用。 + +## **使用限制** + +### **模型调用限流是怎么计算的？** + +每秒或每分钟消耗的 Token 数（TPS/TPM）超过模型限流阈值时会触发限流。限流按主账号维度计算，账号下全部 RAM 子账号、业务空间和 API Key 的调用量合并计算。 + +### **高峰期性能如何？** + +团队版基于多租户隔离架构，调用高峰期间不排队。 + +## **购买、续费与退订** + +### **支持 RAM 子账号购买吗？** + +支持。RAM 子账号使用前，需主账号完成以下授权： + +1. 在 RAM 控制台为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略。 + +2. 在百炼控制台账号管理页面，为该 RAM 用户分配管理员或订阅套餐权限。 + + +### **可以升配吗？升配后额度怎么算？** + +支持坐席升配。升配后立即生效，限额按新坐席类型执行。升配需补缴差价（按剩余天数折算）。 + +### **可以降配吗？** + +不支持降配。如需使用更低坐席类型，可在当前订阅到期后重新订阅。 + +### **自动续费怎么取消？** + +登录[费用中心 > 续费管理](https://usercenter2.aliyun.com/finance/renew-manage)，找到 Token Plan 团队版订单，关闭自动续费。 + +### **续费时可以更换订阅时长吗？** + +不可以。续费仅支持按原订阅时长续费。如需更换订阅时长，可在订阅到期后重新购买。 + +### **限时优惠的计费规则是什么？** + +限时优惠仅适用于包月订阅的新购、续费和自动续费，包年订阅和升级坐席不参与。 + +加购坐席时按剩余时长折算费用，实际收费取折算金额与限时价中的较低值。 + +**示例**：标准坐席原价 ¥198/月，限时价 ¥150/月。 + +- 加购时按剩余时长折算为 ¥99，低于限时价，实际收取 ¥99。 + +- 加购时按剩余时长折算为 ¥165，高于限时价，实际收取 ¥150。 + + +## **与现有产品的关系** + +### **已有个人版，再买团队版会冲突吗？** + +不冲突。两者可以同时持有，各自独立计费。使用时根据 API Key 自动匹配对应套餐。 + +## **接入报错** + +### **常见报错及解决方案** + +**报错信息** + +**可能原因** + +**解决方案** + +401 InvalidApiKey: No API-key provided. + +请求头中未携带 API Key + +生成 API Key 并在工具中完成配置 + +401 InvalidApiKey: Invalid API-key provided. + +误用了按量计费的 API Key 或 Coding Plan 的 Key；订阅过期；Key 复制不完整 + +确认使用 Token Plan 团队版专属 API Key，确保完整且无空格 + +404 model 'xxx' not found or not supported + +模型名称拼写错误或不在支持列表 + +确认模型名称区分大小写，与套餐支持的模型 ID 一致 + +401 invalid access token or token expired + +误用了 Coding Plan 或其他计费模式的 Base URL + +使用 Token Plan 团队版专属 Base URL + +401 Incorrect API key provided + +误用了百炼通用 Base URL（dashscope.aliyuncs.com） + +使用 Token Plan 团队版专属 Base URL + +400 Range of input length should be \[1, xxx\] + +输入内容超出模型最大上下文长度 + +新建会话清空历史，或使用工具的上下文压缩命令 + +429 Requests rate limit exceeded + +短时间内请求过于密集，触发模型限流 + +等待一分钟后重试，降低请求频率 + +429 Allocated quota exceeded + +坐席月度额度用尽 + +购买共享用量包或等待下月额度重置 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md new file mode 100644 index 00000000..bc5280b9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md @@ -0,0 +1,183 @@ +# 团队管理 + +在 Token Plan 控制台或管理平台中添加和管理团队成员、分配和回收席位、监控 Credits 用量。 + +## **访问入口** + +**阿里云主账号或 RAM 用户**：登录 Token Plan 控制台，在**我的订阅**页面进行团队管理：在**团队版**卡片点击**设置**修改组织名称、登录方式（SSO/钉钉）等；在**订阅明细**区域点击**分配座席**进入成员管理。 + +**说明** + +RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + +1. 在 RAM 控制台为该 RAM 用户授予以下系统策略：`AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理），用于使用 Token Plan，权限范围按实际需要选择；同时授予 `AliyunBSSFullAccess`（费用中心权限），用于查看**我的订阅**页面中主账号已订阅的 Token Plan 套餐与用量。缺少费用中心权限时，RAM 用户登录后**我的订阅**页面将显示为空。 + +2. 在百炼控制台**账号管理**页面，为该 RAM 用户分配**管理员**或**订阅套餐**权限。 + + +通过 **SSO 或钉钉**加入的成员：通过管理员分发的**管理平台地址**登录管理平台。管理平台地址可在**设置**页面的**基本信息**区域获取。 + +## **角色与权限** + +**角色** + +**权限** + +所有者 + +添加或移除成员、分配或回收席位、修改成员角色、查看全部成员和模型的用量 + +管理员 + +权限范围与所有者相同。管理员由所有者授予，可被移除或降级。 + +成员 + +使用管理员分配的 API Key 和 Base URL 调用模型 + +## **成员管理** + +### **添加成员** + +- **手动添加**（不能登录管理平台，仅供 API 调用）：在**成员管理**页面点**添加成员**，在弹窗中填用户名（仅支持英文字母、数字、下划线）和角色，可选择同时分配席位。在该成员的操作列点**分配席位**选择席位版本，分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 + +- **SSO 或钉钉登录**（成员可登录管理平台，自管席位和 API Key）：完成 SAML 或钉钉接入配置后，成员从登录页对应入口登录即自动加入。 + + +### **修改成员角色** + +在**成员管理**页面，找到目标成员，点击操作列的**修改角色**，在弹窗中选择新角色（管理员或成员）后保存。所有者角色不可修改。 + +### **重置 API Key** + +在**成员管理**页面，找到目标成员，点击操作列的**重置**。重置后原 API Key 立即失效，新 API Key 需重新发给成员。 + +### **移出成员** + +在**成员管理**页面，找到目标成员，点击操作列的**移出组织**。移出后席位自动回收、API Key 立即失效。 + +## **SAML 接入** + +通过标准 SAML 2.0 对接企业 IdP，对应登录页的**SSO**入口。配置后，成员用 IdP 账号登录管理平台即自动加入组织。 + +### **SAML 配置** + +**前提条件**：组织内有成员时无法编辑 SSO 配置，需先全部移出。 + +1. 登录 Token Plan 控制台，在**我的订阅**页面的**团队版**卡片点击**设置**，找到**SSO 配置**区域。 + +2. 从企业 IdP 获取 IdP 信息（IdP Entity ID、IdP SSO URL、IdP Certificate）。点击**编辑**，填入自定义的 SP Entity ID 和上述 IdP 信息，保存。 + +3. 保存后系统自动生成 ACS URL。将 SP Entity ID 和 ACS URL 填入企业 IdP 的 SSO 应用配置中。 + +4. 在**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 + + +**SP 信息**（百炼侧） + +**参数** + +**说明** + +SP Entity ID + +百炼在 SSO 流程中的唯一标识，自定义填写，需同步填入企业 IdP 的 SSO 应用配置。 + +ACS URL + +IdP 认证成功后回传响应的地址，保存 SSO 配置后由百炼自动生成，需填入企业 IdP。 + +SP Certificate + +百炼侧 SAML 签名证书，保存 SSO 配置后由系统自动生成，用于企业 IdP 验证百炼侧响应签名，需同步配置到企业 IdP 信任链。 + +**IdP 信息**（企业侧，需从企业 IdP 获取后填入） + +**参数** + +**说明** + +IdP Entity ID + +企业身份提供商的唯一标识。 + +IdP SSO URL + +企业 IdP 的登录入口地址，成员登录时跳转到此地址进行认证。 + +IdP Certificate + +企业 IdP 的签名证书，百炼用来验证响应确实来自该企业。 + +### **配置示例：阿里云 IDaaS** + +**前提条件**：已开通阿里云 IDaaS EIAM 实例。 + +1. **在 IDaaS 中创建 SAML 应用**：登录 IDaaS 实例管理平台，进入**应用管理**，点击**添加应用**，选择**标准 SAML 2.0**应用模板，填写应用名称后创建。 + +2. **从 IDaaS 获取 IdP 信息并填入百炼**：在 IDaaS 应用的**单点登录**页面底部的**应用配置信息**区域，复制 IdP 信息，填入 Token Plan 控制台的 SSO 配置区域。 + +3. **将 SP Entity ID 和 ACS URL 填入 IDaaS**：保存后百炼 SSO 配置区域显示自动生成的 ACS URL，在 IDaaS 应用的**登录访问** > **单点登录**页面填入对应参数。 + +4. **在 IDaaS 中创建账户并授权**：在 IDaaS 的**账户管理**中为团队成员创建账户，在 SAML 应用的**登录访问** > **应用授权**页面添加授权。 + +5. **分享管理平台地址给成员**：在**设置**页面的**基本信息**区域复制**管理平台地址**，成员访问该地址选择 SSO 登录即可加入组织。 + + +## **钉钉接入** + +使用钉钉作为身份系统的企业可直接接入，对应登录页的**钉钉**入口。配置后，成员用钉钉账号登录管理平台即自动加入组织。 + +**前提条件**：已在钉钉中创建企业，并将待登录的成员加入企业。 + +1. **创建钉钉企业内部应用**：登录钉钉开发者后台，进入**应用开发** > **企业内部应用** > **钉钉应用**，点击**创建应用**。 + +2. **获取应用凭证**：在应用详情页的**基础信息** > **凭证与基础信息**，记录 Client ID 和 Client Secret。 + +3. **配置回调域名**：在**开发配置** > **安全设置**，在**重定向 URL（回调域名）**中填入 `https://account-enterprise.bailian.aliyunportal.com/api/v1/auth/dingtalk/callback`。 + +4. **开通通讯录读权限**：在应用的**权限管理**页面，开通**通讯录个人信息读权限**。 + +5. **发布应用**：在**应用发布** > **版本管理与发布**页面，创建新版本并发布。 + +6. **在 Token Plan 管理平台填入凭证**：在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在**SSO 配置**区域切换到**钉钉**选项卡，填入配置名称、钉钉应用 AppKey 和 AppSecret。 + +7. **把管理平台地址分享给成员**：在**设置**页面的**基本信息**区域复制**管理平台地址**，成员访问该地址选择钉钉登录即可加入组织。 + + +## **席位操作** + +### **查看席位状态** + +在控制台**我的订阅**页面的**团队座席**区域查看各档位席位的已分配数量和总数；在**订阅明细**区域可查看每个席位的状态和到期时间。 + +### **分配席位** + +1. 在**成员管理**页面，找到目标成员，点击操作列的**分配席位**。 + +2. 在弹窗中选择席位版本，点击**确定**。 + + +分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 + +### **回收席位** + +在**成员管理**页面，找到目标成员，点击操作列的**回收席位**，在确认弹窗中点**确定**。回收后席位转为未分配，原成员将无法使用席位的 Credits；重新分配后系统会为新成员生成新的 API Key。 + +### **加购席位** + +在 Token Plan 控制台的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。新加席位与现有订阅统一到期，费用按剩余时长折算。 + +### **升级席位** + +在 Token Plan 控制台的**我的订阅**页面，在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。升级按剩余时长补缴差价。 + +## **用量分析** + +用量分析可从 Token Plan 控制台的我的订阅页面或管理平台进入。在**用量分析**页面，所有者可查看： + +- **用量趋势**：近 1、7、30 天的 Credits 消耗趋势图。 + +- **模型用量**：组织内各模型的 Credits 消耗。 + +- **成员用量**：各成员的 Credits 消耗。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md new file mode 100644 index 00000000..4b4358d0 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md @@ -0,0 +1,314 @@ +# 概述 + +Token Plan 团队版是阿里云百炼推出的 AI 大模型订阅服务，以 Credits 统一计量，支持文本生成、图像生成与视频生成模型，兼容主流 AI 编程与智能体工具，提供团队管理后台、数据安全保障，调用平稳运行。 + +**说明** + +Token Plan 团队版目前仅支持**华北2（北京）**地域。 + +## **产品简介** + +Token Plan 团队版整合千问和三方模型，支持文本生成、图像生成与视频生成。通过 Credits 统一计量，同一订阅可在多种 AI 工具中使用。 + +- **多模型灵活切换**：支持多模型按需切换，按 Credits 统一抵扣。 + +- **兼容多种工具**：适配多种主流编程工具及热门 Agent 工具。控制台提供快速接入 AI 工具入口，支持 Qwen Code、Claude Code、Qoder、Qoder CN、OpenClaw 等工具接入。 + +- **多档位套餐**：提供标准坐席、高级坐席、尊享坐席多档位套餐，匹配不同使用强度。 + +- **团队管理**：提供管理后台，支持席位分配与回收、成员用量分析等团队管理能力。 + +- **预算可控**：支持按月或按年订阅，预算可控。 + +- **数据安全**：承诺不使用对话数据进行模型训练，满足企业级数据隐私要求。 + +- **平稳运行**：多租户隔离架构，调用高峰期间不排队。 + + +Token Plan 团队版提供套餐专属 Base URL，兼容 OpenAI、Anthropic 接口标准，具体地址可在控制台**我的订阅**的 API Key 区域查看。 + +## **支持的模型** + +**重要** + +团队版支持 qwen3.8-max-preview 预览模型，享有以下限时权益： + +1. **预览版**：qwen3.8-max-preview 当前为预览版本，**预览期间模型能力会持续迭代升级**。预览结束后该模型会下线或替换成正式版本。 + +2. **限时加量 10 倍**：限时活动期间，模型调用 Credits 消耗低至 1 折，相当于增加 10 倍用量。 + + +阿里云百炼有权根据运营情况对活动进行变更或调整，包括不限于活动内容和有效期等，请以页面最新内容或阿里云通知为准。 + +**品牌** + +**模型 ID（Model ID）** + +**模型能力** + +千问 + +qwen3.8-max-preview + +推理模型、视觉理解、文本生成 + +qwen3.7-max + +推理模型、文本生成 + +qwen3.7-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-plus + +推理模型、视觉理解、文本生成 + +qwen3.6-flash + +推理模型、视觉理解、文本生成 + +qwen-image-2.0 + +图片生成 + +qwen-image-2.0-pro + +图片生成 + +万相 + +wan2.7-image + +图片生成 + +wan2.7-image-pro + +图片生成 + +DeepSeek + +deepseek-v4-pro + +推理模型、文本生成 + +deepseek-v4-flash + +推理模型、文本生成 + +deepseek-v3.2 + +推理模型、文本生成 + +月之暗面 + +kimi-k2.7-code + +推理模型、视觉理解、文本生成 + +kimi-k2.6 + +推理模型、视觉理解、文本生成 + +kimi-k2.5 + +推理模型、视觉理解、文本生成 + +智谱 AI + +glm-5.2 + +推理模型、文本生成 + +glm-5.1 + +推理模型、文本生成 + +glm-5 + +推理模型、文本生成 + +MiniMax + +MiniMax-M2.5 + +推理模型、文本生成 + +HappyHorse + +happyhorse-1.1-i2v + +视频生成 + +happyhorse-1.1-t2v + +视频生成 + +happyhorse-1.1-r2v + +视频生成 + +## **套餐与定价** + +前往 Token Plan 团队版购买页面选择坐席类型、数量和订阅周期，完成订阅。主账号和 RAM 账号均可订阅。订阅周期支持按月购买、按年购买、连续包月包年。 + +### **限时活动** + +即日起至 2026 年 7 月 22 日 23:59（UTC+8），qwen3.7-max 模型 Credits 消耗减半，同时支持隐式缓存。 + +### **Token Plan 团队版** + +提供标准坐席、高级坐席、尊享坐席三个档位，匹配不同使用强度。 + +席位（坐席）是 Token Plan 团队版的最小订阅单位，代表一个团队成员的使用名额。管理员在团队管理中将席位分配给成员后，系统自动为该成员生成专属的 API Key。每个席位绑定一个成员、对应一个 API Key，不可共享。 + +**坐席类型** + +**价格** + +**额度** + +**适用场景** + +标准坐席 + +原价 ¥198/坐席/月 +限时 **¥150/坐席/月** + +25,000 Credits/坐席/月 + +轻度使用 AI 辅助的团队成员 + +高级坐席 + +原价 ¥698/坐席/月 +限时 **¥550/坐席/月** + +100,000 Credits/坐席/月 + +日常高频使用 AI 编程或办公的团队成员 + +尊享坐席 + +¥1,398/坐席/月 + +250,000 Credits/坐席/月 + +重度依赖 AI 的核心开发者或高强度使用者 + +**重要** + +限时优惠仅适用于包月订阅的新购、续费和自动续费，包年订阅和升级坐席不参与。加购坐席时按剩余时长折算费用，实际收费取折算金额与限时价中的较低值。 + +### **Token Plan 团队版 - 共享用量包** + +跨坐席共享的弹性用量包，当个别坐席用量超出套餐额度时，可从共享用量包中抵扣。每个共享用量包有效期为 1 个月，到期未使用的额度自动清零。持有多个共享用量包时，优先抵扣最近到期的用量包。 + +**档位** + +**价格** + +**额度** + +Token Plan 团队版 - 共享用量包 + +¥5,000/个 + +625,000 Credits/个 + +## **Credits 计费机制** + +### **计费说明** + +单次消耗的 Credits 由模型类型、Token 用量、思考模式及工具调用等动态决定，实际消耗以控制台订阅页用量明细为准。 + +### **计算示例** + +以 qwen3.6-plus 为例，预估单次请求消耗明细如下（不同模型的单价不同，实际以账单为准）： + +**Token 类型** + +**数量** + +**消耗 Credits** + +输入 tokens + +8,349 + +1.67 + +缓存 tokens + +40,794 + +0.82 + +输出 tokens + +573 + +0.69 + +**合计** + +**约 3.18 Credits** + +### **抵扣顺序** + +1. 优先从坐席套餐的月度额度中抵扣。 + +2. 坐席额度用尽后，从共享用量包中抵扣。持有多个共享用量包时，优先抵扣最近到期的用量包。 + +3. 全部额度用尽后，服务将暂停至下一计费周期或购买共享用量包补充额度。 + + +## **查看额度消耗情况** + +**通过控制台**：登录 Token Plan 控制台，在**我的订阅**页面查看总额度使用百分比、重置时间、团队席位分配情况，以及各席位与共享用量包的状态和到期时间。 + +**通过团队管理平台**：在**用量分析**页面，可查看近 1、7、30 天的 Credits 消耗趋势、各模型用量，以及每个成员的消耗明细。详见团队管理。 + +## **订阅管理** + +### **加购席位** + +在 Token Plan 控制台的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。新加席位与现有订阅统一到期，费用和 Credits 额度均按剩余时长折算。 + +### **升级席位** + +在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。升级按剩余时长补缴差价。 + +### **续费** + +- 点击**续费**按钮，续费周期与订阅时一致（按月订阅则按月续费，按年订阅则按年续费），到期前完成可避免服务中断。 + +- **自动续费**：点击**开启自动续费**，确认后次日生效，到期前 9 天系统按订阅周期自动扣款续费。如需关闭，点击**关闭自动续费**并确认。 + + +### **退订席位** + +在**订阅明细**中点击席位的**退订**，按席位维度退订；已有用量消耗的席位不可退订。也可勾选多个席位后点击**批量退订**。退款原路退回支付账户，预计 1-3 个工作日到账。 + +### **加购共享用量包** + +在**共享用量包**区域点击**前往购买**。 + +### **注意事项** + +- 退订重购后，API Key 和 Base URL 会发生变更，需在工具中重新配置。新 API Key 和 Base URL 均可在控制台[**我的订阅**](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)页面的 API Key 区域获取。 + +- 续费仅延长订阅有效期，不会叠加补充至当前计费周期的额度。 + +- 团队版与个人版可同时持有，各自独立计费。 + + +## **使用细则** + +1. **使用范围**：仅限在兼容的 AI 编程和智能体工具中交互式使用，不可用于自动化脚本或应用后端。违规使用可能导致订阅暂停或 API Key 封禁。 + +2. **数据安全**：Token Plan 团队版不会使用对话数据训练模型。 + +3. **账号规范**：API Key 仅限已分配席位的成员本人使用，不可共享或公开泄露。 + +4. **服务地域**：Token Plan 团队版目前仅在特定地域提供服务，如需从海外调用，请确认符合当地法律法规要求。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md similarity index 69% rename from skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md rename to skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md index 4486ad53..86f23ecf 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-quickstart.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md @@ -4,16 +4,18 @@ ## **步骤一：订阅 Token Plan 团队版** -访问 [Token Plan 团队版购买页面](https://common-buy.aliyun.com/token-plan/)，选择坐席类型、数量和订阅周期（按月或按年）并完成订阅，主账号和 RAM 账号均可订阅。 +访问 [Token Plan 团队版购买页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)，选择套餐档位和订阅周期并完成订阅，主账号和 RAM 账号均可订阅。 -购买须知： - -- RAM 子账号订阅前，需主账号在 RAM 控制台授予 AliyunBailianFullAccess 权限。 +- **RAM 用户授权**：RAM 用户使用 Token Plan 前，需由主账号完成以下授权： + 1. 在 [RAM 控制台](https://ram.console.aliyun.com/)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，同时授予 `AliyunBSSReadOnlyAccess` 系统策略。 + + 2. 在百炼控制台[账号管理](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)页面，为该 RAM 用户分配管理员或订阅套餐权限。 + ## **步骤二：获取 API Key 和 Base URL** -- **API Key**：在 Token Plan 控制台的[成员管理页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)或管理平台创建成员账号，分配席位后，为成员生成 API Key。详见[团队管理](https://help.aliyun.com/zh/model-studio/token-plan-team)。 +- **API Key**：在 Token Plan 控制台的成员管理页面或管理平台创建成员账号，分配席位后，为成员生成 API Key。详见团队管理。 API Key 使用须知： @@ -53,6 +55,8 @@ Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离， ## **步骤三：接入 AI 工具** +将 API Key 和 Base URL 配置到 AI 工具中，即可开始使用。 + [**OpenClaw**开源、自托管个人 AI 助手](https://help.aliyun.com/zh/model-studio/openclaw) [**Hermes Agent**开源 AI 代理框架，内置自学习循环](https://help.aliyun.com/zh/model-studio/hermes-agent) @@ -83,17 +87,10 @@ Token Plan、Coding Plan 和按量付费的 API Key 与 Base URL 完全隔离， [··· **更多工具**其他编程工具](https://help.aliyun.com/zh/model-studio/more-tools) -## **可选：接入图像生成模型** +## **可选：接入多模态生成模型** -Token Plan 团队版支持图像生成模型（qwen-image-2.0、wan2.7-image 等）。图像生成模型使用独立的接口，需要通过 AI 工具的 Skill 或扩展机制接入。具体配置方法请参见[接入多模态生成模型](https://help.aliyun.com/zh/model-studio/token-plan-multimodal-gen)。 +Token Plan 团队版支持图像生成模型（qwen-image-2.0、wan2.7-image 等）。图像生成模型使用独立的接口，需要通过工具的 Skill 或扩展机制接入。详见[接入多模态生成模型](https://help.aliyun.com/zh/document_detail/6546109.html)。 -## **可选：工具调用** - -通过接入工具调用，模型可以在对话中调用联网搜索、代码解释器等扩展能力。 - -- qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash：内置联网搜索、代码解释器、网页抓取、以图搜图、文搜图 5 个工具，通过 Responses API 直接调用。内置工具不额外收费，产生的 token 消耗统一从套餐 Credits 中抵扣。 - -- 其他模型：通过 MCP 服务接入工具。 - +## **可选：接入 Harness 工具** -详细说明请参见[工具调用](https://help.aliyun.com/zh/model-studio/token-plan-tool)。 +部分模型支持通过 Responses API 调用联网搜索、代码解释器等扩展能力。详见[接入 Harness 工具](https://help.aliyun.com/zh/document_detail/6528494.html)。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md deleted file mode 100644 index 9e83ccff..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/token-plan-guide/token-plan-team.md +++ /dev/null @@ -1,272 +0,0 @@ -# 团队管理 - -在 Token Plan 控制台或管理平台中添加和管理团队成员、分配和回收席位、监控 Credits 用量。 - -## **访问入口** - -**阿里云主账号或 RAM 用户**：登录[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)，在**我的订阅**页面进行团队管理：在**团队版**卡片点击**设置**修改组织名称、登录方式（SSO/钉钉）等；在**订阅明细**区域点击**分配座席**进入成员管理。 - -**说明** - -RAM 用户使用 Token Plan 前，需由主账号完成以下授权： - -1. 在[RAM 控制台](https://ram.console.aliyun.com/users)为该 RAM 用户授予 `AliyunTokenPlanReadOnlyAccess`（只读）或 `AliyunTokenPlanFullAccess`（管理）系统策略，具体权限范围按实际需要选择。 - -2. 在百炼控制台[**账号管理**](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/user_management/user_management)页面，为该 RAM 用户分配**管理员**或**订阅套餐**权限。 - - -通过 **SSO 或钉钉**加入的成员：通过管理员分发的**管理平台地址**登录管理平台。管理平台地址可在**设置**页面的**基本信息**区域获取。 - -## **角色与权限** - -**角色** - -**权限** - -所有者 - -添加或移除成员、分配或回收席位、修改成员角色、查看全部成员和模型的用量 - -管理员 - -权限范围与所有者相同。管理员由所有者授予，可被移除或降级。 - -成员 - -使用管理员分配的 API Key 和 Base URL 调用模型 - -## **成员管理** - -### **添加成员** - -- **手动添加**（不能登录管理平台，仅供 API 调用）：在**成员管理**页面点**添加成员**，在弹窗中填用户名（仅支持英文字母、数字、下划线）和角色，可选择同时分配席位（默认不分配，需加购可点弹窗内加购更多席位）。在该成员的操作列点**分配席位**选择席位版本，分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 - -- **SSO 或钉钉登录**（成员可登录管理平台，自管席位和 API Key）：完成 [SAML](#tp05-sso) 或[钉钉](#tp05-dingtalk)接入配置后，成员从登录页对应入口登录即自动加入。 - - -### **修改成员角色** - -在**成员管理**页面，找到目标成员，点击操作列的**修改角色**，在弹窗中选择新角色（管理员或成员）后保存。所有者角色不可修改。 - -### **重置 API Key** - -在**成员管理**页面，找到目标成员，点击操作列的**重置**。重置后原 API Key 立即失效，新 API Key 需重新发给成员。 - -### **移出成员** - -在**成员管理**页面，找到目标成员，点击操作列的**移出组织**。移出后席位自动回收、API Key 立即失效。 - -## **SAML 接入** - -通过标准 SAML 2.0 对接企业 IdP，对应登录页的**SSO**入口。配置后，成员用 IdP 账号登录管理平台即自动加入组织。 - -### **SAML 配置** - -**前提条件**：组织内有成员时无法编辑 SSO 配置，需先全部移出。 - -1. 登录 Token Plan 控制台，在**我的订阅**页面的**团队版**卡片点击**设置**，找到**SSO 配置**区域。 - -2. 从企业 IdP 获取 IdP 信息（IdP Entity ID、IdP SSO URL、IdP Certificate）。点击**编辑**，填入自定义的 SP Entity ID 和上述 IdP 信息，保存。 - -3. 保存后系统自动生成 ACS URL。将 SP Entity ID 和 ACS URL 填入企业 IdP 的 SSO 应用配置中。 - -4. 在**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 - - -**参数说明** - -**SP 信息**（百炼侧） - -**参数** - -**说明** - -SP Entity ID - -百炼在 SSO 流程中的唯一标识，自定义填写，需同步填入企业 IdP 的 SSO 应用配置。 - -ACS URL - -IdP 认证成功后回传响应的地址，保存 SSO 配置后由百炼自动生成，需填入企业 IdP。 - -SP Certificate - -百炼侧 SAML 签名证书，保存 SSO 配置后由系统自动生成，用于企业 IdP 验证百炼侧响应签名，需同步配置到企业 IdP 信任链。 - -**IdP 信息**（企业侧，需从企业 IdP 获取后填入） - -**参数** - -**说明** - -IdP Entity ID - -企业身份提供商的唯一标识（Entity ID）。 - -IdP SSO URL - -企业 IdP 的登录入口地址，成员登录时跳转到此地址进行认证。 - -IdP Certificate - -企业 IdP 的签名证书，百炼用来验证响应确实来自该企业。 - -### **配置示例：**[**阿里云 IDaaS**](https://www.aliyun.com/product/idaas) - -**前提条件**：已开通阿里云 IDaaS EIAM 实例。如未开通，登录[IDaaS 控制台](https://yundun.console.aliyun.com/?p=idaas)，在 EIAM 云身份服务实例列表页点击**免费创建实例**。 - -1. **在 IDaaS 中创建 SAML 应用** - - 登录 IDaaS 实例管理平台，进入**应用管理**，点击**添加应用**，选择**标准 SAML 2.0**应用模板，填写应用名称后创建。 - -2. **从 IDaaS 获取 IdP 信息并填入百炼** - - 在 IDaaS 应用的**单点登录**页面底部的**应用配置信息**区域，复制 IdP 信息。然后在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在 **SSO 配置**区域，点击**编辑**，填入自定义的 SP Entity ID 和以下对应的 IdP 信息后保存： - - **IDaaS 应用配置信息** - - **填入百炼的字段** - - **说明** - - IdP 唯一标识（IdP Entity ID） - - IdP Entity ID - - IDaaS 在应用中的唯一标识 - - IdP SSO 地址（IdP Sign-in URL） - - IdP SSO URL - - 成员 SSO 登录时跳转到的认证地址 - - 公钥证书（Certificate） - - IdP Certificate - - 点击“复制证书内容”获取完整证书 - -3. **将 SP Entity ID 和 ACS URL 填入 IDaaS** - - 保存后，百炼 SSO 配置区域显示自动生成的 ACS URL。在 IDaaS 应用的**登录访问** > **单点登录**页面，填入以下对应参数： - - **百炼 SP 信息** - - **填入 IDaaS 的字段** - - **说明** - - SP Entity ID - - 应用唯一标识（SP Entity ID） - - 填入在百炼中自定义的 SP Entity ID - - ACS URL - - 单点登录地址（ACS URL） - - 填入百炼自动生成的 ACS URL - -4. **在 IDaaS 中创建账户并授权** - - 在 IDaaS 实例管理平台的**账户管理**中，为需要使用 Token Plan 的团队成员创建 IDaaS 账户。 - - 然后在 SAML 应用的**登录访问** > **应用授权**页面，点击**添加授权**，选择需要通过 SSO 登录的账户、组或组织机构。后续添加用户重复此步骤即可，无需重新配置 SSO 参数。 - -5. **分享管理平台地址给成员** - - 在**设置**页面的**基本信息**区域复制**管理平台地址**，分享给已授权的成员。成员访问该地址，在登录页选择 SSO 登录方式即可加入组织。 - - -## **钉钉接入** - -使用钉钉作为身份系统的企业可直接接入，对应登录页的**钉钉**入口。配置后，成员用钉钉账号登录管理平台即自动加入组织。 - -**前提条件**：已在钉钉中创建企业，并将待登录的成员加入企业。 - -1. **创建钉钉企业内部应用** - - 登录[钉钉开发者后台](https://open-dev.dingtalk.com/)，进入**应用开发** > **企业内部应用** > **钉钉应用**，点击**创建应用**。填写应用名称和描述后保存。 - -2. **获取应用凭证** - - 在应用详情页，进入**基础信息** > **凭证与基础信息**，记录 Client ID 和 Client Secret（即钉钉应用 AppKey 与 AppSecret），后续在 Token Plan 管理平台中使用。 - -3. **配置回调域名** - - 在应用详情页，进入**开发配置** > **安全设置**，在**重定向 URL（回调域名）**中填入 `https://account-enterprise.bailian.aliyunportal.com/api/v1/auth/dingtalk/callback`，保存。 - -4. **开通通讯录读权限** - - 进入应用的**权限管理**页面，开通**通讯录个人信息读权限**。该权限用于在登录时识别成员的钉钉账号。 - -5. **发布应用** - - 进入**应用发布** > **版本管理与发布**页面，创建新版本并发布。 - -6. **在 Token Plan 管理平台填入凭证** - - 在 Token Plan 控制台**我的订阅**页面的**团队版**卡片点击**设置**，在**SSO 配置**区域，切换到**钉钉**选项卡，填入配置名称、**钉钉应用 AppKey**和**钉钉应用 AppSecret**（步骤 2 中获取），保存。 - -7. **把管理平台地址分享给成员** - - 在**设置**页面的**基本信息**区域复制**管理平台地址**，分享给团队成员。成员访问该地址，在登录页选择钉钉登录方式即可加入组织。 - - -## **席位操作** - -### **查看席位状态** - -在控制台**我的订阅**页面的**团队座席**区域查看各档位席位的已分配数量和总数；在**订阅明细**区域可查看每个席位的状态和到期时间。 - -### **分配席位** - -1. 在**成员管理**页面，找到目标成员，点击操作列的**分配席位**。 - -2. 在弹窗中选择席位版本，点击**确定**。 - - -分配后系统自动生成 API Key，连同 Base URL 发给成员即可调用模型。 - -### **回收席位** - -在**成员管理**页面，找到目标成员，点击操作列的**回收席位**，在确认弹窗中点**确定**。回收后席位转为未分配，原成员将无法使用席位的 Credits；重新分配后系统会为新成员生成新的 API Key。 - -### **加购席位** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面，点击**加购座席**，选择席位档位和数量后提交订单。 - -新加席位与现有订阅统一到期，费用按剩余时长折算。例如订阅周期剩余 15 天，加购一个标准席位，则只需支付 15 天对应的费用。 - -### **升级席位** - -在[Token Plan 控制台](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan)的**我的订阅**页面，在**订阅明细**中找到目标席位，点击**升级**，选择更高档位后提交订单。需要批量操作时，勾选多个席位后点击**批量升级**。 - -升级按剩余时长补缴差价。例如将标准席位升级为高级席位，只需补缴剩余时长内两个档位的差价。 - -## **用量分析** - -用量分析可从 Token Plan 控制台的我的订阅页面或管理平台进入。在**用量分析**页面，所有者可查看： - -- **用量趋势**：近 1、7、30 天的 Credits 消耗趋势图。 - -- **模型用量**：组织内各模型的 Credits 消耗。 - -- **成员用量**：各成员的 Credits 消耗。 - - -## **常见问题** - -### **席位到期后 API Key 是否仍可使用** - -席位到期后，对应的 API Key 将无法调用模型。续订套餐并重新分配席位后即可恢复使用。 - -### **回收席位后能否分配给其他成员** - -可以。回收后席位转为未分配，重新分配给其他成员后系统会生成新的 API Key。 - -### **能否为同一成员更换席位** - -每个成员同一时间只能持有一个席位。如需更换，先回收当前席位，再分配新的席位即可。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md index fe1e76b9..f56aec75 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md @@ -4,7 +4,7 @@ **重要** -deepseek-v3、deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp、deepseek-r1、deepseek-r1-0528、deepseek-r1-distill-qwen-7b/14b/32b 将于**2026年7月9日**下架。推荐转用：[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-plus)、[qwen3.7-max](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max)、[qwen3.6-flash](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.6-flash)。 +deepseek-v3、deepseek-v3.1、deepseek-v3.2、deepseek-v3.2-exp、deepseek-r1、deepseek-r1-0528、deepseek-r1-distill-qwen-7b/14b/32b 将于**2026年10月10日**下架。推荐转用：[qwen3.7-plus](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-plus)、[qwen3.7-max](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.7-max)、[qwen3.6-flash](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/detail/qwen3.6-flash)。 ## **服务接入地址** @@ -514,11 +514,9 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/apps/anthropic/v ## **推理强度（reasoning\_effort）** -deepseek-v4-pro 和 deepseek-v4-flash 默认开启思考模式。通过`reasoning_effort`参数可以调整推理强度，可选值为`high`和`max`，默认为`high`。 +deepseek-v4-pro 和 deepseek-v4-flash 默认开启思考模式。通过`reasoning_effort`参数可以调整推理强度，可选值为`low`、`medium`、`high`、`xhigh`和`max`，默认为`high`。 -**说明** - -设为`low`或`medium`时会映射为`high`，设为`xhigh`时会映射为`max`。 +其中，`low`和`medium`的效果等同于`high`；`xhigh`的效果等同于`max`。 ## **OpenAI兼容** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md index 23cc13c0..f58696f1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md @@ -24,22 +24,28 @@ - 如果通过SDK调用，需要[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk#8833b9274f4v8) -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5 均支持输入文本、图像或视频。kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型（`enable_thinking` 始终为 true，无法设置为 false）。kimi/kimi-k2.6、kimi/kimi-k2.5 可通过 `enable_thinking` 参数控制思考模式，默认开启思考模式： +kimi 系列模型均支持输入文本、图像或视频： -- **思考模式**（`enable_thinking: true`）：模型会输出详细的推理过程（`reasoning_content`） +1. kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。 -- **非思考模式**（`enable_thinking: false` 或不设置）：直接输出结果，不包含推理过程 +2. kimi/kimi-k3、kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型。 +3. kimi/kimi-k2.6、kimi/kimi-k2.5 可通过 `enable_thinking` 参数控制思考模式，默认开启思考模式： + + - **思考模式**（`reasoning_effort: "max"`）：模型会输出详细的推理过程（`reasoning_content`） + + - **非思考模式**（`enable_thinking: false` 或不设置）：直接输出结果，不包含推理过程 + -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6 支持通过 `preserve_thinking` 参数在多轮对话中传递思考过程，详情请参见[传递思考过程](https://help.aliyun.com/zh/model-studio/deep-thinking#jln7docdq5et5)。 +除 kimi/kimi-k2.5外，其他模型均支持通过 `preserve_thinking` 参数在多轮对话中传递思考过程，详情请参见[传递思考过程](https://help.aliyun.com/zh/model-studio/deep-thinking#jln7docdq5et5)。 -以下示例演示如何调用思考模式的 kimi-k2.6 模型进行文本生成。 +以下示例演示如何调用思考模式的 kimi/kimi-k3 模型进行文本生成。 ## OpenAI兼容 **说明** -`enable_thinking`非 OpenAI 标准参数，OpenAI Python SDK 通过 `extra_body`传入，Node.js SDK 作为顶层参数传入。 +`reasoning_effort`非 OpenAI 标准参数，OpenAI Python SDK 通过 `extra_body`传入，Node.js SDK 作为顶层参数传入。 ## Python @@ -53,10 +59,10 @@ client = OpenAI( ) completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[{"role": "user", "content": "1+1等于多少？"}], - # 通过 extra_body 设置 enable_thinking 开启思考模式 - extra_body={"enable_thinking": True} + # 通过 extra_body 设置 reasoning_effort 开启思考模式 + extra_body={"reasoning_effort": "max"} ) msg = completion.choices[0].message @@ -109,9 +115,9 @@ const messages = [ ]; const response = await client.chat.completions.create({ - model: "kimi/kimi-k2.6", + model: "kimi/kimi-k3", messages, - extra_body: { enable_thinking: true }, + extra_body: { reasoning_effort: "max" }, }); const msg = response.choices[0].message; @@ -156,7 +162,7 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completi --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ --header 'Content-Type: application/json' \ --data '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages":[ { "role": "system", @@ -167,13 +173,13 @@ curl --location 'https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completi "content": "1+1等于多少？" } ], - "enable_thinking": true + "reasoning_effort": "max" }' ``` ## **多模态调用示例** -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5不仅支持纯文本对话，还具备强大的多模态理解能力。本章节将介绍如何让模型理解图像和视频内容。 +Kimi 系列模型不仅支持纯文本对话，还具备强大的多模态理解能力。本章节将介绍如何让模型理解图像和视频内容。 **重要** @@ -198,7 +204,7 @@ client = OpenAI( # 单图传入示例（开启思考模式） completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[ { "role": "user", @@ -213,7 +219,7 @@ completion = client.chat.completions.create( ] } ], - extra_body={"enable_thinking":True} # 开启思考模式 + extra_body={"reasoning_effort":"max"} # 开启思考模式 ) # 输出思考过程 @@ -227,7 +233,7 @@ print(completion.choices[0].message.content) # 多图传入示例（开启思考模式，取消注释使用） # completion = client.chat.completions.create( -# model="kimi/kimi-k2.6", +# model="kimi/kimi-k3", # messages=[ # { # "role": "user", @@ -244,7 +250,7 @@ print(completion.choices[0].message.content) # ] # } # ], -# extra_body={"enable_thinking":True} +# extra_body={"reasoning_effort":"max"} # ) # # # 输出思考过程和回复 @@ -266,7 +272,7 @@ const openai = new OpenAI({ // 单图传入示例（开启思考模式） const completion = await openai.chat.completions.create({ - model: 'kimi/kimi-k2.6', + model: 'kimi/kimi-k3', messages: [ { role: 'user', @@ -281,7 +287,7 @@ const completion = await openai.chat.completions.create({ ] } ], - enable_thinking: true // 开启思考模式 + reasoning_effort: "max" // 开启思考模式 }); // 输出思考过程 @@ -313,7 +319,7 @@ console.log(completion.choices[0].message.content); // ] // } // ], -// enable_thinking: true +// reasoning_effort: "max" // }); // // // 输出思考过程和回复 @@ -332,7 +338,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H "Content-Type: application/json" \ -d '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages": [ { "role": "user", @@ -350,7 +356,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions ] } ], - "enable_thinking": true + "reasoning_effort": "max" }' # 多图输入示例（取消注释使用） @@ -358,7 +364,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions # -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ # -H "Content-Type: application/json" \ # -d '{ -# "model": "kimi/kimi-k2.6", +# "model": "kimi/kimi-k3", # "messages": [ # { # "role": "user", @@ -382,7 +388,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions # ] # } # ], -# "enable_thinking": true +# "reasoning_effort": "max" # }' ``` @@ -404,7 +410,7 @@ client = OpenAI( ) completion = client.chat.completions.create( - model="kimi/kimi-k2.6", + model="kimi/kimi-k3", messages=[ { "role": "user", @@ -440,7 +446,7 @@ const openai = new OpenAI({ async function main() { const response = await openai.chat.completions.create({ - model: "kimi/kimi-k2.6", + model: "kimi/kimi-k3", messages: [ { role: "user", @@ -500,7 +506,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ -H 'Content-Type: application/json' \ -d '{ - "model": "kimi/kimi-k2.6", + "model": "kimi/kimi-k3", "messages": [ { "role": "user", @@ -560,7 +566,7 @@ curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions [结构化输出](https://help.aliyun.com/zh/model-studio/qwen-structured-output) -kimi/kimi-k2.7-code-highspeed +kimi/kimi-k3 支持 @@ -574,22 +580,34 @@ kimi/kimi-k2.7-code-highspeed 支持 +kimi/kimi-k2.7-code-highspeed + kimi/kimi-k2.7-code kimi/kimi-k2.6 kimi/kimi-k2.5 -- kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5支持上下文缓存（隐式缓存，自动开启），kimi/kimi-k2.7-code-highspeed命中缓存的输入Token按输入价格的20.0%计费，kimi/kimi-k2.7-code命中缓存的输入Token按输入价格的20.0%计费，kimi/kimi-k2.6命中缓存的输入Token按输入价格的16.9%计费，kimi/kimi-k2.5命中缓存的输入Token按输入价格的17.5%计费。 +以上模型支持上下文缓存（隐式缓存，自动开启）： + +- kimi/kimi-k3命中缓存的输入Token按输入价格的10%计费 -- 在思考模式下，使用 kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code、kimi/kimi-k2.6、kimi/kimi-k2.5 进行工具调用时：必须在每轮 assistant 消息中保留 `reasoning_content` 字段，`tool_choice` 也仅支持 `"auto"`（默认）和 `"none"`），否则会报错。 +- kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 命中缓存的输入Token按输入价格的20.0%计费， + +- kimi/kimi-k2.6命中缓存的输入Token按输入价格的16.9%计费 + +- kimi/kimi-k2.5命中缓存的输入Token按输入价格的17.5%计费。 ## **参数默认值** **模型** -**stream\_options** +**tool\_choice** + +**preserve\_thinking** + +**reasoning\_effort** **temperature** @@ -599,33 +617,33 @@ kimi/kimi-k2.5 **presence\_penalty** -**tool\_choice** +**stream\_options** -**top\_k** +kimi/kimi-k3 -**preserve\_thinking** +auto -kimi/kimi-k2.7-code-highspeed +默认关闭 -仅支持设置为`true` +max -1.0 +1 0.95 0.0 -0.0 +\- -auto +仅支持设置为`true` -\- +kimi/kimi-k2.7-code-highspeed -默认开启 +auto -kimi/kimi-k2.7-code +默认开启 -仅支持设置为`true` +\- 1.0 @@ -635,33 +653,37 @@ kimi/kimi-k2.7-code 0.0 +仅支持设置为`true` + +kimi/kimi-k2.7-code + auto +默认开启 + \- -默认开启 +1.0 -kimi/kimi-k2.6 +0.95 + +0.0 + +0.0 仅支持设置为`true` +kimi/kimi-k2.6 + +思考模式/非思考模式：auto + +默认关闭 + +\- + 思考模式：1.0 非思考模式：0.6 - - - - - - - - - - - - - - 思考模式/非思考模式：0.95 @@ -669,27 +691,25 @@ kimi/kimi-k2.6 思考模式/非思考模式：0.0 -思考模式/非思考模式：auto - -\- - -默认关闭 +仅支持设置为`true` kimi/kimi-k2.5 \- +\- + - `stream_options`仅支持设置为`true`，`temperature`、`top_p`、`repetition_penalty`、`presence_penalty`不支持设置为其他值； -- 在思考模式下，不支持强制调用某个工具，`tool_choice`仅支持设置为`auto`（默认值）和`none`。 +- kimi/kimi-k3 支持 `reasoning_effort` 参数，唯一支持值为 `max`。 + +- 在思考模式下，使用 Kimi 模型进行工具调用时：必须在每轮 assistant 消息中保留 `reasoning_content` 字段；对于`tool_choice`参数，`kimi-k3` 支持 `auto` / `none` / `required` 三档；其他模型不支持 `required`，传入会报错。`kimi-k3` 支持动态加载工具，详细用法请参见[动态加载工具](https://platform.kimi.com/docs/guide/use-dynamic-tool-loading)。 - ”-”表示没有默认值，也不支持设置。 ## **模型列表与计费** -kimi/kimi-k2.7-code-highspeed、kimi/kimi-k2.7-code 为仅思考模型（`enable_thinking` 始终为 true，无法设置为 false）。kimi/kimi-k2.7-code-highspeed 与 kimi/kimi-k2.7-code 功能完全一致，速度提升5~6倍。kimi/kimi-k2.6、kimi/kimi-k2.5属于混合思考模型，通过`enable_thinking`参数控制是否开启思考模式（注意：均无法通过`thinking_budget`限制思考长度）。 - 模型上下文长度与价格信息请参见[百炼控制台](https://bailian.console.aliyun.com/cn-beijing?tab=model#/model-market/all)。 按照模型的输入与输出 Token 计费。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md index 51c73192..ab820eb4 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md @@ -1,6 +1,6 @@ # Chatbox -Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Chatbox 是一款跨平台 AI 客户端应用，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **下载安装 Chatbox** @@ -12,6 +12,8 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,24 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**API 主机** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**模型** + +填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，如 `qwen3.8-max-preview`。 + ### Token Plan 团队版 **配置项** @@ -27,7 +47,7 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **API 主机** @@ -100,4 +120,6 @@ Chatbox 是一款跨平台 AI 客户端应用，可以通过Token Plan 团队版 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md index af83a91a..162cb9f1 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md @@ -1,6 +1,6 @@ # Cherry Studio -Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Cherry Studio** @@ -12,6 +12,8 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,24 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**API 地址** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**模型** + +可用模型请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + ### Token Plan 团队版 **配置项** @@ -27,7 +47,7 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **API 地址** @@ -96,7 +116,9 @@ Cherry Studio 是一款开源 AI 桌面客户端，可以通过 Token Plan 团 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 The value of the enable\_thinking parameter is restricted to True diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md index 00895664..b0b0c475 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md @@ -1,6 +1,6 @@ # Claude Code -Claude Code 是 Anthropic 推出的命令行 AI 编程助手。通过阿里云百炼，可以使用按量计费、Coding Plan 或 Token Plan 团队版接入 Claude Code。 +Claude Code 是 Anthropic 推出的命令行 AI 编程助手。通过阿里云百炼，可以使用按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入 Claude Code。 ## **安装 Claude Code** @@ -47,6 +47,36 @@ npm install -g @anthropic-ai/claude-code 新建 `~/.claude/settings.json`（Windows 路径：`C:\Users\<用户名>\.claude\settings.json`），写入对应套餐的配置。 +### Token Plan 个人版 + +将 YOUR\_API\_KEY 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro。完整说明参见 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "env": { + "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", + "ANTHROPIC_BASE_URL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", + "ANTHROPIC_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3.6-flash", + "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.8-max-preview", + "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max", + "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "983616" + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- **thinking**：始终开启，不支持关闭。 + +- **temperature**：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- **reasoning\_effort**：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Token Plan 团队版 将 YOUR\_API\_KEY 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型参见 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -56,15 +86,27 @@ npm install -g @anthropic-ai/claude-code "env": { "ANTHROPIC_AUTH_TOKEN": "YOUR_API_KEY", "ANTHROPIC_BASE_URL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", - "ANTHROPIC_MODEL": "qwen3.7-max", + "ANTHROPIC_MODEL": "qwen3.8-max-preview", "ANTHROPIC_DEFAULT_HAIKU_MODEL": "qwen3.6-flash", - "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.7-max", - "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.7-max", - "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max" + "ANTHROPIC_DEFAULT_SONNET_MODEL": "qwen3.8-max-preview", + "ANTHROPIC_DEFAULT_OPUS_MODEL": "qwen3.8-max-preview", + "CLAUDE_CODE_SUBAGENT_MODEL": "qwen3.7-max", + "CLAUDE_CODE_MAX_CONTEXT_TOKENS": "983616" } } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- **thinking**：始终开启，不支持关闭。 + +- **temperature**：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- **reasoning\_effort**：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan 将 YOUR\_API\_KEY 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型参见 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -169,6 +211,14 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 **配置信息** + Token Plan 个人版 + + 供应商名称：百炼-Token Plan 个人版 + + API Key：[控制台获取](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + + 请求地址：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + Token Plan 团队版 供应商名称：百炼-Token Plan @@ -208,35 +258,23 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 ### 接入 Claude Code 桌面版 +Claude Code 桌面版（Claude Desktop）与 Claude Code CLI 是两个独立入口，在 CC Switch 中分别对应 **Claude Code** 与 **Claude Desktop** 面板。桌面版通过 CC Switch 本地网关访问百炼：网关地址与鉴权令牌均由 CC Switch 自动写入桌面版配置，**无需在桌面版中手动填写百炼 API Key**——百炼 API Key 只在 CC Switch 供应商配置中填写，由本地路由转发时自动注入。 + +**重要** + +请勿在桌面版的第三方推理配置中手动填写百炼 API Key。桌面版对 CC Switch 本地网关（地址 `http://127.0.0.1:15721/claude-desktop`）的鉴权令牌由 CC Switch 自动生成并写入，手动填入百炼 API Key 会因令牌不匹配导致鉴权失败。桌面版第三方配置写入目前仅支持 macOS、Windows。 + 1. 从 [Claude 下载页](https://claude.ai/download)安装 Claude Code 桌面版。 -2. 顶部菜单 **Help** → **Troubleshooting** → **Enable Developer Mode**，重启后顶部出现 **Developer** 菜单。 - -3. **Developer** → **Configure Third-Party Inference**，**Connection** 选 **Gateway**，按下表填写后点击 **Apply locally**： - - **字段** - - **填写** - - Gateway base URL - - CC Switch 路由监听地址，默认 `http://127.0.0.1:15721`，如已修改则与下一步保持一致。 +2. 在 CC Switch 左侧应用切换器切换到 **Claude Desktop** 面板。若未显示该入口，前往**设置 → 通用 → 应用可见性**确认 Claude Desktop 未被隐藏。 - Gateway API key +3. 添加百炼供应商：若已在 **Claude Code** 面板配置过百炼供应商，可点击**将 Claude Code 中已有的供应商导入**一键复用；也可点击右上角 **+** 新增。由于百炼模型 ID（如 `qwen3.7-max`）不是 Claude Desktop 识别的 `claude-sonnet-* / claude-opus-* / claude-haiku-*` 三档角色 ID，需开启**需要模型映射**，为 Sonnet、Opus、Haiku 三档分别填写实际请求的百炼模型（如 Sonnet → qwen3.7-max）。 - 百炼API Key +4. 开启本地路由：前往**设置 → 路由 → 本地路由**，打开**在主页面显示本地路由开关**；回到 Claude Desktop 面板，打开 **Claude Desktop 本地路由**开关，监听地址默认 `127.0.0.1:15721`。 - Gateway auth scheme +5. 在供应商卡片点击**启用**，CC Switch 会自动将第三方推理配置写入 Claude Code 桌面版。 - bearer - - Model list - - Model ID须为 Anthropic 风格，如 `claude-opus-4.7`。实际调用模型由 CC Switch 路由决定，对应关系即供应商[高级选项中的模型映射](#ccswitch-add-li2)。Display name 仅影响下拉显示。 - -4. CC Switch 左上角设置 → **路由**，开启**路由总开关**，监听地址默认 `127.0.0.1:15721`，如需修改请同步上一步。 - -5. 在桌面版模型下拉中选择已配置的模型ID即可使用。 +6. 保持 CC Switch 运行，**完全退出并重启** Claude Code 桌面版后生效，在模型菜单中选择已配置的模型即可使用。 ## **Claude Code IDE 插件** @@ -295,7 +333,9 @@ Claude Code 默认使用 200K 上下文窗口。如果需要处理大型代码 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 启动 Claude Code 后，界面显示"Unable to connect to Anthropic services. Failed to connect to api.anthropic.com: ERR\_BAD\_REQUEST" diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md index 0d220bf9..eb7551f0 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cline.md @@ -1,6 +1,6 @@ # Cline -Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、Coding Plan或按量计费接入阿里云百炼。 +Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Cline** @@ -15,6 +15,8 @@ Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、 阿里云百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -22,6 +24,28 @@ Cline 是一款 VSCode 智能编程插件，可以通过 Token Plan 团队版、 - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**配置项** + +**说明** + +API Provider + +选择 **OpenAI Compatible**。 + +Base URL + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +API Key + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。 + +Model ID + +填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，如 `qwen3.8-max-preview`。 + ### Token Plan 团队版 **配置项** @@ -42,7 +66,7 @@ API Key Model ID -填入 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)，如 `qwen3.7-max`。 +填入 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)，如 `qwen3.8-max-preview`。 ### Coding Plan @@ -133,7 +157,9 @@ Model ID - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 401 Incorrect API key provided diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md index 76a25932..45e0c023 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/codex.md @@ -1,6 +1,6 @@ # Codex -Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 +Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 Codex** @@ -21,19 +21,199 @@ Codex 是 OpenAI 推出的终端 AI 编程助手。可通过 Token Plan 团队 ## **配置接入凭证** -接入需要编辑配置文件`~/.codex/config.toml`并配置环境变量`OPENAI_API_KEY`。根据所选计费方案替换对应值，阿里云百炼提供三种计费方案： +接入需要编辑配置文件`~/.codex/config.toml`并配置环境变量`OPENAI_API_KEY`。根据所选计费方案替换对应值，阿里云百炼提供以下计费方案： + +### 配置模型元数据 + +使用自定义模型（如 qwen3.8-max-preview）时，需要配置模型元数据文件，使 Codex 正确识别模型的上下文窗口、推理深度等参数。 + +1. 新建文件 `~/.codex/model-catalog.local.json`，写入以下内容： + + ``` + { + "models": [ + { + "slug": "qwen3.8-max-preview", + "display_name": "qwen3.8-max-preview", + "description": "DashScope model: qwen3.8-max-preview", + "default_reasoning_level": "xhigh", + "supported_reasoning_levels": [ + { + "effort": "low", + "description": "Fast responses with lighter reasoning" + }, + { + "effort": "high", + "description": "Greater reasoning depth for complex problems" + }, + { + "effort": "xhigh", + "description": "Extra high reasoning depth for complex problems" + } + ], + "context_window": 983616, + "effective_context_window_percent": 95, + "supports_parallel_tool_calls": false, + "supports_image_detail_original": true, + "input_modalities": ["text", "image"], + "shell_type": "default", + "visibility": "list", + "supported_in_api": true, + "priority": 1, + "base_instructions": "", + "support_verbosity": false, + "supports_reasoning_summaries": false, + "experimental_supported_tools": [], + "truncation_policy": { + "mode": "bytes", + "limit": 10000 + } + } + ] + } + ``` + +2. 在 `~/.codex/config.toml` 中添加以下配置，指向元数据文件： + + ``` + model_catalog_json = "~/.codex/model-catalog.local.json" + ``` + + +### Token Plan 个人版 + +`model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)，可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro。将`OPENAI_API_KEY`环境变量设置为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。 + +#### Responses API（qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） + +qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 + +``` +model_provider = "Model_Studio_Token_Plan_Personal" +model = "qwen3.8-max-preview" +[model_providers.Model_Studio_Token_Plan_Personal] +name = "Model_Studio_Token_Plan_Personal" +base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +env_key = "OPENAI_API_KEY" +wire_api = "responses" +``` + +#### Chat/Completions API（其他模型） + +其他模型需通过 Chat/Completions API 接入，需安装旧版本 Codex，如 0.80.0： + +``` +npm install -g @openai/codex@0.80.0 +``` +``` +model_provider = "Model_Studio_Token_Plan_Personal" +model = "glm-5" +[model_providers.Model_Studio_Token_Plan_Personal] +name = "Model_Studio_Token_Plan_Personal" +base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" +env_key = "OPENAI_API_KEY" +wire_api = "chat" +``` + +#### 配置环境变量 + +将`OPENAI_API_KEY`环境变量设置为 Token Plan 个人版专属 API Key。 + +## macOS + +1. 在终端中执行以下命令，查看默认 Shell 类型。 + + ``` + echo $SHELL + ``` + +2. 根据 Shell 类型设置环境变量： + + ## Zsh + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + echo 'export OPENAI_API_KEY="YOUR_API_KEY"' >> ~/.zshrc + ``` + + ## Bash + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + echo 'export OPENAI_API_KEY="YOUR_API_KEY"' >> ~/.bash_profile + ``` + +3. 执行以下命令使环境变量生效。 + + ## Zsh + + ``` + source ~/.zshrc + ``` + + ## Bash + + ``` + source ~/.bash_profile + ``` + + +## Windows + +## CMD + +1. 在 CMD 中运行以下命令，设置环境变量。 + + ``` + REM 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + setx OPENAI_API_KEY "YOUR_API_KEY" + ``` + +2. 打开一个新的 CMD 窗口，运行以下命令检查环境变量是否生效。 + + ``` + echo %OPENAI_API_KEY% + ``` + + +## PowerShell + +1. 在 PowerShell 中运行以下命令，设置环境变量。 + + ``` + # 将 YOUR_API_KEY 替换为 Token Plan 个人版 API Key + [Environment]::SetEnvironmentVariable("OPENAI_API_KEY", "YOUR_API_KEY", [EnvironmentVariableTarget]::User) + ``` + +2. 打开一个新的 PowerShell 窗口，运行以下命令检查环境变量是否生效。 + + ``` + echo $env:OPENAI_API_KEY + ``` + + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + ### Token Plan 团队版 `model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。将`OPENAI_API_KEY`环境变量设置为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 -#### Responses API（qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） +#### Responses API（qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash） -qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 +qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-plus 和 qwen3.6-flash 支持 Responses API，可使用最新版 Codex。 ``` model_provider = "Model_Studio_Token_Plan" -model = "qwen3.7-max" +model = "qwen3.8-max-preview" [model_providers.Model_Studio_Token_Plan] name = "Model_Studio_Token_Plan" base_url = "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" @@ -135,6 +315,17 @@ wire_api = "chat" ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan `model`请选择[支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan-overview)。将`OPENAI_API_KEY`环境变量设置为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。 @@ -368,11 +559,11 @@ codex ### **第三方工具提示“不支持国内模型”或“检查被拒 / Bad request (400)”怎么办？** -**原因**：部分第三方管理工具（如 CC-Switch）在切换供应商时会发起“健康检查/连接测试”探测请求，该探测请求的格式与 Codex 实际调用的请求格式不同，百炼网关可能因此返回 400 Bad request 并提示“检查被拒”，工具据此显示“不支持国内模型”。此提示仅代表健康检查探测未通过，**并不代表百炼不支持国内模型，也不影响 Codex 的实际使用。** +**原因**：部分第三方管理工具（如 CC-Switch）在切换供应商时会发起“健康检查/连接测试”探测请求，该探测请求的格式与 Codex 实际调用的请求格式不同，百炼网关可能因此返回 400 Bad request 并提示“检查被拒”，工具据此显示“不支持国内模型”。此提示仅代表健康检查探测未通过，**并不代表百炼不支持中国内地模型，也不影响 Codex 的实际使用。** -**说明**：百炼支持通过 Codex 使用 qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、glm-5 等国内模型，配置方式详见上文[配置接入凭证](#cdx-config)。 +**说明**：百炼支持通过 Codex 使用 qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、glm-5 等中国内地模型，配置方式详见上文[配置接入凭证](#cdx-config)。 -**解决方案**：建议参照上文配置接入凭证，直接在`~/.codex/config.toml`中完成配置，无需依赖第三方工具的健康检查结果；配置完成后参照[验证配置](#cdx-verify)启动 Codex，若能正常进入对话界面即表示可正常使用国内模型。 +**解决方案**：建议参照上文配置接入凭证，直接在`~/.codex/config.toml`中完成配置，无需依赖第三方工具的健康检查结果；配置完成后参照[验证配置](#cdx-verify)启动 Codex，若能正常进入对话界面即表示可正常使用中国内地模型。 ### **报错 wire\_api 配置问题怎么办？** @@ -394,7 +585,7 @@ codex **原因**： -- 误用了其他方案的 API Key（Token Plan 团队版、Coding Plan 和按量计费的 API Key 互不相通） +- 误用了其他方案的 API Key（Token Plan 个人版、Token Plan 团队版、Coding Plan 和按量计费的 API Key 互不相通） - 订阅过期 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md index f903bf87..56e8a317 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/cursor.md @@ -1,6 +1,6 @@ # Cursor -Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Cursor** @@ -12,6 +12,8 @@ Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Toke 阿里云百炼提供三种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -19,6 +21,20 @@ Cursor 是一款 AI 编程 IDE，可以通过按量计费、Coding Plan 或 Toke - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +**API Key** + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +**Base URL** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +**可用模型** + +Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview) + ### Token Plan 团队版 **API Key** @@ -110,7 +126,9 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 在 Cursor 中无法调用已添加的模型 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md index 247ca4b6..85968ae8 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md @@ -1,6 +1,6 @@ # Hermes Agent -Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Hermes Agent** @@ -31,6 +31,8 @@ Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding 通过 `hermes config set` 命令配置接入参数，根据所选方案填入对应的 Base URL 和 API Key： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -44,6 +46,42 @@ Hermes Agent 是一款终端 AI 编程工具，可以通过按量计费、Coding 除命令行版外，Hermes Agent 还提供桌面版（Hermes Desktop）。可从 [Hermes 官网](https://hermes-agent.nousresearch.com/) 下载安装包，或在命令行版安装完成后运行 `hermes desktop` 启动。桌面版与命令行版共用同一份 `~/.hermes/config.yaml` 配置文件，接入参数与本文一致；在桌面版中以自定义端点（Custom Endpoint）方式接入时，请使用上述 OpenAI 兼容 Base URL。 +### Token Plan 个人版 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +hermes config set model.provider custom +hermes config set model.base_url https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic +hermes config set model.api_mode anthropic_messages +hermes config set model.api_key YOUR_API_KEY +hermes config set model.default qwen3.8-max-preview +``` + +以上命令将配置写入 `~/.hermes/config.yaml`。也可以直接编辑该文件，写入以下内容： + +config.yaml 配置示例 + +``` +model: + default: qwen3.8-max-preview + provider: custom + base_url: https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic + api_mode: anthropic_messages + api_key: YOUR_API_KEY +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Token Plan 团队版 将 `YOUR_API_KEY` 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -53,7 +91,7 @@ hermes config set model.provider custom hermes config set model.base_url https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic hermes config set model.api_mode anthropic_messages hermes config set model.api_key YOUR_API_KEY -hermes config set model.default qwen3.7-max +hermes config set model.default qwen3.8-max-preview ``` 以上命令将配置写入 `~/.hermes/config.yaml`。也可以直接编辑该文件，写入以下内容： @@ -62,13 +100,24 @@ config.yaml 配置示例 ``` model: - default: qwen3.7-max + default: qwen3.8-max-preview provider: custom base_url: https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic api_mode: anthropic_messages api_key: YOUR_API_KEY ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -150,4 +199,6 @@ hermes chat -m qwen3.7-max - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md index aec56898..2bb2f1bd 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md @@ -1,6 +1,6 @@ # Kilo CLI -Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Kilo CLI** @@ -23,6 +23,8 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding 使用文本编辑器打开配置文件 `~/.config/kilo/config.json`，根据所选方案写入对应配置： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -30,6 +32,89 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +需先购买 Token Plan 个人版套餐且套餐处于有效期内。可在[Token Plan 个人版页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)购买套餐。 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型：qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "$schema": "https://kilo.ai/config.json", + "provider": { + "bailian-token-plan-personal": { + "npm": "@ai-sdk/openai-compatible", + "name": "Alibaba Cloud Model Studio", + "options": { + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "apiKey": "YOUR_API_KEY" + }, + "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "reasoning": true, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 262144 + } + } + }, + "qwen3.7-max": { + "name": "Qwen3.7 Max", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.7-plus": { + "name": "Qwen3.7 Plus", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-plus": { + "name": "Qwen3.6 Plus", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-flash": { + "name": "Qwen3.6 Flash", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + } + } + } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Token Plan 团队版 需先购买 Token Plan 团队版套餐且套餐处于有效期内。可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)购买套餐。 @@ -48,6 +133,18 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding "apiKey": "YOUR_API_KEY" }, "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "reasoning": true, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 262144 + } + } + }, "qwen3.7-max": { "name": "Qwen3.7 Max", "options": { @@ -156,6 +253,17 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan-overview#b01f82a4218kx)。 @@ -305,4 +413,6 @@ Kilo CLI 是 Kilo Code 的命令行客户端，可以通过按量计费、Coding - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md index addc7866..c144d889 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md @@ -1,6 +1,6 @@ # Qoder CN（原 Lingma） -Qoder CN（原 Lingma）是阿里云智能编码助手，提供独立 IDE，可以通过 Token Plan、Coding Plan 或按量付费接入阿里云百炼。 +Qoder CN（原 Lingma）是阿里云智能编码助手，提供独立 IDE，可以通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量付费接入阿里云百炼。 **说明** @@ -31,7 +31,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -41,7 +41,9 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -55,7 +57,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 ## 了解更多 -如需进一步了解 Qoder CN 的智能体、MCP、Skills 等扩展能力，请参考 [Qoder CN 官方文档](https://help.aliyun.com/zh/lingma/product-overview/introduction-of-lingma)。 +如需进一步了解 Qoder CN 的智能体、MCP、Skills 等扩展能力，请参考 [Qoder CN 官方文档](https://help.aliyun.com/zh/lingma/introduction-of-lingma)。 ## 常见问题 @@ -65,7 +67,9 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) - 按量计费：[错误码](https://help.aliyun.com/zh/model-studio/error-code) @@ -85,7 +89,7 @@ Qoder CN 个人社区版和个人专业版均支持接入百炼，企业版不 - **提供商或类型与实际套餐不一致**：在 Qoder CN 模型配置中，**提供商**与**类型**需与所购套餐保持一致。例如使用 Token Plan 团队版的 API Key，但**类型**选成了 Coding Plan。 -- **选用了套餐不支持的模型**：仅支持当前套餐覆盖的文本生成模型。例如，Token Plan 团队版的支持模型列表可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)查看。 +- **选用了套餐不支持的模型**：仅支持当前套餐覆盖的文本生成模型。例如，Token Plan 团队版的支持模型列表可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/token-plan/enterprise)查看。 - **临时网络或服务波动**：稍后重试。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md index d3f45eae..1a607972 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md @@ -1,9 +1,31 @@ # 更多工具 -除已列出的工具外，阿里云百炼还支持接入兼容 OpenAI / Anthropic API 协议且支持自定义服务端点的第三方编程工具。可通过按量计费、Coding Plan 或 Token Plan 团队版接入。 +除已列出的工具外，阿里云百炼还支持接入兼容 OpenAI / Anthropic API 协议且支持自定义服务端点的第三方编程工具。可通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入。 ## **配置接入凭证** +### Token Plan 个人版 + +**API 协议** + +**Base URL** + +**API Key** + +**支持模型** + +OpenAI + +`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)（仅文本生成类） + +Anthropic + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + ### Token Plan 团队版 **API 协议** @@ -97,7 +119,7 @@ Trae 支持接入自定义模型，无需安装插件即可直接配置上述任 ## **不支持的工具类型** -Token Plan 团队版和 Coding Plan 仅限在 AI 编程工具和 OpenClaw 类型 Agent 中使用，以下类型的工具**不支持**接入： +Token Plan 个人版、Token Plan 团队版和 Coding Plan 仅限在 AI 编程工具和 OpenClaw 类型 Agent 中使用，以下类型的工具**不支持**接入： - **工作流/自动化平台**：如 Dify、n8n、Coze 等。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md index 6f73f11e..05ae329d 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md @@ -1,6 +1,6 @@ # OpenClaw -OpenClaw 是一个开源的个人 AI 助手平台，支持通过多种消息渠道与 AI 交互。通过配置可接入阿里云百炼平台上的 AI 模型，支持按量付费、Coding Plan、Token Plan 团队版三种接入方式。 +OpenClaw 是一个开源的个人 AI 助手平台，支持通过多种消息渠道与 AI 交互。通过配置可接入阿里云百炼平台上的 AI 模型，支持按量付费、Coding Plan、Token Plan 个人版、Token Plan 团队版四种接入方式。 ## **安装 OpenClaw** @@ -84,6 +84,142 @@ How do you want to hatch your bot? ## **配置接入凭证** +### **Token Plan 个人版** + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，完整列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +**API Key** + +Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) + +**Base URL** + +`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` + +**可用模型** + +Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview) + +配置文件位于 `~/.openclaw/openclaw.json`，OpenClaw 启动时会自动读取。 + +**说明** + +示例禁用了网关鉴权（`auth.mode: none`），仅适合单机本地使用。如需共享或远程访问，请运行 `openclaw doctor --fix` 启用 token 鉴权。 + +**首次配置**：复制以下内容到配置文件，将 `YOUR_API_KEY` 替换为 Token Plan 个人版 API Key。 + +**已有配置**：若需保留已有配置，请勿直接全量替换，详见[已有配置如何安全修改](#cp-openclaw-faq-safe-modify)。 + +``` +{ + "meta": { + "lastTouchedVersion": "2026.2.1", + "lastTouchedAt": "2026-02-03T08:20:00.000Z" + }, + "models": { + "mode": "merge", + "providers": { + "bailian-token-plan": { + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic", + "apiKey": "YOUR_API_KEY", + "api": "anthropic-messages", + "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.7-max", + "name": "qwen3.7-max", + "reasoning": false, + "input": ["text"], + "contextWindow": 1000000, + "maxTokens": 65536, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.7-plus", + "name": "qwen3.7-plus", + "reasoning": false, + "input": ["text", "image"], + "contextWindow": 1000000, + "maxTokens": 65536, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "qwen3.6-flash", + "name": "qwen3.6-flash", + "reasoning": false, + "input": ["text", "image"], + "contextWindow": 1000000, + "maxTokens": 32768, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "glm-5.2", + "name": "glm-5.2", + "reasoning": false, + "input": ["text"], + "contextWindow": 1000000, + "maxTokens": 16384, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, + { + "id": "deepseek-v4-pro", + "name": "deepseek-v4-pro", + "reasoning": false, + "input": ["text"], + "contextWindow": 163840, + "maxTokens": 32768, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 } + } + ] + } + } + }, + "agents": { + "defaults": { + "model": { + "primary": "bailian-token-plan/qwen3.8-max-preview" + }, + "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, + "bailian-token-plan/qwen3.7-max": {}, + "bailian-token-plan/qwen3.7-plus": {}, + "bailian-token-plan/qwen3.6-flash": {}, + "bailian-token-plan/glm-5.2": {}, + "bailian-token-plan/deepseek-v4-pro": {} + } + } + }, + "gateway": { + "mode": "local", + "auth": { "mode": "none" } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### **Token Plan 团队版** 将 `YOUR_API_KEY` 替换为 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -134,6 +270,16 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, { "id": "qwen3.7-max", "name": "qwen3.7-max", @@ -278,9 +424,10 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "agents": { "defaults": { "model": { - "primary": "bailian-token-plan/qwen3.7-plus" + "primary": "bailian-token-plan/qwen3.8-max-preview" }, "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, "bailian-token-plan/qwen3.7-max": {}, "bailian-token-plan/qwen3.7-plus": {}, "bailian-token-plan/qwen3.6-plus": {}, @@ -350,6 +497,16 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "apiKey": "YOUR_API_KEY", "api": "anthropic-messages", "models": [ + { + "id": "qwen3.8-max-preview", + "name": "qwen3.8-max-preview", + "reasoning": true, + "input": ["text", "image"], + "contextWindow": 983616, + "maxTokens": 131072, + "cost": { "input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0 }, + "compat": { "thinkingFormat": "openai" } + }, { "id": "qwen3.7-max", "name": "qwen3.7-max", @@ -494,9 +651,10 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to "agents": { "defaults": { "model": { - "primary": "bailian-token-plan/qwen3.7-plus" + "primary": "bailian-token-plan/qwen3.8-max-preview" }, "models": { + "bailian-token-plan/qwen3.8-max-preview": {}, "bailian-token-plan/qwen3.7-max": {}, "bailian-token-plan/qwen3.7-plus": {}, "bailian-token-plan/qwen3.6-plus": {}, @@ -526,6 +684,17 @@ Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/to 先单击 **Save** 按钮将配置写入磁盘，再单击 **Apply** 按钮重启网关使配置生效。 +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### **Coding Plan** 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)（格式为 `sk-sp-xxxxx`）。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -1215,6 +1384,14 @@ Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-pla 以上配置中 `dmPolicy` 和 `groupPolicy` 均设为 `open`，适用于测试或个人使用场景。生产环境中建议设为 `allowlist`，通过白名单限制可访问的用户和群组，降低安全风险。 +切换到 `allowlist` 模式时，将 `dmPolicy` 和 `groupPolicy` 改为 `"allowlist"`，并添加 `allowFrom` 字段，填入允许访问的工号和群 ID。示例： + +``` +"dmPolicy": "allowlist", +"groupPolicy": "allowlist", +"allowFrom": ["你的工号", "群ID"] +``` + #### 步骤四：测试 1. 执行以下命令重启网关。 @@ -1790,7 +1967,7 @@ Skill 是可扩展的能力模块，Agent 会根据请求自动匹配并加载 ### 接入 MCP 服务 -OpenClaw 支持通过 MCP（Model Context Protocol）插件扩展 Agent 的工具调用能力，例如联网搜索、网页抓取等。具体案例可以参考[添加联网搜索MCP](https://help.aliyun.com/zh/model-studio/web-search-for-coding-plan)。 +OpenClaw 支持通过 MCP（Model Context Protocol）插件扩展 Agent 的工具调用能力，例如联网搜索、网页抓取等。具体案例可以参考[添加联网搜索MCP](https://help.aliyun.com/zh/model-studio/web-search-mcp)。 ## 常见问题 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md index 2f2b6d2a..7ec72a7b 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/opencode.md @@ -1,6 +1,6 @@ # OpenCode -OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 OpenCode** @@ -30,6 +30,8 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla 根据所选方案写入对应配置： +- **Token Plan 个人版**：个人订阅，按 token 消耗抵扣 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -37,6 +39,102 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +需先购买 Token Plan 个人版套餐且套餐处于有效期内。可在[Token Plan 个人版页面](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)购买套餐。 + +将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview)。可用模型包括 qwen3.8-max-preview、qwen3.7-max、qwen3.7-plus、qwen3.6-flash、glm-5.2、deepseek-v4-pro，详细列表请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +``` +{ + "$schema": "https://opencode.ai/config.json", + "provider": { + "bailian-token-plan-personal": { + "npm": "@ai-sdk/anthropic", + "name": "Alibaba Cloud Model Studio", + "options": { + "baseURL": "https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic/v1", + "apiKey": "YOUR_API_KEY" + }, + "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 99072 + }, + "temperature": 0.6, + "reasoning": true + } + }, + "qwen3.7-max": { + "name": "Qwen3.7 Max", + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.7-plus": { + "name": "Qwen3.7 Plus", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-plus": { + "name": "Qwen3.6 Plus", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + }, + "qwen3.6-flash": { + "name": "Qwen3.6 Flash", + "modalities": { + "input": ["text", "image"], + "output": ["text"] + }, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 8192 + } + } + } + } + } + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Token Plan 团队版 需先购买 Token Plan 团队版套餐且套餐处于有效期内。可在[Token Plan 团队版页面](https://bailian.console.aliyun.com/?tab=plan#/efm/subscription/overview)购买套餐。 @@ -55,6 +153,19 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla "apiKey": "YOUR_API_KEY" }, "models": { + "qwen3.8-max-preview": { + "name": "Qwen3.8 Max Preview", + "contextWindow": 983616, + "maxOutputTokens": 131072, + "options": { + "thinking": { + "type": "enabled", + "budgetTokens": 99072 + }, + "temperature": 0.6, + "reasoning": true + } + }, "qwen3.7-max": { "name": "Qwen3.7 Max", "options": { @@ -187,6 +298,17 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan 将 `YOUR_API_KEY` 替换为 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan)。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -291,9 +413,9 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla `baseURL` 按地域设置，API Key 需与所选地域对应： -- 华北2（北京）：`https://dashscope.aliyuncs.com/apps/anthropic/v1` +- 华北2（北京）：`https://dashscope.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/apps/anthropic/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) +- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`，请将`WorkspaceId`替换为真实的[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu) ``` @@ -301,10 +423,10 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla "$schema": "https://opencode.ai/config.json", "provider": { "bailian-payg": { - "npm": "@ai-sdk/anthropic", + "npm": "@ai-sdk/openai-compatible", "name": "Alibaba Cloud Model Studio", "options": { - "baseURL": "https://dashscope.aliyuncs.com/apps/anthropic/v1", + "baseURL": "https://dashscope.aliyuncs.com/compatible-mode/v1", "apiKey": "YOUR_API_KEY" }, "models": { @@ -368,4 +490,6 @@ OpenCode 是一款终端 AI 编程工具，可以通过按量计费、Coding Pla - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md index 1de85109..4540e5fb 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md @@ -1,6 +1,6 @@ # Qoder -Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 JetBrains 插件，可以通过按量付费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 JetBrains 插件，可以通过按量付费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## Qoder IDE @@ -27,7 +27,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -37,7 +37,9 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -99,7 +101,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 1. 在对话框中输入 `/model`，通过 Tab 键切换至 `Custom`。 -2. 回车选择 Add custom model，提供商选择 Alibaba Cloud Model Studio - China，类型根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费**。 +2. 回车选择 Add custom model，提供商选择 Alibaba Cloud Model Studio - China，类型根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费**。 3. 选择模型后输入对应方案的专属 API Key，确认后等待配置生效。 @@ -135,7 +137,7 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 类型 - 根据计费方案选择 **Token Plan**、**Coding Plan** 或 **按量付费** + 根据计费方案选择 **Token Plan**（个人版或团队版）、**Coding Plan** 或 **按量付费** 模型 @@ -145,7 +147,9 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 填写对应方案的专属 API Key： - - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) + - Token Plan 个人版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal) + + - Token Plan 团队版：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise) - Coding Plan：[获取 API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) @@ -195,7 +199,11 @@ Qoder 是面向软件开发的 Agentic 编码平台，支持桌面 IDE、CLI 和 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- 按量计费：[错误码](https://help.aliyun.com/zh/model-studio/error-code) ### 为什么在 Qoder 设置中找不到模型选项？ diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md index 7ca6ebb6..23488022 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md @@ -1,6 +1,6 @@ # Qwen Code -Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan 或 Token Plan 团队版接入阿里云百炼。 +Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Plan、Token Plan 个人版或 Token Plan 团队版接入阿里云百炼。 ## **安装 Qwen Code** @@ -33,8 +33,10 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl ## **配置接入凭证** -启动 Qwen Code 后输入 `/auth` 命令进行可视化配置。阿里云百炼提供三种计费方案，根据需要选择： +启动 Qwen Code 后输入 `/auth` 命令进行可视化配置。阿里云百炼提供四种计费方案，根据需要选择： +- **Token Plan 个人版**：按 token 消耗抵扣个人 Credits。 + - **Token Plan 团队版**：按坐席订阅，按 token 消耗抵扣 Credits。 - **Coding Plan**：固定月费订阅，按模型调用次数计量。 @@ -42,6 +44,98 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl - **按量计费**：按实际调用量后付费。 +### Token Plan 个人版 + +启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Token Plan**，输入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 即可完成配置。可用模型请参考 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + +高级配置：通过 settings.json 配置文件 + +编辑或新建 `settings.json` 文件，将 `YOUR_API_KEY` 替换为 Token Plan 个人版专属 API Key。文件路径如下： + +- macOS/Linux：`~/.qwen/settings.json` + +- Windows：`C:\Users\\.qwen\settings.json` + + +``` +{ + "env": { + "BAILIAN_TOKEN_PLAN_API_KEY": "YOUR_API_KEY" + }, + "modelProviders": { + "openai": [ + { + "id": "qwen3.8-max-preview", + "name": "[Token Plan 个人版] qwen3.8-max-preview", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.7-max", + "name": "[Token Plan 个人版] qwen3.7-max", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.7-plus", + "name": "[Token Plan 个人版] qwen3.7-plus", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "qwen3.6-flash", + "name": "[Token Plan 个人版] qwen3.6-flash", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, + { + "id": "glm-5.2", + "name": "[Token Plan 个人版] glm-5.2", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY" + }, + { + "id": "deepseek-v4-pro", + "name": "[Token Plan 个人版] deepseek-v4-pro", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY" + } + ] + } +} +``` + +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Token Plan 团队版 启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Token Plan**，输入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list) 即可完成配置。可用模型请参考 Token Plan 团队版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-overview)。 @@ -62,6 +156,17 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl }, "modelProviders": { "openai": [ + { + "id": "qwen3.8-max-preview", + "name": "[Token Plan 团队版] qwen3.8-max-preview", + "baseUrl": "https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", + "envKey": "BAILIAN_TOKEN_PLAN_API_KEY", + "generationConfig": { + "extra_body": { + "enable_thinking": true + } + } + }, { "id": "qwen3.7-max", "name": "[Token Plan 团队版] qwen3.7-max", @@ -207,12 +312,23 @@ Qwen Code 是一款终端 AI 编程工具，可以通过按量计费、Coding Pl "region": "china" }, "model": { - "name": "qwen3.7-plus" + "name": "qwen3.8-max-preview" }, "$version": 3 } ``` +**重要** + +**qwen3.8-max-preview 思考模式说明**： + +- thinking：始终开启，不支持关闭。 + +- temperature：思考模式下默认值为 0.6；传入值小于 0.6 时自动调整为 0.6。 + +- reasoning\_effort：控制推理深度，可选 xhigh、high、low，默认 xhigh。 + + ### Coding Plan 启动 Qwen Code 后输入 `/auth`，依次选择 **订阅计划** > **阿里云百炼 Coding Plan**，选择 Coding Plan 区域（china），输入 Coding Plan 专属 [API Key](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/coding_plan) 即可完成配置。可用模型请参考 Coding Plan [支持的模型](https://help.aliyun.com/zh/model-studio/coding-plan)。 @@ -547,7 +663,7 @@ Qwen Code 支持在 VS Code 中以插件方式使用，在 IDE 中提供 AI 编 4. 输入以下内容安装 skill。 ``` - 查看我是否有find skills，没有就直接帮我安装：npx skills add https://github.com/vercel-labs/skills --skill find-skills -y -a qwen-code，然后帮我安装 web-component-design 到当前目录qwen code skills中。 + 查看我是否有find skills，没有就直接帮我安装：npx skills add https://github.com/vercel-labs/skills --skill find-skills -y -a qwen-code，然后从 wshobson/agents 帮我安装 web-component-design 到当前目录：npx skills add https://github.com/wshobson/agents --skill web-component-design -y ``` 5. 下载[website.png](https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20260318/ymehla/website.png)到项目目录，输入以下内容，将自动识别截图的布局、样式，生成网页代码。 @@ -593,7 +709,9 @@ Qwen Code 支持在 VS Code 中以插件方式使用，在 IDE 中提供 AI 编 - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### **如何切换模型？** diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md index 00dddafd..53491eff 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md @@ -1,6 +1,6 @@ # QwenPaw -QwenPaw（原 CoPaw）是 AgentScope 团队开源的个人 AI 助手，支持本地或云端部署，可通过 Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 +QwenPaw（原 CoPaw）是 AgentScope 团队开源的个人 AI 助手，支持本地或云端部署，可通过 Token Plan 个人版、Token Plan 团队版、Coding Plan 或按量计费接入阿里云百炼。 ## **安装 QwenPaw** @@ -52,6 +52,22 @@ qwenpaw app 在 Console 点击 **设置** > **模型**，根据计费方案配置对应的提供商。 +### Token Plan 个人版 + +进入内置的 **Aliyun Token Plan** 提供商**设置**页面，填入 API Key。 + +**配置项** + +**说明** + +**API 密钥** + +填入 Token Plan 个人版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/personal)。 + +**模型** + +已预设常用模型。新增模型点击**添加模型**，**模型 ID**填入 Token Plan 个人版[支持的模型](https://help.aliyun.com/zh/model-studio/token-plan-personal-overview)。 + ### Token Plan 团队版 进入内置的 **Aliyun Token Plan** 提供商**设置**页面，填入 API Key。 @@ -62,7 +78,7 @@ qwenpaw app **API 密钥** -填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/uac-admin/organization/members/list)。 +填入 Token Plan 团队版专属 [API Key](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/token-plan/enterprise)。 **模型** @@ -125,7 +141,9 @@ qwenpaw app - Coding Plan：[Coding Plan 常见问题](https://help.aliyun.com/zh/model-studio/coding-plan-faq) -- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-faq) +- Token Plan 个人版：[Token Plan 个人版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) + +- Token Plan 团队版：[Token Plan 团队版常见问题](https://help.aliyun.com/zh/model-studio/token-plan-team-faq) ### 报错 401 Incorrect API key provided diff --git a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md index 74caf7e8..bb974955 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md @@ -1,88 +1,64 @@ # 3d generation -百炼平台基于 Tripo 模型提供 3D 资产生成能力，支持文生 3D、单图生 3D 与多图生 3D 三种输入方式，产出带贴图的 PBR 材质 GLB 模型或无贴图基础模型。由于生成耗时较长，API 采用[异步调用](../concepts/async-invocation.md)，整体流程为「创建任务 → 轮询获取结果」。详细接口与参数见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 适用范围 - -- 仅适用于**华北2（北京）**地域，且必须使用该地域的 [API Key](../concepts/api-key.md)。 -- 需先在百炼控制台模型市场搜索「Tripo」并开通服务、完成授权，再配置好 [API Key](../concepts/api-key.md) 环境变量。具体开通与配置步骤见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 调用流程 - -API 仅支持[异步调用](../concepts/async-invocation.md)，包含两个步骤： - -1. **创建任务**：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` -2. **轮询查询结果**：`GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}` - -创建任务时必须携带 `X-DashScope-Async: enable` 请求头，否则会报错 `current user api does not support synchronous calls`。成功创建后返回 `task_id`，有效期 24 小时，**请勿重复创建任务**，轮询获取即可。 - -轮询建议间隔约 15 秒，任务状态流转为 `PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED` / `FAILED`。查询接口默认 RPS 为 20，如需更高频查询或事件通知建议配置异步任务回调。 - -## 支持的模型 - -| 模型名 | 定位 | 输出面数 | 对应官方 API 版本 | -| --- | --- | --- | --- | -| `Tripo/Tripo-H3.1` | 高精度生成 | 最高 200 万面 | `v3.1-20260211` | -| `Tripo/Tripo-P1.0` | 专业生成，速度更快 | 最高 2 万面 | `P1-20260311` | - -## 输入方式 - -`input` 中 `prompt`、`image`、`images` 三者**互斥**，只能选其一，同时传多个将报错。 - -- **文生 3D**：`prompt` 必填，支持中英文等多语言，每个字符计 1 个字符，最大 1024 字符。 -- **单图生 3D**：`image` 必填，传入单张图像公网 URL。图像格式限 JPEG/PNG，宽高范围 [20, 6000] 像素（建议边长大于 256），文件不超过 20MB，支持 HTTP/HTTPS。 -- **多图生 3D**：`images` 必填，数组长度固定为 4，对应视角顺序为**前、左、后、右**；不需要的视角传空对象 `{}`。实际有效图片数为 2~4 张，多张图像的分辨率和宽高比不要求一致。每个对象含 `type`（`jpeg` 或 `png`）与 `file_token`（公网 URL）字段。 - -## 关键参数（parameters） - -| 参数 | 适用模型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `texture_quality` | 全部 | `standard` | 贴图质量，可选 `standard`（标清）/`detailed`（高清） | -| `geometry_quality` | `Tripo/Tripo-H3.1` | `standard` | 几何精度，`standard` 最高 150 万面，`ultra` 最高 200 万面 | -| `pbr` | 全部 | `true` | 是否生成 PBR 材质模型。设为 `true` 时强制启用贴图，返回 `pbr_model_url` | -| `texture` | 全部 | `true` | 是否生成贴图。生成无贴图模型需**同时**将 `texture` 和 `pbr` 设为 `false`，返回 `base_model_url` | - -## 响应与产物 - -成功响应的 `output.results` 仅在 `task_status` 为 `SUCCEEDED` 时返回，包含以下字段： - -- `pbr_model_url`：PBR 材质模型（GLB）下载 URL，当 `pbr` 为 `true`（默认）时返回。 -- `base_model_url`：无贴图基础模型（GLB）下载 URL，当 `texture` 与 `pbr` 均为 `false` 时返回。 -- `rendered_image_url`：3D 模型预览渲染图（1 张）URL。 - -> **注意**：以上下载链接有效期均为 **2 小时**，请及时下载。 - -`usage` 字段记录任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）、生成数量 `count`、贴图质量与几何精度，仅对成功结果计数。任务状态 `task_status` 的完整枚举为 `PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `CANCELED` / `UNKNOWN`，其中 `UNKNOWN` 表示任务不存在或超过 24 小时有效期。更多响应字段说明见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 - -## 限制与注意事项 - -- 仅限北京地域 [API Key](../concepts/api-key.md) 调用，地域不匹配将无法使用。 -- `task_id` 查询有效期 24 小时，超时返回 `UNKNOWN` 且无法再查询。 -- 查询接口默认 RPS 限制为 20，建议通过异步任务回调获取更高频通知。 -- 产物下载链接有效期仅 2 小时。 -- 调用失败时响应中会返回 `code` 与 `message`，可参照百炼错误码文档排查。 +百炼平台的 3D 生成能力基于 Tripo 模型提供文生3D、单图生3D 和多图生3D 三种模式，支持带贴图/PBR 材质或无贴图的基础模型输出。该能力为异步任务，需通过 `task_id` 轮询获取结果，适用于华北2（北京）地域。详细实现细节请参考 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 + +## 支持的模型/功能 + +- **支持模型**： + - `Tripo/Tripo-P1.0`：专业版，最高输出 2 万面，速度快，适用于快速原型与轻量级应用。 + - `Tripo/Tripo-H3.1`：高精度版，最高输出 200 万面，支持 `geometry_quality: "ultra"`，适用于对几何精度要求高的场景。 +- **输入方式（互斥）**： + - 文生3D：通过 `input.prompt` 输入文本描述（最大 1024 字符）。 + - 单图生3D：通过 `input.image` 提供单张 JPEG/PNG 图像（分辨率 20–6000px，≤20MB）。 + - 多图生3D：通过 `input.images` 提供长度为 4 的数组，按「前、左、后、右」顺序排列；缺失视角用 `{}` 占位，实际有效图数需 ≥2。 +- **输出类型**： + - 默认返回 PBR 材质模型（`pbr_model_url`，GLB 格式）及预览图（`rendered_image_url`）。 + - 无贴图模型需显式设置 `"texture": false, "pbr": false`，此时返回 `base_model_url`。 +- 全部功能均在 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中定义并验证。 + +## 关键参数 + +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `model` | string | ✓ | 固定为 `Tripo/Tripo-P1.0` 或 `Tripo/Tripo-H3.1` | +| `input.prompt` / `input.image` / `input.images` | string / object / array | ✓（三选一） | 仅允许一种输入方式，同时传入将报错 | +| `parameters.texture_quality` | string | ✗ | `"standard"`（默认）或 `"detailed"`；仅对带贴图任务生效 | +| `parameters.geometry_quality` | string | ✗ | 仅 `Tripo/Tripo-H3.1` 支持；`"standard"`（≤150万面）或 `"ultra"`（≤200万面） | +| `parameters.pbr` | boolean | ✗ | 默认 `true`；设为 `false` 时需同步设 `texture: false` 才能获得无贴图模型 | +| `parameters.texture` | boolean | ✗ | 默认 `true`；与 `pbr` 联动，详见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) | + +> **注意**：`pbr` 和 `texture` 的组合逻辑存在隐式依赖——当 `pbr=true` 时，系统强制启用贴图（即忽略 `texture=false`）。因此，**唯一生成无贴图模型的方式是同时设置 `"texture": false, "pbr": false`**。 + +## 使用方式 + +1. **开通与配置** + - 在[百炼控制台（华北2）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)搜索并开通 Tripo 模型。 + - 配置环境变量 `DASHSCOPE_API_KEY`，确保使用北京地域的 API Key（参见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)）。 + +2. **创建异步任务** + - `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + - 请求头必须包含： + - `Content-Type: application/json` + - `Authorization: Bearer $DASHSCOPE_API_KEY` + - `X-DashScope-Async: enable`（**缺失将报错**） + - 响应中提取 `task_id`（有效期 24 小时）。 + +3. **轮询查询结果** + - `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + - 建议间隔 ≥15 秒轮询，状态流转为 `PENDING → RUNNING → SUCCEEDED/FAILED`。 + - 成功响应中 `output.results[0]` 包含 `pbr_model_url` 或 `base_model_url`（链接有效期 2 小时，需及时下载）。 + +## 限制和注意事项 + +- **地域限制**：仅支持华北2（北京）地域，其他地域 URL 不可用。 +- **异步强制性**：所有调用必须启用 `X-DashScope-Async: enable`，不支持同步模式。 +- **task_id 生命周期**：创建后 24 小时内有效，超时查询返回 `task_status: "UNKNOWN"`。 +- **RPS 限制**：任务查询接口默认限流 20 QPS；高频轮询建议配置[异步回调](https://help.aliyun.com/zh/model-studio/async-task-api)。 +- **图像约束**：单图/多图输入均要求公网可访问 URL（HTTP/HTTPS），格式为 JPEG/PNG，单图 ≤20MB，多图各图独立校验。 +- **错误处理**：失败任务返回 `code` 和 `message`，需结合[统一错误码文档](https://help.aliyun.com/zh/model-studio/error-code)定位原因。 ## 来源文档 - [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md index eb343a80..375aaaf5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md @@ -1,151 +1,69 @@ # application call -阿里云百炼平台提供两套 API 来调用智能体和工作流应用：**OpenAI 兼容的 Responses API** 和 **DashScope API**。两者均支持同步/[异步调用](../concepts/async-invocation.md)、多轮对话、[流式输出](../concepts/streaming.md)等核心能力，开发者可根据生态兼容性和功能需求选择合适的接入方式。调用前需先获取 APP ID（以及子[业务空间](../concepts/workspace.md)场景下的 Workspace ID）和 [API Key](../concepts/api-key.md)。 - -## 前置准备 - -### 获取凭证 - -通过 API 调用应用时，必须提供 **APP ID** 来指定目标应用。如果应用位于子[业务空间](../concepts/workspace.md)，还需提供 **Workspace ID**。详细获取方式参见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 - -- **APP ID**：在控制台「应用管理」页面的应用卡片上复制。 -- **Workspace ID**：在调用子[业务空间](../concepts/workspace.md)下的应用或特定地域（德国、华北2、新加坡、日本）的模型时必须提供，可通过控制台右上角图标查看。 - -> **注意**：目前只能通过控制台手动获取 APP ID 和 Workspace ID，不支持通过 API 或 CLI 查询。 - -### 其他前提 - -- 已获取 [API Key](../concepts/api-key.md) 并配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已创建并发布百炼应用（智能体或工作流）。 -- 如使用 SDK 调用，需安装对应的 SDK（OpenAI SDK 或 [DashScope SDK](../concepts/dashscope-sdk.md)）。 - -## 两套 API 对比 - -| 维度 | Responses API（OpenAI 兼容） | DashScope API | -|------|---------------------------|---------------| -| Endpoint | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK | OpenAI Python/Java SDK | DashScope Python/Java SDK | -| 多轮对话 | 通过 `input` 数组传递完整历史消息 | 通过 `session_id` 或 `messages` 维护上下文 | -| [异步调用](../concepts/async-invocation.md) | 设置 `background=true` | 暂不支持（仅 Responses API 提供） | -| 适用地域 | 仅华北2（北京） | 仅华北2（北京） | - -## Responses API（OpenAI 兼容模式） - -### 同步调用 - -适用于需要即时获取结果的实时交互场景。完整参数说明参见 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 - -**核心请求参数：** - -| 参数 | 类型 | 必选 | 说明 | -|------|------|------|------| -| `input` | string / array | 是 | 请求输入，可为简单字符串或包含多轮对话历史的消息数组 | -| `stream` | boolean | 否 | 是否[流式输出](../concepts/streaming.md)，默认 `false` | -| `background` | boolean | 否 | 是否异步执行，默认 `false` | - -**Python 示例（单轮对话）：** - -```python -from openai import OpenAI -import os - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url=f'https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/' -) - -response = client.responses.create(input="你是谁？") -print(response.model_dump_json(indent=2)) -``` - -**多轮对话**需在 `input` 中传递完整的消息历史数组，每条消息包含 `role`（user/assistant/system）和 `content` 字段。 - -### [多模态](../concepts/multimodal.md)输入 - -Responses API 支持在 `content` 数组中混合多种输入类型： - -- **图像输入**：通过 `input_image` 类型传入图片 URL。[智能体应用](../concepts/agent-application.md)需选用通义千问 VL 系列模型并将文件处理方式设为「自定义处理」。 -- **文件输入**：通过 `input_file` 类型传入文件 URL。仅[智能体应用](../concepts/agent-application.md)支持，需配置「全文引用」或「切片检索」处理方式。 - -### [异步调用](../concepts/async-invocation.md) - -对于耗时较长的任务（如生成报告、多步骤工具调用），可设置 `background=true` 开启异步模式，避免请求超时。详细流程参见 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 - -核心流程： - -1. **创建任务**：请求中设置 `background=true`，API 立即返回任务 ID。 -2. **轮询状态**：通过 `client.responses.retrieve(task_id)` 定期查询任务状态。 -3. **处理结果**：当状态变为 `completed`、`failed` 或 `cancelled` 时获取最终结果。 - -> **注意**：异步任务暂不支持[流式输出](../concepts/streaming.md)（`stream=true`）。 - -### [流式输出](../concepts/streaming.md) - -设置 `stream=true` 可边生成边输出，适用于需要实时展示生成内容的场景。若应用类型为工作流，需在结束节点或流程输出节点中启用「[流式输出](../concepts/streaming.md)」开关并重新发布。 - -## DashScope API - -DashScope API 提供更全面的功能支持，适合需要深度集成百炼平台能力的场景。支持 Python、Java、PHP、Node.js、C#、Go 等多种语言的 HTTP 调用，以及 Python/Java SDK。详细参数参见 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 - -**Python 示例：** - -```python -from dashscope import Application -import os - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -**多轮对话**通过 `session_id` 维护上下文：首次请求无需传入，响应中会返回 `session_id`；后续请求携带该值即可延续对话。`session_id` 在最后一次请求后 1 小时内有效。 - -新版[智能体应用](../concepts/agent-application.md)（Agent 2.0）的调用方式与上述基本一致，参见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 - -## 参数传递 - -### 自定义参数 - -工作流应用中定义的自定义参数，通过请求体中的 `biz_params` 传递，参数名和类型需与应用内配置保持一致。 - -```python -response = await client.responses.create( - input="你好", - extra_body={"biz_params": {"city": "北京"}}, - background=True -) -``` - -### 插件参数 - -智能体或工作流中配置的插件工具参数，同样通过 `biz_params` 传递。工作流需在开始节点创建自定义参数并将其传入插件节点的输入参数中。 - -## 限制与注意事项 - -- 两套 API 目前均**仅适用于华北2（北京）地域**。 -- Responses API 的多轮对话暂不支持基于 `pre_response_id` 或 `conversation_id` 的上下文功能，需每次传递完整对话历史。 -- [异步调用](../concepts/async-invocation.md)仅 Responses API 支持，且不能与[流式输出](../concepts/streaming.md)同时使用。 -- RAM 子账号查看[业务空间](../concepts/workspace.md)管理页面需要超级管理员权限（`AliyunBailianFullAccess` 或 `AliyunBailianControlFullAccess`）。 -- [API Key](../concepts/api-key.md) 不建议硬编码到代码中，应通过环境变量配置以降低泄露风险。 +`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可选择 OpenAI 兼容的 Responses API 或原生 DashScope API 两种方式发起同步或异步请求，支持文本、图像、文件等[多模态](../concepts/multi-modal.md)输入，并可通过 `session_id` 或完整消息历史维护对话上下文。所有调用均需提供有效的 APP ID 及认证凭证。 + +## 支持的模型/功能 + +- **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流应用；其中文件输入仅限智能体应用，且需在应用配置中启用“全文引用”或“切片检索”[新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 +- **[多模态](../concepts/multi-modal.md)能力**： + - 图像输入：需选用通义千问 VL 系列模型，并在应用中正确配置图像处理方式 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； + - 文件输入：仅智能体应用支持，依赖应用内文件处理方式配置； + - [流式输出](../concepts/streaming-output.md)：仅同步调用支持，且工作流应用需在结束节点启用“[流式输出](../concepts/streaming-output.md)”开关并重新发布 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 +- **会话管理**： + - DashScope API 通过 `session_id` 维护上下文，有效期为最后一次请求后 1 小时； + - Responses API 当前不支持 `pre_response_id` 或 `conversation_id`，需在每次请求中传递完整消息历史 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 + +> **注意**：文档 4 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 4 明确限定为“新版智能体应用”，而文档 5 泛指“智能体、工作流应用”。实际调用时，该接口对两类应用均有效，但功能支持（如 `session_id` 行为、参数结构）以应用类型和发布配置为准，建议优先参考 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 + +## 关键参数 + +| 参数名 | 类型 | 必选 | 说明 | +|--------|------|------|------| +| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/#/app-center)页面获取。若应用位于子业务空间，还需传入 `workspace_id` [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 | +| `input` / `prompt` | string 或 array | 是 | 核心输入内容：
- DashScope API 使用 `prompt` 字符串；
- Responses API 支持字符串（单轮）或消息数组（多轮/[多模态](../concepts/multi-modal.md)），数组元素含 `role`（`user`/`system`/`assistant`）与 `content`（支持 `input_text`/`input_image`/`input_file`）[同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 | +| `stream` | boolean | 否 | 仅 Responses API 支持。设为 `true` 启用[流式输出](../concepts/streaming-output.md)，需应用端配合启用流式开关 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 | +| `background` | boolean | 否 | 仅 Responses API 支持。设为 `true` 进入异步模式，立即返回任务 ID，后续通过 `retrieve` 查询结果 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | +| `biz_params` | object | 否 | Responses API 中用于传递工作流或智能体应用内定义的自定义参数，键名需与应用配置完全一致 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | +| `session_id` | string | 否 | DashScope API 多轮对话必需。首次调用不传，响应中返回；后续调用需携带上一轮返回的 `session_id` [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 | + +## 使用方式 + +### 1. 认证与端点 +- **API Key**：通过[密钥管理](https://bailian.console.aliyun.com/?tab=app#/api-key)获取，并推荐配置为环境变量 `DASHSCOPE_API_KEY`。 +- **Base URL / Endpoint**： + - Responses API（OpenAI 兼容）：`https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/`（同步/异步共用）； + - DashScope API（原生）：`https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`。 + +### 2. 调用示例 +- **同步调用（Responses API）**： + ```python + from openai import OpenAI + client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url=f"https://dashscope.aliyuncs.com/api/v2/apps/agent/{app_id}/compatible-mode/v1/") + response = client.responses.create(input="你好") + ``` +- **异步调用（Responses API）**： + 设置 `background=True`，获取 `task_id` 后轮询 `retrieve` 接口 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 +- **DashScope SDK 调用**： + ```python + from dashscope import Application + response = Application.call(api_key=..., app_id=..., prompt="你好") + ``` + +## 限制和注意事项 + +- **地域限制**：Responses API（同步/异步）与 DashScope API 均**仅支持华北2（北京）地域**，其他地域（如德国法兰克福、新加坡）调用需显式传入 `workspace_id` 并确认 Base URL [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **异步限制**：异步调用不支持 `stream=true`，且暂无流式输出能力 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 +- **凭证获取**：APP ID 和 Workspace ID **仅支持控制台手动获取**，不提供 API 或 CLI 查询接口 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **权限要求**：查询所有业务空间 ID 需主账号或具备 `AliyunBailianFullAccess` 权限的 RAM 子账号，普通子账号仅能查看已加入的业务空间 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 +- **参数兼容性**：`biz_params` 仅在 Responses API 中生效；DashScope API 的自定义参数需通过 `parameters` 字段传递（文档未明确示例，以 SDK 实际行为为准）。 ## 来源文档 - [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) -- [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) +- [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md index 65de9baf..76e8a143 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md @@ -1,188 +1,73 @@ # application component api reference -百炼平台应用组件 API（`bailian/2023-12-29`）提供了数据连接、知识库、Prompt 模板、长期记忆等核心能力的 OpenAPI 接口，采用 ROA 签名风格。开发者可通过阿里云百炼 SDK 直接调用，也可使用自签名方式对接。所有接口均需传入 `WorkspaceId`（[业务空间](../concepts/workspace.md) ID），RAM 子账号需要先获取对应权限策略并加入[业务空间](../concepts/workspace.md)后才能调用。 - -## 服务接入点与鉴权 - -当前支持两个地域的接入点： - -| 地域 | 地域 ID | 公网接入地址 | VPC 接入地址 | -|------|---------|-------------|-------------| -| 华北2（北京） | cn-beijing | bailian.cn-beijing.aliyuncs.com | bailian-vpc.cn-beijing.aliyuncs.com | -| 新加坡 | ap-southeast-1 | bailian.ap-southeast-1.aliyuncs.com | bailian-vpc.ap-southeast-1.aliyuncs.com | - -调用前需准备 AccessKey，建议使用 RAM 用户而非主账号以降低安全风险。RAM 权限策略的 RamCode 为 `sfm`，授权粒度为操作级。大多数写操作需要 `AliyunBailianDataFullAccess` 策略，部分只读接口（如 DescribeFile、GetIndexJobStatus）也支持 `AliyunBailianDataReadOnlyAccess`。详见[授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)。 - -## 数据连接（原应用数据） - -数据连接相关 API 用于管理类目、文件、解析设置、表格和连接器，是构建知识库的数据基础。 - -### 类目管理 - -| API | 说明 | 限流 | 幂等性 | -|-----|------|------|--------| -| AddCategory | 在[业务空间](../concepts/workspace.md)中新建类目，每空间最多 500 个 | 5 次/秒 | 否 | -| ListCategory | 查询类目列表，支持分页 | 5 次/秒 | 是 | -| DeleteCategory | 永久删除指定类目 | 5 次/秒 | 是 | - -> **注意**：当前不支持通过 API 查询或新增数据表，数据表操作请通过控制台完成。 - -### 文件管理 - -文件上传采用两步流程：先调用 ApplyFileUploadLease 获取上传租约，使用返回的 URL 上传文件后，再调用 AddFile 将文件导入百炼。也可通过 AddFilesFromAuthorizedOss 直接从已授权的 OSS Bucket 导入。详见[ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| ApplyFileUploadLease | 申请上传租约（知识库文件或会话交互文件） | 10 次/秒 | -| AddFile | 将临时存储文件导入数据连接 | 10 次/秒 | -| AddFilesFromAuthorizedOss | 从已授权 OSS Bucket 批量导入文件 | 5 次/秒 | -| DescribeFile | 查询文件基本信息（名称、类型、状态等） | 10 次/秒 | -| ListFile | 分页查询指定类目下的文件列表 | 5 次/秒 | -| UpdateFileTag | 更新单个文件的标签 | 5 次/秒 | -| BatchUpdateFileTag | 批量更新文件标签 | 5 次/秒 | -| DeleteFile | 删除单个文件 | 5 次/秒 | -| DeleteFiles | 批量删除文件 | 5 次/秒 | - -AddFile 接口的 `Parser` 参数支持以下解析器类型： -- `DOCMIND`（智能文档解析） -- `DOCMIND_DIGITAL`（电子文档解析） -- `DOCMIND_LLM_VERSION`（大模型文档解析） -- `DASH_QWEN_VL_PARSER`（Qwen VL 解析） -- `DOCMIND_LLM_VERSION_MEDIA`（音视频解析） -- `AUTO_SELECT`（自动选择解析器） - -### 解析设置 - -| API | 说明 | -|-----|------| -| GetParseSettings | 获取类目的解析设置 | -| GetAvailableParserTypes | 获取指定文件支持的解析器类型列表 | -| ChangeParseSetting | 修改类目的解析设置 | - -### 表格与连接器 - -| API | 说明 | -|-----|------| -| AddTable | 添加表格 | -| UpdateTableFromAuthorizedOss | 从已授权 OSS Bucket 更新表格 | -| AddConnector | 新增连接器 | -| GetConnector | 获取连接器信息（当前仅支持文件连接器） | -| UpdateConnector | 编辑连接器名称和描述 | - -连接器的 `StorageType` 支持 `OSS_CUSTOM`（自有 OSS 存储）和 `OSS_PLATFORM`（平台 OSS 存储）。 - -## Prompt 工程 - -Prompt 模板 API 支持对 Prompt 模板的完整 CRUD 操作。模板内容支持变量占位符（如 `${theme}`），系统会自动提取变量列表。详见[CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md)。 - -| API | 方法 | 说明 | -|-----|------|------| -| CreatePromptTemplate | POST | 创建模板（暂不支持文生图模板） | -| GetPromptTemplate | GET | 按模板 ID 获取详情 | -| UpdatePromptTemplate | PATCH | 增量更新模板名称或内容 | -| DeletePromptTemplate | DELETE | 按模板 ID 删除 | -| ListPromptTemplates | GET | 分页查询模板列表，支持按名称和类型（System/Custom）过滤 | - -## 知识库 - -知识库 API 是百炼 RAG 能力的核心，覆盖知识库的创建、数据导入、检索、文件与切片管理全流程。 - -### 知识库生命周期 - -创建知识库的典型流程为：CreateIndex -> SubmitIndexJob -> 轮询 GetIndexJobStatus 直到完成。详见[CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md)。 - -| API | 说明 | 限流 | -|-----|------|------| -| CreateIndex | 创建知识库（非结构化或结构化），不具幂等性 | 10 次/秒 | -| SubmitIndexJob | 提交知识库创建任务，必须在 CreateIndex 后调用 | 10 次/秒 | -| SubmitIndexAddDocumentsJob | 向已有知识库追加文件（不支持数据查询/图片问答类） | 10 次/秒 | -| GetIndexJobStatus | 查询任务状态，调用间隔建议 5 秒以上 | - | -| UpdateIndex | 更新知识库配置（名称、描述、检索参数等） | - | -| ListIndices | 分页查询[业务空间](../concepts/workspace.md)下的知识库列表 | 10 次/秒 | -| DeleteIndex | 永久删除知识库（不可逆，不删除源文件） | 10 次/秒 | -| GetIndexMonitor | 获取知识库监控数据 | - | - -> **注意**：CreateIndex 仅初始化知识库，必须后续调用 SubmitIndexJob 才能完成创建，否则将得到空知识库。CreateIndex 不具幂等性，重复调用会创建多个同名知识库。 - -UpdateIndex 支持调整检索参数： -- `DenseSimilarityTopK`：向量检索 Top K，范围 [0-100]，默认 100 -- `SparseSimilarityTopK`：关键词检索 Top K，范围 [0-100]，默认 100 -- 两者之和不超过 200 -- `RerankMinScore`：排序最低分数，范围 [0-1] -- `PipelineCommercialType`：知识库规格（standard / enterprise） - -### 知识库检索 - -Retrieve 接口用于在指定知识库中检索信息，支持通过百炼 SDK（AccessKey 鉴权）或 Spring AI Alibaba（API-Key 鉴权）调用。接口具有幂等性，但因包含复杂检索逻辑，响应时间可能较长，建议合理设置超时和重试策略。详见[Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md)。 - -### 文件与切片管理 - -| API | 说明 | -|-----|------| -| ListIndexFileDetails | 查询知识库中文件的详细信息，支持按状态和名称过滤 | -| ListIndexDocuments | 查询知识库中文件的概要信息 | -| DeleteIndexDocument | 从知识库中删除指定文件 | -| ListChunks | 查询文件的切片列表（文档搜索类查指定文件，数据查询类查全部） | -| UpdateChunk | 修改切片内容和标题（仅支持文档搜索类知识库） | -| DeleteChunk | 删除指定切片 | - -文件导入状态包括：`RUNNING`（构建中）、`FINISH`（成功）、`INSERT_ERROR`（导入失败）、`PARSE_FAILED`（解析失败）、`DOC_PARSING`（解析中）、`DELETED`（已删除）。 - -## 长期记忆 - -长期记忆 API 用于管理智能体的记忆能力，包括记忆体（Memory）和记忆片段（MemoryNode）两个层级。 - -| API | 说明 | -|-----|------| -| CreateMemory | 创建长期记忆体 | -| GetMemory | 获取记忆体详情 | -| UpdateMemory | 更新记忆体 | -| DeleteMemory | 删除记忆体 | -| ListMemories | 查询记忆体列表 | -| CreateMemoryNode | 创建记忆片段 | -| GetMemoryNode | 获取记忆片段详情 | -| UpdateMemoryNode | 更新记忆片段 | -| DeleteMemoryNode | 删除记忆片段 | -| ListMemoryNodes | 查询记忆片段列表 | - -## 其他 - -| API | 说明 | -|-----|------| -| ApplyTempStorageLease | 申请临时文件上传许可 | -| GetAlipayTransferStatus | 查询支付宝打赏状态 | -| GetAlipayUrl | 获取支付宝打赏 URL | - -## 通用注意事项 - -- 所有接口均需 `WorkspaceId` 路径参数，获取方式参见[业务空间](../concepts/workspace.md)文档 -- 建议使用官方 SDK 调用而非自签名，自签名对接复杂度高（约需 5 个工作日） -- 分页查询使用 `NextToken` / `MaxResults` 模式（部分接口使用 `PageNumber` / `PageSize`） -- 各接口限流频率为 5-15 次/秒不等，触发限流后需等待后重试 -- 版本变更历史可查看[版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md)，近期变更包括 CreateIndex 入参调整、UpdateIndex 新增、GetIndexMonitor 新增等 +本 API 参考文档面向开发者，系统性地描述了百炼平台 Application Component（应用组件）提供的核心 OpenAPI 能力，涵盖数据连接（应用数据）、知识库、[Prompt 工程](../concepts/prompt-engineering.md)及辅助功能等模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，支持多语言 SDK 封装调用。开发者需通过 RAM 子账号配合最小权限策略进行安全接入。 + +## 支持的模型/功能 + +Application Component 提供以下四类核心能力： + +- **数据连接（原应用数据）**：管理非结构化文件与结构化表格。支持类目（Category）增删查、文件上传（`ApplyFileUploadLease` + `AddFile`）、OSS 批量导入（`AddFilesFromAuthorizedOss`）、解析器配置（`GetAvailableParserTypes`, `ChangeParseSetting`）及连接器管理（`AddConnector`, `GetConnector`）。注意：API 不支持直接操作数据表（如新增/删除表），该功能仅限控制台，详见 [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) 文档说明。 +- **知识库（Knowledge Base）**：支持创建（`CreateIndex`）、更新（`UpdateIndex`）、删除（`DeleteIndex`）知识库；向知识库追加文档（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）、查询文件列表（`ListIndexDocuments`）及删除知识库内文件（`DeleteIndexDocument`）。知识库类型包括文档/音视频（非结构化）和数据查询/图片问答（结构化）两类。 +- **[Prompt 工程](../concepts/prompt-engineering.md)**：提供完整的 Prompt 模板生命周期管理，包括创建（`CreatePromptTemplate`）、获取（`GetPromptTemplate`）、更新（`UpdatePromptTemplate`）、删除（`DeletePromptTemplate`）及列表查询（`ListPromptTemplates`）。 +- **辅助功能**：包含支付宝打赏状态查询（`GetAlipayTransferStatus`）、临时存储租约申请（`ApplyTempStorageLease`）等场景化能力。 + +> **注意**：文档 33 (`AddChunk`) 明确指出“目前尚不支持对音视频搜索类（multimedia）知识库进行相关操作”，而文档 34 (`CreateIndex`) 则将知识库类型划分为“基于文档或音视频的非结构化知识库”与“用于数据查询或图片问答的结构化知识库”。二者存在表述矛盾——实际能力以 `CreateIndex` 的分类为准，`AddChunk` 接口当前仅适用于 `document`、`table` 和 `image` 类型知识库，不支持 `multimedia` 类型。 + +## 关键参数 + +- **通用路径参数**：几乎所有接口均需 `WorkspaceId`（业务空间 ID），用于标识资源归属。其值可通过控制台或 `ListWorkspace` 接口获取。 +- **身份认证**：所有请求必须携带有效的 AccessKey（建议使用 RAM 子账号并遵循最小权限原则），签名方式为 ROA。详细准备流程见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 +- **分页与幂等**： + - 分页接口（如 `ListCategory`, `ListFile`, `ListPromptTemplates`）统一使用 `MaxResults`（每页最大条数）和 `NextToken`（下一页凭证）实现。 + - 幂等性接口（如 `ListCategory`, `DescribeFile`, `Retrieve`）可安全重试；非幂等接口（如 `AddCategory`, `ApplyFileUploadLease`）重复调用可能导致重复资源创建或失败。 +- **关键业务参数**： + - 文件操作：`CategoryId`（类目 ID）、`FileId`（文件 ID）、`LeaseId`（上传租约 ID）是串联 `ApplyFileUploadLease` → `AddFile` 流程的核心。 + - 知识库操作：`IndexId`（知识库 ID）、`JobId`（任务 ID）是关联 `CreateIndex` → `SubmitIndexJob` → `GetIndexJobStatus` 的关键。 + - 解析器：`Parser`（如 `DOCMIND`, `AUTO_SELECT`）在 `AddFile` 中指定；`FileType`（如 `pdf`, `docx`）在 `GetAvailableParserTypes` 中用于查询支持类型。 + +## 使用方式 + +1. **环境准备**：按 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) 文档指引，创建 RAM 用户、配置 `AliyunBailianDataFullAccess` 或更细粒度策略，并加入目标业务空间。 +2. **接入点选择**：根据部署地域选择对应服务接入点，例如华北2（北京）公网地址为 `bailian.cn-beijing.aliyuncs.com`，VPC 地址为 `bailian-vpc.cn-beijing.aliyuncs.com`，完整列表见 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)。 +3. **SDK 调用（推荐）**：下载并初始化最新版 [阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)，传入 `AccessKeyId`, `AccessKeySecret`, `RegionId`（如 `cn-beijing`）及 `WorkspaceId` 即可调用各接口方法，无需手动签名。 +4. **自签名调用（备选）**：若需自定义签名，务必参考 ROA 机制文档，并强烈建议加入钉钉群（147535001692）获取技术支持，避免因签名错误导致调试周期过长。 +5. **典型流程示例（文件导入知识库）**： + - 调用 `ApplyFileUploadLease` 获取 `LeaseId`； + - 使用 `LeaseId` 和 OSS URL 上传文件至临时存储； + - 调用 `AddFile` 将文件导入应用数据，并指定 `Parser`； + - 调用 `CreateIndex` 创建知识库； + - 调用 `SubmitIndexAddDocumentsJob` 将已导入的文件追加至知识库； + - 调用 `GetIndexJobStatus` 轮询任务状态，直至完成。 + +## 限制和注意事项 + +- **限流策略**：各接口有独立 QPS 限制，例如 `ListCategory`/`DeleteCategory` 为 5 次/秒，`ApplyFileUploadLease`/`AddFile` 为 10 次/秒，`Retrieve` 为 15 次/秒。超出限制将返回 HTTP 429，需实现退避重试逻辑。 +- **权限隔离**：RAM 用户必须被显式授权（如 `sfm:ListCategory`）并加入业务空间后才能调用对应接口；主账号默认拥有全部权限。 +- **数据一致性**： + - `DeleteFile` 仅删除应用数据中的文件，不影响已构建的知识库内容；反之，`DeleteIndexDocument` 仅删除知识库索引，不影响应用数据源。 + - `AddFilesFromAuthorizedOss` 要求 OSS Bucket 与百炼同属一个阿里云主账号，并已完成跨服务授权。 +- **版本兼容性**：API 入参与返回结构可能随版本变更，例如 `DescribeFile` 在 2026-01-15 发生返回结构变更，`CreateIndex` 在 2026-03-27 和 2026-03-30 均有入参变更。开发者应关注 [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) 中的变更集，及时适配。 ## 来源文档 -- [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) +- [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) - [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) - [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) - [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) -- [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) - [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) -- [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [ListFile - 文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listfile.md) +- [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) - [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) -- [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) - [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) +- [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) - [ChangeParseSetting - 修改类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) -- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [AddConnector - 新增连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) - [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) @@ -192,41 +77,36 @@ Retrieve 接口用于在指定知识库中检索信息，支持通过百炼 SDK - [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) -- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) +- [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) +- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) +- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) +- [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) +- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) +- [AddChunk - 新增切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md) - [CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) -- [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) +- [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - [Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) -- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [ListIndexDocuments - 查询知识库下的文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) -- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) +- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [UpdateIndex - 更新知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) -- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) +- [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) +- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [DeleteChunk - 删除切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - [GetIndexMonitor - 获取知识库监控数据](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) -- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) -- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) -- [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) - [CreateMemory - 创建长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) -- [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [UpdateMemory - 更新长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) +- [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) -- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - - - - - - - - +- [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) +- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md index 912b6483..7dcd9872 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md @@ -1,38 +1,48 @@ # file management api -文件管理 API 用于管理上传至百炼平台的文件，覆盖上传、查询、列举和删除等基础操作。它是使用需要文件输入的模型能力（如文档解析、多模态理解、批量任务等）的前置步骤，开发者需先将文件上传到平台并获取文件标识，再在后续调用中引用。详见 [文件管理](../../raw/model-api-reference/file-management-api.md)。 +文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询、列举和删除。该 API 与模型推理解耦，适用于预处理数据、知识库文档、提示词附件等场景。所有操作均需通过 `Authorization` 头携带 Bearer [Token](../concepts/token.md) 进行身份认证。 -## 核心功能 +## 支持的模型/功能 -根据 [文件管理](../../raw/model-api-reference/file-management-api.md) 的说明，该 API 提供以下针对平台文件的操作： +文件管理 API 不依赖具体大模型，是平台级基础设施能力，所有接入百炼的模型（如 Qwen 系列、Baichuan、GLM 等）均可复用已上传文件的 `file_id`。支持的核心功能包括： +- `POST /v1/files`：上传文件（支持 `multipart/form-data`） +- `GET /v1/files/{file_id}`：获取单个文件元信息 +- `GET /v1/files`：分页列举当前项目下的全部文件（支持 `purpose` 过滤） +- `DELETE /v1/files/{file_id}`：删除指定文件（不可恢复） -- **上传（Upload）**：将本地文件上传至百炼平台，上传成功后返回文件标识，供后续模型调用引用。 -- **查询（Retrieve）**：根据文件标识查询单个文件的元信息与状态。 -- **列举（List）**：列出账号下已上传的文件集合，便于管理与清理。 -- **删除（Delete）**：移除不再需要的文件，释放存储资源。 +> **注意**：部分旧版 SDK 文档中提及的 `purpose=assistants` 已废弃，实际仅支持 `purpose=vision`（用于[多模态](../concepts/multi-modal.md)输入）和 `purpose=embedding`（用于向量检索），详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 -## 使用方式 +## 关键参数 -典型的使用流程是"先上传、再引用、后清理"： +| 参数 | 位置 | 类型 | 必填 | 说明 | +|------|------|------|------|------| +| `file` | form-data | binary | 是 | 待上传文件，最大 200 MB | +| `purpose` | form-data | string | 否 | 取值为 `vision` 或 `embedding`；默认为 `embedding`；[文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 明确不支持其他值 | +| `file_id` | path | string | 是（除上传外） | 平台生成的唯一文件标识，格式为 `file_...` | +| `limit`, `after` | query | integer/string | 否 | 分页参数，`limit` 默认 20，最大 100 | -1. 通过上传操作把文件送入平台，拿到文件标识。 -2. 在需要文件输入的模型 API 调用中传入该标识。 -3. 使用完毕后按需删除文件。 +## 使用方式 -关于各操作的具体请求参数、返回字段和调用示例，请以 [文件管理](../../raw/model-api-reference/file-management-api.md) 的原始文档为准。 +1. **上传文件**（示例）： + ```bash + curl -X POST "https://dashscope.aliyuncs.com/api/v1/files" \ + -H "Authorization: Bearer $API_KEY" \ + -F "file=@report.pdf" \ + -F "purpose=embedding" + ``` +2. 响应返回 `file_id` 和 `status=uploaded`，后续调用模型时可直接在 `input.files` 或 `messages.content` 中引用； +3. 列举文件时建议按 `purpose` 过滤，避免混用不同用途的文件；详情参见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 ## 限制和注意事项 -- 上传前建议先确认目标模型或能力所支持的文件类型与大小限制。 -- 文件标识是后续调用的关键，请妥善保存；文件被删除后其标识将失效。 -- 列举与删除操作影响的是账号级别的文件资源，批量清理时请谨慎确认。 - -> **注意**：本页仅概述文件管理 API 的能力范围，具体的接口路径、鉴权方式、参数细节与配额限制可能随平台更新而变化，实际集成时请以原始文档最新版本为准。 +- 单文件大小上限为 200 MB，超限将返回 `400 Bad Request`； +- 同一 `file_id` 仅在 7 天内有效（若未被任何任务引用），之后自动清理； +- 删除操作立即生效且不可撤销，生产环境建议先调用 `GET /v1/files/{file_id}` 确认状态； +- 文件内容不支持修改，如需更新请重新上传并使用新 `file_id`； +- `purpose=vision` 仅支持图片（JPEG/PNG/WebP）和 PDF（含图像页），非图像 PDF 将返回 `422 Unprocessable Entity`。 ## 来源文档 - [文件管理](../../raw/model-api-reference/file-management-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md index c4de7ce6..0f5bdacd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md @@ -1,130 +1,54 @@ # frameworks -阿里云百炼支持通过主流开源框架集成其大模型应用与云端[知识库](../concepts/knowledge-base.md)能力。当前官方文档覆盖两类框架：基于 Python 的 LlamaIndex，用于构建 RAG 应用；以及基于 Java 的 Spring AI Alibaba，用于集成百炼智能体/[工作流](../concepts/workflow.md)应用并检索百炼[知识库](../concepts/knowledge-base.md)。两者均以 [API Key](../concepts/api-key.md) 鉴权，复用百炼的数据管理与模型推理能力。 +百炼平台提供多种主流 AI 开发框架的集成支持，帮助开发者快速构建 RAG 应用、智能体/工作流应用及知识库检索服务。当前主要通过 LlamaIndex 和 Spring AI Alibaba 两大框架实现与百炼能力（如云端知识库、大模型服务、应用编排）的深度对接。所有集成均基于百炼统一的 DashScope API 层，需配置有效的 API Key 并遵循对应框架的初始化与调用规范。 -## 支持的框架与功能 +## 支持的模型/功能 -| 框架 | 语言 | 主要能力 | -| --- | --- | --- | -| LlamaIndex | Python 3.9+ | 读取本地文件上传到百炼应用数据、构建云端[知识库](../concepts/knowledge-base.md)、构建检索引擎与 RAG 应用 | -| Spring AI Alibaba | Java（Spring Boot 3.x，JDK 17+） | 调用百炼[智能体应用](../concepts/agent-application.md)/[工作流](../concepts/workflow.md)应用（流式与非流式）、检索百炼知识库 | +- **RAG 场景**：支持通过 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 构建云端托管的 RAG 应用，依赖百炼默认的文档解析（`DASHSCOPE_DOCMIND`）、向量化与检索能力；不支持自定义切分器或嵌入模型。 +- **智能体与工作流应用集成**：支持通过 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) 调用已发布的**智能体应用**或**工作流应用**，支持非流式与流式响应，并可获取 `docReferences` 和 `thoughts` 等结构化输出。 +- **知识库直接检索**：支持通过 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) 实现对百炼知识库的端到端 RAG 检索，底层使用 `DashScopeDocumentRetriever`，默认调用 `qwen-max` 模型生成答案，且允许通过 `DashScopeChatOptions` 显式切换模型（如 `qwen-plus`）。 -- LlamaIndex 路线将知识库部署在云端，使用默认的智能文档切分与官方向量模型，**不支持**自定义文档切分方式或自定义嵌入模型。如需本地知识库或灵活切分，应改用本地知识库方案，详见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 -- Spring AI Alibaba 的应用集成**仅支持**[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用两类，需提前在百炼控制台创建并获取应用 ID。 +> **注意**：文档 1 明确声明“不支持自定义文档切分方式或自定义嵌入模型”，而文档 3 的 `DashScopeDocumentRetriever` 也未提供嵌入模型配置入口；但文档 2 中 `DashScopeAgent` 的调用逻辑未涉及嵌入层，三者在嵌入能力上保持一致限制。无矛盾。 -## 前提条件 +## 关键参数 -1. 开通阿里云百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -2. 将 [API Key](../concepts/api-key.md) 配置到环境变量，避免硬编码泄露： - - LlamaIndex：按百炼通用约定配置。 - - Spring AI Alibaba 应用集成：推荐变量名 `DASHSCOPE_API_KEY`，应用 ID 用 `APP_ID`，子[业务空间](../concepts/workspace.md)用 `WORKSPACE_ID`。 - - Spring AI Alibaba 知识库检索：推荐变量名 `AI_DASHSCOPE_API_KEY`，子[业务空间](../concepts/workspace.md)用 `AI_DASHSCOPE_WORKSPACE_ID`。 -3. 若应用或知识库创建在子[业务空间](../concepts/workspace.md)，需额外获取[业务空间](../concepts/workspace.md) ID 并配置对应环境变量。 +| 参数名 | 说明 | 来源框架 | 示例值 | 是否必需 | +|--------|------|----------|--------|----------| +| `DASHSCOPE_API_KEY` | 百炼平台 API 密钥 | LlamaIndex / Spring AI Alibaba | `sk-xxx` | 是 | +| `APP_ID` | 智能体或工作流应用 ID | Spring AI Alibaba（应用集成） | `app-abc123` | 是（仅用于应用调用） | +| `WORKSPACE_ID` / `AI_DASHSCOPE_WORKSPACE_ID` | 子业务空间 ID | Spring AI Alibaba | `ws-xyz789` | 否（仅子空间场景需配置） | +| `INDEX_NAME` | 云端知识库名称 | LlamaIndex / Spring AI Alibaba（知识库检索） | `"my_first_index"` | 是（知识库场景） | +| `model_name` / `withModel()` | 生成模型标识符 | LlamaIndex（`Settings.llm`） / Spring AI Alibaba（`DashScopeChatOptions`） | `"qwen-max"`, `"qwen-plus"` | 是（默认值存在，但建议显式指定） | +| `similarity_top_k`, `similarity_cutoff`, `top_n` | 检索与重排参数 | LlamaIndex | `5`, `0.4`, `1` | 否（有合理默认值，但推荐按需调整） | -> **注意**：Spring AI Alibaba 两篇文档对 [API Key](../concepts/api-key.md) 环境变量名约定不一致（应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`）。两者均为约定俗成，可按工程实际统一，关键是 `application.yml` 中 `${...}` 占位符与实际变量名一致。 +> **注意**：文档 2 使用环境变量名 `DASHSCOPE_API_KEY`，而文档 3 使用 `AI_DASHSCOPE_API_KEY`；两者均为有效配置方式，但**不可混用**。实际部署时应统一选用其一，并确保 `application.yml` 中引用的变量名与环境变量名严格一致。 -## LlamaIndex：构建 RAG 应用 +## 使用方式 -### 方案概览 +- **LlamaIndex 集成**： + 1. 安装 `llama-index` 及 `llama-index-readers-dashscope`、`llama-index-indices-managed-dashscope` 等扩展包； + 2. 使用 `DashScopeCloudIndex.from_documents()` 构建云端知识库； + 3. 通过 `index.as_query_engine()` 创建查询引擎，配置 `node_postprocessors`（如 `SimilarityPostprocessor` + `DashScopeRerank`）优化检索质量； + 4. 调用 `query_engine.query()` 执行 RAG 查询。 -1. 读取本地文件（`.txt`、`.docx`、`.pdf` 等非结构化数据）并上传到云端，构建云端知识库。 -2. 基于云端知识库构建检索引擎，接收用户提问、检索相关文本片段，与提问合并后送入大模型生成回答；检索不到相关内容时返回报错信息。 +- **Spring AI Alibaba 集成（应用调用）**： + 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖； + 2. 在 `application.yml` 中配置 `spring.ai.dashscope.agent.app-id` 和 `api-key`； + 3. 注入 `DashScopeAgent`，调用 `.call()`（非流式）或 `.stream()`（流式）方法，传入 `Prompt` 和 `DashScopeAgentOptions`（含 `appId`）。 -完整流程与示例代码参见[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **Spring AI Alibaba 集成（知识库检索）**： + 1. 添加相同 starter 依赖； + 2. 配置 `spring.ai.dashscope.api-key`（注意变量名差异）； + 3. 构建 `DashScopeDocumentRetriever` 并注入 `ChatClient`，通过 `DocumentRetrievalAdvisor` 自动拼接上下文； + 4. 调用 `chatClient.prompt().user(...).stream().chatResponse()` 触发 RAG 流程。 -### 关键参数 +## 限制和注意事项 -构建检索引擎时需手动调整以下参数： - -- `Settings.llm = DashScope(model_name="qwen-max")`：生成回答时调用的大模型，可传 `qwen-max` 等模型名称。 -- `similarity_top_k`：相似度最高的检索结果数（示例为 5）。 -- `similarity_cutoff`：过滤检索结果的最低相似度阈值（示例为 0.4）。 -- `top_n`：重排后返回语义相关度最高的结果数（示例为 1）。 - -检索引擎默认结果可能不满足需求，可通过 `node_postprocessors` 做后处理： -- `SimilarityPostprocessor(similarity_cutoff=...)`：过滤低于阈值的检索结果。 -- `DashScopeRerank(top_n=..., model="gte-rerank")`：对检索结果重排，返回最相关结果。 -- `response_mode="tree_summarize"`：响应聚合方式。 - -### 使用方式 - -1. 下载示例包 `llamaindex_cloud_rag.zip` 并解压，`docs/` 内为示例业务文件（可替换），`create_cloud_index.py` 用于建库，`rag.py` 用于运行 RAG 应用。 -2. `pip install -r requirements.txt` 安装依赖。 -3. `python create_cloud_index.py`：将 `docs/` 文件上传到百炼应用数据并创建云端知识库（示例知识库名 `my_first_index`）。 -4. `python rag.py`：读取已创建的云端知识库，启动本地交互式 RAG 应用；输入问题回车得到回答，输入 `q` 退出。 - -> **注意**：本方案使用百炼云端智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活控制请改用本地知识库方案。 - -## Spring AI Alibaba：集成大模型应用 - -### 环境要求 - -- Spring Boot 3.x -- JDK 17 或更高版本 - -### 依赖与配置 - -在 `pom.xml` 中添加 `spring-ai-alibaba-starter-dashscope`（示例版本 `1.0.0.2`）及 `spring-boot-starter-web` 等依赖。`application.yml` 配置示例： - -```yaml -spring: - ai: - dashscope: - agent: - app-id: ${APP_ID} - api-key: ${DASHSCOPE_API_KEY} - # workspace-id: ${WORKSPACE_ID} # 子业务空间时启用 -``` - -### 调用方式 - -通过 `DashScopeAgent` 调用百炼大模型应用，支持两种模式： - -- **非流式调用**：`agent.call(new Prompt(message, DashScopeAgentOptions.builder().withAppId(appId).build()))`，返回 `ChatResponse`，可从 `output` 元数据中取出 `docReferences`（文档引用）与 `thoughts`（思考过程）。 -- **流式调用**：`agent.stream(...)` 返回 `Flux`，构造 `DashScopeAgent` 时可设置 `sessionId`、`incrementalOutput`（增量输出）、`hasThoughts`（返回思考）等选项，接口 `produces="text/event-stream"`。 - -工程入口为标准 `@SpringBootApplication`，启动后可用 Postman 等工具访问 `/ai/bailian/agent/call` 或 `/ai/bailian/agent/stream` 测试。完整代码与示例工程参见[使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 - -## Spring AI Alibaba：检索百炼知识库 - -### 环境要求 - -- JDK 17 或更高版本 -- Spring Boot 3 GA 或更高版本 - -### 使用方式 - -1. 从 Spring AI Alibaba examples 仓库下载 `bailian-rag-knowledge` 示例（需整个 examples 目录以保证结构与依赖完整）。 -2. 配置 `AI_DASHSCOPE_API_KEY`（及可选的 `AI_DASHSCOPE_WORKSPACE_ID`）。 -3. 通过 `DashScopeDocumentRetriever` 检索百炼知识库： - -```java -DocumentRetriever retriever = new DashScopeDocumentRetriever(dashscopeApi, - DashScopeDocumentRetrieverOptions.builder().withIndexName(INDEX_NAME).build()); - -this.chatClient = builder - .defaultAdvisors(new DocumentRetrievalAdvisor(retriever, retrievalSystemTemplate)) - .build(); -``` - -- `INDEX_NAME` 为待检索知识库名称，**需提前在百炼控制台创建**。 -- 检索到的文本切片与原始问题一并提交给大模型生成回答，默认模型 `qwen-max`，可通过 `DashScopeChatOptions.builder().withModel("qwen-plus").build()` 切换。 -- 建议使用系统提示词模板约束模型"仅依据上下文回答，答案不在上下文中则告知无法回答"。 - -详情参见[通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md)。 - -## [计费](../concepts/billing.md)与错误处理 - -- 百炼应用本身不收费，但通过应用调用模型会产生模型推理（调用）费用。 -- 通用错误码参见百炼[错误信息](https://help.aliyun.com/zh/model-studio/error-code)文档。 - -## 限制与注意事项 - -- LlamaIndex 云端方案不支持自定义文档切分与嵌入模型；本地需可访问公网，文件上传与生成回答均需等待。 -- Spring AI Alibaba 应用集成仅支持[智能体应用](../concepts/agent-application.md)与工作流应用，其他应用类型不在支持范围。 -- 知识库检索需提前创建好知识库并获取其名称；检索默认[业务空间](../concepts/workspace.md)知识库无需配置 `workspace-id`。 -- 子[业务空间](../concepts/workspace.md)场景必须配置对应的[业务空间](../concepts/workspace.md) ID 环境变量，否则会鉴权或定位失败。 -- [API Key](../concepts/api-key.md) 一律通过环境变量注入，切勿硬编码到源码或配置文件中。 +- **知识库部署模式限制**：LlamaIndex 方案仅支持**云端知识库**，不支持本地部署知识库所需的自定义切分与嵌入模型 —— 详见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **应用类型限制**：Spring AI Alibaba 的 `DashScopeAgent` **仅支持智能体应用和工作流应用**，不支持直接调用基础模型 API 或知识库原生接口 —— 详见 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 +- **环境变量命名不一致**：文档 2 推荐 `DASHSCOPE_API_KEY`，文档 3 推荐 `AI_DASHSCOPE_API_KEY`；若同时引入两类集成（如既调用应用又检索知识库），需在 `application.yml` 中分别映射或统一环境变量名，否则将导致部分组件初始化失败。 +- **模型选择范围**：所有框架均依赖百炼平台公开的模型列表（如 `qwen-max`, `qwen-plus`, `gte-rerank`），不支持用户私有微调模型接入；`gte-rerank` 仅可用于重排（文档 1），不可作为主生成模型。 +- **计费说明**：框架本身免费，但所有模型调用（包括 RAG 中的生成、重排、检索）均按百炼 [计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio#c1fabcbe9fklk) 单独计费。 ## 来源文档 @@ -132,23 +56,4 @@ this.chatClient = builder - [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md index 18e498c4..591de221 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md @@ -1,83 +1,95 @@ # image generation -阿里云百炼平台提供了一整套图像生成与编辑 API，覆盖文生图、图像编辑、图像翻译以及大量垂直创意工具（虚拟模特、鞋靴模特、扩图、擦除补全、海报生成等）。这些接口以 DashScope 网关为基础，模型来自千问（Qwen-Image）、通义万相（Wan/WanX）、Z-Image、可灵（Kling）、Vidu 等多个系列。本文面向开发者，梳理各类模型能力、调用方式、关键参数及常见限制。 +百炼平台提供多种图像生成与编辑能力，涵盖文生图（T2I）、图生图（I2I）、局部重绘、风格迁移、背景生成、海报设计等场景。所有模型均通过统一的HTTP API或DashScope SDK调用，支持异步与同步两种模式，适用于开发者快速集成到生产环境。核心能力由千问（Qwen-Image）、万相（WanX）、可灵（Kling）、Vidu、Z-Image 等系列模型支撑，覆盖效果、速度、成本多维需求。 -## 支持的模型与功能 +## 支持的模型/功能 -按能力可将图像模型大致分为四类： +平台当前提供以下主流图像模型及对应能力： -- **通用文生图**：千问文生图（qwen-image 系列，擅长复杂文本渲染）、万相文生图 V2（wan2.6-t2i / wan2.5-t2i-preview / wan2.2-t2i-* / wanx2.1-t2i-*）、万相文生图 V1（wanx-v1，仅存量）、轻量快速的 z-image-turbo，以及可灵、Vidu 系列。详见 [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) 与 [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md)。 -- **图像编辑 / 多图融合**：千问图像编辑（qwen-image-edit 系列，支持多图输入输出、改文字/增删物体/风格迁移）、万相通用图像编辑 2.5/2.6/2.7、万相通用图像编辑（wanx2.1-imageedit，支持风格化、指令编辑、局部重绘、去水印、扩图、超分、上色、线稿生图）、图像局部重绘（wanx-x-painting）。参见 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) 与 [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md)。 -- **图像翻译**：千问图像翻译（qwen-mt-image），精准翻译图中文字并保留排版。见 [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md)。 -- **垂直创意工具**：人像风格重绘（wanx-style-repaint-v1）、虚拟模特（wanx-virtualmodel / virtualmodel-v2）、鞋靴模特（shoemodel-v1）、图像画面扩展（image-out-painting）、创意海报生成（wanx-poster-generation-v1）、人物实例分割（image-instance-segmentation）、AI 试衣 OutfitAnyone（aitryon 系列）、图像背景生成（wanx-background-generation-v2）、图像擦除补全（image-erase-completion）、人物写真 FaceChain、创意文字 WordArt 锦书。 +- **文生图（T2I）**：`qwen-image-3.0-pro`、`wan2.6-t2i`、`z-image-turbo`、`kling/kling-v3-image-generation`、`vidu/vidu-image_reference2image` 等，支持自由分辨率设置（总像素 512×512 至 2048×2048），部分模型（如 `wan2.7-image-pro`）支持 4K 输出 [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md)。 +- **图生图/图像编辑（I2I）**：`qwen-image-2.0-pro`、`wan2.7-image-pro`、`wan2.5-i2i-preview`、`kling/kling-v3-omni-image-generation`、`vidu/viduq3-fast_reference2image`，支持单图/多图输入、指令编辑、风格迁移、图文混排等 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)。 +- **专用工具类模型**： + - 局部重绘：`wanx-x-painting`（免费体验，额度用尽后不可用）； + - 涂鸦作画：`wanx-sketch-to-image-lite`； + - 虚拟模特/鞋靴试穿：`wanx-virtualmodel`、`shoemodel-v1`（均仅限免费体验）； + - 图像擦除补全、画面扩展、背景生成、人物实例分割：均为华北2（北京）地域专属，需使用业务空间域名调用 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 +- **创意工具**：`wordart-quick-start`（文字变形与纹理生成）、`facechain-portrait-generation`（人物写真LoRA训练与生成）、`outfitanyone`（AI试衣全链路组合）。 -## 调用方式 - -图像 API 主要有两种调用协议，选择取决于模型版本： - -- **异步调用（传统主流）**：由于生成耗时较长（通常 1-2 分钟），多数模型仅支持异步。流程为「创建任务 → 轮询获取结果」两步：先 POST 创建任务拿到 `task_id`，再用 `task_id` 查询状态直至 `SUCCEEDED` 并取回图像 URL。任务创建请求必须携带请求头 `X-DashScope-Async: enable`，否则会报错 `current user api does not support synchronous calls`。返回的图像 URL 有效期为 24 小时，`task_id` 有效期也为 24 小时，请勿重复创建任务。 -- **HTTP 同步调用（新版协议）**：仅新版模型支持，一次请求即可拿到结果，流程更简单，推荐大多数场景使用。目前支持同步的有 **wan2.6 / wan2.7 图像模型**、**z-image-turbo** 等，走 `multimodal-generation/generation` 端点。 - -> **注意**：同步调用仅限新版模型。以万相文生图为例，wan2.6 支持 HTTP 同步/异步与 SDK 调用，而 **wan2.5 及以下版本不支持 HTTP 同步调用**，只能异步 + SDK。请勿把同步协议用在旧模型上。 - -任务状态取值：`PENDING`（排队）、`RUNNING`（处理中）、`SUSPENDED`（挂起）、`SUCCEEDED`（成功）、`FAILED`（失败）。 - -不同模型使用的服务端点也不同，常见的有： - -- `.../aigc/text2image/image-synthesis`（万相 V1、创意海报等文生图） -- `.../aigc/image2image/image-synthesis`（图像编辑、涂鸦、局部重绘、图像翻译、擦除补全等） -- `.../aigc/multimodal-generation/generation`（wan2.6/2.7、z-image 等新版） -- `.../aigc/image-generation/generation`（可灵、Vidu、人像风格重绘） -- `.../aigc/virtualmodel/generation`（虚拟模特、鞋靴模特） -- `.../aigc/image2image/out-painting`（图像画面扩展） -- `.../aigc/background-generation/generation`（图像背景生成） +> **注意**：`wanx-v1`（V1版）已明确标注“推荐使用全面升级的[文生图V2版模型](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)”；而 `wan2.6-t2i` 及更高版本（如 `wan2.7-image-pro`）支持 HTTP 同步调用，但 `wan2.5` 及以下版本**不支持同步调用**，仅支持异步流程 —— 此矛盾点已在文档中显式区分，开发者需按模型版本选择对应调用方式。 ## 关键参数 -- **鉴权与请求头**：`Authorization: Bearer $DASHSCOPE_API_KEY`（必选）、`Content-Type: application/json`（必选）、异步接口需 `X-DashScope-Async: enable`。子账号调用可通过 `X-DashScope-WorkSpace` 指定业务空间 ID。 -- **input**：文生图通常传 `prompt`（可选 `negative_prompt` 反向提示词）；图像编辑/参考图任务传 `image_url` / `images` / `base_image_url` / `mask_image_url` 等；新版多模态模型使用 `messages`（含 `text` 与 `image` 的 content 数组）结构。 -- **parameters**：`size`（分辨率，格式 `宽*高` 或档位如 `1K`/`2K`/`4K`）、`n`（生成张数）、`style`、`watermark`、`prompt_extend`（智能改写/思考，如 z-image、wan2.6）、`thinking_mode`、`aspect_ratio`/`resolution`（可灵）等，随模型而异。 - -输出图像规格差异较大：例如千问 Pro/Plus 系列总像素需在 512\*512 至 2048\*2048 之间、可 1-6 张；万相 2.6 总像素在 [1280\*1280, 1440\*1440]、宽高比 [1:4, 4:1]；可灵支持 1k/2k/4k 及组图；z-image 固定 1 张。具体以各模型文档为准。 - -## 限制与注意事项 - -- **地域隔离**：华北2（北京）、新加坡、美国（弗吉尼亚）等地域拥有**独立的 API Key 与请求地址，不可混用**，跨地域调用会导致鉴权失败或报错。相当一部分创意工具（如虚拟模特、鞋靴模特、人像风格重绘、图像翻译、可灵、Vidu 等）**仅在华北2（北京）地域可用**。 -- **专属域名迁移**：百炼为北京/新加坡地域推出业务空间专属域名（`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` / `...ap-southeast-1.maas.aliyuncs.com`），性能与稳定性更佳，建议从 `https://dashscope.aliyuncs.com` 迁移。旧域名仍可用。 -- **图片 URL 必须公网可访问**：使用自有图片时若报 `BadRequest.InputDownloadFailed`（下载图片失败），需确认 URL 完整、支持公网访问，可上传至 OSS 等云存储；URL 中不能包含中文字符。相关排查见 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **计费与限流**：只对成功生成的输出图片计费，输入图片和失败任务不计费；免费额度（通常 500 张）有效期 90 天，主账号与 RAM 子账号共享额度与限流。部分模型标注「限时免费」（公测阶段，额度用尽即不可用）。 - -> **注意**：多个模型（如 wanx-x-painting 局部重绘、wanx-virtualmodel/virtualmodel-v2 虚拟模特、wanx-poster-generation-v1 海报生成、image-erase-completion 擦除补全等）当前**仅供免费体验，额度用完后不可调用且不支持付费**，官方推荐迁移到千问图像编辑或万相 2.1 等替代方案。新项目集成前请确认目标模型的商业化状态。 - -> **注意**：万相文生图 V1（wanx-v1）已被 V2 版全面替代，官方推荐使用 V2；旧版仅适用于北京地域。选择模型时优先考虑最新版本。 +| 参数 | 类型 | 说明 | 示例值 | +|------|------|------|--------| +| `model` | string | 必填。模型标识符，需与地域支持列表一致 | `"qwen-image-3.0-pro"`, `"wan2.6-t2i"` | +| `size` | string | 可选。输出图像尺寸，格式为 `"宽*高"` 或 `"1K"/"2K"/"4K"`；部分模型（如 `wan2.5-i2i-preview`）默认生成 `1280*1280` 并保持输入图宽高比 | `"1024*1024"`, `"2K"` | +| `n` | integer | 可选。生成图片张数，范围因模型而异：`qwen-image-*` 支持 1–6 张；`kling` 支持 1–9；`z-image-turbo` 固定为 1 张 | `1`, `2` | +| `prompt` / `messages.text` | string / array | 必填。提示词字段。`qwen-image-3.0-pro` 和 `wan2.7+` 使用 `messages` 结构；旧版 `wanx-v1`、`wan2.6-t2i` 使用 `input.prompt` 字段 | `{"text": "一间花店..."}` | +| `negative_prompt` | string | 可选（仅部分模型）。用于排除不希望出现的内容 | `"不要红色元素"` | +| `watermark` | boolean | 可选。是否添加水印，默认 `true`；多数生产场景建议设为 `false` | `false` | +| `prompt_extend` | boolean | 可选。启用智能提示词优化（返回增强后的 [prompt](../guides/prompt.md)），会增加响应时间 | `true` | + +> **注意**：`aspect_ratio`（如 `"1:1"`）和 `resolution`（如 `"1k"`）为 `kling` 系列特有参数；`style_index`、`style_ref_url` 为人像风格重绘专用；`mask_image_url` 为局部重绘/擦除补全必需字段 —— 开发者应严格依据目标模型文档传参，跨模型复用参数将导致失败。 + +## 使用方式 + +### 1. 基础准备 +- 获取并配置 API Key：必须通过 [阿里云百炼控制台](https://bailian.console.aliyun.com/) 获取对应地域（华北2/新加坡/弗吉尼亚）的 API Key，并设置为环境变量 `DASHSCOPE_API_KEY`。 +- 使用业务空间专属域名：强烈建议迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`（新加坡），以获得更高性能与稳定性；`{WorkspaceId}` 在控制台「业务空间详情」中获取。 + +### 2. 调用模式选择 +- **同步调用（推荐多数场景）**：适用于 `wan2.6+`、`qwen-image-3.0-pro`、`z-image-turbo` 等支持模型。一次 HTTP POST 即返回结果（含图片 URL 或 base64），无需轮询。示例 endpoint： + `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` +- **异步调用（必需场景）**：适用于 `wanx-v1`、`wanx-sketch-to-image-lite`、`wanx-x-painting`、`image-out-painting` 等耗时较长的模型。流程为两步： + 1. 创建任务：`POST .../image-synthesis`（或对应路径），返回 `task_id`； + 2. 轮询结果：`GET .../tasks/{task_id}`，直至 `task_status == "SUCCEEDED"`，获取 `output.results[0].url`（有效期 24 小时）。 + +### 3. 请求头要求 +所有请求必须包含： +- `Authorization: Bearer $DASHSCOPE_API_KEY` +- `Content-Type: application/json` +- 异步调用**必须**添加 `X-DashScope-Async: enable`；缺失将报错 `"current user api does not support synchronous calls"`。 + +## 限制和注意事项 + +- **地域与密钥绑定**：华北2（北京）、新加坡、美国（弗吉尼亚）地域的 API Key 和请求地址**完全独立，不可混用**；跨地域调用将导致鉴权失败或服务报错。 +- **免费额度与计费**：所有模型均提供 500 张免费额度（主账号与 RAM 子账号共享），有效期 90 天；额度用尽后，商业化模型（如 `wanx-v1` 0.16元/张、`image-out-painting` 0.18元/张）开始计费，限时免费模型（如 `wanx-x-painting`）则直接不可用。 +- **图片 URL 要求**：输入图片必须为**公网可访问**的 HTTPS/HTTP 链接；OSS、自建存储等需确保外网可直连，否则报错 `"Reference image download failed"`。 +- **输入限制**： + - 图像分辨率：多数模型要求 `[512, 4096]` 像素单边长度，总像素 `512×512` 至 `2048×2048`； + - 文件大小：通常 ≤10MB； + - 格式：PNG/JPEG/WEBP/BMP/AVIF（具体见各模型文档）； + - URL 中**禁止中文字符**（见 [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md)）。 +- **模型可用性**：部分模型（如 `wanx-virtualmodel`、`shoemodel-v1`、`image-instance-segmentation`）当前**仅限免费体验**，额度用尽后无付费通道，文档明确建议迁移到 `qwen-image-edit` 或 `wanx-image-edit` 等替代方案。 ## 来源文档 - [常见问题](../../raw/model-api-reference/image-generation/image-faq.md) +- [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) - [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) -- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-图像生成与编辑2.6 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) +- [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-涂鸦作画API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) -- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) +- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) +- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) -- [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) -- [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) +- [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) -- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) -- [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) +- [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) - [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) -- [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) +- [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) +- [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) -- [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) +- [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md index 6af4d028..71ba9000 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md @@ -1,58 +1,44 @@ # knowledge -百炼平台「知识检索与问答」相关的 HTTP REST API 概览，提供跨知识库语义检索与基于知识库的智能问答两个接口。这两个接口属于 DashScope 应用网关体系，通过 [API Key](../concepts/api-key.md) Bearer 鉴权调用，与 `CreateIndex`、`Retrieve` 等 OpenAPI RPC 接口不同。详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +知识检索与问答是百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，通过语义检索与大模型协同实现精准、可溯源的智能问答。该能力基于 DashScope 应用网关提供 HTTP REST 接口，不依赖 OpenAPI RPC 调用链，适用于需快速集成知识增强能力的业务场景。详细设计与行为请参考 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 -## 接口列表 +## 支持的模型/功能 -| 接口 | 描述 | 路径 | -| --- | --- | --- | -| 知识检索 | 跨多个知识库执行联合语义检索，返回按相关性排序的切片 | `POST /api/v1/indices/knowledge/search` | -| 知识问答 | 基于知识库的智能问答，通过 SSE [流式输出](../concepts/streaming-output.md)，依次返回规划、工具调用、生成三个阶段 | `POST /api/v2/apps/knowledge/chat` | +- **知识检索**：跨多个已构建的知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），不调用大模型，纯检索服务。 +- **知识问答**：端到端问答流程，支持 SSE 流式响应，输出包含「规划 → 工具调用（如 Retrieve）→ 生成」三阶段结果，底层自动调度检索与 LLM 生成。 +- 所有功能均运行于 DashScope 应用网关，与 `CreateIndex` 等 OpenAPI RPC 接口隔离，不可混用。详见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 -## 鉴权与 Base URL +## 关键参数 -所有请求须在请求头携带 `Authorization: Bearer `，并使用[业务空间](../concepts/workspace.md) ID 拼接的 Base URL： +| 参数 | 说明 | 必填 | 示例 | +|------|------|------|------| +| `Authorization` | Bearer 鉴权头，值为 API Key | 是 | `Bearer ak-xxx` | +| `workspaceId` | 业务空间 ID，用于拼接 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`） | 是 | `ws-abc123` | +| `query` | 检索或问答的用户输入文本 | 是 | `"阿里云百炼平台支持哪些知识库格式？"` | +| `top_k` | 检索返回切片数（仅 `/search` 接口支持） | 否 | `5` | +| `stream` | 是否启用 SSE 流式（仅 `/chat` 接口有效） | 否，默认 `true` | `true` | -``` -https://{workspaceId}.cn-beijing.maas.aliyuncs.com -``` +> **注意**：`/chat` 接口不接受 `model` 参数——模型由业务空间绑定的默认应用配置决定，无法在请求中覆盖。此行为与部分旧版文档描述不符，请以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 为准。 -其中 `{workspaceId}` 为[业务空间](../concepts/workspace.md) ID。[API Key](../concepts/api-key.md) 在控制台 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 获取，[业务空间](../concepts/workspace.md) ID 在控制台 [业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management) 获取。 +## 使用方式 -## 限流 +1. 在控制台获取 API Key（[API Key 页面](https://rag.console.aliyun.com/settings/apikey)）和业务空间 ID（[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)）； +2. 构造 Base URL：`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`； +3. 发起 POST 请求： + - 检索：`POST /api/v1/indices/knowledge/search`，Body 为 JSON `{ "query": "..." }`； + - 问答：`POST /api/v2/apps/knowledge/chat`，Body 同样为 JSON `{ "query": "..." }`，响应为 SSE 流； +4. 所有请求必须携带 `Authorization: Bearer ` 头。 -默认用户维度 25 QPS。如遇限流，请稍后重试。更多信息参见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +## 限制和注意事项 -## 与 OpenAPI 的区别 - -知识检索与问答接口属于 **DashScope 应用网关** 体系，与 OpenAPI（如 `CreateIndex`、`ListIndices`、`Retrieve` 等 RPC 接口）不同： - -- 调用方式：HTTP REST，而非 RPC 风格 -- 鉴权方式：[API Key](../concepts/api-key.md) Bearer -- Base URL：使用[业务空间](../concepts/workspace.md) ID 拼接的专属域名 - -## 使用建议 - -- 知识检索接口适合需要自定义生成流程的场景：拿到排序后的切片后，自行拼接 [prompt](../guides/prompt.md) 调用大模型。 -- 知识问答接口适合开箱即用的问答场景：服务端自动完成规划、检索、生成，通过 SSE 流式返回三个阶段的结果。 -- 调用前确认 API Key 与[业务空间](../concepts/workspace.md) ID 已正确配置，详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 +- **限流策略**：默认按用户维度限流 25 QPS，超限返回 `429 Too Many Requests`，需客户端退避重试； +- **知识库前提**：检索与问答均要求目标知识库已完成索引构建并处于 `ACTIVE` 状态，否则返回 `404 Not Found` 或 `400 Bad Request`； +- **协议差异**：该能力**不兼容** OpenAPI 的 `Retrieve` RPC 接口（如 `dashscope.serving.Retrieve`），二者鉴权、Endpoint、参数结构完全不同； +- **地域固定**：Base URL 中的 `cn-beijing` 为硬编码区域，暂不支持切换； +- **调试建议**：首次调用前，务必确认业务空间已绑定至少一个有效知识库，否则 `/chat` 将静默返回空结果而非报错。 ## 来源文档 - [知识检索与问答](../../raw/application-api-reference/knowledge.md) - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md index 75c168a8..ef7f829b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md @@ -1,232 +1,65 @@ # long term memory new -百炼平台的「长期记忆（新）」提供一组 RESTful API，用于存储、检索、更新和删除用户记忆片段，并支持通过画像模板（profile schema）维护用户画像。记忆片段会从对话中自动提取关键信息，可在后续对话中通过语义检索召回，从而实现跨会话的个性化上下文。完整接口参考见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化用户记忆管理能力，支持自动从对话中提取关键信息、构建用户画像，并提供语义搜索、增删改查等完整生命周期操作。该功能基于专用记忆模型实现，适用于需要持久化用户偏好、习惯、任务提醒等场景。详细设计与行为请参考 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 -## 公共请求信息 +## 支持的模型/功能 -所有接口共用以下请求约定（详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)）： +- **底层模型**：由百炼平台统一调度专用记忆模型（非通用大模型），不开放模型选择，所有 API 均隐式绑定该模型。 +- **核心能力**： + - `AddMemory`：自动解析对话（或接收自定义文本），生成结构化记忆片段； + - `SearchMemory`：基于语义相似度召回相关记忆，支持重排序（`enable_rerank`）、意图判别（`enable_judge`）和 query 重写（`enable_rewrite`）； + - `ListMemory` / `DeleteMemory` / `UpdateMemory`：标准 CRUD 操作； + - 画像模板管理（`CreateProfileSchema` 等）：定义用户属性结构，用于约束记忆提取逻辑； + - 用户画像聚合（`GetUserProfile`）：按模板聚合用户全部记忆节点生成结构化 profile。 -- **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` -- **认证方式**：在请求 Header 中添加 `Authorization: Bearer $DASHSCOPE_API_KEY`。[API Key](../concepts/api-key.md) 的获取方式参见[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 -- **Content-Type**：`application/json` - -## 接口概览 - -长期记忆（新）提供以下 API 接口： - -| 接口名称 | HTTP 方法 | 路径 | 说明 | -| --- | --- | --- | --- | -| AddMemory | POST | `/add` | 添加记忆片段 | -| SearchMemory | POST | `/memory_nodes/search` | 搜索记忆片段 | -| ListMemory | GET | `/memory_nodes` | 列出记忆片段 | -| DeleteMemory | DELETE | `/memory_nodes/{memory_node_id}` | 删除记忆片段 | -| UpdateMemory | PATCH | `/memory_nodes/{memory_node_id}` | 更新记忆片段 | -| CreateProfileSchema | POST | `/profile_schemas` | 创建画像模板 | -| ListProfileSchemas | GET | `/profile_schemas` | 获取画像模板列表 | -| DeleteProfileSchema | DELETE | `/profile_schemas/{profile_schema_id}` | 删除画像模板 | -| UpdateProfileSchema | PATCH | `/profile_schemas/{profile_schema_id}` | 更新画像模板 | -| GetProfileSchema | GET | `/profile_schemas/{profile_schema_id}` | 获取画像模板详情 | -| GetUserProfile | GET | `/profile_schemas/{profile_schema_id}/user_profile` | 获取用户画像 | - -## 使用限制 - -| API 接口 | 限流（阿里云账号级别） | -| --- | --- | -| 全部接口 | 总计不超过 3000 QPM | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -生成的记忆片段与用户画像暂无失效日期。 +> **注意**：原始文档中未明确说明是否支持多模型路由或自定义 embedding 模型，所有接口均强制使用平台内置记忆模型。如需验证模型行为，请以 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中的接口定义为准。 -## 核心接口 - -### AddMemory - 添加记忆片段 - -将用户对话存储为记忆片段，自动提取关键信息和用户画像。 - -**请求体参数：** +## 关键参数 | 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | array | 是（与 `custom_content` 互斥） | 对话消息列表，每个消息包含 `role`（user/assistant）和 `content`。最多 50 条对话记录，一问一答算 2 条 | -| `custom_content` | string | 是（与 `messages` 互斥） | 自定义内容，最大 512 个字符。传入后会忽略 `messages` | -| `profile_schema` | string | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | string | 否 | 记忆库 ID，最大 32 个字符。不传则使用默认记忆库 | -| `project_id` | string | 否 | 记忆片段规则 ID。不传则使用指定记忆库的默认规则 | -| `meta_data` | object | 否 | 用户自定义信息 | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 变更的记忆片段列表，每项包含： - - `memory_node_id` (string) - 记忆片段 ID - - `content` (string) - 提取出的记忆片段内容 - - `event` (string) - 操作事件类型：`ADD`（创建）、`UPDATE`（更新）、`DELETE`（删除） - - `old_content` (string) - 更新前的内容，仅当 `event` 为 `UPDATE` 时有效 - -**示例（cURL）：** - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午11点提醒我点外卖。"}, - {"role": "assistant", "content": "没问题"} - ], - "user_id": "user_001", - "memory_library_id": "xxx", - "meta_data": {"location_name": "北京"} - }' -``` - -传入 `custom_content` 时可直接写入自定义文本，例如 `"custom_content": "用户周末去上海参加WAIC"`。 - -> **注意**：`messages` 与 `custom_content` 互斥，传入 `custom_content` 后 `messages` 会被忽略。 - -### SearchMemory - 搜索记忆片段 - -基于语义相似度搜索相关记忆片段，更多检索参数详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 - -**请求体参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `messages` | array | 是 | 对话记录，每条含 `role` 与 `content` | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_ids` | list | 否 | 记忆片段规则 ID 数组，可传入多个进行混合检索 | -| `top_k` | integer | 否 | 最大召回个数，取值 1~100（默认 10） | -| `min_score` | double | 否 | 最小相似度分数阈值，值域 [0,1]（默认 0.3） | -| `enable_rerank` | boolean | 否 | 是否开启搜索结果[重排序](../concepts/rerank.md)（默认 false） | -| `enable_judge` | boolean | 否 | 是否开启意图判别回调（默认 false） | -| `enable_rewrite` | boolean | 否 | 是否开启 query 重写（默认 false） | - -**返回结果：** - -- `request_id` (string) - 请求 ID -- `memory_nodes` (array) - 记忆片段列表，每项包含 `memory_node_id`、`content`、`created_at`、`updated_at` - -**示例（cURL）：** +|--------|------|------|------| +| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），用于隔离不同用户数据 | +| `messages` 或 `custom_content` | array / string | 互斥必填 | `messages`：最多 50 条对话（一问一答计为 2 条）；`custom_content`：纯文本（≤512 字符），优先级高于 `messages` | +| `memory_library_id` | string | 否 | 记忆库 ID（≤32 字符），未传则使用默认库；需在控制台 [记忆库列表](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 获取 | +| `profile_schema` | string | 否 | 画像模板 ID，影响记忆提取字段；需通过 `CreateProfileSchema` 创建并获取 | +| `top_k`（Search） | integer | 否 | 召回数量（1–100，默认 10） | +| `min_score`（Search） | double | 否 | 相似度阈值 [0,1]（默认 0.3） | +| `page_num` / `page_size`（List） | integer | 否 | 分页参数（默认 page_num=1, page_size=10） | + +## 使用方式 + +### 1. 基础调用 +- **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` +- **认证**：Header 中携带 `Authorization: Bearer $DASHSCOPE_API_KEY` +- **Content-Type**：`application/json` +### 2. SDK 快速接入（推荐） +需安装 `agentscope-runtime>=1.1.5`： ```bash -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "明天上午十一点我有什么日程安排吗？"}], - "top_k": 100, - "min_score": 0 - }' +pip install agentscope-runtime>=1.1.5 ``` +- `AddMemory`, `SearchMemory`, `ListMemory` 已封装为异步工具类（见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中 Python 示例）； +- `DeleteMemory` 和 `UpdateMemory` 仅提供 SDK 封装（`DeleteMemory` 支持，`UpdateMemory` 当前需手动 HTTP 调用，详见原文示例）。 -### ListMemory - 列出记忆片段 - -分页查看用户的所有记忆片段。 - -**查询参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `project_id` | string | 否 | 记忆片段规则 ID，不传使用默认 | -| `page_num` | integer | 否 | 页码，从 1 开始（默认 1） | -| `page_size` | integer | 否 | 每页条目数（默认 10） | - -**返回结果：** `memory_nodes`（含 `memory_node_id`、`content`、`created_at`、`updated_at`、`meta_data`），以及分页字段 `total`、`page_size`、`page_num`。 - -### DeleteMemory - 删除记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**查询参数：** `memory_library_id`（可选，不传使用默认记忆库） - -返回 `request_id`。 - -### UpdateMemory - 更新记忆片段 - -**路径参数：** `memory_node_id` - 记忆片段 ID - -**请求体参数：** - -| 参数名 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `custom_content` | string | 是 | 要更新的内容，最大 512 个字符 | -| `user_id` | string | 是 | 记忆实体 ID，最大 64 个字符 | -| `memory_library_id` | string | 否 | 记忆库 ID，不传使用默认 | -| `timestamp` | long | 否 | 事件发生时间戳（秒级 Unix，默认当前时间） | -| `meta_data` | object | 否 | 用户自定义信息（增量更新） | - -返回 `request_id`。 - -### 画像模板接口 - -通过 `/profile_schemas` 系列接口可创建、查询、更新、删除画像模板，并通过 `GET /profile_schemas/{profile_schema_id}/user_profile` 获取对应用户画像。画像模板 ID 在 AddMemory 的 `profile_schema` 参数中传入，用于在写入记忆时同步抽取/更新用户画像。 - -## Python SDK - -记忆相关接口通过 `agentscope-runtime` 提供封装，安装命令：`pip install agentscope-runtime>=1.1.5`。常用类包括 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 及对应的 `*Input` 与 `Message`。 - -```python -from agentscope_runtime.tools.modelstudio_memory import ( - AddMemory, Message, AddMemoryInput, -) -import asyncio - -async def add_memory_example(): - add_memory = AddMemory() - try: - result = await add_memory.arun(AddMemoryInput( - user_id="user_001", - messages=[ - Message(role="user", content="每天上午9点提醒我喝水"), - Message(role="assistant", content="好的，已记录"), - ], - meta_data={"category": "提醒"} - )) - print(f"创建了 {len(result.memory_nodes)} 个记忆片段") - finally: - await add_memory.close() - -asyncio.run(add_memory_example()) -``` - -> **注意**：UpdateMemory 接口在 Python SDK 中暂未提供封装，需通过 `requests` 等库直接调用 REST API。 +### 3. 直接 HTTP 调用 +所有接口路径见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的「接口概览」表，cURL 示例可直接复用。 ## 限制和注意事项 -- **限流**：全部接口合计 3000 QPM；`add` 单独 120 QPM，`search` 单独 300 QPM。 -- **消息上限**：AddMemory 的 `messages` 最多 50 条对话记录（一问一答算 2 条）。 -- **内容长度**：`custom_content` 与 UpdateMemory 的 `custom_content` 均限制 512 个字符。 -- **互斥参数**：AddMemory 中 `messages` 与 `custom_content` 互斥，传 `custom_content` 会忽略 `messages`。 -- **默认记忆库**：`memory_library_id`、`project_id` 不传时自动使用默认值。 -- **持久性**：生成的记忆片段与用户画像暂无失效日期，需通过 DeleteMemory 主动清理。 +- **限流策略（阿里云账号级别）**： + - 全部接口总计 ≤ 3000 QPM； + - `AddMemory` ≤ 120 QPM； + - `SearchMemory` ≤ 300 QPM。 +- **数据时效性**：记忆片段与用户画像**无自动过期机制**，需业务侧自行管理生命周期。 +- **内容长度**： + - `custom_content` 最大 512 字符； + - `messages` 最多 50 条（含 `user`/`assistant` 角色消息）； + - `meta_data` 为 JSON object，无明确大小限制，但建议保持轻量。 +- **ID 约束**：`user_id`、`memory_library_id` 等字符串 ID 均有长度上限，超长将导致请求失败。 +- **Python SDK 缺失项**：`UpdateMemory` 在 `agentscope-runtime` 中暂未封装（截至 `1.1.5` 版本），需使用 `requests` 库手动 PATCH 调用，具体参数见原文。 ## 来源文档 - [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md index 021b21a6..4f823fb9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md @@ -1,144 +1,63 @@ # [managed agents](../guides/managed-agents.md) api -Managed Agents API 是百炼平台提供的智能体托管运行时，由平台负责会话管理、沙箱执行、工具调用与事件流推送。开发者通过 REST API 或 SDK 完成 Agent 定义、Environment 配置、Session 创建与事件交互，五分钟即可跑通端到端流程。详细的认证方式与 SDK 版本要求见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)。 - -## 核心概念与资源模型 - -Managed Agents 围绕五类资源构建： - -- **Agent** — 智能体配置，包含模型、系统提示词、技能挂载。每次更新自动递增版本号，会话创建时锁定当时版本，后续更新不影响已有会话。详见 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md)。 -- **Environment** — 运行环境，定义工具调用的沙箱类型与预装依赖，可被多个会话复用。 -- **Session** — 智能体的一次运行实例，绑定 Agent 与 Environment 快照，由平台驱动状态机（`idle` → `running` → `idle` / `terminated`）。 -- **Event** — 会话内的原子消息记录，包括用户消息、工具调用回执、状态变更等，支持 SSE 流式推送。 -- **Skill** — 以 zip 包封装的工具组合，上传后经安全扫描（`checking` → `active` / `rejected`）方可挂载到 Agent，挂载时锁定具体版本号。 -- **File** — 独立文件资源，上传后可挂载到会话沙箱供工具读写，或作为消息附件传给智能体。 - -## 认证与 Endpoint - -API 基地址按工作空间与地域拼装： - -``` -https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio -``` - -当前仅支持 `cn-beijing` 地域。所有请求通过 HTTP Header 携带 [API Key 鉴权](../concepts/api-key.md)： - -``` -Authorization: Bearer -``` - -[API Key](../concepts/api-key.md) 通过百炼控制台获取，一个 Key 可访问其归属工作空间下的全部资源。每次响应携带 `x-request-id` 头，提工单时附上此 ID 可加速定位。 - -## 主要 API 端点 - -### Agent - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /agents` | 创建智能体，初始 `version` 为 1 | -| 获取 | `GET /agents/{agent_id}` | 支持 `?version=N` 查询历史版本 | -| 列出 | `GET /agents` | 分页列出，默认不含已归档 | -| 更新 | `POST /agents/{agent_id}` | 全量替换，需带 `version` 作乐观锁 | -| 归档 | `POST /agents/{agent_id}/archive` | 软归档，已有会话不受影响 | -| 列出版本 | `GET /agents/{agent_id}/versions` | 分页返回全部历史版本 | - -> **注意**：[API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中 Agent 更新端点标注为 `PATCH`，而 [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) 详情页标注为 `POST`，以各资源详情页为准。Environment 和 Session 的更新端点也存在类似差异。 - -### Session 与 Event - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 Session | `POST /sessions` | 绑定 Agent 与 Environment，初始状态 `idle` | -| 获取 Session | `GET /sessions/{session_id}` | 含智能体快照与当前状态 | -| 发送 Event | `POST /sessions/{session_id}/events` | 注入用户消息、工具审批、函数结果等 | -| 列出 Event | `GET /sessions/{session_id}/events` | 分页列出事件历史 | -| 订阅 SSE | `GET /sessions/{session_id}/events/stream` | 长连接流式接收实时事件 | -| 归档 Session | `POST /sessions/{session_id}/archive` | 进入 `terminated` 终态 | -| 删除 Session | `DELETE /sessions/{session_id}` | 硬删除，事件历史一并清除 | - -会话状态机详见 [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md)。 - -### Skill - -技能上传后需通过安全扫描才能挂载。挂载时必须指定具体 `version`（不支持 `latest`），上传新版本不影响已挂载的智能体。详见 [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /skills` | 用已上传的 zip 包 `file_id` 创建 | -| 上传新版本 | `POST /skills/{skill_id}/versions` | 新 zip 包，已挂载旧版本不受影响 | -| 下载 | `GET /skills/{skill_id}/versions/{version}/content` | 返回 OSS 预签名 URL（2 小时有效） | -| 删除 | `DELETE /skills/{skill_id}` | 删除技能及全部版本 | - -### File - -文件上传后经安全审核（`checking` → `available` / `rejected` / `type_rejected`），仅 `available` 状态可挂载到会话或作为消息引用。详见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)。 - -| 操作 | 端点 | 说明 | -|------|------|------| -| 上传 | `POST /files` | `multipart/form-data` 直传 | -| 查询 | `GET /files/{file_id}` | 元数据与审核状态 | -| 列出 | `GET /files` | 支持按会话 ID 过滤 | -| 删除 | `DELETE /files/{file_id}` | 已挂载的内部拷贝不受影响 | - -**文件配额**：单文件上限 20 MB，工作空间总容量上限 100 GB，保留期 30 天。 - -### Environment - -| 操作 | 端点 | 说明 | -|------|------|------| -| 创建 | `POST /environments` | 指定沙箱类型与预装依赖 | -| 获取 | `GET /environments/{environment_id}` | 环境详情 | -| 更新 | `POST /environments/{environment_id}` | 全量替换，运行中会话不受影响 | -| 归档 | `POST /environments/{environment_id}/archive` | 软归档，已绑定会话仍可用 | -| 删除 | `DELETE /environments/{environment_id}` | 硬删除，不可恢复 | - -## 典型调用流程 - -一次完整的任务执行分五步，详见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)： - -1. **创建 Agent** — 定义模型与系统提示词，得到 `agent_xxx`（通常只创建一次，长期复用） -2. **创建 Environment** — 定义运行沙箱，得到 `env_xxx`（通常只创建一次，长期复用） -3. **创建 Session** — 绑定 Agent 与 Environment，得到 `sesn_xxx` -4. **发送 Event** — 向 Session 提交用户消息，触发 Agent 进入 `running` -5. **订阅 SSE** — 流式接收执行结果，直至 Session 回到 `idle` - -## SDK 支持 - -Managed Agents 模块通过 [DashScope SDK](../concepts/dashscope-sdk.md) 接入，版本要求： - -| 语言 | 包名 | 最低版本 | -|------|------|----------| -| Python | `dashscope` | v1.26.2 | -| Java | `dashscope-sdk-java` | v2.22.24 | - -## 分页约定 - -列表端点统一支持分页参数：`limit`（默认 20，最大 100）和 `page`（首次不传，后续传上一次响应的 `next_page`）。响应不含 `next_page` 表示已是末页。 - -## 关键设计要点 - -- **版本锁定**：Agent 更新采用乐观锁（请求体带 `version`，不一致返回 409）；会话创建时锁定 Agent 版本，Skill 挂载锁定具体版本号，确保运行中会话不受配置变更影响。 -- **软归档 vs 硬删除**：Agent、Session、Environment 均支持软归档（`archived_at` 标记），归档后不影响已有会话；File 和 Environment 支持硬删除（不可恢复）。 -- **安全扫描**：Skill 和 File 上传后均需经过安全扫描/审核，仅通过后方可使用。 -- **沙箱隔离**：文件挂载到会话时服务端做内部拷贝，生成独立 `file_id`，仅对应会话可见。 +Managed Agents API 是百炼平台提供的智能体托管运行时服务，负责会话生命周期管理、沙箱环境调度、工具执行协调与事件流分发。开发者通过 REST 或 SDK 调用，可快速构建具备工具调用、多轮交互与状态感知能力的智能体应用。所有资源均按工作空间隔离，需配合 API Key 与地域化 Endpoint 使用。 + +## 支持的模型与功能 + +- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中的 `model.id` 字段示例）；不支持自定义模型或外部模型接入。 +- **核心功能模块**： + - **Agent**：封装模型、系统提示词、工具集与技能，支持版本化管理与软归档； + - **Environment**：定义沙箱类型（如 `"cloud"`）、预装依赖与网络策略，独立于 Agent 生命周期； + - **Session**：绑定 Agent 快照与 Environment 实例，驱动 `idle → running → idle/terminated` 状态机； + - **Event**：支持用户消息、工具回填、中断指令等原子事件，提供 SSE 流式订阅； + - **File**：上传后经安全审核（`checking` → `available`），可用于消息内容或挂载至沙箱； + - **Skill**：以 zip 包形式封装工具组合，上传后需通过安全扫描（`checking` → `active`）方可挂载，挂载时必须指定具体版本号。 + +> **注意**：文档 2 的快速开始示例中使用 `model: "qwen-plus"` 作为字符串传入，而文档 1 的 API 总览中 `model` 字段结构为 `{"id": "qwen-plus"}`。实际请求体应严格遵循文档 1 的嵌套对象格式，否则将返回 400 错误 —— 此处以 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 为准。 + +## 关键参数 + +| 参数 | 位置 | 类型 | 必填 | 说明 | +|------|------|------|------|------| +| `Authorization` | Header | string | 是 | `Bearer `，从控制台获取并配置为环境变量 | +| `workspace_id` | Endpoint path | string | 是 | 工作空间 ID（如 `ws_xxxxxxxxxxxx`），见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) | +| `region` | Endpoint path | string | 是 | 当前仅支持 `cn-beijing` | +| `agent.id` | Session 创建体 | string | 是 | Agent ID，创建时生成；会话锁定其 `version` 快照 | +| `environment_id` | Session 创建体 | string | 是 | Environment ID，会话绑定其快照 | +| `input` | `/sessions/{id}/events` 请求体 | array | 是 | 符合 ChatML 格式的 message 数组，如 `[{"role":"user","type":"message","content":[{"type":"text","text":"..."}]}]` | +| `limit` / `page` | 列表端点 Query | integer | 否 | 分页参数，默认 `limit=20`，最大 `100`；响应含 `next_page` 表示可继续翻页 | + +## 使用方式 + +1. **初始化**：导出 `DASHSCOPE_API_KEY` 与 `AGENTSTUDIO_URL`（形如 `https://.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio`）； +2. **资源准备**： + - 创建 Agent（`POST /agents`），指定 `model.id`、`system` 等； + - 创建 Environment（`POST /environments`），配置 `config.type`（如 `"cloud"`）； +3. **会话启动**： + - 创建 Session（`POST /sessions`），传入 `agent` 和 `environment_id`； + - 发送 Event（`POST /sessions/{id}/events`）触发执行； +4. **结果消费**： + - 订阅 SSE 事件流（`GET /sessions/{id}/events/stream`），监听 `session_status` 变更及 `message` 内容； + - 或轮询事件历史（`GET /sessions/{id}/events`）。 + +SDK 使用需满足最低版本要求：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24 —— 具体安装与初始化方式参见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)。 + +## 限制和注意事项 + +- **配额限制**：单文件直传上限 **20 MB**，工作空间总容量上限 **100 GB**，文件保留期 **30 天**（[File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； +- **版本与快照**：Agent 更新采用全量替换 + 乐观锁（需带 `version` 字段），会话创建即锁定 Agent 与 Environment 快照，后续更新不影响运行中会话； +- **状态终态**：Session 归档（`POST /sessions/{id}/archive`）使其进入 `terminated` 终态，不可恢复；删除（`DELETE /sessions/{id}`）则彻底清除事件历史； +- **安全约束**：Skill 上传后必须通过安全扫描（`status: active`）才可挂载；File 仅 `available` 状态可被引用或挂载； +- **错误排查**：所有响应携带 `x-request-id`，提工单时务必提供该值以便定位问题。 ## 来源文档 -- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) -- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) - [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) +- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) +- [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) - [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) -- [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) +- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) - [File](../../raw/application-api-reference/managed-agents-api/files-api.md) -- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - - - - - - - - - +- [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md index f2c1d6e0..981d9381 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md @@ -1,33 +1,44 @@ # model production -百炼平台提供模型生产相关的 API，覆盖从模型微调训练到部署上线的完整流程。开发者可以通过[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)接口定制专属模型，再通过[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)接口将其发布为在线推理服务。 +model production 是百炼平台中用于模型定制化与服务化的关键能力集合，涵盖模型微调、部署及生命周期管理。它面向开发者提供标准化 API 接口，支持从训练到上线的端到端流程。所有操作均通过 RESTful API 或 SDK 调用，需配合百炼平台认证体系使用。 -## 模型调优 +## 支持的模型/功能 -模型调优（Fine-tuning）允许开发者通过微调训练定制专属模型，以适配特定业务场景。调优流程通常包括： +- **微调（Fine-tuning）**：支持基于预训练大语言模型（如 Qwen 系列）进行监督微调，适配下游任务（如指令遵循、领域问答）。输入为结构化 JSONL 格式数据集，支持 LoRA 等高效微调方法。 +- **部署（Deployment）**：支持将微调完成的模型或直接导入的兼容格式模型（如 GGUF、ONNX 导出模型）发布为高可用推理服务，自动分配 endpoint 并支持流量路由与扩缩容。 +- **模型版本管理**：每个微调任务生成唯一 `job_id`，对应产出模型可被多次部署；部署实例绑定 `model_id` 与 `version_id`，确保可追溯性。 +详见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 和 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md)。 -- **创建调优任务**：指定基础模型、训练数据集和超参数，提交微调训练任务 -- **查询任务状态**：轮询或监听训练任务进度，获取训练指标 -- **管理调优产物**：训练完成后获取调优模型，用于后续部署或评估 +## 关键参数 -详细的接口定义和参数说明请参考[模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)文档。 +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `model` | string | 是 | 基座模型 ID（如 `qwen2.5-7b`），必须在 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 支持列表中 | +| `training_file` | string | 是（微调） | OSS 或 S3 URI，指向训练数据集（JSONL 格式） | +| `endpoint_name` | string | 是（部署） | 全局唯一标识符，长度 3–63 字符，仅含小写字母、数字和连字符 | +| `instance_type` | string | 否 | 默认 `gpu-a10`；部署时可选 `gpu-v100`、`cpu-small`（仅限测试） | -## 模型部署 +> **注意**：文档 2 中提及“支持导入 ONNX 模型”，但当前版本（v2.4+）实际仅支持 ONNX 的 *推理兼容验证*，不支持 ONNX 模型直接部署；完整支持计划见 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 的“未来特性”章节（该内容已过时，以控制台 API 文档为准）。 -模型部署将微调或导入的模型发布为在线推理服务，使其可通过 API 调用进行推理。部署流程通常包括： +## 使用方式 -- **创建部署**：选择调优完成的模型或外部导入的模型，配置推理资源和服务参数 -- **管理部署实例**：查看部署状态、调整资源配置、启停服务 -- **调用推理服务**：部署成功后，通过标准 API 端点发送推理请求 +1. **微调流程**： + - POST `/api/v1/fine_tuning_jobs`，携带 `model`、`training_file` 等参数； + - 轮询 `GET /api/v1/fine_tuning_jobs/{job_id}` 直至 `status == "succeeded"`； + - 提取响应中的 `fine_tuned_model_id` 用于后续部署。 -详细的接口定义和参数说明请参考[模型部署](../../raw/model-api-reference/model-production/deployments-api.md)文档。 +2. **部署流程**： + - POST `/api/v1/deployments`，传入 `model_id`（来自微调结果）、`endpoint_name`、`instance_type`； + - 部署成功后，`endpoint_url` 可立即用于 `POST /v1/chat/completions` 请求。 -## 典型工作流 +完整示例代码与错误码说明参见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)。 -1. 准备训练数据集 -2. 通过调优 API 提交微调训务，等待训练完成 -3. 通过部署 API 将调优产物部署为在线服务 -4. 调用部署后的模型端点进行推理 +## 限制和注意事项 + +- 单次微调最大训练时长为 72 小时，超时任务自动终止且不计费； +- 每个账号默认最多同时运行 3 个微调任务、5 个部署实例，配额可通过工单申请提升； +- 微调数据集须经敏感信息过滤（如 PII），平台不承担未脱敏数据导致的合规风险； +- 部署实例启动后需 2–5 分钟完成初始化，期间 `health_check` 返回 `503`，请实现重试逻辑。 ## 来源文档 @@ -35,12 +46,3 @@ - [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md index b2130305..f857ae5a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md @@ -1,111 +1,54 @@ # [more](more.md) about models -阿里云百炼在模型调用的核心流程之外，提供了一系列辅助能力，涵盖安全认证、异步任务管理、文件上传、子[业务空间](../concepts/workspace.md)隔离以及高并发场景下的连接优化。本文汇总这些进阶用法的关键要点，帮助开发者在生产环境中安全、高效地使用模型服务。 +百炼平台提供多种模型调用机制与配套能力，涵盖同步/异步任务处理、多业务空间隔离、文件上传、连接优化等关键场景。本文面向开发者，系统梳理模型服务的核心能力、参数配置、使用方式及限制条件，帮助您高效、安全地集成模型能力。 -## 临时 [API Key](../concepts/api-key.md) +## 支持的模型/功能 -在浏览器或移动端等不可信环境中，直接暴露永久 [API Key](../concepts/api-key.md) 存在安全风险。百炼提供了[生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)的接口，通过后端服务生成有限时效的临时凭证。 +百炼支持两类主要模型调用模式： +- **同步模型**（如 `qwen-plus`、`qwen-max`）：适用于文本生成类请求，响应快、链路简单，直接返回结果； +- **异步模型**（如图像生成 `wanx2.1-t2i-turbo`、视频生成 `wanx2.1-kf2v-plus`、语音识别 `paraformer-16k-1`）：因处理耗时长，需通过任务 ID 分两步完成（提交 → 查询），并支持批量状态查询与取消 [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)。 -**请求方式**： +此外，部分[多模态](../concepts/multi-modal.md)模型（如 `qwen-vl-plus`）需传入文件 URL，平台提供免费临时 OSS 存储能力，上传后获得 `oss://` 格式 URL（有效期 48 小时）[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 +> **注意**：文档 3 中提到“文生图、文生视频提供了 SDK，SDK 已实现轮询”，但文档 2 明确指出异步任务接口本身**不内置轮询逻辑**，SDK 实现属封装层行为；实际调用仍需按文档 2 的接口规范自行轮询或接入事件通知。 -``` -POST https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds= -``` +## 关键参数 -- `expire_in_seconds`：有效期，范围 [1, 1800] 秒，默认 60 秒。 -- 返回的 `token` 字段即为临时 [API Key](../concepts/api-key.md)，`expires_at` 为 UNIX 过期时间戳。 -- 临时 API Key 继承生成它的永久 API Key 的全部权限，到期后自动失效，无法手动删除。 +| 参数 | 说明 | 典型值/范围 | 注意事项 | +|------|------|-------------|----------| +| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒 | 默认 60 秒，超时自动失效，不可手动删除 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | +| `task_id` | 异步任务唯一标识 | UUID 字符串 | 必须用于查询或取消任务；任务完成后保留约 24 小时（具体以各模型文档为准） | +| `model_name` | 模型名称 | 如 `qwen-plus`, `wanx2.1-t2i-turbo` | 文件上传时必须指定且与后续调用模型一致；子业务空间调用需确保该空间已授权该模型 | +| `X-DashScope-OssResourceResolve` | 启用 OSS 资源解析 | `enable` | 使用 `oss://` URL 时**必须显式设置**此 Header，否则调用失败 | +| 连接池参数（Java/Python） | 控制 HTTP 连接复用 | 如 `connectionPoolSize=256`, `limit=100` | 高并发场景下需调优，避免阻塞或资源浪费 [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | -> **注意**：各地域的 API Key 不同，新加坡地域需将 Endpoint 中的 WorkspaceId 替换为实际值。 +## 使用方式 -## 异步任务管理 +### 1. 调用环境准备 +- 所有调用均需有效 API Key，并推荐配置为环境变量 `DASHSCOPE_API_KEY`； +- 子业务空间调用必须使用**该空间专属的 API Key**，且需提前在控制台为其授予对应模型权限 [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)； +- 临时 API Key 适用于前端/移动端等不可信环境，由后端安全生成并透传 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)。 -图像生成、视频生成等耗时较长的模型采用[异步调用](../concepts/async-invocation.md)机制。百炼提供了三个通用的[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)： +### 2. 异步任务处理 +- **轮询模式**：调用 `/api/v1/tasks/{task_id}` 查询状态（20 QPS 限流），支持 `PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED` 等状态判断； +- **事件驱动模式**：通过事件总线（EventBridge）配置 HTTP 回调或 RocketMQ 接收 `dashscope:System:AsyncTaskFinish` 事件，避免轮询 [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)； +- **批量操作**：使用 `/api/v1/tasks/` 接口按时间、状态、模型名等条件批量查询任务；仅 `PENDING` 状态任务可取消。 -### 查询单个任务 +### 3. [多模态](../concepts/multi-modal.md)文件处理 +- 上传前调用 `GET /api/v1/uploads?action=getPolicy&model={model_name}` 获取凭证； +- 使用凭证直传 OSS，获得 `oss://` URL； +- 在模型请求中传入该 URL，并在 Header 中添加 `X-DashScope-OssResourceResolve: enable`。 -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id} -``` +### 4. SDK 连接优化 +- **Java**：通过 `Constants.connectionConfigurations` 全局配置连接池参数（如 `connectionPoolSize`, `readTimeout`）； +- **Python**：同步调用传入 `requests.Session`，异步调用传入 `aiohttp.ClientSession`，复用底层 TCP 连接。 -返回 `output.task_status` 标识任务状态：`PENDING` / `RUNNING` / `SUCCEEDED` / `FAILED` / `UNKNOWN`。已完成任务通常保留 24 小时后自动清理。 +## 限制和注意事项 -### 批量查询任务状态 - -``` -GET https://dashscope.aliyuncs.com/api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx -``` - -支持按时间范围、模型名称、任务状态等条件组合过滤，单次查询时间跨度不超过 24 小时。 - -### 取消任务 - -``` -POST https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}/cancel -``` - -仅支持取消 `PENDING` 状态的任务，已开始处理的任务无法取消。 - -以上三个接口的流量限制均为 20 QPS（主账号维度）。 - -## 异步任务完成通知 - -频繁轮询任务结果接口会浪费资源且可能触发限流。百炼已接入阿里云事件总线 EventBridge，支持在任务完成后主动推送通知。详见[通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)。 - -两种接入方案： - -| 方案 | 适用场景 | 特点 | -|------|----------|------| -| HTTP 回调 URL | 通用场景 | 需要公网或 VPC 可达的 HTTP 接口，接入较简单 | -| RocketMQ | 消息可靠性要求高的场景 | 保证无丢失、支持失败重试，需额外开通 RocketMQ 实例 | - -事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`。事件体中 `data.task_status` 和 `data.task_id` 是关键字段，收到通知后只需调用一次查询接口即可获取结果。 - -## 子[业务空间](../concepts/workspace.md)的模型调用 - -默认[业务空间](../concepts/workspace.md)的 API Key 拥有所有模型的调用权限。如需按业务线隔离权限或分账，可使用[子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)。 - -**使用要点**： - -- 必须使用子[业务空间](../concepts/workspace.md)自身的 API Key 进行调用。 -- 调用标准模型（如 `qwen-plus`）前，需为该空间设置模型调用权限。 -- 调用在百炼上调优并部署的模型无需额外授权，但仅能由其所在空间的 API Key 调用。 -- 支持 OpenAI 兼容方式和 DashScope 方式调用，但调优后模型仅支持 DashScope 方式。 - -## 上传本地文件获取临时 URL - -[多模态](../concepts/multimodal.md)、图像、视频、音频模型调用时通常需要传入文件 URL。百炼提供了免费的临时存储空间，支持上传本地文件并获取 `oss://` 前缀的临时 URL，详见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 - -**关键限制**： - -- 文件有效期 48 小时，过期自动清理。 -- 上传时必须指定模型名称，且与后续调用的模型一致，不同模型无法共享文件。 -- 上传与调用的 API Key 必须属于同一阿里云主账号。 -- 单文件不超过 1GB，上传凭证接口限流 100 QPS。 -- 使用 `oss://` 形式的 URL 调用模型时，HTTP 请求头中必须添加 `X-DashScope-OssResourceResolve: enable`。 - -> **注意**：临时 URL 不适用于生产环境。生产环境建议使用阿里云 OSS 等稳定存储方案。 - -上传方式包括 Python/Java 代码上传和 DashScope 命令行工具（`dashscope oss.upload`）上传。 - -## [DashScope SDK](../concepts/dashscope-sdk.md) 连接复用配置 - -高并发场景下，频繁创建连接会导致超时和资源消耗过大。[DashScope SDK](../concepts/dashscope-sdk.md) 支持连接复用来优化性能，详见[DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md)。 - -**Java SDK** 内置连接池，默认启用，核心配置参数： - -| 参数 | 默认值 | 说明 | -|------|--------|------| -| `connectionPoolSize` | 32 | 连接池最大连接数 | -| `maximumAsyncRequests` | 32 | 最大并发请求数（需 <= 连接数） | -| `connectTimeout` | 120s | 建立连接超时 | -| `readTimeout` | 300s | 读取数据超时 | -| `connectionIdleTimeout` | 300s | 空闲连接超时 | - -**Python SDK** 通过传入自定义 Session 实现连接复用： - -- 异步场景：使用 `aiohttp.ClientSession` 配合 `aiohttp.TCPConnector`，可配置 `limit`（总连接数，默认 100）和 `limit_per_host`（单主机连接数）。 -- 同步场景：使用 `requests.Session`，同一 Session 内多次请求自动复用 TCP 连接。 +- **临时存储限制**：`oss://` URL 有效期严格为 **48 小时**，过期即失效；文件大小上限 **1GB**；上传接口限流 **100 QPS（主账号+模型维度）**，**严禁用于生产环境或压测**，生产环境应使用阿里云 OSS [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)； +- **地域隔离**：API Key、Endpoint、临时 [Token](../concepts/token.md) 均按地域（北京/新加坡/弗吉尼亚）独立，跨地域调用将失败； +- **权限继承**：临时 API Key 继承其生成者 API Key 的全部权限（含模型/知识库访问限制），无额外管控能力 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)； +- **子空间约束**：在子业务空间调优部署的模型**仅能被该空间的 API Key 调用**，且不支持 OpenAI 兼容方式；标准模型调用则需显式授权； +- **异步任务清理**：已完成任务数据约保留 24 小时，超时后无法查询，需及时获取结果。 ## 来源文档 @@ -117,12 +60,3 @@ POST https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}/cancel - [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md index 65fa010b..b623e0f3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md @@ -1,143 +1,82 @@ # [more](more.md) models -本页汇总百炼平台上除通义千问主对话模型之外的一组专用模型的 API 参考，涵盖法律、意图理解、深度研究、翻译、OCR、界面交互等场景。这些模型大多通过 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)或 [DashScope SDK](../concepts/dashscope-sdk.md) 调用，但各模型在地域、协议、请求参数和调用流程上存在差异，使用前需对照本文确认。 +百炼平台提供一系列面向垂直场景的专用模型，覆盖法律、翻译、深度研究、OCR、GUI自动化和意图理解等能力。这些模型基于通义千问系列基座模型精调或增强，支持通过 DashScope SDK 或 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)调用。所有模型均需配置业务空间专属域名以获得最佳性能与稳定性。 -## 支持的模型与功能 +## 支持的模型/功能 -| 模型名称 | 用途 | 调用方式 | 地域 | -| --- | --- | --- | --- | -| `farui-plus` | 法律行业大模型，支持法律咨询、文书生成、案情分析 | [DashScope SDK](../concepts/dashscope-sdk.md)（Python/Java） | 默认地域 | -| `tongyi-intent-detect-v3` | 意图理解，同时输出意图与[函数调用](../concepts/function-calling.md)信息 | OpenAI 兼容 / DashScope | 默认地域 | -| `qwen-deep-research` | 深度研究，两阶段（反问确认 + 深入研究） | 仅 Python [DashScope SDK](../concepts/dashscope-sdk.md)，仅华北2（北京） | 华北2（北京） | -| `qwen-mt-plus` | 翻译，支持术语干预、翻译记忆、领域提示 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `qwen3.5-ocr` | 图像文字提取（OCR）与结构化字段抽取 | OpenAI 兼容 / DashScope | 北京 / 新加坡 / 美国（弗吉尼亚） | -| `gui-plus-2026-02-26` | 界面交互专用模型，通过 `computer_use` 工具操控桌面 GUI | OpenAI 兼容 | 华北2（北京） | +- **通义法睿（`farui-plus`）**：法律行业专用大模型，支持法律咨询、文书生成、案情分析、合同审查等，详见 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md)。 +- **Qwen-MT（`qwen-mt-plus`）**：机器翻译模型，支持术语干预、翻译记忆、领域提示等高级功能，兼容 OpenAI 接口，详见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md)。 +- **Qwen-Deep-Research（`qwen-deep-research`）**：支持两阶段交互式深度研究（反问确认 + 网络检索增强），**仅限华北2（北京）地域且仅支持 Python DashScope SDK**，不支持 Java SDK 或 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，详见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)。 +- **Qwen-OCR（`qwen3.5-ocr`）**：[多模态](../concepts/multi-modal.md) OCR 模型，支持图像输入与结构化文本提取（如车票信息），支持流式与非[流式输出](../concepts/streaming-output.md)，详见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md)。 +- **GUI-Plus（`gui-plus-2026-02-26`）**：界面交互专用模型，可调用 `computer_use` 工具执行鼠标/键盘操作并解析截图，适用于自动化 GUI 测试与桌面任务，详见 [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md)。 +- **意图理解（`tongyi-intent-detect-v3`）**：毫秒级意图识别模型，支持两种模式：① 输出结构化工具调用（需 `INTENT_MODE` system [prompt](../guides/prompt.md)）；② 仅输出预定义意图标签（支持单 [Token](../concepts/token.md) 响应优化），详见 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 -> **注意**：`qwen-deep-research` 当前**仅支持通过 Python [DashScope SDK](../concepts/dashscope-sdk.md) 调用，暂不支持 Java SDK 与 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**，且仅支持华北2（北京）地域。如需使用，必须使用该地域的 [API Key](../concepts/api-key.md)。详见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)。 +> **注意**：文档 2 和文档 4 均重复列出新加坡/美国地域的 `base_url` 配置两次，属冗余描述，实际使用时按地域选择唯一正确地址即可；文档 5 中 GUI-Plus 的 `vl_high_resolution_images` 参数在文档 4（Qwen-OCR）中亦有相同用法，但未在文档 5 明确说明其作用，建议开发者参考 Qwen-OCR 文档中关于 `min_pixels`/`max_pixels` 的图像预处理逻辑进行适配。 ## 关键参数 -### 通用参数 - -- `model`（string，必选）：模型名称，取值见上表。 -- `messages`（array，必选）：对话消息列表，按顺序排列。 -- `stream`（bool，可选）：是否[流式输出](../concepts/streaming-output.md)。`qwen-deep-research` 第一步反问阶段需设为 `true`。 - -### Qwen-MT 专属参数（`translation_options`） - -Qwen-MT 通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制翻译行为，详见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md)： - -- `source_lang`（string）：源语言，可填 `auto` 自动识别。 -- `target_lang`（string）：目标语言。 -- `terms`（array）：术语干预，元素为 `{"source": "...", "target": "..."}`，强制指定术语译法。 -- `tm_list`（array）：翻译记忆，元素为 `{"source": "...", "target": "..."}`，提供历史译文供模型参考，提升一致性。 -- `domain_prompt`（string）：领域提示，向模型注入领域上下文（如 IT、金融）。 - -### Qwen-OCR 专属参数 - -Qwen-OCR 的 `messages.content` 为[多模态](../concepts/multimodal.md)数组，图像元素支持： - -- `min_pixels`（int）：图像最小像素阈值，小于该值会放大，示例 `32 * 32 * 3`（即 3072）。 -- `max_pixels`（int）：图像最大像素阈值，超过该值会缩小，示例 `32 * 32 * 8192`（即 8388608）。 -- `text` 段：可传入自定义 Prompt；未传入时使用默认 Prompt `Please output only the text content from the image without any additional descriptions or formatting.`。 - -详见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md)。 - -### GUI-Plus 专属参数 - -- `vl_high_resolution_images`（bool）：通过 `extra_body` 传入，启用高分辨率图像处理。 -- `computer_use` 工具：在 system [prompt](../guides/prompt.md) 中定义，`action` 枚举包括 `key`、`type`、`mouse_move`、`left_click`、`left_click_drag`、`right_click`、`middle_click`、`double_click`、`triple_click`、`scroll`、`hscroll`、`wait`、`terminate`、`answer`、`interact`。屏幕分辨率固定为 1000x1000。 +| 参数 | 类型 | 说明 | 必填 | +|------|------|------|------| +| `model` | string | 模型名称，如 `"farui-plus"`、`"qwen-mt-plus"` 等 | ✅ | +| `messages` | array | 对话消息列表，含 `role`（`user`/`system`/`assistant`）与 `content`；OCR/GUI-Plus 支持 `image_url` 类型内容 | ✅ | +| `result_format` / `response_format` | string | DashScope SDK 使用 `result_format='message'`；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)默认为 `chat.completions` 格式 | ❌（默认） | +| `stream` | boolean | 启用[流式输出](../concepts/streaming-output.md)（需配合 `stream_options={"include_usage": true}` 获取 token 统计） | ❌（默认 false） | +| `extra_body` | object | OpenAI 兼容接口扩展字段：
• `qwen-mt-plus`: `{"translation_options": {...}}`
• `gui-plus-*`: `{"vl_high_resolution_images": true}`
• `tongyi-intent-detect-v3`: 无特殊字段，依赖 system [prompt](../guides/prompt.md) 控制行为 | ❌（按需） | +| `output_format` | string | 仅 `qwen-deep-research` 支持：`"model_detailed_report"`（默认）或 `"model_summary_report"` | ❌（默认） | ## 使用方式 -### 地域与域名 - -多数模型推荐使用[业务空间](../concepts/workspace.md)专属域名以获得更好性能与稳定性： - -- 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` -- 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -其中 `{WorkspaceId}` 为[业务空间](../concepts/workspace.md) ID，可在百炼控制台「[业务空间](../concepts/workspace.md)详情」页面查看。现有域名（如 `https://dashscope.aliyuncs.com`）仍可正常使用。 - -> **注意**：各地域的 [API Key](../concepts/api-key.md) 不同，切换地域时需同时更换 [API Key](../concepts/api-key.md) 与 `base_url`。 - -### 前提条件 - -- 已开通百炼服务并[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，建议配置到环境变量 `DASHSCOPE_API_KEY`。 -- 已[安装 DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)（Python/Java）或 OpenAI SDK（Python/Node.js，需 Node.js v18+ 且在 ES Module 环境运行）。 - -### 调用示例 - -**通义法睿单轮对话**（DashScope Python SDK）： - -```python -import dashscope -messages = [{'role': 'system', 'content': 'You are a helpful assistant.'}, - {'role': 'user', 'content': '我哥欠我10000块钱，给我生成起诉书。'}] -response = dashscope.Generation.call(model="farui-plus", messages=messages) -``` - -**Qwen-MT 基础翻译**（OpenAI 兼容）： - -```python -from openai import OpenAI -client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1") -completion = client.chat.completions.create( - model="qwen-mt-plus", - messages=[{"role": "user", "content": "我看到这个视频后没有笑"}], - extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English"}}) -``` - -**Qwen-OCR 字段抽取**（OpenAI 兼容）： - -```python -completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[{"role": "user", "content": [ - {"type": "image_url", "image_url": {"url": "https://..."}, - "min_pixels": 32*32*3, "max_pixels": 32*32*8192}, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站..."}]}]) -``` - -**GUI-Plus 界面交互**：在 system [prompt](../guides/prompt.md) 中注入 `computer_use` 工具定义，user 消息中传入截图与指令（如「帮我打开浏览器」），模型返回工具调用 JSON，由调用方执行并截图回传，循环直到 `action=terminate`。 +1. **环境准备** + - 获取并配置 API Key 到环境变量 `DASHSCOPE_API_KEY`（[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)）； + - 安装对应 SDK：`pip install dashscope`（DashScope）或 `pip install openai`（OpenAI 兼容）； + - **必须配置业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），旧域名（`dashscope.aliyuncs.com`）虽仍可用，但性能与稳定性较低。 + +2. **调用示例（通用流程）** + ```python + # DashScope SDK（以 farui-plus 为例） + import dashscope + dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" + response = dashscope.Generation.call( + model="farui-plus", + messages=[{"role": "user", "content": "生成一份离婚协议书"}], + result_format="message" + ) + ``` + + ```python + # OpenAI 兼容接口（以 qwen-mt-plus 为例） + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + completion = client.chat.completions.create( + model="qwen-mt-plus", + messages=[{"role": "user", "content": "我看到这个视频后没有笑"}], + extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English"}} + ) + ``` + +3. **特殊模型注意事项** + - `qwen-deep-research`：必须分两步调用（先反问确认，再传入 assistant 回复 + user 补充指令）； + - `gui-plus-*`：system [prompt](../guides/prompt.md) 必须包含完整 `` 定义与 `Response format` 规则； + - `tongyi-intent-detect-v3`：意图识别模式由 system prompt 决定——含 `INTENT_MODE` 则输出 ``/ 块；否则仅输出纯标签字符串。 ## 限制和注意事项 -- **限流**：各模型有独立的限流条件，`farui-plus` 等模型限流详见[限流](https://help.aliyun.com/zh/model-studio/rate-limit)。 -- **上下文与[计费](../concepts/billing.md)**：`farui-plus` 上下文 12k、最大输入 12k、最大输出 2k，输入成本 20 元/百万 [Token](../concepts/token.md)；`tongyi-intent-detect-v3` 上下文 8,192、最大输入 8,192、最大输出 1,024，输入 0.4 元、输出 1 元/百万 [Token](../concepts/token.md)，开通后 90 天内赠送 100 万 [Token](../concepts/token.md) 免费额度。 -- **协议限制**：`qwen-deep-research` 仅支持 Python [DashScope SDK](../concepts/dashscope-sdk.md)；`gui-plus-2026-02-26` 仅在华北2（北京）地域提供。 -- **SDK 线程安全**：DashScope Java SDK 的 `Generation` 等对象非线程安全，需自行管理同步机制或及时关闭进程。 -- **图像像素阈值**：Qwen-OCR 的 `min_pixels`/`max_pixels` 影响识别精度与耗时，过小会放大、过大会缩小，建议按示例值设置。 -- **翻译记忆与术语**：Qwen-MT 的 `terms` 强制覆盖译法，`tm_list` 仅作参考；两者可叠加使用以提升专业领域一致性。 +- **地域限制**：`qwen-deep-research` 仅支持华北2（北京）地域；其他模型（如 `qwen-mt-plus`、`qwen3.5-ocr`、`gui-plus-*`）在华北2、新加坡、美国（弗吉尼亚）三地均可用，但需使用对应地域的 `base_url` 和独立 API Key。 +- **SDK 支持差异**：`qwen-deep-research` **不支持 Java SDK 与 OpenAI 兼容接口**（文档明确说明），仅支持 Python DashScope SDK；其余模型均支持两种调用方式。 +- **输入格式约束**：OCR 与 GUI-Plus 模型要求 `messages[0].content` 为数组，内含 `image_url` 和 `text` 对象；普通文本模型（如 `farui-plus`、`tongyi-intent-detect-v3`）则接受字符串 `content`。 +- **成本与限流**：各模型按输入/输出 token 计费（如 `farui-plus` 输入 20元/百万 token），具体见各模型文档表格；全局限流策略参见 [限流](https://help.aliyun.com/zh/model-studio/rate-limit)，未在原始文档中统一说明。 +- **安全实践**：强烈建议将 `DASHSCOPE_API_KEY` 配置为环境变量，避免硬编码或日志泄露（[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)）。 ## 来源文档 - [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) -- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) -- [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) +- [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) - [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) - - - - - - - - - - - - - - - - - - - +- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more.md b/skills/bailian-docs-llm-wiki/wiki/api/more.md index 43d4533d..e43728f1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more.md @@ -1,112 +1,53 @@ # more -本主题汇总百炼平台在应用接入与数据检索中常用的几项辅助能力：临时 [API Key](../concepts/api-key.md) 生成、服务关联角色（SLR）管理，以及[知识库](../concepts/knowledge-base.md) Retrieve 接口的 SearchFilters 过滤语法。它们分别覆盖安全鉴权、跨云服务授权与结构化数据检索过滤三个场景，详细说明可参见 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)、[服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) 与 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +`more` 是百炼平台面向高级用例提供的扩展能力集合，涵盖临时凭证管理、服务权限委托和知识库精细化检索三大核心方向。它不构成独立 API 服务，而是作为模型调用、工作流编排和 RAG 场景的支撑性机制，需结合具体功能模块（如 `Retrieve`、函数计算节点、安全存储空间等）协同使用。开发者应根据实际场景选择对应能力，并严格遵循权限最小化原则。 -## 生成临时 [API Key](../concepts/api-key.md) +## 支持的模型/功能 -在浏览器、移动 App 等不可信环境中调用模型服务时，应通过后端服务生成临时 [API Key](../concepts/api-key.md)，避免永久 [API Key](../concepts/api-key.md) 泄露。临时 [API Key](../concepts/api-key.md) 继承生成它的永久 [API Key](../concepts/api-key.md) 的全部权限（包括对特定模型或[知识库](../concepts/knowledge-base.md)的访问限制），到期后自动失效，无法提前删除。 +`more` 本身不提供模型推理能力，但为以下关键功能提供底层支持： -**前提条件**：在百炼密钥管理页面（北京 / 新加坡 / 弗吉尼亚）创建永久 [API Key](../concepts/api-key.md)，并将其配置为环境变量 `DASHSCOPE_API_KEY`。 +- **临时 API Key 生成**：用于在浏览器、移动端等不可信环境安全调用模型服务（如 `qwen-max`、`qwen-plus` 等所有支持 DashScope 协议的模型），避免永久密钥泄露 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 +- **服务关联角色（SLR）**：为百炼工作流、数据管理、安全存储空间、知识库、用量监控等模块自动创建并托管 RAM 角色，实现对 FC、OSS、ADB-PG、MNS、SLS 等云服务的安全访问授权 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- **知识库 SearchFilters**：在 `Retrieve` 接口调用中对语义检索结果进行结构化过滤，支持单值、多值、范围、模糊及标签查询，显著提升结构化数据（如员工表、产品目录）的召回精度 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 -**请求**： +> **注意**：文档 2 中列出的 `AliyunServiceRoleForSFMAccessFC` 权限仅包含 `fc:ListFunctions` 和 `fc:InvokeFunction`，但实际工作流调用 FC 函数可能还需 `fc:GetFunction` 等元数据权限；建议以控制台实际授予策略为准，而非仅依赖文档描述。 -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` +## 关键参数 -**关键参数**： +| 参数/字段 | 所属能力 | 类型 | 说明 | 示例 | +|-----------|----------|------|------|------| +| `expire_in_seconds` | 临时 API Key | integer | TTL 有效期，单位秒，取值范围 `[1, 1800]` | `1800`（30 分钟） | +| `searchFilters` | 知识库检索 | array of object | 过滤条件数组，每个元素为一个子分组（AND 语义），支持 `{"字段名": "值"}` 或高级语法如 `{"年龄": {"gte": 20, "lte": 30}}` | `[{"姓名": "张三"}, {"岗位": "技术员"}]` | +| `token` | 临时 API Key 响应 | string | 生成的短期凭证，格式为 `st-***` | `st-9a8b7c6d...` | +| `expires_at` | 临时 API Key 响应 | number | UNIX 时间戳，表示过期时间 | `1744080369` | -| 参数 | 说明 | -| --- | --- | -| `expire_in_seconds` | 临时 Key 有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 | +## 使用方式 -**正常响应**： +### 临时 API Key +1. 在后端服务中配置永久 `DASHSCOPE_API_KEY` 环境变量； +2. 向 `https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800` 发起 POST 请求（北京地域）或对应地域 Endpoint； +3. 将响应中的 `token` 作为 `Authorization: Bearer ` 用于后续模型调用。 -```json -{ - "token": "st-****", - "expires_at": 1744080369 -} -``` +### 服务关联角色 +- **无需手动创建**：当首次在控制台启用对应功能（如添加函数计算节点、配置 OSS 数据源）时，系统自动创建 SLR； +- **权限验证**：可在 [RAM 控制台](https://ram.console.aliyun.com/) 查看角色及绑定策略； +- **删除前提**：必须先解除该角色所依赖的所有业务配置（如删除函数计算节点、断开 OSS 连接等），否则删除失败。 -`token` 为生成的临时 [API Key](../concepts/api-key.md)，`expires_at` 为过期 UNIX 时间戳（秒）。错误响应包含 `code`、`message`、`request_id` 三段，常见如 `InvalidApiKey`。 +### SearchFilters +- 在 `RetrieveRequest` 请求体中直接传入 `searchFilters` 字段； +- 每个子分组内支持多种查询语法： + - 单值：`{"姓名": "张三"}` + - 范围：`{"年龄": {"gte": 25, "lte": 35}}` + - 模糊：`{"岗位": {"like": "技%员"}}` + - 多值（需 JSON 序列化）：`{"姓名": "[\"张三\",\"李四\"]"}` +- 注意：子分组间为 AND 关系，不可更改；标签查询仅适用于文档/音视频类知识库。 -> **注意**：各地域（北京 / 新加坡 / 弗吉尼亚）的 [API Key](../concepts/api-key.md) 不互通，请求时需使用对应地域的 Endpoint 与永久 Key。 +## 限制和注意事项 -## 服务关联角色（SLR） - -百炼在实现特定功能时，需通过服务关联角色（Service Linked Role, SLR）访问其他云服务（如 FC、OSS、ADB-PG、MNS、内容安全、SLS、CMS、DTS、CPFS 等）。当您首次在百炼中开通相关功能（如函数计算节点、OSS 数据导入、安全存储空间等）时，系统会**自动创建**对应的 SLR，无需手动创建。所有 SLR 可在 [RAM 控制台](https://ram.console.aliyun.com/) 的角色管理页面查看。 - -**主要 SLR 与用途**： - -| 服务关联角色 | 用途 | -| --- | --- | -| `AliyunServiceRoleForSFMAccessFC` | [工作流](../concepts/workflow.md)应用 / 流程编排访问函数计算（FC）资源 | -| `AliyunServiceRoleForSFMDataHubOSSImport` | 数据管理从 OSS 导入数据 | -| `AliyunServiceRoleForAccessOSS` | 安全存储空间访问 OSS | -| `AliyunServiceRoleForSFMAccessADB` | [知识库](../concepts/knowledge-base.md) / 安全存储空间访问 ADB-PG 实例 | -| `AliyunServiceRoleForSFMAccessingMNS` | 数据管理访问 MNS 队列中的 OSS 变更消息 | -| `AliyunServiceRoleForSFMTelemetry` | 用量监控与性能分析访问 OpenTelemetry 实例 | -| `AliyunServiceRoleForSFMAccessingCIP` | 百炼应用访问内容安全服务 | -| `AliyunServiceRoleForSFMAccessSLS` | 模型监控访问 SLS 资源 | -| `AliyunServiceRoleForSFMAccessCMS` | 模型监控访问 CMS 资源 | -| `AliyunServiceRoleForAccessCusOss` | 百炼平台托管操作用户 OSS 文件 | -| `AliyunServiceRoleForSFMConnectorAccessDTS` | 创建和管理 DTS 任务，从数据源接入数据 | -| `AliyunServiceRoleForSFMFineTuning` | [模型调优](../concepts/fine-tuning.md) / 数据管理访问 CPFS 与 OSS | - -每个 SLR 关联一个固定的系统策略（如 `AliyunServiceRolePolicyForSFMAccessFC`），策略中通过 RAM 条件（`ram:ServiceName`）限定只能由百炼服务使用，请勿修改或授予其他 RAM 身份。 - -**删除前注意事项**：删除 SLR 会导致对应功能不可用，须先清理依赖资源。例如删除 `AliyunServiceRoleForSFMAccessFC` 前须先删除所有已发布[工作流](../concepts/workflow.md)应用和流程中的函数计算节点并重新发布；删除 `AliyunServiceRoleForAccessOSS` 前须在安全存储空间中断开所有 OSS 连接；删除 `AliyunServiceRoleForSFMDataHubOSSImport` 前须确保没有进行中的 OSS 数据导入任务。具体删除步骤参见 [服务关联角色](https://help.aliyun.com/zh/ram/user-guide/service-linked-roles)。 - -## 知识库 SearchFilters - -在调用知识库 [Retrieve](https://help.aliyun.com/zh/model-studio/api-bailian-2023-12-29-retrieve) 接口时，若返回结果包含较多与 Query 无关的干扰信息（尤其适合结构化数据场景），可在请求体中传入 `searchFilters` 对语义检索结果做进一步过滤。 - -**效果对比**：未传入 `searchFilters` 时，Retrieve 可能返回多条低相关切片（如查询「张三」却返回李四、王五）；传入后可仅保留命中过滤条件的切片。 - -**语法**：`searchFilters` 是一个数组，每个元素是一个由 Key-Value 键值对组成的**子分组**。子分组之间默认采用 **AND** 语义且不可更改。 - -```json -{ - "searchFilters": [ - { "姓名": "张三", "性别": "男" }, - { "岗位": "技术员" } - ] -} -``` - -**支持的查询类型**： - -| 查询类型 | 适用字段类型 | 说明 | -| --- | --- | --- | -| 单值查询 | 数值（long/double）、字符串（string） | 字段等于某个值 | -| 多值查询 | 纯数值数组或纯字符串数组 | 字段命中数组中任一值；多值需用 `json.dumps` 序列化后传入 | -| 范围查询-等值 | 数值、字符串 | 支持 `eq`（等于）、`neq`（不等于）；一个字段不可配多个值 | -| 范围查询-区间 | 数值（long/double） | 支持 `gt`/`gte`/`lt`/`lte` | -| 模糊查询 | 字符串 | 支持 `like` 属性，`%` 匹配任意字符（含零个） | -| 标签（Tag）查询 | 仅文档搜索、音视频搜索类知识库 | `tags` 字段，多个标签之间为 OR 关系 | - -**前置条件**：子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](../concepts/workspace.md)（主账号可操作所有[业务空间](../concepts/workspace.md)），获取[业务空间](../concepts/workspace.md) ID，安装百炼 SDK（2023-12-29 版本）并配置 `ALIBABA_CLOUD_ACCESS_KEY_ID` / `ALIBABA_CLOUD_ACCESS_KEY_SECRET` 环境变量。 - -**调用示例（Python，单值查询）**： - -```python -retrieve_request = bailian_20231229_models.RetrieveRequest() -retrieve_request.query = '公司中叫张三的员工' -retrieve_request.index_id = '请传入实际的知识库ID' -retrieve_request.search_filters = [{"姓名": "张三"}] -resp = client.retrieve('请传入实际的业务空间ID', retrieve_request) -``` - -多值、范围、模糊、标签查询的写法类似，区别在于将字段的值替换为 `json.dumps` 序列化后的对象（如 `{"like": "技%员"}`、`{"gte": 20, "lte": 27}`、`["张三", "李四"]`）。完整 Python/Java 示例参见 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 - -## 限制与注意事项 - -- 临时 [API Key](../concepts/api-key.md) 无法手动删除，只能等 TTL 到期自动失效；各地域 [API Key](../concepts/api-key.md) 不互通。 -- 服务关联角色由百炼自动创建并绑定固定系统策略，不可修改、不可授予其他 RAM 身份；删除前必须先解除对应功能的依赖资源，否则相关功能将不可用。 -- SearchFilters 子分组之间为 AND 语义且不可更改；多值 / 范围 / 模糊 / 标签查询的值需通过 `json.dumps` 序列化为字符串后传入；标签查询仅支持文档搜索与音视频搜索类知识库。 -- 子账号只能操作已加入[业务空间](../concepts/workspace.md)中的知识库，主账号可操作所有[业务空间](../concepts/workspace.md)。 +- **临时 API Key**：无法提前撤销，到期自动失效；继承父密钥全部权限，**不得用于高权限操作场景**；各地域 Endpoint 不互通，需按实际部署地域调用 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 +- **服务关联角色**：删除前必须满足前置清理条件（如文档 2 中明确要求“删除所有已发布的工作流应用中的函数计算节点”），否则操作被拒绝；`AliyunServiceRoleForSFMAccessingMNS` 明确禁止用户修改或删除 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- **SearchFilters**：仅对已索引字段生效，未在知识库配置中启用“参与检索”的字段无法过滤；多值查询需将数组 JSON 序列化为字符串传入（见文档 3 Python 示例）；模糊查询 `like` 仅支持 `%` 通配符，不支持正则表达式 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +- **通用限制**：所有 `more` 相关能力均受百炼配额与计费规则约束，临时 Key 调用计入调用者配额；SLR 权限变更可能影响已有工作流执行，请在生产环境变更前充分测试。 ## 来源文档 @@ -114,23 +55,4 @@ resp = client.retrieve('请传入实际的业务空间ID', retrieve_request) - [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md index 9a40d452..b57b1443 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md @@ -1,179 +1,89 @@ # omni realtime api -Qwen-Omni-Realtime API 是阿里云百炼平台提供的实时[多模态](../concepts/multimodal.md)交互接口，基于 WebSocket 协议实现低延迟的音视频对话。该 API 支持语音输入/输出、图像输入、语音活动检测（VAD）、工具调用（Function Calling）、联网搜索及声音复刻等功能，适用于智能客服、语音助手等实时对话场景。 +Qwen-Omni Realtime API 是阿里云百炼平台提供的低延迟、[多模态](../concepts/multi-modal.md)实时交互接口，支持语音/音视频输入与文本/音频输出的流式双向通信。它基于 WebSocket 协议，内置 VAD（语音活动检测）、ASR（语音识别）、LLM 推理与 TTS（语音合成）全链路能力，适用于智能客服、虚拟助手、实时会议等场景。开发者可通过 Python 或 Java SDK 快速集成，无需自行编排模型调用流程。 -## 支持的模型 +## 支持的模型与功能 -| 模型系列 | 模型名称 | 特性 | -| --- | --- | --- | -| Qwen3.5-Omni-Realtime | qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime | 支持 semantic_vad、联网搜索、工具调用、idle_timeout_ms | -| Qwen3-Omni-Flash-Realtime | qwen3-omni-flash-realtime | 支持 smooth_output 参数 | -| Qwen-Omni-Turbo-Realtime | qwen-omni-turbo-realtime | 大部分生成参数不支持修改 | +当前支持以下 Qwen-Omni 实时系列模型，各模型能力存在差异，需按需选型： -各模型的默认音色不同：Qwen3.5-Omni-Realtime 系列为 `Tina`，Qwen3-Omni-Flash-Realtime 为 `Cherry`，Qwen-Omni-Turbo-Realtime 为 `Chelsie`。 +- **`qwen3.5-omni-realtime`**：基础旗舰版，支持 `semantic_vad`、联网搜索（`enable_search`）和工具调用（`tools`），是唯一同时支持三者的模型。 +- **`qwen3.5-omni-plus-realtime` 与 `qwen3.5-omni-flash-realtime`**：增强与轻量变体，支持 `idle_timeout_ms` 等高级 VAD 参数，但**不支持 `semantic_vad`**；仅 `plus` 版支持声音复刻驱动（见 [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)）。 +- **`qwen3-omni-flash-realtime`**：侧重响应速度，支持 `smooth_output` 口语化控制，但**不支持联网搜索与工具调用**。 +- **`qwen-omni-turbo-realtime`**：极致轻量版，参数（如 `temperature`、`top_p`、`max_tokens` 等）**完全不可修改**，仅支持基础对话。 -## 交互模式 +> **注意**：文档 1 和文档 6 均称 `qwen3.5-omni-realtime` 支持 `semantic_vad`，而文档 2 明确指出该能力“仅 `qwen3.5-omni-realtime` 系列模型支持”，但文档 4 的 `session.created` 示例中 `turn_detection.type` 字段注释却写为“取值为 `server_vad` 或 `semantic_vad`（仅 `qwen3.5-omni-realtime` 支持）”，存在表述冗余。以文档 1 和文档 2 的明确限定为准：`semantic_vad` 为 `qwen3.5-omni-realtime` 独占特性。 -根据[实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)，API 支持两种交互模式： +所有模型均支持： +- [多模态](../concepts/multi-modal.md)输入：纯音频、音视频（`append_audio` + `append_video`） +- [多模态](../concepts/multi-modal.md)输出：文本（`TEXT`）与音频（`AUDIO`）组合 +- 实时语音转录（ASR）：固定使用 `qwen3-asr-flash-realtime` 模型，不可替换 +- 声音复刻音色接入：需确保复刻时指定的 `target_model` 与实时对话模型严格一致（详见 [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)） -### VAD 模式（默认） +## 关键参数 -将 `session.turn_detection` 设为 `server_vad` 或 `semantic_vad`。服务端自动检测语音起止并触发模型响应，适用于持续音频流场景。支持语音打断。 +参数分为连接级（构造时设置）与会话级（`update_session` 时设置），部分参数模型间行为不同： -### Manual 模式 +| 参数 | 类型 | 说明 | 模型兼容性 | +|------|------|------|------------| +| `model` | `str`/`String` | 模型名称，如 `"qwen3.5-omni-realtime"` | 所有模型 | +| `url` | `str`/`String` | WebSocket 地址，**必须使用业务空间专属域名**：
`wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime`（北京）
`wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime`（新加坡） | 所有模型，[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 与 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) 均强调此迁移要求 | +| `output_modalities` | `list[MultiModality]`/`List` | 输出模态，`[TEXT]` 或 `[TEXT, AUDIO]`（默认） | 所有模型 | +| `voice` | `str`/`String` | 音色名，如 `"Tina"`；自定义音色需通过声音复刻获取 | 所有模型 | +| `turn_detection_type` | `str`/`String` | VAD 类型：`"server_vad"`（默认）或 `"semantic_vad"`（仅 `qwen3.5-omni-realtime`） | 见上文注意项 | +| `enable_search` | `bool`/`Boolean` | 启用联网搜索，**与 `tools` 互斥** | 仅 `qwen3.5-omni-realtime` | +| `tools` | `list[dict]`/`List>` | 工具定义列表，**与 `enable_search` 互斥** | 仅 `qwen3.5-omni-realtime` | +| `smooth_output` | `bool`/`Boolean` | 口语化开关，`true`/`false`/`null`，**仅 `qwen3-omni-flash-realtime` 支持** | 仅 `qwen3-omni-flash-realtime` | +| `temperature` / `top_p` / `top_k` / `max_tokens` 等采样参数 | 各自类型 | 控制生成多样性与长度，详见各文档默认值表 | `qwen-omni-turbo-realtime` 系列**全部不可修改** | -将 `session.turn_detection` 设为 `null`。客户端通过 `input_audio_buffer.commit` + `response.create` 手动控制对话节奏，适用于按下即说场景。 +> **注意**：`repetition_penalty` 默认值在文档 1 中写为“其他模型：1.05”，而文档 2 写为“`qwen3-omni-flash-realtime` 系列：1.05；`qwen-omni-turbo-realtime` 系列：1.05”，文档 4 的 `session.created` 示例中亦为 `1.05`。文档 6 未提及其他模型默认值，仅重复 `qwen3.5-omni-realtime` 为 `1.0`。此处以文档 2 的完整列表为准。 -## 连接地址 +## 使用方式 -``` -wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime # 北京地域 -wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime # 新加坡地域 -``` - -将 `{WorkspaceId}` 替换为[业务空间](../concepts/workspace.md) ID。建议使用[业务空间](../concepts/workspace.md)专属域名以获得更好的性能和稳定性。 - -## 客户端事件 - -详细参数说明参见[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 - -| 事件 | 用途 | -| --- | --- | -| `session.update` | 更新会话配置（模态、音色、VAD、工具等） | -| `input_audio_buffer.append` | 追加音频数据（Base64 编码） | -| `input_audio_buffer.commit` | 提交音频缓冲区（Manual 模式必需） | -| `input_audio_buffer.clear` | 清空音频缓冲区 | -| `input_image_buffer.append` | 追加图像数据（JPG/JPEG，Base64 编码） | -| `response.create` | 触发模型生成响应 | -| `response.cancel` | 取消正在进行的响应 | -| `conversation.item.create` | 回传工具调用结果 | - -## 服务端事件 - -详细参数说明参见[服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md)。 - -| 事件 | 含义 | -| --- | --- | -| `session.created` | 连接建立，返回默认配置 | -| `session.updated` | 会话配置更新成功 | -| `error` | 错误信息 | -| `input_audio_buffer.speech_started` | VAD 检测到语音开始 | -| `input_audio_buffer.speech_stopped` | VAD 检测到语音结束 | -| `input_audio_buffer.committed` | 音频缓冲区已提交 | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `response.done` | 响应完成 | -| `response.function_call_arguments.done` | 工具调用参数完成 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | - -## 关键会话参数 - -通过 `session.update` 事件配置： - -| 参数 | 说明 | 默认值 | -| --- | --- | --- | -| `modalities` | 输出模态：`["text"]` 或 `["text","audio"]` | `["text","audio"]` | -| `voice` | 音色名称 | 因模型而异 | -| `input_audio_format` | 输入音频格式，仅支持 `pcm`（16kHz） | `pcm` | -| `output_audio_format` | 输出音频格式，仅支持 `pcm`（24kHz） | `pcm` | -| `instructions` | 系统消息 | - | -| `turn_detection.type` | VAD 类型：`server_vad` / `semantic_vad` | `server_vad` | -| `turn_detection.threshold` | VAD 灵敏度，范围 [-1.0, 1.0] | 0.5 | -| `turn_detection.silence_duration_ms` | 静音触发时间（ms），范围 [200, 6000] | 800 | -| `turn_detection.idle_timeout_ms` | 静默超时（ms），范围 [5000, 30000]，仅 qwen3.5 系列 | - | -| `enable_search` | 联网搜索，仅 Qwen3.5-Omni-Realtime | `false` | -| `tools` | 工具定义列表，仅 Qwen3.5-Omni-Realtime | `[]` | -| `smooth_output` | 口语化风格，仅 Qwen3-Omni-Flash-Realtime | `true` | - -> **注意**：`tools` 和 `enable_search` 不兼容，不可同时开启。 - -### 生成参数 - -| 参数 | Qwen3.5-Omni-Realtime | Qwen3-Omni-Flash-Realtime | Qwen-Omni-Turbo-Realtime | -| --- | --- | --- | --- | -| `temperature` | 0.7 | 0.9 | 1.0（不可改） | -| `top_p` | 0.8 | 1.0 | 0.01（不可改） | -| `top_k` | 20 | 50 | 20（不可改） | -| `repetition_penalty` | 1.0 | 1.05 | 1.05（不可改） | -| `presence_penalty` | 1.5 | 0.0 | 0.0（不可改） | - -> **注意**：`qwen-omni-turbo` 系列模型的生成参数不支持修改。 - -## SDK 使用 - -### Python SDK - -需要 [DashScope SDK](../concepts/dashscope-sdk.md) >= 1.25.17。核心类为 `OmniRealtimeConversation`，通过 `from dashscope.audio.qwen_omni import OmniRealtimeConversation` 引入。详见 [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md)。 +API 交互基于 WebSocket，核心流程分两种模式： +### 1. VAD 模式（推荐，默认） +服务端自动检测语音起止并触发响应，客户端只需持续 `append_audio`（及可选 `append_video`）： ```python -from dashscope.audio.qwen_omni import MultiModality, OmniRealtimeCallback, OmniRealtimeConversation - -conv = OmniRealtimeConversation(model="qwen3.5-omni-plus-realtime", callback=callback, url=url) +conv = OmniRealtimeConversation(model="qwen3.5-omni-realtime", callback=cb, url=url) conv.connect() -conv.update_session( - output_modalities=[MultiModality.AUDIO, MultiModality.TEXT], - voice="Tina", - enable_turn_detection=True -) -conv.append_audio(audio_base64) -conv.close() +conv.update_session(enable_turn_detection=True) # 启用 server_vad +# 循环：mic.read() → conv.append_audio(base64) +# 无需手动 commit 或 create_response ``` +- 事件流：`input_audio_buffer.speech_started` → `input_audio_buffer.speech_stopped` → `input_audio_buffer.committed` → `response.*` +- 工具调用时，服务端发送 `response.function_call_arguments.done` 后，客户端执行工具并调用 `conversation.item.create`，服务端**自动**生成最终响应（见 [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)）。 -### Java SDK - -需要 DashScope Java SDK >= 2.22.15。核心类为 `OmniRealtimeConversation`，通过 `OmniRealtimeParam` 和 `OmniRealtimeConfig` 配置参数。详见 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md)。 - +### 2. Manual 模式 +客户端完全控制节奏，需显式提交与触发： ```java -OmniRealtimeParam param = OmniRealtimeParam.builder() - .model("qwen3.5-omni-plus-realtime") - .url(url) - .build(); -OmniRealtimeConversation conversation = new OmniRealtimeConversation(param, callback); -conversation.connect(); conversation.updateSession(OmniRealtimeConfig.builder() - .modalities(Arrays.asList(OmniRealtimeModality.AUDIO, OmniRealtimeModality.TEXT)) - .voice("Tina") - .enableTurnDetection(true) + .enableTurnDetection(false) // 关闭 VAD .build()); +// ... appendAudio ... +conversation.commit(); // 提交音频缓冲区 +conversation.createResponse(null, Arrays.asList(AUDIO, TEXT)); // 触发响应 ``` +- 适用场景：聊天软件“按住说话”、离线音频文件处理。 +- 工具调用时，客户端在收到 `response.function_call_arguments.done` 后，需**手动再次调用 `createResponse`** 触发最终响应（见 [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)）。 -> **注意**：Java SDK 中 `instructions`、`smooth_output`、`enable_search`、`search_options`、`tools` 及生成参数（temperature/top_p/top_k 等）需通过 `OmniRealtimeConfig` 的 `parameters` 方法设置。 - -## 工具调用（Function Calling） - -仅 Qwen3.5-Omni-Realtime 模型支持。流程如下： - -1. 通过 `session.update` 配置 `tools` 列表 -2. 服务端识别到需要调用工具时，通过 `response.function_call_arguments.done` 返回函数名和参数 -3. 客户端执行工具函数，通过 `conversation.item.create` 回传结果（`type: "function_call_output"`） -4. VAD 模式下服务端自动生成响应；Manual 模式下需额外发送 `response.create` - -## 声音复刻 - -通过 `qwen-voice-enrollment` 模型创建自定义音色，然后在实时对话中使用。音频要求：WAV/MP3/M4A 格式，10-20 秒，采样率 >= 24kHz，单声道，文件 < 10MB。创建音色时指定的 `target_model` 必须与后续对话使用的模型一致。 +## 限制和注意事项 -支持的驱动模型：qwen3.5-omni-plus-realtime、qwen3.5-omni-flash-realtime。 - -## 输入限制 - -- 音频输入：16kHz 采样率 PCM，音频缓冲区最大 15MiB -- 图像输入：JPG/JPEG 格式，建议 480p-720p（最高 1080p），Base64 编码后不超过 256KB，建议 1 帧/秒 -- 图像需在至少一次 `input_audio_buffer.append` 之后发送，通过 `input_audio_buffer.commit` 与音频一起提交 +- **域名迁移强制要求**：华北2（北京）与新加坡地域必须使用 `wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 或 `wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，旧域名 `dashscope.aliyuncs.com` 将逐步下线（[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 与 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) 均明确提示）。 +- **音视频格式约束**： + - 输入音频：`PCM_16000HZ_MONO_16BIT`（Python）或 `PCM_16000HZ_MONO_16BIT`（Java），Base64 编码。 + - 输入视频：JPG/JPEG 格式，分辨率建议 480P–720P（≤1080P），单图 Base64 后 ≤256KB。 + - 输出音频：固定 `PCM_24000HZ_MONO_16BIT`，不可自定义。 +- **并发与资源**：单个 WebSocket 连接对应一个会话；`append_audio` 单次数据块无明确上限，但 `commit` 前总缓冲区建议 ≤15 MiB（[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 提示）。 +- **互斥配置**：`enable_search` 与 `tools` 不可同时启用，否则返回 `invalid_request_error`（[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 明确说明）。 +- **错误处理**：所有服务端错误均以 `error` 事件返回，含 `type`、`code`、`message` 和 `param`（见 [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md)）。 ## 来源文档 -- [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) -- [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md) - [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) -- [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) +- [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) - [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) +- [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md) - [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - - - - - - - +- [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md index b4840377..daa3fa5d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md @@ -1,73 +1,54 @@ # preparations -本页汇总在阿里云百炼平台调用模型 API 前的准备工作，涵盖获取鉴权凭证（API Key）、安装官方或兼容 SDK、使用百炼 CLI 快速集成，以及常见错误码的排查思路。面向开发者，帮助你从零完成环境搭建并稳定发起第一次调用。 +在调用阿里云百炼平台的模型或应用前，开发者需完成基础环境准备，包括获取并安全配置 API Key、安装合适的 SDK 或 CLI 工具、理解关键参数约束及常见限制。这些步骤是所有 API 调用和本地开发的前提，直接影响服务可用性、安全性与调试效率。 -## 获取并配置 API Key +## 支持的模型/功能 -调用模型或应用前，需先获取 API Key 作为鉴权凭证。需使用主账号，或具备 `管理员` / `API-Key` 页面权限的子账号，在[阿里云百炼控制台](https://bailian.console.aliyun.com/)对应地域的 **API Key** 页面创建。详见 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。 +百炼平台支持多类模型与能力，涵盖文本生成（如 `qwen3-max`、`qwen3-235b-a22b-instruct-2507`）、图像生成（`qwen-image-2.0`）、视频生成（`happyhorse-1.0-t2v`）、语音合成（`cosyvoice-v3-flash`）、语音识别（`paraformer-real-time`）、向量嵌入（`text-embedding-v3`）、排序（`text-rerank-v3`）及全模态理解（`qwen3.5-omni-plus`）。部分模型具备特定能力约束，例如： +- 思考模式（`enable_thinking=true`）仅适用于指定模型（如 `qwen3-235b-a22b-thinking-2507`），且强制要求 `stream=true` 与 `incremental_output=true`； +- 结构化输出（`response_format={"type": "json_object"}`）不支持与思考模式共用； +- 联网搜索（`enable_search=true`）仅限明确标注支持该能力的模型； +- 工具调用（`tools` 参数）仅被 Qwen 和 DeepSeek 系列模型支持，纯文本模型（如 `qwen3-max`）若传入含 `image_url` 的 `messages` 将报错 [原文标题](../../raw/model-api-reference/preparations/error-code.md)。 -创建时的关键选项： +## 关键参数 -- **归属业务空间**：决定该 Key 的调用权限。同一空间内的 Key 权限相同，无需为不同模态（文生文、文生图、语音等）分别创建。默认业务空间的 Key 可调用所有标准模型及默认空间内的应用；子业务空间的 Key 只能调用已授权的模型及本空间应用。 -- **权限**：可选 **全部**（调用所有模型与应用），或 **自定义**（配置 IP 白名单最多 20 个 IPv4/IPv6 地址或网段，以及可访问的模型/应用范围）。 +调用时需注意以下核心参数的合法范围与互斥关系（详见 [原文标题](../../raw/model-api-reference/preparations/error-code.md)）： +- `temperature`：必须在 `[0.0, 2.0)` 区间； +- `top_p`：必须在 `(0.0, 1.0]` 区间； +- `max_tokens`：不得超过模型文档中声明的最大输出 [Token](../concepts/token.md) 数； +- `n`：取值范围为 `[1, 4]`； +- `seed`：DashScope 协议下需为 `[0, 9223372036854775807]` 内整数； +- `thinking_budget`：须为正整数且不超过模型最大思维链长度； +- `stop`：仅接受 `str`、`list[str]`、`list[int]` 或 `list[list[int]]` 类型，且列表内元素类型必须一致； +- `messages`：纯文本模型要求 `content` 为字符串；[多模态](../concepts/multi-modal.md)模型要求 `content` 数组中每个元素为合法对象（`type` 仅限 `text`/`image_url`/`video_url` 等）； +- `response_format`：结构化输出必须设为 `{"type": "json_object"}`，且提示词中需包含 `json` 关键词。 -> **注意**：百炼已对按量付费 API Key 做安全升级（美国（弗吉尼亚）地域除外）。升级后新建的 Key 以 `sk-ws` 开头，且**仅在创建时展示一次明文**，关闭弹窗后无法再次查看，务必立即复制保存；升级前 `sk-` 开头的旧 Key 仍可正常使用。此外，Token Plan / Coding Plan 使用以 `sk-sp-` 开头的专属 Key，不同于本文的按量付费 Key。 +> **注意**：文档 3 中“`The value of the enable_thinking parameter is restricted to True`”与文档 1 中“API Key 权限说明”存在隐含矛盾——前者指出部分模型强制开启思考模式，后者未提及该限制对权限配置的影响。实际调用时应以模型文档为准，而非仅依赖 API Key 权限设置。 -推荐将 API Key 配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄漏。各系统配置方式（`~/.bashrc`、`~/.zshrc`、`~/.bash_profile`、Windows 系统属性 / `setx` / PowerShell）参见原文。调用时除 API Key 外，还需指定**服务端点** `base_url`（即创建弹窗中的 API Host），且 OpenAI 兼容协议与 Anthropic 兼容协议的 `base_url` 不同、随地域变化，请以对应接口文档为准。 +## 使用方式 -除控制台外，百炼还提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）以编程方式管理 Key，调用需使用阿里云账号 AccessKey 签名认证并具备相应 RAM 权限。 +### API Key 获取与配置 +需使用主账号或具备 `管理员`/`API-Key` 页面权限的子账号，在对应地域（如华北2、新加坡、美国弗吉尼亚）的 [API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建密钥。新创建的密钥以 `sk-ws` 开头，明文仅显示一次，务必立即保存 [原文标题](../../raw/model-api-reference/preparations/get-api-key.md)。推荐将 `DASHSCOPE_API_KEY` 配置为环境变量（Linux/macOS/Windows 均有详细步骤），避免硬编码。 -## 安装 SDK +### SDK 安装 +- **Python**：可选 `openai`（OpenAI 兼容协议）或 `dashscope`（原生协议）SDK，均需 `pip install -U `； +- **Java/Node.js/Go**：DashScope 提供官方 Java SDK；OpenAI SDK 支持多语言（Java/Node.js/Go），其中 Go 需 `Go 1.22+` 并建议配置阿里云镜像代理； +- **CLI 工具**：通过 `npm install -g bailian-cli` 安装百炼 CLI（要求 Node.js ≥ 22.12.0），支持 `bl text chat`、`bl image generate` 等命令行调用 [原文标题](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 -百炼同时支持官方 **DashScope SDK**（Python、Java）与通过 **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)**调用的多语言 SDK。详见 [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md)。 +### 协议与端点 +调用时除 API Key 外，**必须指定服务端点（API Host）**，其值取决于所选协议与地域： +- OpenAI 兼容协议：`base_url` 为 `https://dashscope.aliyuncs.com/v1`（中国站）或对应国际站地址； +- Anthropic 兼容协议：`base_url` 为 `https://dashscope.aliyuncs.com/anthropic/v1`； +- 不同地域的端点不同，务必以控制台创建 API Key 时弹窗显示的 `API Host` 为准。 -- **Python**（需 `python >= 3.8`）：`pip install -U openai` 或 `pip install -U dashscope` -- **Java**：DashScope 用 `com.alibaba:dashscope-sdk-java`；OpenAI 用 `com.openai:openai-java`（需 Java 8+，推荐 `3.5.0`），均通过 Maven / Gradle 引入。 -- **Node.js**：`npm install --save openai`（或 `yarn add openai`）；安装失败可配置镜像源 `npm config set registry https://registry.npmmirror.com/`。 -- **Go**（需 `Go 1.22+`）：`go get 'github.com/openai/openai-go/v3'`；超时可设 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 +## 限制和注意事项 -安装后即可调用文本生成、图像生成、视频生成、语音合成/识别、向量、排序等模型。 - -## 使用百炼 CLI - -百炼 CLI（npm 包 `bailian-cli`，命令 `bl` / `bailian`）是面向 AI Agent 的命令行工具，可将平台能力集成到各类 AI 工具中。安装前置要求 **Node.js ≥ 22.12.0**，且**仅支持 npm 安装**（勿用 pnpm / yarn 安装该包）。详见 [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 - -```bash -# 1. 安装 CLI -npm install -g bailian-cli -# 2. 安装 Skills（注册能力描述文件到各 Agent） -npx skills add modelstudioai/cli --all -g -# 3. 验证 -bl --version -``` - -**认证方式**（可组合使用，互不覆盖）： - -| 方式 | 命令 | 适用场景 | -| --- | --- | --- | -| 控制台登录（推荐） | `bl auth login --console` | 模型调用 + 应用管理（浏览器 OAuth） | -| API Key | `bl auth login --api-key sk-xxx` | 模型调用；会先校验 Key 有效性 | -| 环境变量 | 配置 API Key 环境变量 | CI/CD、无界面环境 | -| 配置文件 | `bl config set --key api_key --value sk-xxx` | 持久化，**不校验** Key 有效性 | -| 临时传入 | `bl text chat --api-key sk-xxx ...` | 单次调用，不落盘 | - -常用全局参数：`--region `（默认 cn）、`--base-url`、`--output `、`--non-interactive`（Agent/CI）、`--dry-run`、`--concurrent ` 等。子命令覆盖文本对话（`bl text chat`）、全模态（`bl omni`）、图像（`bl image generate/edit`）、视频（`bl video generate/edit/ref`）、视觉理解（`bl vision describe`）、语音合成（`bl speech synthesize`）等。 - -> **注意**：CLI 文档中示例默认模型（如 `qwen3.7-max`、`qwen3.5-omni-plus`、`qwen-image-2.0`、`happyhorse-1.0-t2v` 等）为工具内置默认值，可能随版本变化；实际可用模型请以模型列表 / 控制台为准。安全约束上，禁止将真实 API Key 写入仓库、日志、Skill 或聊天记录的可公开部分。 - -## 常见错误码与排查 - -调用过程中的报错多为 **400-InvalidParameter** 类的参数问题，可对照错误信息定位。完整清单见 [错误码](../../raw/model-api-reference/preparations/error-code.md)，以下为高频场景： - -- **思考模式相关**：思考模式模型需 `enable_thinking=true` 时配合[流式输出](../concepts/streaming.md)，并设 `incremental_output=true`、`result_format="message"`；部分模型（如 `qwen3-235b-a22b-thinking-2507`）不允许将 `enable_thinking` 设为 `false`。 -- **参数取值范围**：`temperature` ∈ [0.0, 2.0)、`top_p` ∈ (0.0, 1.0]、`top_k` ≥ 0、`presence_penalty` ∈ [-2.0, 2.0]、`n` ∈ [1, 4]；`max_tokens` 与输入长度上限以模型列表为准。 -- **模型不存在（Model not exist）**：核对 `model` 名称大小写与空格，勿混用开源社区名与百炼模型 ID（用 `qwen3-235b-a22b-instruct-2507` 而非 `Qwen/Qwen3-235B-A22B-Instruct-2507`）。 -- **content 类型错误**：纯文本模型的 `content` 必须为字符串，不能传数组或图片等多模态元素；需要图片输入请改用 Qwen-VL / Qwen3-VL 等多模态模型。 -- **结构化输出**：使用 `response_format` 的 `json_object` 时，提示词须包含 `json` 关键词，且不能同时开启思考模式。 -- **文件类（Qwen-Long）**：仅支持纯文本格式（TXT/DOCX/PDF/EPUB/MOBI/MD），单文件 < 150 MB、< 15000 页，file-id 数量 < 100。 -- **账号状态（Arrearage）**：账号欠费会导致访问被拒绝，需在费用与成本页面充值后等待系统更新。 - -排障时可借助[阿里云 AI 助理](https://www.aliyun.com/ai-assistant/)，直接粘贴报错信息即可获得原因与解决方案。 +- **API Key 安全**：`sk-` 开头旧密钥仍可用，但新密钥统一为 `sk-ws` 格式，且不可再次查看明文。美国（弗吉尼亚）地域不支持禁用/重置操作。 +- **地域隔离**：API Key 与模型服务绑定地域，跨地域调用需对应地域的 API Key 和端点。 +- **IP 白名单**：仅北京、新加坡等部分地域支持自定义 IP 白名单（最多 20 个 IPv4/IPv6 地址或网段），美国（弗吉尼亚）地域不支持。 +- **文件限制**：Qwen-Long 模型仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 纯文本文件，单文件大小 ≤ 150 MB、页数 ≤ 15000、内容非空；图片/扫描件需先用 Qwen-VL 提取文本。 +- **[Token](../concepts/token.md) 限制**：输入总长度（含 messages、[prompt](../guides/prompt.md)、file content）不得超过模型最大上下文窗口；纯文本模型不支持[多模态](../concepts/multi-modal.md) `content`，否则触发 `Unexpected item type in content` 错误。 +- **CLI 环境约束**：百炼 CLI 严格依赖 npm（非 pnpm/yarn）且要求 Node.js ≥ 22.12.0；认证方式中，`bl auth login --console` 推荐用于交互式场景，`--api-key` 适用于 CI/CD 或无浏览器环境。 ## 来源文档 @@ -77,4 +58,3 @@ bl --version - [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md index f4dc1072..69ec2ecd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md @@ -1,37 +1,44 @@ # qwen api reference -百炼平台为文本生成模型提供了多种调用接口，开发者可根据迁移成本、功能完整度和生态兼容性选择合适的入口。当前共有四类接口：OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages 以及百炼原生的 DashScope 接口。详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +Qwen API 提供多种调用方式，支持文本生成、工具调用、联网搜索等能力，开发者可根据技术栈兼容性与功能需求选择合适接口。所有接口均基于 Qwen 系列大模型（如 Qwen2、Qwen2.5、Qwen3）提供服务，需通过百炼平台鉴权访问。详细参数说明与行为差异请参考 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 -## 支持的接口 +## 支持的模型与功能 -百炼针对不同的接入场景提供了以下四种接口，功能定位各有侧重： +- **基础文本生成**：支持 `qwen-max`、`qwen-plus`、`qwen-turbo` 等多档位模型，适用于通用对话、摘要、创作等场景。 +- **增强能力接口**： + - OpenAI 兼容 Chat Completions：适合已有 OpenAI 生态集成的应用快速迁移； + - OpenAI 兼容-Responses：自动启用联网搜索、代码解释器、网页提取等工具链，无需手动管理工具调用流程； + - Anthropic 兼容 Messages：支持 `tool_use`、`thinking` 等结构化输出，适配 Anthropic 工作流； + - DashScope 原生接口：提供最全参数控制（如 `top_k`、`repetition_penalty`、`enable_search`），是调试与高阶定制的首选。 -- **OpenAI 兼容 Chat Completions**：与 OpenAI 客户端库直接兼容，迁移现有应用或接入第三方工具的成本最低。适合已经基于 OpenAI SDK 构建的应用平滑迁移。 -- **OpenAI 兼容 Responses**：内置联网搜索、代码解释器和网页内容提取工具，并自动管理对话历史，无需手动维护上下文。 -- **Anthropic 兼容 Messages**：兼容 Anthropic Messages API，支持思考（thinking）和工具调用（tool use）。适合基于 Anthropic 生态构建的应用接入。 -- **DashScope**：百炼原生接口，提供最完整的功能集和参数支持，是需要使用平台全部能力时的首选。 +> **注意**：`qwen-max` 在 DashScope 接口中默认启用思考模式（`enable_thinking=true`），但在 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)中该参数不可设；此行为差异已在 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 中明确标注，使用时需注意一致性。 -以上接口的完整清单与说明参见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +## 关键参数 -## 如何选择 +| 参数名 | 类型 | 说明 | 是否必需 | 备注 | +|--------|------|------|----------|------| +| `model` | string | 模型标识符，如 `qwen-max`、`qwen-plus` | 是 | 不同接口对模型命名格式要求一致，详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) | +| `messages` | array | 对话历史，格式为 `[{ "role": "...", "content": "..." }]` | 是（Chat Completions / Messages） | DashScope 接口额外支持 `system` 角色和 `tools` 字段 | +| `tools` | array | 工具定义列表（JSON Schema 格式） | 否 | 仅 DashScope 和 Anthropic Messages 接口原生支持；OpenAI 兼容-Responses 的工具由服务端自动注入，不开放显式传参 | +| `enable_search` | boolean | 是否启用联网搜索（DashScope 专属） | 否 | 默认 `false`；启用后将自动触发搜索并融合结果 | -- 追求**最低迁移成本**、已有 OpenAI 应用：选择 OpenAI 兼容 Chat Completions。 -- 需要**内置工具（联网搜索/代码解释器/网页提取）与自动对话管理**：选择 OpenAI 兼容 Responses。 -- 处于 **Anthropic 生态**、需要思考与工具调用：选择 Anthropic 兼容 Messages。 -- 需要**最完整的功能与参数**、使用平台全部能力：选择 DashScope 原生接口。 +## 使用方式 -## 使用方式与注意事项 +1. **认证**：使用百炼平台颁发的 `API Key`，通过 `Authorization: Bearer ` 请求头传递； +2. **Endpoint 示例**： + - DashScope：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation` + - OpenAI 兼容：`POST https://dashscope.aliyuncs.com/v1/chat/completions` +3. **SDK 调用**：推荐使用官方 `dashscope` Python SDK（v1.20.0+）或 `openai` 客户端（v1.0+），配置 `base_url` 指向百炼 OpenAI 兼容地址即可复用现有逻辑。 -- [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)可直接复用官方 OpenAI 客户端库，仅需替换 base URL 和 API Key，改动量小。 -- 若依赖联网搜索、代码解释器等内置工具，需使用 Responses 接口，而非普通的 Chat Completions。 -- 不同接口在参数集合和功能覆盖上存在差异：DashScope 参数最全，OpenAI/Anthropic 兼容接口以对应生态的字段约定为准，跨接口迁移时需核对参数映射。 +## 限制和注意事项 -> **注意**：本页仅为文本生成模型各接口的入口索引，具体的请求参数、字段格式与调用示例请查阅对应接口的专属文档；随着平台迭代，接口能力可能变化，请以 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 为准。 +- 单次请求 `messages` 总长度（token 数）上限为 32768（Qwen3 模型）或 8192（旧版模型），具体以实际模型文档为准； +- 工具调用（如 `code_interpreter`）在 OpenAI 兼容-Responses 接口中为全自动模式，**不支持用户自定义工具函数**，与 DashScope 接口的可控性存在本质差异； +- 流式响应（`stream=true`）在所有接口中均支持，但字段结构不同：DashScope 返回 `output.text`，OpenAI 兼容返回 `choices[0].delta.content`； +- > **注意**：原始文档中 OpenAI 兼容-Responses 的“自动管理对话历史”描述与实际行为存在偏差——当启用 `enable_search` 时，历史会被截断以预留上下文空间，该限制未在 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 中明示，建议在长对话场景下主动控制 `max_tokens` 与历史长度。 ## 来源文档 - [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md new file mode 100644 index 00000000..81c8f2b8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md @@ -0,0 +1,74 @@ +# realtime api user guide + +Realtime API 是一套面向低延迟、[多模态](../concepts/multi-modal.md) AI 交互场景的实时通信协议栈，支持 WebSocket、WebRTC 和 AOQ（AI over QUIC）三种传输协议，分别适配服务端集成、浏览器端互动和移动端原生应用等不同技术栈与网络环境。开发者需根据目标平台、延迟要求、弱网适应性及数据类型选择合适协议，并配合对应 SDK 或标准 Web API 实现接入。 + +## 支持的模型/功能 + +Realtime API 当前支持以下核心模型与应用类型，但协议支持存在明确差异： + +- **实时全模态模型**（如 `qwen3.5-omni-plus-realtime`、`qwen3.5-omni-flash-realtime`、`qwen3.5-livetranslate-flash-realtime`）：三协议均支持，是唯一在 WebSocket、WebRTC 和 AOQ 上完全可用的模型类别。 +- **[多模态](../concepts/multi-modal.md)开发套件**（`multimodal-dialog`）：仅支持 WebRTC 和 WebSocket，[不支持 AOQ](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md)。 +- **实时语音识别**（Fun-ASR 系列）、**实时语音合成**（CosyVoice 系列）、**实时语音对话**（`qwen-audio-3.0-realtime-plus` 等）：**仅支持 WebSocket 协议**，[WebRTC 和 AOQ 均不支持](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md)。 + +> **注意**：文档 4 与文档 5 均以 WebRTC 接入 `multimodal-dialog` 和 `qwen3.5-omni-plus-realtime` 为示例，但文档 1 明确指出 `multimodal-dialog` 不支持 AOQ；而文档 7 的 AOQ 示例仅覆盖 `qwen3.5-omni-plus-realtime`，未提及 `multimodal-dialog`。因此，`multimodal-dialog` 的 AOQ 支持状态以文档 1 的表格为准，属明确不支持项，非过时信息。 + +## 关键参数 + +### 鉴权参数 +- `Authorization: Bearer `：所有协议建连阶段必需的 HTTP Header。AOQ 协议中该 Key 仅用于服务端向百炼网关发起 `allocate` 请求，客户端使用返回的 `aoqTokenForClient` 连接，避免密钥暴露 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)。 + +### 协议特有参数 +- **AOQ**：`x-dashscope-rtc-transport: moq`（必须）、`clientIp`（选填，用于 Relay 节点优化）。 +- **WebRTC**：SDP 交换请求中 `Content-Type: application/sdp`，且 `model` 参数需显式指定（如 `?model=qwen3.5-omni-plus-realtime`）。 +- **WebSocket**：无特殊 Header，依赖标准 WebSocket 握手，模型通过 URL query 参数或初始消息体指定。 + +### 会话配置参数（通过 `session.update` 事件发送） +- `modalities`: 指定输出模态，如 `["text", "audio"]`。 +- `voice`: 输出音色 ID（如 `"Ethan"`）。 +- `input_audio_format` / `output_audio_format`: 当前仅支持 `"pcm"`。 +- `turn_detection`: VAD 配置对象，`type` 可选 `"server_vad"` 或 `"semantic_vad"`（推荐后者），含 `threshold` 和 `silence_duration_ms`。 + +## 使用方式 + +### 协议选择与接入路径 +- **WebSocket**：适用于服务端或快速原型验证，使用 DashScope SDK（参见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)），接入成本最低，但弱网对抗能力差。 +- **WebRTC**：适用于浏览器端，需自行管理 `RTCPeerConnection`、媒体流与 DataChannel，内置回声消除与降噪，[通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) 提供完整 JS 示例。 +- **AOQ**：适用于 Android/iOS/HarmonyOS 原生应用，需集成 [AOQ SDK](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md)，具备极致弱网对抗与混合数据传输能力，[通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) 包含各平台集成指南。 + +### 核心流程共性 +1. **获取凭证**：WebSocket 直接使用 API Key；WebRTC 通过 SDP 交换携带 Key；AOQ 由 AppServer 调用百炼 `allocate` 接口获取 `sid` 与 `aoqTokenForClient`。 +2. **建立连接**：WebSocket 直连；WebRTC 完成 Offer/Answer 协商；AOQ 调用 `engine.connect(config)`。 +3. **会话初始化**：连接成功后，发送 `session.update` 事件配置模态、音色、VAD 等。 +4. **媒体流控制**：AOQ 必须在收到 `session.updated` 后调用 `enableSendMediaStream(.audio, true)` 开启发送；WebRTC 需在收到 `session.created` 后解除媒体门控；WebSocket 通常由 SDK 自动处理。 + +### 媒体流高级控制（AOQ 专属） +- **自定义音频采集/播放**：通过 `isExternal=true` 关闭内部设备，使用 `addAudioExternalStream` + `pushAudioExternalStreamData` 或 `setAudioFrameObserver` 实现 TTS 注入或 ASR 处理 [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md)。 +- **自定义视频输入**：支持原始帧（I420/NV12/BGRA）或编码帧（JPEG）推送，需先 `startVideoCapture(isExternal=true)` [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md)。 + +## 限制和注意事项 + +- **浏览器兼容性**：WebRTC 原生支持所有现代浏览器；AOQ 不支持浏览器，仅限原生平台；WebSocket 兼容性最广。 +- **建连与媒体发送时机**：AOQ 和 WebRTC 均要求严格遵循“先建连 → 收到服务端确认（`session.updated` 或 `session.created`）→ 再开启媒体发送”流程，否则模型可能无法接收数据。此逻辑在 [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) 中有明确强调。 +- **CORS 限制**：WebRTC 的 SDP 交换在浏览器端直连百炼服务受 CORS 限制，[文档 4 和 5 均明确指出 Demo 需通过 curl 或业务后端代理完成](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md)，生产环境必须由 AppServer 代理。 +- **Opus 编解码**：AOQ SDK 使用插件化 Opus，下载 SDK 时必须同步获取并集成 `libPluginOpus`（Android/iOS/HarmonyOS 各平台均有对应包）。 +- **连接状态管理**：AOQ SDK 提供明确的状态机（Connecting → Connected → Failed → Disconnected），业务需监听 `onConnectionStatusChange` 回调处理状态迁移，`Failed` 为瞬态，SDK 会自动进入 `Disconnected`，无需手动 `disconnect` [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md)。 + +## 来源文档 + +- [Realtime API简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) +- [SDK下载](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) +- [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) +- [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) +- [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) +- [实现接通模型/应用](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) +- [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) +- [AOQ SDK简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) +- [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) +- [音频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) +- [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) +- [自定义音频播放](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) +- [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) +- [视频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) +- [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md index fad0f724..83599b34 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md @@ -1,111 +1,82 @@ # toolkits and [frameworks](frameworks.md) -阿里云百炼的通义千问等模型提供了一套与 OpenAI 高度兼容的接口体系，覆盖 Chat Completions、Responses、Completions、Embedding、文件、Batch、Conversations 等能力，并可直接接入 LangChain/LangChain4j 等主流框架。对于已有 OpenAI 应用，通常只需替换 `api_key`、`base_url` 与 `model` 三项即可完成迁移，无需改动业务逻辑。 - -## 迁移三要素与服务地址 - -将 OpenAI 应用迁移到百炼的核心是配置以下三项（详见 [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)）： - -- **`api_key`**：替换为[百炼 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。**各地域的 API Key 不同**，切换地域时需同步更换。建议配置到环境变量 `DASHSCOPE_API_KEY` 以降低泄露风险。 -- **`base_url`**：OpenAI SDK 调用统一使用 `/compatible-mode/v1` 路径；HTTP 调用在其后追加具体资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。 -- **`model`**：替换为百炼支持的模型名称。 - -各地域 SDK `base_url`： - -| 地域 | base_url | -| --- | --- | -| 华北2（北京） | `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | -| 新加坡 | `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 日本（东京） | `https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` | -| 德国（法兰克福） | `https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` | -| 美国（弗吉尼亚） | `https://dashscope-us.aliyuncs.com/compatible-mode/v1` | - -其中 `{WorkspaceId}` 为业务空间 ID，可在百炼控制台**业务空间详情**页面查看。 - -> **注意**：百炼为北京、新加坡地域推出了业务空间专属域名，性能与稳定性更佳，建议从旧域名迁移：北京 `https://dashscope.aliyuncs.com` → `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`；新加坡 `https://dashscope-intl.aliyuncs.com` → `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`。现有域名仍可正常使用。 - -> **注意**：Responses 与 Conversations 接口的旧版路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将停止维护，请尽快迁移至新版 `/compatible-mode/v1/...` 路径。 - -## 各兼容接口一览 - -### Chat Completions（对话补全） - -最常用的兼容接口，支持非流式、流式（`stream=True`，配合 `stream_options={"include_usage": True}` 返回 Token 统计）与 function call（工具调用）。支持模型广泛：Qwen 大语言模型（商业版/开源版）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及 DeepSeek、Kimi、GLM、MiniMax 等三方模型。 - -> **注意**：三方直供模型仅在中国站的中国内地地域可用，调用前需先在百炼控制台开通对应服务。Qwen-Audio 不支持 OpenAI 兼容协议，仅支持 DashScope 协议。 - -### Responses（智能体原生接口） - -作为 Chat Completions 的演进版本，Responses API 内置联网搜索、网页抓取、代码解释器、文搜图/图搜图等工具，输入更灵活（可直接传字符串），并通过 `previous_response_id` 自动管理多轮上下文，无需手动拼接消息历史。详见 [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 - -- 支持模型示例：`qwen3-max`、`qwen3.7-plus`、`qwen-plus`、`qwen-flash`、`qwen3-coder-plus` 等。 -- `previous_response_id` 需传入上一轮响应的顶层 `id`（`resp_xxx`），而非 `output` 数组内消息的 `id`；当前响应 `id` 有效期为 **7 天**。 - -### Conversations（会话管理） - -提供会话的创建、查询、更新、删除及消息项管理。配合 Responses API 可自动注入历史上下文，实现跨设备、跨会话的对话延续。初始消息项 `items` 最多 20 条，`metadata` 最多 16 对键值对（key ≤ 64 字符、value ≤ 512 字符）。删除会话时其消息项不会被删除。 - -### Completions（文本补全） - -专为代码补全、内容续写设计，当前仅支持 `qwen-coder-turbo`，且**仅适用于中国内地（北京地域）**。通过 `<|fim_prefix|>...<|fim_suffix|>...<|fim_middle|>` 模板可实现「前缀生成后续」或「前缀+后缀生成中间」两种补全（暂不支持仅凭后缀生成前缀）。关键参数包括 `max_tokens`、`temperature`、`top_p`、`stop`、`seed`、`presence_penalty` 等，详见 [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 - -### Embedding（文本向量） - -兼容 OpenAI Embedding 规范，支持 `text-embedding-v1/v2/v3/v4`。其中 v3、v4 支持通过 `dimensions` 参数指定向量维度（v4 可选 64~2048 多档，默认 1024）。 - -> **注意**：多模态 Embedding 模型（如 qwen3-vl-embedding、tongyi-embedding-vision 系列）不支持 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，需改用[多模态向量接口](https://help.aliyun.com/zh/model-studio/multimodal-embedding-api-reference)。 - -### Vision（视觉理解） - -Qwen-VL、QVQ、Qwen-OCR 兼容 OpenAI Chat 接口，通过 `content` 数组中的 `image_url` 传入图片。各地域支持的模型有差异。QVQ 模型仅支持[流式输出](../concepts/streaming.md)。 - -### 文件接口与 Batch - -文件上传接口（`client.files.create`）通过 `purpose` 区分用途，详见 [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)： - -| purpose | 用途 | 单文件大小上限 | -| --- | --- | --- | -| `file-extract` | Qwen-Long / Qwen-Doc-Turbo 文档问答与数据提取 | 150 MB | -| `batch` | 批量推理输入（jsonl） | 500 MB | -| `fine-tune` | 模型调优数据集（jsonl） | 300 MB | - -百炼存储空间上限为 10000 个文件、总计 100 GB，达到任一上限后新上传会失败，需删除文件释放配额。上传返回的文件 ID（如 `file-batch-xxx`）可重复使用。 - -百炼提供两种批量推理方式，费用均约为实时调用的 **50%**： - -- **Batch（文件输入）**：上传 jsonl 文件异步批处理，适合大批量、时效性要求不高的场景（数据分析、模型评测）。可先用测试模型 `batch-test-model` 做全链路验证（文件 ≤ 1 MB、≤ 100 行、最大并行 2 个任务，不产生推理费用）。 -- **Batch Chat**：保持与实时 API 一致的同步调用方式，仅需将 `base_url` 改为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1`，单次仅支持一个请求；默认等待超时 3600 秒（可设 60~3600 秒）。 - -> **注意**：Batch 场景下 `enable_thinking` 须作为请求 body 的顶层参数（与 `model` 同级）传入，不能放在 `extra_body` 中；`qwen3.7`/`qwen3.6`/`qwen3.5` 系列默认开启思考模式，会产生额外思考 Token 成本，建议显式设置。 - -## 框架集成（LangChain） - -百炼可通过两条路径接入 LangChain（Python / JavaScript / Java），详见 [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)： - -- **OpenAI 兼容路径**：使用 `langchain_openai.ChatOpenAI`（JS 为 `@langchain/openai`，Java 为 `langchain4j-open-ai`），配置 `base_url` 指向 `compatible-mode/v1`。**仅支持 OpenAI 兼容模式覆盖的部分模型**。 -- **DashScope 原生路径**：使用 `ChatTongyi`（`langchain-community` + `dashscope`）或 JS 的 `ChatAlibabaTongyi`，**支持百炼所有文本生成模型（含部署后的模型）**。 - -> **注意**：LangChain4j 1.0.0-beta3 需要 Java 17 及以上版本，使用 Java 11 编译会报 `Unsupported class file major version 61` 错误。 - -## 限制与注意事项 - -- **地域隔离**：API Key 与 `base_url` 均按地域区分，跨地域调用必须成对更换；不同接口/模型在各地域的可用性存在差异，以[百炼控制台](https://bailian.console.aliyun.com/)为准。 -- **协议差异**：并非所有模型都支持 OpenAI 兼容协议（如 Qwen-Audio、多模态 Embedding），此类模型需使用 DashScope 原生协议。 -- **端点区别**：普通请求走各地域 `compatible-mode/v1`，而 Batch Chat 使用独立的 `batch.dashscope.aliyuncs.com` 域名。 -- 调用失败时请参考[错误码](https://help.aliyun.com/zh/model-studio/error-code)排查。 +阿里云百炼平台提供多种 OpenAI 兼容的工具包与框架接口，支持开发者快速迁移现有应用。核心能力覆盖文本生成（Chat、Completions、Responses）、[多模态](../concepts/multi-modal.md)理解（Vision）、向量化（Embedding）、批量处理（Batch）、会话管理（Conversations）及文件操作（Files），并兼容主流开发框架如 LangChain。所有接口均基于统一的 `compatible-mode/v1` 路径设计，通过调整 `base_url`、`api_key` 和 `model` 即可完成集成。 + +## 支持的模型/功能 + +百炼支持的 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)按功能划分如下： + +- **Chat Completions**：适用于标准对话场景，支持 Qwen 系列（`qwen-plus`、`qwen-flash` 等）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及第三方直供模型（DeepSeek、Kimi、GLM、MiniMax）[原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **Completions**：专为代码补全与内容续写设计，当前仅支持 `qwen-coder-turbo` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 +- **Responses**：作为 Chat Completions 的演进版，内置联网搜索、网页抓取等智能体原生工具，支持 `qwen3.7-plus`、`qwen3.6-flash` 等新一代 Qwen3 系列模型，并在华北2（北京）、新加坡、弗吉尼亚等多地部署 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 +- **Vision**：支持视觉理解任务，兼容 `qwen3-vl-plus`、`QVQ`、`qwen-vl-ocr` 等模型，支持图像 URL 与 Base64 输入 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 +- **Embedding**：提供 `text-embedding-v4`、`v3`、`v2`、`v1` 四代文本向量模型，支持多语种及可选维度（如 `dimensions=1024`），但[多模态](../concepts/multi-modal.md) Embedding 模型（如 `qwen3-vl-embedding`）**不支持** [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 +- **Batch**：分为两种模式： + - **文件批量（Batch File）**：通过 JSONL 文件异步提交请求，支持 `qwen3.7-max`、`qwen3-vl-plus`、`text-embedding-v4` 等模型，费用为实时调用的 50% [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md)； + - **同步 Batch Chat**：单请求同步等待返回，适用于数据标注等非实时场景，端点为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md)。 +- **Conversations**：用于跨设备/长时间会话状态管理，配合 Responses API 自动注入历史上下文，支持创建、查询、更新、删除会话及添加消息项 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 +- **Files**：支持上传文件用于文档问答（`purpose="file-extract"`）、批量推理（`purpose="batch"`）或模型调优（`purpose="fine-tune"`），最大单文件 150 MB（extract）、500 MB（batch）、300 MB（fine-tune） [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 + +> **注意**：文档 1 和文档 5 均提及旧域名迁移建议（如 `dashscope.aliyuncs.com` → `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），但文档 6 的“前提条件”中仍列出 `https://dashscope.aliyuncs.com/compatible-mode/v1` 为中国内地服务端点，与文档 1 的推荐实践存在不一致。实际生产环境应优先采用业务空间专属域名以保障性能与稳定性。 + +## 关键参数 + +| 参数 | 类型 | 必选 | 说明 | 适用接口 | +|------|------|------|------|----------| +| `base_url` | string | 是 | 接口根地址，需按地域和功能选择（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。`{WorkspaceId}` 须替换为控制台获取的实际 ID | 所有 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) | +| `model` | string | 是 | 模型名称，必须从各接口支持列表中选取（如 `qwen-plus`、`text-embedding-v4`） | 所有 OpenAI 兼容接口 | +| `stream` | boolean | 否 | 是否启用[流式输出](../concepts/streaming-output.md)，默认 `false`；流式响应中可通过 `stream_options={"include_usage": true}` 在末尾返回 token 统计 | Chat、Completions、Vision、Responses | +| `max_tokens` | integer | 否 | 最大生成 token 数，超限将截断输出（不影响模型内部生成逻辑） | Chat、Completions、Responses | +| `temperature` / `top_p` | float | 否 | 控制生成多样性，二者互斥，建议只设其一（`temperature ∈ [0,2)`，`top_p ∈ (0,1]`） | Chat、Completions、Responses | +| `enable_thinking` | boolean | 否 | Batch 场景下控制思考模式开关（`true`/`false`），影响 token 成本；须与 `model` 同级传入，不可置于 `extra_body` 内 | Batch Chat、Batch File | +| `dimensions` | integer | 否 | Embedding 接口专用，指定向量维度（仅 `text-embedding-v3`/`v4` 支持） | Embedding | +| `purpose` | string | 是（Files） | 文件上传用途：`file-extract`（文档分析）、`batch`（批量输入）、`fine-tune`（调优数据集） | Files | + +## 使用方式 + +### 基础 SDK 配置（Python 示例） +```python +from openai import OpenAI +import os + +client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), # 强烈建议配置至环境变量 + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" # 替换 {WorkspaceId} +) +``` + +### 各接口典型调用 +- **Chat**：`client.chat.completions.create(model="qwen-plus", messages=[...])` +- **Completions**：`client.completions.create(model="qwen-coder-turbo", prompt="{code_prefix}")` +- **Responses**：`client.responses.create(model="qwen3.7-plus", input="你好！")` +- **Vision**：`client.chat.completions.create(model="qwen3-vl-plus", messages=[{"role":"user","content":[{"type":"text","text":"这是什么"},{"type":"image_url","image_url":{"url":"..."}}]}])` +- **Embedding**：`client.embeddings.create(model="text-embedding-v4", input="文本", dimensions=1024)` +- **Batch Chat**：使用 `base_url="https://batch.dashscope.aliyuncs.com/compatible-mode/v1"`，调用方式与 Chat 完全一致 +- **Conversations**：`client.conversations.create(items=[{"role":"system","content":"..."}])` +- **Files**：`client.files.create(file=Path("doc.pdf"), purpose="file-extract")` + +LangChain 集成详见 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)，推荐 `langchain_openai.ChatOpenAI`（部分模型）或 `langchain_community.chat_models.tongyi.ChatTongyi`（全模型支持）。 + +## 限制和注意事项 + +- **地域与模型绑定**：并非所有模型在所有地域可用。例如 `qwen3.7-max` 在北京、新加坡、弗吉尼亚、法兰克福、东京均支持，但 `qwen3.5-397b-a17b` 仅在北京、新加坡、法兰克福提供；`qwen3.7-plus` 在东京仅支持日本部署范围 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 +- **三方模型可用性**：DeepSeek、Kimi 等第三方直供模型**仅在中国站的中国内地地域可用**，且需在控制台单独开通服务 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **协议限制**：`Qwen-Audio` 不支持 OpenAI 兼容协议，仅支持 DashScope 原生协议；`QVQ` 模型仅支持[流式输出](../concepts/streaming-output.md) [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 +- **文件配额**：百炼文件存储上限为 **10,000 个文件** 或 **100 GB 总大小**，任一达到即拒绝新上传 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 +- **Batch 超时**：Batch Chat 默认等待 3600 秒（1 小时），超时后连接断开并返回错误；Batch File 任务最长等待时间为 `completion_window`（如 `"24h"`），需主动轮询状态 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md)。 +- **API Key 隔离**：不同地域（如北京 vs 新加坡）的 API Key **不可混用**，需分别获取并配置 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 ## 来源文档 - [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) -- [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) +- [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) - [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) -- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) +- [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) - [OpenAI Conversations接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md) - [在LangChain中使用阿里云百炼](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md index 2a8da565..38e16587 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md @@ -1,96 +1,59 @@ # vector and sort -百炼平台围绕"向量化"与"排序"提供了一整套模型 API，覆盖通用文本向量、多模态向量与文本/多模态重排序三大能力。它们共同服务于语义搜索、推荐、聚类、分类与 RAG 检索：向量模型负责把文本、图片、视频编码为同一语义空间中的数值向量，排序（rerank）模型则在召回阶段之后对候选结果做二次精排，提升最终相关性。 +`vector and sort` 是百炼平台提供的核心向量化与排序能力集合，涵盖文本、[多模态](../concepts/multi-modal.md)内容的向量生成（embedding）以及跨模态/纯文本的语义相关性重排序（rerank）。该能力支撑语义搜索、RAG、推荐系统、聚类等典型AI应用，支持同步、异步及OpenAI兼容调用方式，适用于从单条文本到百万级批量数据的不同场景。 -## 能力与模型总览 +## 支持的模型/功能 -按用途可分为三类接口，分别对应不同的 endpoint 与调用方式： +### 文本向量模型（Embedding） +- **通用文本向量**：支持 `qwen3.7-text-embedding`、`text-embedding-v4`、`text-embedding-v3`、`text-embedding-v2`、`text-embedding-v1` 等系列模型，提供 64–2560 维可选向量，覆盖 201 种语种 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **批处理文本向量**：`text-embedding-async-v2`（最大 100,000 行/请求，单行 ≤2,048 [Token](../concepts/token.md)）和 `text-embedding-async-v1`，专为大规模离线向量化设计 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **[多模态](../concepts/multi-modal.md)向量**：支持文本、图像、视频统一语义空间编码，包括 `qwen3-vl-embedding`（支持独立/融合向量）、`tongyi-embedding-vision-plus-2026-03-06`（支持多分辨率 `res_level` 和视频帧数控制 `max_video_frames`）等 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 -- **通用文本向量（同步）**：将字符串 / 字符串列表 / 文件转为向量，实时返回。支持 `qwen3.7-text-embedding`、`text-embedding-v4/v3/v2/v1`。详见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 -- **通用文本向量（批处理）**：面向大规模离线向量化，仅支持异步模式，通过文件 URL 输入。支持 `text-embedding-async-v2/v1`。详见 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **多模态向量**：将文本、图片、视频编码到同一语义空间，支持跨模态检索与融合表征。支持 `qwen3-vl-embedding`、`qwen2.5-vl-embedding`、`tongyi-embedding-vision-plus/flash`（含 `2026-03-06` 快照版）、`multimodal-embedding-v1`。详见 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 -- **文本 / 多模态排序**：对召回文档做精排，支持 `qwen3-rerank`、`qwen3-vl-rerank`（多模态）、`gte-rerank-v2`。详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 +### 排序模型（Rerank） +- **纯文本排序**：`qwen3-rerank`（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，最大 500 文档/请求，单文档 ≤4,000 [Token](../concepts/token.md)），已替代即将下线的 `gte-rerank` 系列 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 +- **[多模态](../concepts/multi-modal.md)排序**：`qwen3-vl-rerank` 支持文本、图片、视频混合查询与文档排序（如“以图搜文”、“以文搜视频”），最大支持 100 文本/40 图片/4 视频文档 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 -> **注意**：`gte-rerank` 模型将于 2026-05-30 下线，官方推荐迁移到 `qwen3-rerank`。新项目请直接选用 `qwen3-rerank` / `qwen3-vl-rerank`。 +> **注意**：文档 4 明确指出 `gte-rerank` 模型将于 2026 年 05 月 30 日下线，新项目应使用 `qwen3-rerank` 或 `qwen3-vl-rerank`，避免依赖已废弃模型。 -## 通用文本向量 +## 关键参数 -### 同步接口 +| 参数 | 适用模型 | 说明 | 是否必选 | +|------|----------|------|----------| +| `model` | 所有 | 模型名称，如 `"text-embedding-v4"`、`"qwen3-rerank"` | 必选 | +| `input` / `query` + `documents` | 所有 | 向量：字符串、字符串列表或文件 URL；排序：`query`（字符串或模态对象）+ `documents`（字符串列表或模态对象数组） | 必选 | +| `dimensions` | `qwen3.7-text-embedding`, `text-embedding-v3/v4`, `qwen3-vl-embedding`, `tongyi-embedding-vision-plus-2026-03-06` 等 | 指定向量维度，值域因模型而异（如 `text-embedding-v4`: 64–2048；`qwen3-vl-embedding`: 256–2560） | 可选（默认值见各模型概览） | +| `top_n` | `qwen3-rerank`, `qwen3-vl-rerank`, `gte-rerank-v2` | 返回前 N 个最相关结果 | 可选 | +| `enable_fusion` | 仅 `qwen3-vl-embedding` | `true` 时将 `contents` 中所有模态融合为 1 个向量；`false`（默认）则各模态独立生成向量 | 可选（仅该模型） | +| `instruct` | `qwen3-rerank`, `qwen3-vl-rerank` | 任务指令（如 `"Retrieve semantically similar text."`），影响排序策略 | 可选 | +| `res_level` / `max_video_frames` | 仅 `tongyi-embedding-vision-plus-2026-03-06` / `tongyi-embedding-vision-flash-2026-03-06` | 分辨率档位（0–3）和视频最大采样帧数（≤64） | 可选 | -- **兼容方式**：提供 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，可用 OpenAI SDK 直连。 - - base_url：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - - endpoint：`POST .../compatible-mode/v1/embeddings` - - 调用前需将 `{WorkspaceId}` 替换为真实业务空间 ID。 -- **关键参数**： - - `model`（必选）：模型名称。 - - `input`（必选）：`string` / `array` / `file` 三种形态。 - - `dimensions`（可选）：仅 `text-embedding-v3/v4`（及 `qwen3.7-text-embedding` 的 2560 维）支持自定义维度，取值 2560/2048/1536/1024/768/512/256/128/64，默认 1024。 - - `encoding_format`（可选）：当前仅支持 `float`。 -- **输入上限（按模型区分）**： - - `qwen3.7-text-embedding`：单条字符串最长 128,000 Token；列表/文件最多 20 条。 - - `text-embedding-v3/v4`：单条 8,192 Token；列表/文件最多 10 条。 - - `text-embedding-v1/v2`：单条 2,048 Token；列表/文件最多 25 条。 +## 使用方式 -> **注意**：`dimensions` 只对部分模型生效——`text-embedding-v1/v2` 为固定维度（分别 1536 / 1536），传入该参数无意义；`v4` 才支持 2048/1536 等高维度。选维度前请对照模型概览表。 +### 同步调用（推荐小批量） +- **文本向量**：使用 OpenAI 兼容 SDK 或 HTTP POST 到 `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings`，支持 `input` 为字符串、列表或文件流 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **排序**：`qwen3-rerank` 使用 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks`；`qwen3-vl-rerank` 使用专用接口 `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank`。 -### 批处理接口 +### 异步调用（推荐大批量） +- **文本向量**：通过 `X-DashScope-Async: enable` 头发起批处理任务，上传含文本的 OSS URL，再轮询 `GET /api/v1/tasks/{task_id}` 获取结果 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **多模态向量/排序**：暂不支持异步模式，需同步调用。 -批处理专用于大批量离线场景，特点是**仅支持异步**： +### SDK 封装 +DashScope SDK 提供 `BatchTextEmbedding`（批向量）、`TextReRank`（排序）等高层封装，自动处理地域配置、认证与响应解析，降低集成复杂度。示例见各文档中 Python/Java SDK 调用片段。 -- endpoint：`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/text-embedding/text-embedding`。 -- HTTP 请求**必须**带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- 输入通过 `input.url` 传入文件 HTTP URL（一行一条），单行最长 2,048 Token、最多 100,000 行、文件不超过 200MB。 -- `parameters.text_type` 可选 `document`（默认）或 `query`；检索类非对称任务建议区分 query / document。 -- 调用两步走：创建任务拿到 `task_id` → `GET .../api/v1/tasks/{task_id}` 轮询结果。任务状态含 PENDING / RUNNING / SUCCEEDED / FAILED / CANCELED / UNKNOWN。 -- **数据时效**：任务结果 URL 仅保留 24 小时，务必及时下载，详见 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **限流**：`text-embedding-async-v2` 任务下发 RPS 为 1，排队+运行作业不超过 50 个，同时并发运行不超过 3 个。 +## 限制和注意事项 -## 多模态向量 - -多模态向量把 text / image / video 编码进**同一语义空间**，可直接用余弦相似度做跨模态匹配（以文搜图、以图搜视频等）。 - -- endpoint：`POST https://dashscope.aliyuncs.com/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding`。 -- 输入通过 `input.contents` 数组传入，每个元素为 `{"模态类型": "值"}`，支持 `text` / `image` / `video` / `multi_images` 四种类型。图片可用 URL 或 Base64 Data URI，视频仅支持公开 URL。 -- **独立向量 vs 融合向量**： - - 独立向量：为每个输入分别生成一个向量，适合逐项对比（以图搜图）。 - - 融合向量：将所有输入融合为 1 个向量，适合整体语义表征（如商品图+描述文本）。 - - `qwen3-vl-embedding` 通过 `enable_fusion=true` 开启融合；`tongyi-embedding-vision-*-2026-03-06` 则通过把 text/image/video 放进同一个 content 对象来生成融合向量（不使用 `enable_fusion`）。 -- **关键参数（在 `parameters` 内）**：`dimension`（不同模型取值不同）、`output_type`（仅 `dense`）、`fps`（视频帧率比例 [0,1]）、`instruct`（任务说明，建议英文）、`res_level`（分辨率档位 0/1/2/3，仅 2026-03-06 版）、`max_video_frames`（最大采样帧，≤64，仅 2026-03-06 版）。 - -> **注意**：各模型的向量类型能力差异明显——`qwen2.5-vl-embedding` **仅**支持融合向量、不支持独立向量与多图；`tongyi-embedding-vision-plus/flash`（非快照版）**仅**支持独立向量；`multimodal-embedding-v1` 与 `tongyi-embedding-vision-plus/flash` 不支持 `dimension` 参数（维度固定）。选型前务必核对 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) 中的"模型能力对照"表。 - -## 文本 / 多模态排序（Rerank） - -排序模型对召回文档二次精排，返回相关性分数。不同模型使用不同接口： - -- `qwen3-rerank`：`POST .../compatible-api/v1/reranks`，且 `query` / `documents` / `top_n` / `instruct` 与 `model` **同层级**（不使用 `input` / `parameters` 包装）。 -- `qwen3-vl-rerank`（多模态）/ `gte-rerank-v2`：`POST .../api/v1/services/rerank/text-rerank/text-rerank`，参数需包装进 `input` 与 `parameters` 对象。 - -关键参数与返回： - -- `query`（必选）：最大 4,000 Token；`qwen3-vl-rerank` 支持 `{"text": ...}` 或 `{"image": ...}` 对象形式。 -- `documents`（必选）：候选文档数组；`qwen3-vl-rerank` 每项可为 `text` / `image` / `video`。 -- `top_n`（可选）：返回前 N 条，默认全部。 -- `return_documents`（可选，默认 `false`）：是否回带原文，仅 `gte-rerank-v2` / `qwen3-vl-rerank` 支持。 -- `instruct`（可选）：仅 `qwen3-rerank` / `qwen3-vl-rerank` 生效，用于切换问答检索 / 语义相似度等排序策略，建议英文。 -- `fps`（可选）：仅 `qwen3-vl-rerank` 支持，控制视频抽帧比例。 -- 返回 `results` 按 `relevance_score`（0.0–1.0）降序排列，`index` 对应输入原始位置。 - -> **注意**：`relevance_score` 是**单次请求内的相对分数**，仅用于本次请求内排序，不可作为跨请求比较的绝对阈值。此外两类接口响应结构不同——`qwen3-rerank` 的 `results` 位于响应顶层且无 `output` 对象，其余模型结果在 `output.results` 内，详见 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -## 通用限制与注意事项 - -- **前提条件**：所有接口都需先[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并配置到环境变量 `DASHSCOPE_API_KEY`；SDK 调用还需安装 DashScope SDK。 -- **地域**：同步向量与 rerank 走 `maas.aliyuncs.com`（需替换 `{WorkspaceId}`），多模态与批处理走 `dashscope.aliyuncs.com`；新加坡地域需将 base_url 换为 `dashscope-intl.aliyuncs.com`。 -- **SDK 与 HTTP 差异**：HTTP 使用嵌套的 `input` / `parameters` 结构，DashScope SDK 多为扁平参数，开发时注意区分。 -- **超长截断**：rerank 中单条超过"单条最大输入 Token"会被截断，API 仅基于截断后内容计算，可能影响排序准确性。 -- **限流与错误码**：触发条件参见平台[限流](https://help.aliyun.com/zh/model-studio/rate-limit)文档，失败响应通过 `code` / `message` 指明原因，对照[错误码](https://help.aliyun.com/zh/model-studio/error-code)排查。 +- **[Token](../concepts/token.md) 与尺寸限制**： + - `qwen3.7-text-embedding` 单文本最长 128,000 Token；`text-embedding-v4` 仅 8,192 Token；`qwen3-vl-embedding` 文本限 32,000 Token，图片 ≤10 MB，视频 ≤50 MB [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 + - `qwen3-rerank` 单次请求总 Token = `Query Tokens × Document 数量 + Document Tokens 总和`，上限 120,000；`qwen3-vl-rerank` 文本文档上限 100 条，图片上限 40 条 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 +- **免费额度与计费**：各模型均有 90 天有效期的免费额度（如 `text-embedding-v4` 100 万 Token），超限后按实际消耗计费；注意 `text-embedding-async-v2` 单价为 0.0007 元/千 Token，而 `text-embedding-v4` Batch 调用为 0.00025 元/千 Token [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **地域与 endpoint 差异**：北京地域使用 `cn-beijing.maas.aliyuncs.com`，新加坡地域需替换为 `ap-southeast-1.maas.aliyuncs.com`；多模态向量统一使用 `dashscope.aliyuncs.com` 公共域名 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- **模型能力差异**：`tongyi-embedding-vision-plus` 固定 1152 维，不支持 `dimension` 参数；`multimodal-embedding-v1` 不支持 `dimension` 且仅支持中英文；`qwen2.5-vl-embedding` 仅支持融合向量，不支持 `enable_fusion` 参数 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 ## 来源文档 - [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md) -- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) -- [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) - [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md) +- [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) +- [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md index f4a55920..50b05f70 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md @@ -1,102 +1,118 @@ # video generation api -阿里云百炼平台提供覆盖多家厂商（万相 Wan、爱诗 PixVerse、Vidu、可灵 Kling、HappyHorse 等）的视频生成 API，支持文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑、数字人、人像驱动、视频超清与对口型等能力。所有视频生成任务均通过统一的异步调用模式完成，开发者先提交任务拿到 `task_id`，再轮询查询结果。 +百炼平台的 Video Generation API 提供多种视频生成与编辑能力，包括文生视频（T2V）、图生视频（I2V）、参考生视频（R2V）、视频编辑、口型替换、风格重绘等。所有接口均采用异步调用模式，任务创建后需轮询 `task_id` 获取结果，典型耗时为 1–5 分钟。开发者需确保模型、Endpoint URL 与 API Key 严格属于同一地域，跨地域调用将失败。 -## 统一调用模式：异步「创建任务 → 轮询获取」 +## 支持的模型/功能 -由于视频生成耗时较长（通常 1-5 分钟，个别统一编辑模型约 5-10 分钟），API 全部采用异步方式，流程分两步： +API 覆盖三大类能力： -1. **创建任务**：向 `video-synthesis` 端点发起 `POST` 请求，请求头必须带 `X-DashScope-Async: enable`（缺少会报错 `current user api does not support synchronous calls`），返回一个 `task_id`。 -2. **轮询获取**：用 `task_id` 发起 `GET https:///api/v1/tasks/{task_id}` 查询任务状态，直到完成并拿到视频 URL。 +- **基础生成类**：支持纯文本输入生成视频（如 `happyhorse-1.1-t2v`、`wan2.7-t2v-2026-06-12`、`vidu/viduq3-turbo_text2video`、`pixverse/pixverse-c1-t2v`），部分模型支持智能分镜或多镜头叙事（如 [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) 中通过 [prompt](../guides/prompt.md) 描述时间戳实现）。 + +- **[多模态](../concepts/multi-modal.md)驱动类**： + - 图生视频：支持首帧（`happyhorse-1.1-i2v`、`wan2.7-r2v-2026-06-12`）、首尾帧（`pixverse/pixverse-c1-kf2v`、`vidu/viduq3-turbo_start-end2video`）及视频续写； + - 参考生视频：支持传入多张图像/视频/音频（如 [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) 和 [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md)）； + - 视频编辑：支持指令式风格迁移（`wan2.7-videoedit`）、局部替换（`happyhorse-1.0-video-edit`）及超清增强（`pixverse/pixverse-upscale`）。 -其余通用约定： +- **人像动画类**：聚焦数字人与表情驱动，包括： + - 唱演/播报：`emo-v1`、`liveportrait`、`wan2.2-s2v`； + - 动作迁移：`animate-anyone-gen2`、`pixverse/pixverse-motioncontrol`； + - 口型替换：`videoretalk`、`pixverse/pixverse-lipsync`； + - 换人/复刻：`wan2.2-animate-mix`、`wan2.2-animate-move`。 -- `task_id` 有效期为 **24 小时**，过期无法查询（返回状态 `UNKNOWN`）；请勿重复创建任务，轮询即可。 -- 请求头 `Content-Type: application/json`、`Authorization: Bearer $DASHSCOPE_API_KEY` 为必填。 -- 新手可参考 [Postman 首次调用指引](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)。 - -> **注意**：绝大多数视频生成模型使用端点路径 `/api/v1/services/aigc/video-generation/video-synthesis`，但部分数字人/换人/图生动作类模型（[万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md)、[万相-图生动作](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md)、[万相-视频换人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md)、[万相2.2-首尾帧](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)）使用的是 `/api/v1/services/aigc/image2video/video-synthesis`。接入时请以对应文档端点为准。 - -## 支持的模型与功能 - -按厂商与任务类型划分，主要能力如下： - -- **万相 Wan（2.7 新版协议）**： - - 文生视频（`wan2.7-t2v-*`），支持通过 `prompt` 自然语言控制单/多镜头。 - - 图生视频（`wan2.7-i2v-*`），支持多模态输入（文本/图像/音频/视频），可完成首帧生视频、首尾帧生视频、视频续写三大任务。 - - 参考生视频（`wan2.7-r2v-*`），多主体参考（图像+视频+音色）。 - - 视频编辑（`wan2.7-videoedit`），指令编辑与视频迁移。 -- **万相 Wan（旧版协议，2.1-2.6）**：文生视频、图生视频-基于首帧、参考生视频（`wan2.6-r2v-flash`）、首尾帧生视频（`wan2.2-kf2v-flash`）、视频编辑统一模型（`wanx2.1-vace-plus`，支持多图参考、视频重绘等 `function`）。 -- **万相人物/数字人系列**：数字人 `wan2.2-s2v`（图片+音频，需先用 `wan2.2-s2v-detect` 检测图片）、图生动作 `wan2.2-animate-move`、视频换人 `wan2.2-animate-mix`（均含 `wan-std`/`wan-pro` 两种模式）。 -- **爱诗 PixVerse**：文生视频、图生视频、首尾帧生视频（`pixverse/pixverse-c1-*`、`-v6-*`、`-v5.6-*`）、参考生视频（`-r2v`）、视频超清（`pixverse/pixverse-upscale`，固定输出 4K）、视频对口型（`pixverse/pixverse-lipsync`，支持音频驱动或 TTS 文本）、视频动作模仿（`pixverse/pixverse-motioncontrol`）。 -- **Vidu**：文生视频、图生视频、首尾帧生视频、参考生视频（`vidu/viduq3-*`、`viduq2_*`）。 -- **可灵 Kling**：一个模型（`kling/kling-v3-video-generation`、`kling/kling-v3-omni-video-generation`）统一支持文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑，并支持智能分镜/多镜头（`multi_shot`、`shot_type`、`multi_prompt`）。 -- **HappyHorse**：文生视频、图生视频-基于首帧、参考生视频（多图像）、视频编辑。 -- **人像驱动系列（两步调用：先检测后生成）**：舞动人像 AnimateAnyone（图生舞蹈）、悦动人像 EMO（图生唱演，`style_level` 控制风格强度）、灵动人像 LivePortrait（图生播报）、表情包 Emoji（预设模板 `driven_id`）、声动人像 VideoRetalk（口型替换）、视频风格重绘 `video-style-transform`（8 种预设风格）。 +> **注意**：文档中存在协议版本冲突。例如，万相系列明确区分“旧版协议”（仅支持 wan2.6 及更早模型，如 [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)）与“新版协议”（仅支持 wan2.7 模型）。混用模型名与旧版 endpoint 将导致调用失败。 ## 关键参数 -请求体主要由 `model`、`input`、`parameters` 三部分构成： - -- `model`：模型名称，决定能力与协议版本。 -- `input`：任务输入。文生类用 `prompt`；图生/参考/编辑类多用 `media` 数组（`type` 可为 `image_url`/`first_frame`/`last_frame`/`video_url`/`audio_url`/`reference_image` 等），部分旧版模型用 `image_url`/`video_url`/`audio_url`/`first_frame_url`/`last_frame_url`/`ref_images_url` 等独立字段。 -- `parameters`：常见有 `resolution`（如 `480P`/`540P`/`720P`/`1080P`）、`size`（如 `1280*720`）、`duration`（秒）、`watermark`、`prompt_extend`（智能改写）、`audio`、`shot_type`/`multi_shot`（分镜）、`seed`、`style`/`style_level` 等，具体取值随模型不同。 - -> **注意**：多镜头控制方式在不同模型间不一致。万相 2.7 与 PixVerse-c1 通过 `prompt` 自然语言描述控制，设置 `shot_type` 不生效；而旧版万相 2.6（见 [万相-文生视频（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)）需显式设置 `shot_type: "multi"` 且 `prompt_extend: true` 才能启用多镜头。接入前务必确认所用模型的具体协议。 - -## 地域与域名 - -- **必须保证模型、Endpoint URL 与 API Key 属于同一地域**，跨地域调用会失败（鉴权失败或服务报错）。 -- 多数第三方模型（PixVerse、Vidu、Kling、数字人等）**仅支持华北2（北京）地域**；万相与 HappyHorse 部分能力还支持新加坡、美国（弗吉尼亚）、德国（法兰克福）等地域。 -- 百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，提供更高性能与稳定性，建议迁移： - - 华北2（北京）：`https://dashscope.aliyuncs.com` → `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - - 新加坡：`https://dashscope-intl.aliyuncs.com` → `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - `{WorkspaceId}` 为业务空间 ID，可在控制台「业务空间详情」查看；现有域名仍可正常使用。 - -## 限制与注意事项 - -- **版本选型**：万相已推出 2.7 新版协议，[万相2.7-图生视频](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) 支持首帧/首尾帧/续写三大任务，官方推荐优先选用；旧版 wan2.6 及更早模型仅支持首帧生视频。新旧协议接口不通用，`wan2.7-*` 只走新版协议。 -- **服务开通**：PixVerse、Vidu、Kling 等第三方模型需先在百炼控制台模型广场搜索并「立即开通」授权后方可调用。 -- **两步式模型**：数字人、AnimateAnyone、EMO、LivePortrait、Emoji 等需先调用对应的 `-detect` 检测模型确认图片合规（如清晰度、单人、正面），再调用生成模型。检测模型多为同步调用（如 `wan2.2-s2v-detect` 0.004 元/张）。 -- **限流**：视频生成模型通常「同时处理中任务数量」限制较低（多为 1，即同一时刻仅 1 个作业运行，其余排队），任务下发接口 RPS/QPS 约为 5，接入时需做好排队与重试。 -- **计费**：多按生成视频时长计费（如 LivePortrait 0.02 元/秒、EMO/VideoRetalk/AnimateAnyone 0.08 元/秒、数字人 720P 0.9 元/秒），`wan-pro` 等专业模式价格高于标准模式。 -- VideoRetalk 目前仅支持 API 调用，不支持控制台在线体验。 +所有请求必须包含以下通用参数： + +- **HTTP 头部（Headers）**： + - `Content-Type: application/json`（必选）； + - `Authorization: Bearer $DASHSCOPE_API_KEY`（必选）； + - `X-DashScope-Async: enable`（必选；同步调用不被支持）。 + +- **请求体（Body）**： + - `model`：精确模型标识符（如 `"wan2.7-t2v-2026-06-12"`），不可省略； + - `input`：根据任务类型结构化： + - 文生视频：`{"prompt": "..."}`； + - 图生视频：`{"media": [{"type": "image_url", "url": "..."}], "prompt": "..."}`； + - 首尾帧：`{"media": [{"type": "first_frame", ...}, {"type": "last_frame", ...}], "prompt": "..."}`； + - 口型替换：`{"media": [{"type": "video_url", ...}, {"type": "audio_url", ...}]}`； + - `parameters`：可选，常见字段包括： + - `duration`（秒，默认 5）； + - `resolution` 或 `size`（如 `"720P"`、`"1280*720"`）； + - `watermark`（布尔值，默认 `true`）； + - `aspect_ratio`（如 `"16:9"`，见 [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md)）； + - `style`（风格重绘专用，整数 0–7）。 + +## 使用方式 + +1. **准备环境**： + - 在百炼控制台开通对应模型服务； + - 获取目标地域的 [API Key](https://help.aliyun.com/zh/model-studio/get-api-key) 并配置至环境变量 `DASHSCOPE_API_KEY`； + - 获取业务空间 ID（WorkspaceId），用于构造专属 endpoint。 + +2. **发起异步任务**： + - 向 `https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` 发送 `POST` 请求； + - 所有模型共用该 endpoint（除少数遗留模型如 `wan2.2-kf2v-fla` 使用 `/image2video/` 路径，见 [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)）； + - 成功响应返回 `{"task_id": "xxx"}`，有效期 24 小时。 + +3. **轮询结果**： + - 使用 `GET https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/tasks/{task_id}` 查询状态； + - 当 `status` 为 `"SUCCESS"` 时，`output.video_url` 即为生成视频地址。 + +## 限制和注意事项 + +- **地域强一致性**：模型、API Key、Endpoint 必须同属一个地域（如华北2北京），否则鉴权失败或返回 `401 Unauthorized`。新加坡、美国、德国等地域 endpoint 格式不同，需严格匹配。 + +- **URL 构造规范**：业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）为推荐路径，旧域名（如 `https://dashscope.aliyuncs.com`）虽仍可用，但性能与稳定性较低。 + +- **并发与限流**： + - 多数模型对单账号 RPS/QPS 有限制（如 `liveportrait` 为 1 QPS，`emo-v1` 为 1 并发任务）； + - 免费额度按模型独立计算（如 `emo-detect-v1` 免费 200 张，`emo-v1` 免费 1800 秒）； + - 详细限流策略请查阅各模型资费文档。 + +- **输入约束**： + - 图像/视频 URL 必须公网可访问且 HTTPS 协议； + - 音频文件需为清晰人声（MP3/WAV），时长建议 ≤ 30 秒； + - Prompt 长度通常 ≤ 512 字符，含敏感词将触发拦截。 + +- **错误处理**：常见错误码包括 `400 Bad Request`（参数缺失或格式错误）、`403 Forbidden`（地域不匹配或配额不足）、`429 Too Many Requests`（超出限流）。建议在轮询逻辑中加入指数退避重试。 ## 来源文档 -- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) +- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) +- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) - [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) -- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) -- [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) -- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) +- [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) - [万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) -- [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) -- [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) - [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) -- [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) -- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) +- [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) -- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) +- [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) +- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) +- [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) +- [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) +- [爱诗-视频对口型API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) +- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) +- [爱诗-视频动作模仿API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) +- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) -- [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) - [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) -- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) -- [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) +- [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) - [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) +- [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) -- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) -- [爱诗-视频对口型API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) -- [爱诗-视频动作模仿API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) +- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md deleted file mode 100644 index c5ca2d6f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/api-invocation-comparison.md +++ /dev/null @@ -1,75 +0,0 @@ -# Qwen API vs 应用调用 vs 托管智能体API对比 - -百炼平台为开发者提供了三种主要的 API 调用方式：直接调用 Qwen 系列大模型（Qwen API）、调用已在控制台编排好的应用（应用调用 API）、以及通过托管智能体运行时管理完整的智能体生命周期（Managed Agents API）。三者在抽象层级、使用复杂度和适用场景上差异显著，本文帮助开发者根据业务需求做出技术选型。 - -## 核心定位 - -- **Qwen API**：直接访问基础大模型能力，开发者完全掌控对话编排与工具集成。 -- **应用调用 API**：调用控制台已配置好的应用（智能体/工作流），平台负责模型选择、提示词和工具编排。 -- **Managed Agents API**：平台托管智能体全生命周期（会话、沙箱、工具执行、事件流），开发者通过 REST 管理资源。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **抽象层级** | 模型层（底层） | 应用层（中层） | 运行时层（高层） | -| **调用对象** | Qwen 系列模型 | 控制台编排的应用（APP ID） | 平台托管的 Agent 实例 | -| **兼容协议** | OpenAI / Anthropic / DashScope 原生 | OpenAI Responses / DashScope 原生 | 百炼专有 REST API | -| **Endpoint 示例** | `POST /chat/completions` | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST /api/v1/agentstudio/sessions/{id}/events` | -| **认证方式** | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md) (DASHSCOPE_API_KEY) | [API Key](../concepts/api-key.md)（Bearer [Token](../concepts/token.md)） | -| **必需标识** | model 名称 | APP ID（+ 可选 Workspace ID） | workspace_id + agent_id | -| **对话历史管理** | 调用方自行维护（Responses 接口除外） | OpenAI Responses 模式自动管理；DashScope 模式需自行维护 | 平台托管，通过 Session/Event 机制自动管理 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其余需自行定义 | 由控制台应用配置决定，调用时无需关心 | Agent 配置挂载 Skill（zip 包）、Environment（沙箱） | -| **流式输出** | 支持 | 支持（stream=True） | SSE 事件流订阅 | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | 支持（background=True） | 原生异步：Session 状态机驱动 | -| **[多模态](../concepts/multimodal.md)** | 取决于具体模型能力 | 支持（OpenAI Responses 模式） | 支持（通过 File 资源挂载） | -| **沙箱/执行环境** | 无 | 无（平台内部处理） | 开发者可创建和管理 Environment | -| **版本控制** | 无（指定模型版本即可） | 无 | Agent 自动版本递增，Session 锁定创建时版本 | -| **SDK 兼容** | OpenAI SDK / Anthropic SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK / [DashScope SDK](../concepts/dashscope-sdk.md) | 需直接 HTTP 调用或自封装 | -| **迁移成本** | 低（直接复用 OpenAI/Anthropic 代码） | 中（需先在控制台配置应用） | 高（专有 API，需学习资源模型） | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -| --- | --- | --- | --- | -| **[计费](../concepts/billing.md)粒度** | [Token](../concepts/token.md) 用量（按模型计价） | [Token](../concepts/token.md) 用量（应用内模型调用） | Token 用量 + 可能的沙箱资源费用 | -| **文件配额** | 无 | 无 | 单文件 20MB，空间总量 100GB，保留 30 天 | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接控制模型参数（temperature、top_p 等）进行精细调优 -- 已有 OpenAI/Anthropic 代码希望低成本迁移到百炼 -- 构建自定义 RAG、Agent 框架，需要底层模型能力 -- 对工具调用逻辑有完全自主的编排需求 - -### 选择应用调用 API - -- 已在百炼控制台完成应用编排（提示词、知识库、插件），希望快速集成到业务系统 -- 团队中有非开发角色负责应用配置，开发者只需调用 -- 需要工作流（多步骤串联）能力但不想自行编排 -- 希望通过 OpenAI SDK 兼容方式接入已编排好的应用 - -### 选择 Managed Agents API - -- 需要平台托管智能体完整生命周期（创建、会话、工具执行、文件管理） -- 有复杂的工具执行需求，需要安全沙箱环境 -- 需要细粒度的会话状态管理和事件流订阅 -- 构建多智能体协作系统，需要独立管理每个 Agent 的版本和配置 -- 希望将工具包（Skill）作为可复用资产跨智能体共享 - -## 选型决策路径 - -1. **是否已在控制台配置好应用？** 是 -> 应用调用 API（最快集成） -2. **是否需要平台托管工具执行沙箱和会话状态机？** 是 -> Managed Agents API -3. **是否需要直接访问模型底层能力并自行编排？** 是 -> Qwen API -4. **从 OpenAI/Anthropic 迁移？** 优先 Qwen API 的兼容接口，迁移成本最低 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md deleted file mode 100644 index b3a96d1b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-eval-vs-model-eval.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供应用评测和模型评测两套独立的评测体系，分别面向不同的评测对象和使用场景。应用评测聚焦于智能体应用和工作流应用的端到端输出质量，覆盖 RAG 链路的各个环节；模型评测则聚焦于大语言模型本身的推理能力，用于模型选型和基础能力基准测试。理解两者的定位差异，有助于开发者在不同阶段选择合适的评测工具。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| **评测对象** | 智能体应用、工作流应用（已发布） | 文本生成类大模型 | -| **核心目标** | 评估应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型与调优验证 | -| **评测方式** | 自动评测（单应用/多应用横向）、手动评测 | 自定义评测（AI/规则/人工）、基线评测 | -| **评测集来源** | 基于知识库自动生成，或手动上传（xls/xlsx/jsonl） | 手动上传评测数据集（Prompt + Completion），或使用公开基线数据集 | -| **评估机制** | 新版：评估器（LLM/Code）+ 标签；旧版：内置评分模型 | 评测维度（大模型评估/规则评估/人工评估） | -| **自动评分方式** | LLM 评估器（语义理解）、Code 评估器（规则判断） | 裁判模型打分（数值型/分类型）、规则评估（ROUGE/BLEU/Cosine 等） | -| **人工标注** | 旧版手动评测；新版通过标签体系支持 | 人工评估维度（Pass/Fail 标注） | -| **横向对比能力** | 多应用横向评测（最多 8 个应用） | 排行榜（相同维度下对比多个模型） | -| **归因分析** | 支持（模型理解有误/重排不佳/检索无效/切片不完整/未获取知识） | 不支持链路归因，仅提供维度评分明细 | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可 | -| **支持的评分模型** | 评测集生成和评估仅支持 qwen-max、qwen-plus | 裁判模型推荐千问-Max，可选其他模型 | -| **基线能力评测** | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag） | -| **版本管理** | 新版评测集支持版本管理 | 不支持评测集版本管理 | -| **操作入口** | 控制台（应用管理模块） | 控制台（模型管理模块） | -| **API/SDK 支持** | 通过应用调用间接支持 | 不提供公开 API/SDK，仅控制台操作 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | - -## 评测数据与评分体系对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **数据集类型** | 旧版：对话分析（xls）、知识问答（jsonl）；新版：智能体/工作流/自定义 | 评测数据集（Prompt + Completion）、推理结果集 | -| **评分输出** | 1-5 分制（正确率 = 得分 >= 4 的占比） | 自定义评分范围（如 0-5）+ 通过阈值 | -| **结果分析** | 总正确率、BadCase 分析、调优建议、RAG 评价 | 综合得分、通过率、分数分布、逐样本明细 | -| **评估器/维度上限** | 每个任务最多 10 个评估器 | 无明确数量限制 | - -## 计费对比 - -| 对比项 | 应用评测 | 模型评测 | -|-------|---------|---------| -| **费用构成** | 评测集生成 [Token](../concepts/token.md) + 评估模型 [Token](../concepts/token.md) | 被评测模型推理 [Token](../concepts/token.md) + 裁判模型评分 Token | -| **免费方式** | 评估器模型当前限时免费 | 规则评估无裁判模型费用；已部署调优模型不额外计费 | -| **成本优化** | 合理控制评测集规模 | 先小规模验证（50-100 条）→ 保存推理结果集复用 → 优先规则评估 | - -## 适用场景建议 - -**选择应用评测的场景：** - -- 已构建完整的 RAG 智能体应用，需要评估端到端回答质量 -- 需要定位问题环节（模型理解、检索、重排、切片、知识库内容） -- 对多个应用版本进行 A/B 对比，验证迭代效果 -- 知识库更新、Prompt 调整、检索策略变更后的回归测试 -- 需要将人工标注经验固化为自动化评估规则（评估器） - -**选择模型评测的场景：** - -- 项目初期进行模型选型，对比多个候选模型的基础能力 -- 使用公开基准数据集（C-Eval、MMLU 等）快速了解模型水平 -- 模型微调后验证调优效果 -- 翻译、摘要、NL2SQL 等有确定性评判标准的任务，优先用规则评估降低成本 -- 需要在排行榜上持续跟踪模型表现 - -**组合使用建议：** - -在实际项目中，两种评测往往互补使用。典型流程是先通过模型评测筛选出基础能力最优的候选模型，再将其集成到应用中，通过应用评测验证端到端效果并持续优化 RAG 链路。模型评测解决"哪个模型更好"的问题，应用评测解决"应用整体表现如何优化"的问题。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md deleted file mode 100644 index 6adae0c7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-evaluation-vs-monitoring.md +++ /dev/null @@ -1,65 +0,0 @@ -# 应用评估与应用监控对比 - -阿里云百炼平台在应用上线运营的不同阶段提供了两类互补能力：**应用评估**负责在质量维度上系统化地衡量并调优应用输出，**应用监控（应用观测）**负责端到端追踪线上应用的调用链路与运行指标。两者常被同时使用，但目标、数据来源和使用方式差异明显。本文从技术选型角度对两者做对比，帮助开发者理清各自定位与协同关系。 - -## 核心定位 - -- **应用评估**：面向**质量**。通过评测集、评估器、标签构建多维度评测闭环，回答「应用回答得好不好、哪里出了问题、如何优化」。 -- **应用监控（应用观测）**：面向**运行时可观测性**。追踪应用内部调用链路、延时与 Token 消耗，回答「应用跑起来发生了什么、慢在哪里、花了多少 Token」。 - -## 关键维度对比 - -| 维度 | 应用评估 | 应用监控（应用观测） | -|------|----------|----------------------| -| 主要目标 | 系统化衡量并优化应用输出质量 | 端到端追踪线上运行状态与性能指标 | -| 支持应用类型 | 智能体应用、工作流应用（自动评测仅面向已发布且已配置知识库的智能体应用） | 智能体应用、工作流应用、高代码应用（暂不支持 Assistant API 创建的智能体；高代码仅观测入口 CHAIN 节点） | -| 数据来源 | 评测集（自动生成 / 手动上传 / 从应用观测导入） | 线上真实调用（自动追踪，分钟级同步） | -| 输入格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：智能体/工作流/自定义评测集 | 无需上传，应用发布并添加到观测列表后自动采集 | -| 输出格式 | 评测报告（总正确率、BadCase 分析、归因分析、调优建议） | Trace 列表、监控统计图表；可导出 JSONL / EXCEL | -| 评估/统计方式 | 自动评测（LLM 评估器）+ 手动评测（人工标注）+ Code 评估器 | 调用次数/失败率、Token 量、平均首 Token 耗时、平均调用时长等性能指标 | -| 评测/观测粒度 | 应用级别的回答质量、单应用或最多 8 个应用横向对比 | 调用链路节点级别（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等），支持 Root/All/Model Span | -| 是否有 API | 有评测任务相关能力（评测过程调用大模型） | **无 API**，仅可通过控制台操作 | -| 数据留存 | 评测集与报告持久化，评测集支持版本管理 | 调用记录最长保留 30 天 | -| 计费方式 | 评测调用大模型产生 Token 费用；Code 评估器无额外费用 | 功能本身不收费，观测数据存储由 OpenTelemetry 服务收费 | -| 前提条件 | 应用已发布 + 已配置知识库 + 开通应用观测 + 相应权限（管理员/应用评测-操作） | 完成应用观测配置（授权链路角色、开通 OpenTelemetry、初始化 LogStore）+ 相应权限 | -| 典型场景 | 版本迭代验证、模型/Prompt 调优、多应用选型、回归评测 | 线上问题排查、性能优化、成本分析、调用链路诊断 | - -## 组件与能力差异 - -### 应用评估的核心组件 - -- **评测模式**：自动评测（大模型基于知识库自动生成评测集并评分，支持单应用与最多 8 个应用横向评测）与手动评测（人工构建评测集并打分，属旧版能力）。 -- **评测集**：旧版分对话分析/知识问答两类；新版分智能体/工作流/自定义三类，支持版本管理与从应用观测导入。 -- **评估器**：预置模板、LLM 评估器、Code 评估器，或基于历史评测任务抽象生成；每个评测任务最多 10 个评估器，建议组合 3-5 个。 -- **归因分析**：将 BadCase 定位到 RAG 具体环节（模型理解有误、重排不佳、检索无效、切片不完整、未获取知识）并给出优化方向。 - -### 应用监控的核心能力 - -- **调用链路追踪**：按节点类型（CHAIN、AGENT、RETRIEVER、REWRITER、EMBEDDING、RERANKER、LLM、TOOL、GUARDRAIL 及工作流专属节点）展示嵌套调用关系。 -- **监控统计**：调用次数与失败率、Token 总量、平均单次请求 Token、平均首 Token 耗时、平均调用时长，支持分钟/小时/天聚合，最长 30 天。 -- **数据筛选与标注**：Span 筛选模式与多条件过滤器；支持对 Span 打标签（布尔/分类/数字/文本）。 -- **数据导出与回流**：导出 JSONL/EXCEL，并可将线上 Span 数据直接加入评测集（最多 50 个字段映射）。 - -## 两者的协同关系 - -应用评估与应用监控并非二选一，而是形成闭环： - -1. **观测提供样本**：应用监控采集的真实线上调用可直接导入评测集，让评测更贴近生产分布。 -2. **共享标签体系**：两者的标签管理统一，观测阶段的标注可复用到评测任务，反之亦然。 -3. **评测的前提依赖观测**：自动评测要求应用已开通应用观测并加入观测列表。 -4. **优化闭环**：监控发现异常（如延时高、失败率上升）→ 评估定位质量问题与归因 → 实施优化 → 再次观测与评测确认改进。 - -## 技术选型建议 - -- **要衡量「答得好不好」并做调优**：选择**应用评估**。尤其在知识库更新、Prompt 调整、模型升级、检索/重排策略变更后，或需要多应用/多版本选型时。 -- **要看「跑得怎么样、慢在哪、花多少 Token」**：选择**应用监控**。用于线上性能诊断、成本分析和调用链路排查。 -- **高代码应用**：只能用应用监控（且仅观测入口 CHAIN 节点），当前不在自动评测的支持范围内。 -- **需要程序化集成**：应用监控无 API，只能控制台操作；若强依赖自动化上报，需结合 AgentScope-AI 的 Tracing 模块与部署时的 `--telemetry enable` 参数。 -- **最佳实践**：先用应用监控开启持续观测积累真实调用数据，再用应用评估以这些数据为样本建立定期回归评测机制，形成「观测 → 评测 → 优化 → 再观测」的持续改进闭环。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md deleted file mode 100644 index c8136a11..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-monitoring-vs-model-monitoring.md +++ /dev/null @@ -1,70 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两种互补的可观测性能力：**应用观测**和**模型监控**。应用观测聚焦于应用内部的端到端调用链路追踪，帮助开发者理解智能体应用、工作流应用的执行过程；模型监控则聚焦于模型调用层面的运行状态与成本管理，提供用量统计、性能指标、日志审计和主动告警。两者分别从"应用维度"和"模型维度"保障系统的可观测性，开发者通常需要同时使用。 - -## 关键维度对比 - -| 对比维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（大语言模型、视觉模型、语音模型、向量模型等） | -| **核心目标** | 追踪应用内部调用链路，优化运营效果与成本 | 监控模型运行状态与用量，保障稳定性与成本可控 | -| **数据维度** | 按应用 + [业务空间](../concepts/workspace.md) | 按模型 + [业务空间](../concepts/workspace.md) | -| **数据刷新频率** | 分钟级 | 普通监控：小时级；高级监控：分钟级 | -| **数据保留期** | 最长 30 天 | 普通用量最长 30 天，更早需查账单 | -| **支持的应用/模型范围** | 智能体应用、工作流应用、高代码应用（不支持 Assistant API 创建的智能体应用） | 所有模型均支持用量统计；普通监控支持全地域全模型；高级监控限北京、新加坡、弗吉尼亚 | -| **链路追踪** | 支持（CHAIN、AGENT、LLM、RETRIEVER、TOOL 等多节点类型，可展开查看嵌套调用） | 不支持应用级链路追踪，仅记录单次模型调用 | -| **关键指标** | 调用次数/失败率、[Token](../concepts/token.md) 总量（输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时、平均调用时长 | 调用总量/失败量/失败率、平均调用时长、平均首包时长、RPM、TPM、限流错误次数(429)、内容安全错误次数 | -| **监控分类** | 无分类，统一展示性能与调用指标 | 四类：安全、成本、性能、错误 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域），可设置超时、Token 消耗突增等阈值 | -| **日志/历史对话** | 支持查看 Prompt 内容、输出、延时等完整调用记录 | 支持推理日志和历史对话记录（仅北京地域部分模型，需手动开通） | -| **Token 消耗追踪** | 按 Span 节点记录 Token 消耗 | 三层管理：汇总统计、单次调用追踪、阈值告警 | -| **数据筛选** | 按状态、Span Name、输入/输出内容、延时、Token 量、标签等多维筛选；支持 Request ID / Trace ID / Span ID 检索 | 按 API-KEY、推理类型、时间范围、时间精度筛选 | -| **数据标注** | 支持（布尔值/分类/数字/文本四种标签类型，与评测共享） | 不支持 | -| **导入评测集** | 支持将 Span 数据直接加入评测集作为评测样本 | 不支持 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出，需通过费用与成本页面查询 | -| **费用管理** | 不提供费用管理功能 | 提供费用概览、账单趋势、免费额度管理及用完即停开关 | -| **计费** | 功能本身免费，OpenTelemetry 存储费用另计 | 功能本身免费，高级监控可能涉及额外费用 | -| **开通方式** | 控制台手动开通（授权 OpenTelemetry 服务角色 + 开通服务 + 初始化 LogStore） | 系统自动采集（普通监控）；高级监控和日志需手动开通 | -| **API 支持** | 无 API，仅控制台操作 | 无专用 API，仅控制台操作 | -| **地域限制** | 无明确地域限制 | 普通监控无限制；高级监控限北京/新加坡/弗吉尼亚；告警限北京/新加坡；日志限北京 | - -## 适用场景建议 - -### 应用观测适用于 - -- **调试应用内部逻辑**：需要查看智能体应用或工作流应用的完整调用链路，定位某个节点（如检索、模型推理、插件调用）的异常或性能瓶颈。 -- **优化 RAG 效果**：通过 RETRIEVER、REWRITER、RERANKER 等节点的详细数据，分析检索召回质量和排序效果。 -- **构建评测数据集**：将线上真实调用数据标注后导入评测集，用于持续优化应用效果。 -- **应用级性能分析**：关注单个应用的整体调用时长、Token 消耗趋势，评估应用的运营效率。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用的失败率、限流错误(429)、内容安全错误等，及时发现异常。 -- **成本核算与控制**：按模型、按[业务空间](../concepts/workspace.md)统计 Token/图片/视频的用量与费用，管理免费额度。 -- **主动告警**：对关键模型设置超时、Token 消耗突增等告警规则，防止静默失败。 -- **合规审计**：通过推理日志记录每次模型调用的输入与输出，满足内容审计需求。 -- **多模型对比**：在监控列表中横向对比不同模型的性能与错误率，辅助模型选型。 - -### 建议同时使用的场景 - -当应用上线后需要全面保障服务质量时，建议同时开启两项能力：用应用观测定位"哪个环节出了问题"，用模型监控回答"模型本身是否正常、成本是否可控"。例如，当应用观测发现某次调用的 LLM 节点延时异常时，可切换到模型监控确认该模型是否存在全局性的性能劣化或限流。 - -## 技术选型参考 - -| 选型考量 | 推荐方案 | -| --- | --- | -| 需要追踪应用内部多节点调用链路 | 应用观测 | -| 需要监控模型全局运行状态和失败率 | 模型监控 | -| 需要对异常指标设置主动告警 | 模型监控 | -| 需要将线上数据导入评测集 | 应用观测 | -| 需要按模型维度统计费用和用量 | 模型监控 | -| 需要对调用数据打标签做质量分析 | 应用观测 | -| 需要审计模型调用的输入输出内容 | 模型监控（推理日志） | -| 需要端到端的可观测性保障 | 两者配合使用 | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md deleted file mode 100644 index 02011a75..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-evaluation.md +++ /dev/null @@ -1,64 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向智能体应用和工作流应用的端到端输出质量评估，**模型评测**面向文本生成类模型的基础能力评估。两者在评测对象、数据集格式、评分机制和使用场景上存在本质差异，开发者需根据自身需求选择合适的评测路径。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -|---------|---------|---------| -| 评测对象 | 智能体应用、工作流应用 | 文本生成类模型（含调优模型） | -| 核心目标 | 评估应用端到端输出质量 | 评估模型推理能力，辅助选型或验证调优效果 | -| 评测方式 | 自动评测（大模型生成评测集并评分）、手动评测（人工逐条标注） | 自定义评测（自有数据集 + 自定义维度）、基线评测（公开标准数据集） | -| 评测集格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：按应用出入参自动生成模板 | 统一格式：Prompt（用户问题）+ Completion（参考答案）两列 | -| 评测集创建 | 支持自动生成（知识问答类型，限 qwen-max/qwen-plus）和手动上传 | 在数据管理模块上传评测集（EvaluationSet）类型数据 | -| 评分机制 | 评估器体系：LLM 评估器（语义理解）、Code 评估器（规则判断）、预置模板 | 评测维度体系：大模型评估（数值型/分类型）、规则评估（文本相似度/字符串匹配）、人工评估 | -| 评分配置单元 | 评估器（最多 10 个/任务，建议 3-5 个组合） | 评测维度（按需配置，类型创建后不可修改） | -| 裁判模型 | 由评估器内部选择模型 | 推荐千问-Max | -| 规则评分算法 | Code 评估器自定义 Python 函数 | 内置 7 种算法（ROUGE-1/2/L、BLEU、Cosine、Fuzzy Match、Accuracy）+ 字符串匹配 | -| 人工标注 | 标签系统（分类/布尔值/数字/文本四种类型），支持快速标注模式 | 人工评估维度（Pass/Fail 分类型） | -| 排行榜 | 不支持 | 支持（相同维度下横向对比多模型，得分 0-100） | -| 横向对比 | 自动评测最多同时评测 8 个应用 | 通过排行榜对比多个模型 | -| 结果下载 | 支持 | 支持（基线评测除外） | -| 基线评测 | 不支持 | 支持（C-Eval、MMLU、ARC、GSM8K、BBH、HellaSwag），仅北京地域 | -| 版本管理 | 新版/旧版两套界面并存 | 统一界面 | -| 操作方式 | 控制台 | 仅控制台（无公开 API/SDK） | -| 前置要求 | 自动评测需已发布应用并配置知识库，需开通应用观测 | 无特殊前置，上传数据集即可 | - -## 评分体系差异 - -应用评测和模型评测虽然都支持"大模型评分"和"规则评分"，但实现方式不同： - -- **应用评测**的评估器是独立可复用的组件，同一评估器可在不同评测任务间共享。LLM 评估器和 Code 评估器各有优势，建议组合使用以覆盖语义理解和精确规则两个层面。 -- **模型评测**的评测维度在创建时即绑定评分方式，类型不可修改。规则评估提供开箱即用的标准算法（ROUGE、BLEU 等），无需编写代码，适合翻译、摘要等有确定性标准的场景。 - -## 计费差异 - -两套评测体系的费用结构相似，均包含推理费用和评分费用，但细节有所不同： - -- **应用评测**：调用大模型的 Token 费用正常计费，Code 评估器无额外费用。 -- **模型评测**：被评测模型推理费用按 Token 计费（使用推理结果集时不产生），大模型评估维度额外产生裁判模型费用，规则评估和人工评估无裁判模型费用。已部署的调优模型评测不额外计费。 - -两者共同的成本优化策略：先小规模验证（50-100 条），确认配置无误后再扩大规模。模型评测还支持保存推理结果集复用，避免重复推理。 - -## 适用场景建议 - -**选择应用评测**： -- 已构建完整的智能体或工作流应用，需要评估端到端输出质量 -- 需要领域专家介入进行人工标注和多维度质量评估 -- 需要结合应用观测进行线上真实数据的持续质量监控 -- 评测关注点是应用整体表现而非底层模型能力 - -**选择模型评测**： -- 处于技术选型阶段，需要在多个候选模型间做横向对比 -- 对模型进行了微调/调优，需要量化对比调优前后的能力变化 -- 需要使用公开基准数据集（C-Eval、MMLU 等）快速了解模型基础能力 -- 评测关注点是模型本身的推理和生成能力 - -**组合使用**：在实际项目中，建议先通过模型评测选定基础模型，再通过应用评测验证集成到应用后的端到端效果，形成"模型选型 → 应用构建 → 应用评测 → 持续监控"的完整闭环。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md deleted file mode 100644 index 291634a5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/app-vs-model-monitoring.md +++ /dev/null @@ -1,66 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的可观测能力：**应用观测**聚焦应用内部调用链路的端到端追踪，帮助开发者理解智能体、工作流等应用的执行过程；**模型监控**则聚焦模型维度的运行指标与成本核算，保障模型调用的稳定性与经济性。两者观测粒度、数据来源和使用场景各有侧重，开发者需要根据排查目标选择合适的工具，也可配合使用以获得从应用到模型的全链路可观测性。 - -## 关键维度对比 - -| 维度 | 应用观测 | 模型监控 | -| --- | --- | --- | -| **观测对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（所有模型，含调优后的自定义模型） | -| **核心目标** | 追踪应用内部调用链路，定位延时瓶颈与逻辑问题 | 监控模型运行状态与用量，保障稳定性与控制成本 | -| **数据粒度** | 单次请求的完整 Trace / Span 链路 | 按"模型 + [业务空间](../concepts/workspace.md)"维度的聚合指标 | -| **数据更新频率** | 分钟级 | 普通监控小时级，高级监控分钟级 | -| **数据保留时长** | 最长 30 天调用记录 | 用量统计保留 30 天，更早数据需到费用与成本页面查询 | -| **关键指标** | 延时、Token 量（输入/输出）、首 Token 耗时、调用次数、失败率 | 调用总量、失败率、平均调用时长、首包时长、RPM、TPM、Token 消耗 | -| **指标分类** | 按节点类型（CHAIN、LLM、RETRIEVER 等）查看 | 按安全、成本、性能、错误四类分类查看 | -| **链路追踪** | 支持，可展开查看完整 Span 树及每个节点的输入输出 | 不支持链路追踪，仅提供模型级聚合数据 | -| **日志能力** | Trace 详情中可查看 Prompt 内容与模型输出 | 需额外开通推理日志，仅北京地域部分模型支持 | -| **告警能力** | 不支持 | 支持主动告警（仅北京、新加坡地域） | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 不支持直接导出（需通过费用与成本页面获取账单数据） | -| **数据标注** | 支持对 Span 数据添加标签（布尔值/分类/数字/文本） | 不支持 | -| **与评测集联动** | 支持将 Span 数据直接加入评测集 | 不支持 | -| **地域限制** | 无特殊地域限制 | 高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，日志限北京 | -| **费用** | 功能免费，OpenTelemetry 存储另计 | 功能免费，高级监控的底层存储可能产生费用 | -| **开通方式** | 需手动配置：授权服务角色 → 开通 OpenTelemetry → 初始化 LogStore | 普通监控自动采集；高级监控和日志需手动开通 | -| **操作方式** | 仅控制台，无 API | 仅控制台，无 API | -| **用量/计费统计** | 不涉及计费统计 | 提供按模型类型的用量统计（Token/张/秒），可用于成本核算 | - -## 适用场景建议 - -### 应用观测适用于 - -- **应用调试与优化**：需要查看智能体或工作流应用内部各环节（检索、重写、模型推理、插件调用等）的执行顺序与耗时，定位性能瓶颈。 -- **Prompt 审查**：需要查看每次调用的具体 Prompt 内容与模型输出，排查回答质量问题。 -- **评测数据积累**：将线上真实调用数据标注后加入评测集，用于持续改进应用效果。 -- **单次请求排障**：通过 Request ID / Trace ID / Span ID 精确定位某一次调用的异常环节。 - -### 模型监控适用于 - -- **模型稳定性保障**：监控模型调用失败率、限流错误（429）、内容安全错误等指标，及时发现服务异常。 -- **成本管理**：按[业务空间](../concepts/workspace.md)和模型维度统计 Token 消耗和调用量，核算各模型的使用成本。 -- **容量规划**：通过 RPM（每分钟请求数）、TPM（每分钟 Token 数）等指标评估负载水平，预判扩容需求。 -- **主动告警**：设置 Token 消耗或错误率阈值，在异常发生时第一时间收到通知，避免静默故障。 -- **合规审计**：开通推理日志后可回溯历史对话的输入输出，满足内容审计需求。 - -### 建议组合使用 - -在生产环境中，建议同时开启应用观测和模型监控：应用观测负责应用层面的链路追踪与调试，模型监控负责基础设施层面的稳定性与成本管理。当模型监控发现某模型失败率上升时，可在应用观测中按时间范围筛选相关 Trace，进一步定位是应用逻辑问题还是模型服务问题。 - -## 技术选型参考 - -| 排查目标 | 推荐工具 | -| --- | --- | -| 某次调用为什么返回了错误答案 | 应用观测（查看 Trace 中各节点的输入输出） | -| 某个模型最近的失败率是否正常 | 模型监控（查看错误类指标） | -| 应用整体响应变慢，瓶颈在哪个环节 | 应用观测（对比各 Span 延时） | -| 本月 Token 消耗是否超出预算 | 模型监控（查看用量统计与费用概览） | -| 需要对线上 bad case 做标注并加入评测集 | 应用观测（数据标注 + 添加到评测集） | -| 模型调用量突增需要告警 | 模型监控（配置告警规则） | -| 回溯某次模型调用的完整输入输出 | 模型监控（推理日志，仅北京地域部分模型） | - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md deleted file mode 100644 index b2325ef4..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-call-vs-bailian-calling.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用调用方式对比 - -阿里云百炼的应用（智能体应用、[工作流](../concepts/workflow.md)应用、新版智能体 Agent 2.0）可通过 API 集成到业务系统中。官方文档中存在两篇高度相关的主题页：一篇偏 **API 参考**（`application call`），重点介绍两套调用 API（OpenAI 兼容 Responses API 与 DashScope 原生 `/completion`）的端点、参数与同步/异步模式；另一篇偏 **使用指南**（`bailian application calling`），重点介绍通过 DashScope SDK / HTTP 调用智能体与[工作流](../concepts/workflow.md)应用的实操步骤、多轮对话与自定义参数透传。本文对两者做维度对比，帮助开发者在技术选型时快速定位所需信息。 - -## 关键维度对比 - -| 维度 | [application call](../api/application-call.md)（API 参考） | bailian [application call](../api/application-call.md)ing（使用指南） | -| --- | --- | --- | -| 文档定位 | API 参考，强调端点、参数、调用模式 | 使用指南，强调 SDK 实操与场景示例 | -| 所属分类 | api | guides | -| 覆盖的 API 模式 | 两套：OpenAI 兼容 Responses API + DashScope 原生 `/completion` | 一套：DashScope 原生 `/completion`（`Application.call` / `POST /apps/{app_id}/completion`） | -| 主 Endpoint | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`（OpenAI 兼容）
`POST /api/v1/apps/{APP_ID}/completion`（DashScope 原生） | `POST /api/v1/apps/{APP_ID}/completion` | -| SDK 推荐 | OpenAI 兼容模式用 OpenAI SDK；DashScope 原生模式用 DashScope SDK | DashScope SDK（Python / Java），Node.js 用 `axios` | -| 支持应用类型 | 智能体、[工作流](../concepts/workflow.md)、新版智能体 Agent 2.0 | 智能体应用、工作流应用（智能体编排应用已被工作流应用替代） | -| 输入格式 | `input` 为 string 或 messages 数组；多模态支持 `input_text` / `input_image` / `input_file` | `input.prompt` 字符串 或 自行管理 `messages` 数组 | -| 多轮对话 | 通过 messages 数组传递完整对话历史；`pre_response_id` / `conversation_id` 上下文后续支持 | 两种方式：`session_id`（云端托管，1 小时 / 50 轮）或自行管理 `messages`（推荐） | -| 同步/异步 | 支持 `stream` 流式、`background` 异步；异步暂不支持流式 | 默认同步；多轮对话通过 `session_id` 或 `messages` 实现 | -| 多模态 | 显式支持图像、文件（`input_image` / `input_file`，文件仅智能体应用支持） | 未专门展开 | -| 自定义参数透传 | 未展开 | 支持 `biz_params.user_defined_params` 透传业务参数到自定义插件 / 工作流插件节点 | -| 业务空间 | 明确说明子业务空间需 Workspace ID，多地域 Base URL 含 Workspace ID | 提及业务空间对插件与应用关联的约束（同一业务空间内） | -| 地域说明 | 明确给出华北2（北京）默认 Endpoint，并列出德国、新加坡、日本等地域需带 Workspace ID | 未专门说明地域差异 | -| 典型代码示例 | Python（OpenAI SDK 同步多轮） | Python / Java / Node.js / curl（基础调用） | -| 凭证准备 | APP ID + Workspace ID + API Key + SDK | API Key + APP_ID + DashScope SDK | - -## 适用场景建议 - -- **选 `application call`（API 参考）**：当你需要复用现有 OpenAI 生态代码库与工具链；需要使用[流式输出](../concepts/streaming-output.md)或异步执行；需要调用新版智能体 Agent 2.0；需要多模态输入（图像、文件）；或部署在非北京地域需要明确 Workspace ID 与 Base URL 拼接规则时。 -- **选 `bailian application calling`（使用指南）**：当你首次接入百炼应用、需要 Python / Java / curl 的最小可运行示例；需要通过 `session_id` 实现云端托管多轮对话；需要向自定义插件或工作流插件节点透传业务参数（`biz_params.user_defined_params`）；或团队已习惯使用 DashScope SDK 时。 - -## 技术选型参考 - -两篇文档并非互斥，而是互补：`application call` 给出"调哪套 API、用什么端点、传哪些字段"的契约层信息，`bailian application calling` 给出"用哪个 SDK、怎么写代码、怎么传业务参数"的实操层信息。建议的选型路径： - -1. 先读 `application call` 确定调用模式（OpenAI 兼容 vs DashScope 原生），明确端点与参数契约； -2. 若选择 DashScope 原生 `/completion`，再读 `bailian application calling` 获取多语言 SDK 示例与多轮、自定义参数等进阶能力； -3. 若选择 OpenAI 兼容 Responses API，则以 `application call` 为主，参考 OpenAI SDK 既有用法，`bailian application calling` 中的 `session_id` / `biz_params` 等能力在该模式下暂不适用。 - -> 注：两篇文档对应用类型的命名略有差异（"新版智能体 Agent 2.0" vs "智能体编排应用已被工作流应用替代"），接入前请以控制台实际应用类型与最新 API 参考为准。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md deleted file mode 100644 index c3c897e5..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-calling-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用调用方式对比：API 直调与百炼应用调用 - -阿里云百炼平台为已编排好的应用（智能体、工作流、新版智能体 Agent 2.0）提供两套对外调用路径：一是面向 OpenAI 生态的 **Responses API（OpenAI 兼容模式）**，二是面向百炼原生的 **DashScope `Application.call` / `/completion` API**。两者底层均指向同一个 `APP_ID`，但在端点形态、SDK 选型、输入结构、多轮与多模态能力、扩展参数等方面存在差异。本文从技术选型视角对比两种方式，帮助开发者根据现有技术栈与功能需求做出取舍。 - -## 关键维度对比 - -| 维度 | OpenAI 兼容 Responses API（API 直调） | DashScope 原生 API（百炼应用调用） | -| --- | --- | --- | -| 调用端点 | `POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` | `POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` | -| SDK 选型 | OpenAI SDK（多语言） | DashScope SDK（Python / Java），或直接 HTTP | -| base_url 配置 | `https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1` | 无需 base_url，SDK 内置或直接 POST | -| 输入格式 | `input` 为字符串或消息数组，`role` 取 `system`/`user`/`assistant`，多模态 `content` 为数组（`input_text`/`input_image`/`input_file`） | `input.prompt` 字符串，或 `messages` 数组（自行管理多轮历史） | -| 输出格式 | OpenAI Responses 结构，`response.output` 等 | `{"output": {"finish_reason","session_id","text"}, "usage":{...}, "request_id":"..."}`，业务侧消费 `output.text` | -| 支持模型 | 智能体、工作流、新版智能体 Agent 2.0 | 智能体应用、工作流应用（[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代） | -| 同步/异步 | 支持 `background` 异步执行，同步默认；流式 `stream=true` | 主要为同步调用，`session_id` 由云端管理历史 | -| [流式输出](../concepts/streaming-output.md) | 支持（`stream=true`），异步暂不支持流式 | 通过 SDK / HTTP 支持（详见调用文档） | -| 多轮对话 | 传递完整 `input` 消息数组；基于 `pre_response_id`/`conversation_id` 的上下文能力后续支持 | 两种方式：`session_id`（云端托管，1 小时有效、最多 50 轮）或自行维护 `messages`（推荐，更灵活） | -| 多模态 | 原生支持文本、图像、文件（`input_file` 仅智能体应用支持） | 通过 `messages` 与应用内编排支持 | -| 自定义参数透传 | 通过 `input`/应用编排间接实现 | `biz_params.user_defined_params` 透传至自定义插件与工作流插件节点 | -| 业务空间 | 默认空间仅需 APP ID；子空间或海外地域需在请求中包含 Workspace ID | 同样需要 APP_ID；子空间按地域 Base URL 处理 | -| 典型场景 | 复用现有 OpenAI 代码库与工具链、多模态交互、统一 OpenAI 协议接入 | 全面功能与更高性能、自定义插件参数透传、Java/Node.js 直接 HTTP 集成 | - -## 适用场景建议 - -**OpenAI 兼容 Responses API 适合：** - -- 已有 OpenAI SDK 代码资产、希望以最小改动接入百炼应用的团队。 -- 需要多模态输入（文本 + 图像 + 文件）的智能体交互场景。 -- 希望统一在 OpenAI 协议生态下做模型/应用切换、保持代码中立。 -- 需要异步执行（`background`）与[流式输出](../concepts/streaming-output.md)能力的实时或长任务交互。 - -**DashScope 原生 `/completion` API 适合：** - -- 追求更全面功能与更高性能，使用百炼原生能力（如自定义插件参数透传 `biz_params`）。 -- Java/Node.js 项目希望直接以 HTTP 方式集成，不引入 OpenAI SDK 依赖。 -- 工作流应用需要通过 `session_id` 让云端托管对话历史，简化多轮实现。 -- 需要在工作流大模型节点中配合 `historyList` 变量精细控制提示词与上下文。 - -## 技术选型建议 - -1. **优先看协议生态**：若团队代码栈已围绕 OpenAI SDK 构建（含观测、重试、流式解析），选 Responses API 可降低迁移与维护成本；若以阿里云/DashScope 体系为主，选原生 API 更顺。 -2. **看扩展能力**：自定义插件参数透传（`biz_params.user_defined_params`）目前是原生 API 的明确能力，需要此能力的场景应选原生 API。 -3. **看多轮管理偏好**：希望云端托管历史、降低客户端状态复杂度，用原生 API 的 `session_id`；希望完全自控历史与上下文，两套 API 都支持 `messages` 数组方式。 -4. **看多模态需求**：图像、文件等多模态输入在 Responses API 中有标准化的 `content` 数组结构，接入更直接；原生 API 需结合应用编排实现。 -5. **看地域与业务空间**：两套 API 均支持默认空间仅凭 APP ID 调用；子业务空间或海外地域需携带 Workspace ID，选型不影响该约束，但需在请求中正确拼装。 -6. **凭证一致**：两套方式都使用同一份 `DASHSCOPE_API_KEY`，无需为不同调用方式分别管理密钥，切换成本主要在 SDK 与请求结构层面。 - -综上，两种方式并非互斥：同一 `APP_ID` 可同时被两套 API 调用，团队可按业务模块分别选型——面向外部生态集成用 Responses API，面向内部能力扩展用原生 API。 - -## 被对比主题页 - -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md deleted file mode 100644 index 044ec3fc..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-evaluation-vs-monitoring.md +++ /dev/null @@ -1,45 +0,0 @@ -# 应用评测与应用监控对比 - -阿里云百炼平台同时提供"应用评测"与"应用监控（应用观测）"两套围绕应用质量的能力：前者面向**离线**场景，用评测集 + 评估器系统化衡量应用输出质量；后者面向**在线**场景，端到端追踪已发布应用的真实调用链路并采集延时、Token 量等运行时指标。两者通过共享的"标签"组件打通——线上 Span 可标注后直接沉淀为评测样本，实现"线上观测 → 离线评测 → [模型调优](../concepts/fine-tuning.md)"的闭环。本文从目的、数据来源、输入输出、计费、典型场景等维度对比两者，帮助开发者做技术选型。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 应用监控（应用观测） | -| --- | --- | --- | -| 核心目的 | 离线评估应用输出质量，产出评分、BadCase、归因与调优建议 | 在线追踪应用真实调用链路，采集延时、Token 量等运行时指标 | -| 数据来源 | 评测集（手动上传或大模型自动生成） | 已发布应用在线接收的真实 Prompt 与调用 | -| 触发方式 | 手动发起评测任务，运行评测集 | 应用添加到观测列表后自动追踪，分钟级同步 | -| 应用范围 | 智能体应用（自动评测需配置知识库）、[工作流](../concepts/workflow.md)应用；新版支持自定义类型 | 智能体应用、[工作流](../concepts/workflow.md)应用、高代码应用；暂不支持 Assistant API 创建的智能体应用 | -| 输入格式 | 旧版：`.xls`/`.xlsx`（对话分析）、`.jsonl`（知识问答）；新版：`.xls`/`.xlsx`，按应用出入参生成模板，单文件 ≤ 20MB，单次最多 10 个文件 | 真实线上请求，无需上传文件；导出支持 JSONL 与 EXCEL | -| 输出形式 | 评分（1-5 / 0-100 / 0-1 等）、评估器结果、BadCase 分析、归因报告、调优建议 | 调用链路树（CHAIN/AGENT/RETRIEVER/LLM 等节点）、延时、Token 量、监控统计图表 | -| 评估器/评分机制 | 预置模板 + LLM 评估器 + Code 评估器；每任务最多 10 个评估器 | 不做评分，仅记录指标；可对 Span 人工打标签 | -| 标签能力 | 分类/布尔值/数字/文本四种类型，与监控共用 | 同一套标签体系，可对每个 Span 标注，自动保存 | -| API 端点 | 通过控制台操作（评测集、评测任务、评估器、标签管理） | 无 API，仅控制台操作；底层依赖 OpenTelemetry 服务 | -| 数据时效 | 任务执行后产出，按版本留存 | 指标分钟级更新，调用记录最长可查 30 天 | -| 计费方式 | 评测任务调用大模型产生 Token 费用，正常计费 | 应用观测功能本身不收费；观测数据存储费用由 OpenTelemetry 服务收取 | -| 关键限制 | 自动评测仅面向已发布且配置知识库的智能体应用，单次最多 8 个应用横向评测；评测任务发起后配置不可修改 | 高代码应用仅能观测到入口 CHAIN 节点，不支持内部链路追踪；暂不支持长期记忆检索过程观测 | -| 版本差异 | 区分"旧版"与"新版"两套界面，评测集类型、关联应用范围、评估器机制不同 | 无新旧版本区分 | - -## 两者协同关系 - -应用监控与评测并非孤立能力，平台在数据层做了打通： - -- **Span → 评测集**：应用观测支持将 Span 数据直接加入评测集（选择目标评测集、导入方式、字段映射，每评测集最多 50 个字段映射），把真实线上调用沉淀为评测样本。 -- **共享标签**：标签管理为评测与监控共用，线上 Span 标注的标签可在评测任务中复用，反之亦然，便于跨阶段质量分析。 - -## 适用场景建议 - -- **选应用评测** when：需要系统化、可重复地评估应用输出质量；要对响应做相关性、有害性、幻觉等语义判断；有领域专家介入做端到端标注；发布前做回归验证或横向对比多个应用（最多 8 个）。 -- **选应用监控** when：应用已上线，需要排查真实调用的延时、失败率、Token 成本；想查看智能体/[工作流](../concepts/workflow.md)内部检索、重写、重排、LLM 等子节点的执行链路；需要分钟级性能监控与最长 30 天的调用记录检索；想从线上真实数据中沉淀评测样本。 -- **组合使用** when：希望构建"线上观测 → 标注/BadCase 沉淀 → 离线评测 → 调优迭代"的闭环质量运营体系。此时建议先开通应用观测（自动评测的前置条件之一），再用观测到的 Span 数据补充评测集。 - -## 技术选型小结 - -若问题偏"质量好不好"（输出是否准确、完整、合规）→ 用应用评测；若问题偏"跑得稳不稳"（延时、失败、成本、链路追踪）→ 用应用监控；若希望用线上真实数据驱动持续调优，则两者配合使用，通过共享标签与 Span 导入评测集的能力打通闭环。开通自动评测前需先开通应用观测，这也是两者协同的一个隐含前置条件。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md new file mode 100644 index 00000000..77c9f6f3 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md @@ -0,0 +1,69 @@ +# 应用构建框架对比：Managed Agents vs Application Component API vs Frameworks + +## 对比目的与背景 + +在百炼平台生态中，开发者面临多种技术路径来构建 AI 原生应用：从全托管的智能体运行时（Managed Agents），到细粒度的数据与知识能力组合（Application Component API），再到面向主流开发范式的框架级集成（Frameworks）。三者定位不同、抽象层级各异、适用边界清晰。本对比旨在为开发者提供客观、可操作的技术选型参考，帮助其根据**应用形态、控制粒度、团队能力与交付节奏**等核心因素，快速判断最适合的构建路径，避免过度工程化或能力缺失风险。 + +--- + +## 关键维度对比表 + +| 维度 | Managed Agents API | Application Component API | Frameworks（LlamaIndex / Spring AI Alibaba） | +|------|---------------------|----------------------------|-----------------------------------------------| +| **定位与角色** | 智能体（Agent）全生命周期托管运行时，聚焦“会话驱动型”交互式应用 | 底层能力组件化服务，提供数据连接、知识库、Prompt 管理等原子能力，供自主编排 | 主流开源框架的百炼适配层，降低 RAG/智能体/知识检索类应用的接入门槛 | +| **输入格式** | ChatML 格式 message 数组（含 `role`, `type`, `content`），支持[多模态](../concepts/multi-modal.md)文本块；事件驱动模型 | RESTful 请求体（JSON），按接口语义定义（如 `AddFileRequest`, `RetrieveRequest`）；文件上传需先申请租约（Lease） | 框架原生对象（如 LlamaIndex 的 `Document`/`QueryEngine`，Spring AI 的 `Prompt`/`ChatClient`），由 SDK 自动序列化为百炼协议 | +| **输出格式** | SSE 流式事件（`message`, `tool_call`, `session_status` 等）或分页事件历史；结构化程度高，含 `thoughts`、`docReferences` 等语义字段 | JSON 响应体，严格遵循 OpenAPI Schema（如 `ListFilesResponse`, `RetrieveResponse`）；返回字段明确，但无统一语义层封装 | 框架标准返回类型（如 `Response`、`StreamingResponse`、`List`），经适配器映射为百炼能力，部分字段（如 `docReferences`）需显式启用 | +| **支持模型** | 仅限百炼托管模型（如 `qwen-plus`），`model.id` 必须为字符串且严格匹配平台列表；**不支持自定义/外部模型** | **不直接调用大模型**；所有模型能力通过下游组件（如知识库检索、Prompt 渲染）间接使用；知识库检索默认用 `qwen-max`，但不可在 Component API 层切换 | 支持指定生成模型（`qwen-max`, `qwen-plus`）及重排模型（`gte-rerank`）；模型名通过框架配置项传入，**仍受限于百炼公开模型池** | +| **API 端点** | 地域化 MAAS Endpoint（如 `https://.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio`）；强绑定 workspace + region | ROA 风格通用 Endpoint（如 `bailian.cn-beijing.aliyuncs.com`）；按资源类型路由（`/data-connection`, `/knowledge-base`, `/prompt`） | **无独立端点**；复用 DashScope 统一 API（`dashscope.aliyuncs.com`）；框架内部完成鉴权、路由与协议转换 | +| **计费方式** | 按实际调用计费：Session 运行时长（秒）、工具执行次数、文件存储（GB/天）、事件流传输量；**会话空闲期不计费** | 按调用频次（QPS）与资源用量计费：文件解析/索引构建（按页/小时）、知识库检索（次）、Prompt 执行（次）、OSS 数据同步（流量） | **框架本身免费**；所有底层调用（模型推理、知识库检索、重排、文档解析）均按百炼对应计费项单独计费，与直接调用 API 一致 | +| **典型场景** | 客服对话机器人、多步骤任务助手（如订机票+查天气+发邮件）、需沙箱隔离与状态持久化的复杂工作流 | 构建企业级知识中枢（对接 ERP/CRM 文件）、管理 Prompt 版本库、自动化数据导入与索引构建、定制化检索增强流程 | 快速验证 RAG 效果、将现有 LlamaIndex/Spring Boot 应用迁移至百炼、需要框架生态（插件、可观测性、Spring 生态集成）的中大型项目 | +| **状态管理** | 内置完整状态机（`idle → running → terminated`）；Session 自动管理上下文、工具状态、中断恢复；支持 `archive`/`delete` 终态控制 | **无会话状态**；纯无状态 CRUD 接口；状态需由调用方自行维护（如缓存检索上下文、轮询任务状态） | 依赖框架自身状态管理（如 LlamaIndex 的 `Index` 实例、Spring AI 的 `ChatClient` Bean）；百炼侧不维护跨请求状态 | +| **安全与隔离** | 工作空间级资源隔离；沙箱环境（`cloud` 类型）提供网络/依赖隔离；Skill/File 需安全扫描后激活 | RAM 子账号 + 最小权限策略；文件/知识库/Category 均归属 Workspace；OSS 授权需主账号显式配置 | 继承框架运行时安全模型（如 Spring Security）；百炼侧仅校验 `DASHSCOPE_API_KEY`，不感知框架内权限体系 | + +--- + +## 各方案适用场景建议 + +### ✅ 选择 **Managed Agents API** 当： +- 应用核心是**多轮、有状态、带工具调用的对话体验**（如销售顾问、IT 支持助手）； +- 需要开箱即用的**沙箱执行环境**（运行 Python 工具脚本、访问内部 API）； +- 要求**会话级状态自动管理**（上下文延续、中断恢复、超时清理）； +- 团队希望**最小化运维负担**，专注 Agent 设计与 Skill 编排，而非底层基础设施； +- 对模型选择无定制需求，接受百炼托管模型能力边界。 + +### ✅ 选择 **Application Component API** 当： +- 构建**后台数据中枢或知识平台**，需精细控制文件解析、知识库构建、Prompt 版本发布等环节； +- 需要**与现有系统深度集成**（如定时同步数据库表、监听 OSS 事件触发知识更新）； +- 要求**完全自主的状态与流程编排**（例如：自定义重排逻辑、混合检索策略、多知识库路由）； +- 团队具备较强后端开发能力，熟悉 RESTful 设计与幂等性处理； +- 需要规避框架锁定，保持未来技术栈演进灵活性（如迁移到自研调度引擎）。 + +### ✅ 选择 **Frameworks** 当： +- 项目已基于 **LlamaIndex 或 Spring Boot 技术栈**，追求**零改造迁移**至百炼； +- 目标是**快速原型验证或 MVP 上线**，优先保障开发效率而非极致控制； +- 需要利用框架生态能力（如 LlamaIndex 的 Node Postprocessor、Spring AI 的 `Advisor` 机制、Actuator 健康检查）； +- 应用形态明确为 **RAG 检索问答、智能体调用、或知识库增强聊天**，无需沙箱或复杂状态机； +- 团队熟悉对应框架，且接受其抽象带来的约束（如 LlamaIndex 不支持自定义切分器）。 + +--- + +## 技术选型决策指南（面向开发者） + +| 决策维度 | 推荐方案 | 关键判断依据 | +|----------|----------|--------------| +| **应用是否需要“会话”概念？** | Managed Agents API | 若用户交互天然具有上下文依赖（如“上一条说的XX，现在帮我查下相关文档”），且需自动维持状态，则 Agents 是唯一选择。Component API 和 Frameworks 均需自行实现会话管理。 | +| **是否必须运行任意代码（Python/Shell）？** | Managed Agents API | 只有 Managed Agents 提供沙箱环境（`Environment`）支持工具脚本执行；Component API 仅提供数据能力，Frameworks 仅调用百炼已有服务。 | +| **是否已有成熟框架代码基？** | Frameworks | 若已有 LlamaIndex 构建的 RAG 应用或 Spring Boot 项目，直接集成 Frameworks 可节省 80%+ 接入成本；反之，为新项目强行引入框架可能增加学习曲线。 | +| **是否需对接非百炼数据源（如 MySQL、SharePoint）？** | Application Component API | Component API 的 `Connector` 和 `AddFilesFromAuthorizedOss` 支持标准化对接；Frameworks 仅支持百炼知识库；Managed Agents 需将对接逻辑封装为 Skill（开发成本高）。 | +| **是否要求模型完全可控（微调/私有部署）？** | ❌ 三者均不支持 | 百炼当前所有路径均**仅支持平台托管模型**；若需私有模型，请评估百炼 Model Studio 或阿里云 PAI 平台。 | +| **团队是否缺乏全栈 AI 工程经验？** | Managed Agents API 或 Frameworks | Agents 提供最高抽象（拖拽式 Agent 配置 + SDK 调用）；Frameworks 利用社区惯用范式降低认知负荷；Component API 要求理解 ROA、租约、幂等性等细节，适合资深后端。 | + +> **重要提醒**：三者并非互斥。生产实践中常见**组合使用**——例如：用 Application Component API 构建和维护知识库，再通过 Managed Agents API 创建调用该知识库的智能体；或用 Frameworks 快速搭建前端 Demo，后端核心流程用 Component API 实现高可靠性调度。选型应以**端到端交付价值**为最终目标，而非单一技术指标。 + +## 被对比主题页 + +- [managed agents api](../api/managed-agents-api.md) +- [application component api reference](../api/application-component-api-reference.md) +- [frameworks](../api/frameworks.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md deleted file mode 100644 index 2194f12d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-observation-vs-evaluation.md +++ /dev/null @@ -1,59 +0,0 @@ -# 应用监控与应用评测对比 - -阿里云百炼平台围绕[智能体应用](../concepts/agent-application.md)、工作流应用提供两类数据驱动的运营能力：**应用监控（应用观测）**与**应用评测**。两者都服务于"用数据评估应用质量、辅助调优"这一目标，但定位不同——应用监控聚焦**线上真实流量的运行时可观测性**（延时、[Token](../concepts/token.md)、调用链路），应用评测聚焦**离线/预上线数据集的质量评估**（评分、BadCase、归因）。本页通过关键维度对比帮助开发者在不同阶段做出技术选型。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 应用评测 | -| --- | --- | --- | -| 核心定位 | 线上运行时可观测，追踪调用链路与性能指标 | 离线/预上线数据集质量评估，量化输出好坏 | -| 数据来源 | 自动采集应用实际调用（Prompt、输出、延时、[Token](../concepts/token.md)） | 评测集（自动生成或手动上传）+ 被测应用输出 | -| 触发方式 | 开启观测后自动同步，分钟级更新 | 手动发起评测任务，任务完成后产出报告 | -| 支持应用类型 | [智能体应用](../concepts/agent-application.md)、工作流应用、高代码应用（仅入口 CHAIN 节点） | [智能体应用](../concepts/agent-application.md)、工作流应用（自动评测需配置[知识库](../concepts/knowledge-base.md)且已发布） | -| 不支持场景 | 暂不支持 Assistant API 创建的智能体；高代码内部链路不可追踪；无 API | 自动评测仅面向配置[知识库](../concepts/knowledge-base.md)的已发布智能体，单次最多 8 个应用 | -| 输入格式 | 实际线上 Prompt 及调用数据（无需用户准备数据集） | 评测集：旧版（`.xls`/`.xlsx` 对话分析、`.jsonl` 知识问答）；新版（`.xls`/`.xlsx`，单文件 ≤20MB，单次 ≤10 文件） | -| 输出格式 | 调用记录、Trace 列表、监控统计图表；可导出 **JSONL** / **EXCEL** | 评测报告：总正确率、BadCase、归因分析、调优建议；任务详情含数据明细与指标统计 | -| 评估方式 | 不评分，仅呈现延时/[Token](../concepts/token.md)/失败率等指标，支持人工标注标签 | 自动评测用大模型评 1-5 分；手动评测人工打标（较差/一般/较好或 1-5 分）；新版支持 LLM/Code 评估器 | -| 评估器/评分 | 无评分器，依赖指标与人工标注 | 预置模板（通用质量/智能体/文本匹配/相似度/格式校验）+ 自定义 LLM/Code 评估器，每任务最多 10 个 | -| 关键指标 | 调用次数（含失败率）、Token 总量（输入/输出）、平均首 Token 耗时、平均调用时长、节点级延时 | 总正确率（≥4 分占比）、BadCase Top-5、RAG 智能体按问题类型分档得分、各评估器通过率 | -| 调用链路追踪 | 支持，按节点类型展示（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等，可嵌套） | 不追踪链路，仅按评测集条目组织输入输出 | -| 数据时效与保留 | 分钟级更新，调用记录最长查 30 天，支持按 Request ID/Trace ID/Span ID 检索 | 任务结果持久保存，评测集每次发布生成新版本，可选特定版本评测 | -| 标签与标注 | 支持对 Span 标注标签（布尔/分类/数字/文本），与应用评测共享统一管理 | 标签为核心组件，用于数据明细筛选；支持普通模式与快速标注 | -| 与对方联动 | 可将 Span 数据直接加入评测集（选目标集、导入方式、字段映射，最多 50 字段映射） | 评测集可承接来自应用观测的真实线上样本 | -| 前提条件 | 需开通 OpenTelemetry 服务并初始化 LogStore；子账号需 `AliyunBailianFullAccess` + 应用观测权限 + `ram:CreateServiceLinkedRole` | 自动评测需已开通应用观测；评测集需发布后可用；任务配置后不可修改 | -| 计费方式 | 功能本身不收费，数据存储费用由 OpenTelemetry 服务收取 | 评测任务调用大模型产生的 Token 费用正常计费；Code 评估器无额外费用 | -| API 支持 | 无 API，仅控制台操作 | 无 API，仅控制台操作（评测集/任务/评估器/标签均通过控制台管理） | -| 典型场景 | 线上性能监控、故障排查、Token 成本分析、瓶颈节点定位、把真实调用回流为评测样本 | 上线前质量验收、Prompt/检索/[知识库](../concepts/knowledge-base.md)调优验证、多应用横向对比、BadCase 归因与优化建议 | - -## 适用场景建议 - -**优先选择应用监控（应用观测）的场景**： - -- 应用已上线，需要持续观察**真实流量**下的延时、Token 消耗、失败率等运行时指标。 -- 需要**端到端调用链路追踪**，定位某个检索/重排/模型节点是性能瓶颈或错误来源。 -- 需要把线上真实调用**回流为评测样本**，让评测集贴近实际分布。 -- 关注**成本治理**，希望按应用/节点维度统计 Token 用量。 - -**优先选择应用评测的场景**： - -- 应用上线前或迭代中需要**系统化评估输出质量**，得到可量化的评分与归因。 -- 需要用**领域专家或自动评估器**对一批固定用例打分，产出 BadCase 与调优建议。 -- 需要对比多个应用/版本在同一评测集上的**横向表现**（自动评测单次最多 8 个应用）。 -- 需要**可复用的评分逻辑**（LLM 评估器做语义判断，Code 评估器做格式/数值精确校验）。 - -## 技术选型参考 - -两者并非二选一，而是**互补的闭环**：应用监控提供"线上发生了什么"的运行时事实，应用评测提供"这些输出到底好不好"的质量判定。推荐组合使用—— - -1. **开发/调优阶段**：用手动或自动评测在固定评测集上迭代 Prompt、检索配置、知识库切片，借助评估器与归因建议收敛质量。 -2. **上线/运营阶段**：开启应用观测监控真实流量，识别性能瓶颈与成本异常，并用 Span 标注或"加入评测集"把线上问题转化为下一轮评测输入。 -3. **数据回流**：将应用观测中标记为错误或低质的 Span 直接加入评测集，形成"线上 → 评测 → 调优 → 线上"的持续优化闭环。 - -选型时关键判据是**数据来源**：要用真实流量就看监控，要用受控数据集就看评测；要量化"好坏"就用评测，要看"运行状态"就用监控。两者共享标签体系，便于跨阶段贯通标注与筛选。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [application evaluation](../guides/application-evaluation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md deleted file mode 100644 index 9c8ed744..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-vs-managed-agents.md +++ /dev/null @@ -1,56 +0,0 @@ -# 百炼应用与托管 Agent 对比 - -阿里云百炼平台提供两条构建 AI 应用的路径：**[智能体应用](../concepts/agent-application.md)**（含新版 Agent 2.0、旧版 Agent 1.0、工作流、高代码应用）与 **Managed Agents（托管 Agent）**。二者都能组合模型、提示词、工具与 MCP 服务，但在运行模式、状态管理、执行环境和适用任务上有本质差异。本页帮助开发者理解两者定位，做出正确的技术选型。 - -核心区别在于：**[智能体应用](../concepts/agent-application.md)是无状态的应用构建与发布形态**，由应用侧维护上下文，面向问答、对话等交互式场景；而 **Managed Agents 是服务端托管的运行时**，在独立云端沙箱中持久化会话状态并支持中断续接，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 关键维度对比 - -| 对比维度 | [智能体应用](../concepts/agent-application.md)（LLM Application） | Managed Agents（托管 Agent） | -|---------|------------------------------|------------------------------| -| 定位 | 应用构建与发布形态（Agent / 工作流 / 高代码） | 服务端托管的智能体运行时 | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，百炼托管的云端容器 | -| 状态持久化 | 不持久化，依赖调用方传递历史 | 事件历史在服务端持久化 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流（User/Agent/Tool/Tool_output/Error/Model/System） | -| 核心对象 | 应用（Agent/工作流/高代码应用） | 智能体 / 运行环境 / 会话 / 事件（四对象解耦复用） | -| 工具能力 | 内置沙箱工具（bash/read/write/edit/glob/grep/download_file）、知识库、MCP、插件 | 7 个内置工具（bash/read/write/edit/glob/grep/download_file）、MCP、Skill | -| 依赖安装 | 不支持自定义环境依赖 | 环境可通过 `config.packages` 预装 apt / pip 依赖并设网络策略 | -| 文件挂载 | 单会话最多 10 个文件、单文件 ≤10MB（文件问答） | 资源挂载到 `/mnt/session/uploads`，独立于会话、可多会话复用；单文件 ≤10MB | -| 记忆能力 | 新版 Agent 短期记忆 0-30 轮，长期记忆未支持 | 会话状态由状态机管理，天然承载长程上下文 | -| 主要 API 端点 | 各应用发布后经统一 API 集成调用 | `POST /agents`、`POST /environments`、`POST /sessions`、`POST /sessions/{id}/events`、`GET /sessions/{id}/events/stream` | -| 开发方式 | 零代码（Agent）/ 低代码（工作流）/ Python（高代码） | 控制台向导或 API 编排 | -| 计费方式 | 按模型 Token、知识库召回、MCP/插件、高代码资源等分项计费 | 按模型 Token 用量及托管沙箱资源计费 | -| 典型场景 | 问答、对话、RAG 检索、固定流程自动化、轻量任务 | 多步工具调用、代码执行、文件批处理等长时任务 | - -## 适用场景建议 - -### 选择智能体应用 - -- **快速交付交互式问答/对话**：新版智能体（Agent 2.0）以自然语言配置即可上线，适合业务人员和产品经理。 -- **知识库 / RAG 问答**：需要挂载企业私有知识库并做标签过滤检索时，智能体应用的知识库能力开箱即用。 -- **流程固定、精确可控**：意图分类、多步骤审批等确定性流程，用工作流做可视化节点编排更稳定。 -- **需要完整代码控制或企业级部署**：用高代码应用（Serverless Function / K8s），配套网关、可观测、日志等企业能力。 -- **多渠道发布**：需要发布到钉钉、微信公众号等第三方平台时，智能体应用原生支持。 - -### 选择 Managed Agents - -- **长时运行任务**：单次任务需要多轮工具调用、可能被中断后续接，服务端持久化事件历史更可靠。 -- **代码执行与文件处理**：需要在独立沙箱中执行 shell 命令、读写文件、安装依赖（apt/pip）、下载资源等。 -- **智能体自主规划**：让智能体在云端容器中自主决定命令与工具调用顺序，无需应用侧管理复杂上下文。 -- **资源复用与会话隔离**：同一环境 / 挂载资源被多个会话复用，且会话间修改互不影响时。 - -## 技术选型参考 - -- 若你的应用本质是**一问一答或短流程交互**，优先用智能体应用；无旧版依赖时选新版 Agent 2.0。 -- 若任务是**服务端长跑、含代码执行或多步工具编排**，选 Managed Agents，避免在应用侧自行拼接和维护会话状态。 -- 两者并非互斥：可以用智能体应用承载前端交互体验，用 Managed Agents 承接后端的长时执行任务。 -- 注意文档中 Managed Agents 的模型名存在不一致（向导示例 `qwen3.7-plus`、API 示例 `qwen3-max`），实际以控制台可选模型 ID 为准；智能体应用推荐使用工具调用能力强的千问-Max 系列。 -- 两类方案的文件上传上限均为单文件 10MB，超限需走文件上传 API。 - -## 被对比主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents](../guides/managed-agents.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md deleted file mode 100644 index d23dd1fd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/data-sources-for-rag-comparison.md +++ /dev/null @@ -1,61 +0,0 @@ -# 知识库与数据连接对比 - -在阿里云百炼平台中，**知识库**与**数据连接**是为大模型应用补充外部数据的两条主要路径，二者定位互补但常被混淆：知识库面向 RAG 检索增强生成，强调"先建索引、再语义召回"；数据连接面向外部数据源的统一接入与实时访问，强调"按需查询、按需引用"。本文从输入格式、输出形态、支持模型、调用方式、计费模型、典型场景等维度对两者做技术选型对比，帮助开发者根据数据形态与访问模式选择合适方案。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 数据连接（Data Connection） | -| --- | --- | --- | -| 定位 | 基于向量化检索增强生成，为模型补充私有数据与最新信息 | 外部数据源统一入口，安全访问企业数据库、文档系统、对象存储 | -| 数据形态 | 非结构化文档、表格、图片、音视频（向量化后检索） | 非结构化文件、结构化表格、关系型数据库、对象存储、语雀文档 | -| 支持的数据源类型 | 本地上传、OSS、数据连接器导入 | 文件、表格、MySQL、PostgreSQL、PolarDB-X 2.0、语雀、OSS | -| 输入格式 | pdf/docx/pptx/xlsx/txt/markdown/html/png/jpg/mp4/mkv 等 | 文件类同知识库；表格 xlsx/xls；数据库走 SQL；不支持直接导入 JSON/CSV/YAML | -| 处理方式 | 切片 + 向量化 + 召回排序（建立索引） | 文件类解析入库；数据库走流式实时查询；OSS 走向量检索服务 | -| 访问模式 | 语义检索召回切片（TopK 1–20，相似度阈值过滤） | 文件/表格按类目导入与查询；数据库实时执行 SQL；OSS 工具调用 | -| 支持模型 | 千问 QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research、千问 VL 系列、开源版、第三方（DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 等） | 由挂载的应用决定，本身不绑定模型；数据库/SQL 访问依赖应用编排 | -| 向量模型 | text-embedding-v4/v3（512 维）、multimodal-embedding-v1（1024 维） | OSS 连接器需开通向量检索服务；其他类型不强制向量化 | -| API 端点 | 知识库 API（上传租约→文件→类目→索引任务轮询） | 数据连接 API + 应用内调用（searchOSSFile、searchOSSFileByFileName 等工具） | -| 计费方式 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时（1 RCU≈50 QPS） | 文件/表格平台存储限时免费、超额按量；数据库走源实例计费；OSS 走 OSS 计费 | -| 地域限制 | 仅中国站华北2（北京）可开通使用 | 数据库类需私网/公网可达；PolarDB-X 2.0 仅私网；语雀仅公网版本 | -| 数据时效 | 增量导入后需重新解析与建索引，存在更新延迟 | 数据库与语雀类为实时访问；文件/表格/OSS 导入后立即可用 | -| 配额要点 | 业务空间类目 500、文件 10 万、数据表 1 千；标准版 100 GB、旗舰版 9,999 GB | 平台存储 10 万文件、1 TB 免费额度；导入文件仅保留 90 天可查看 | -| 典型场景 | 文档问答、企业知识问答、图文并茂回复、音视频检索问答 | 实时查询业务库、引用语雀文档、检索 OSS 对象、结构化表格问答 | - -## 适用场景建议 - -### 优先选择知识库 - -- 数据以**非结构化文档**为主，需要**语义检索**（即使关键词不匹配也能召回）。 -- 需要图文并茂回复、视觉理解、音视频内容检索等 RAG 场景。 -- 应用形态为智能体应用、工作流应用，或通过 SDK 集成 RAG 检索能力。 -- 对**召回质量**有可调优诉求（相似度阈值、TopK、切片策略、元数据过滤）。 - -### 优先选择数据连接 - -- 数据源是**关系型数据库**（MySQL、PostgreSQL、PolarDB-X 2.0），需要在对话中**实时执行 SQL** 查询最新业务数据。 -- 数据存储在**语雀**或**OSS**，希望按需引用而非预先向量化。 -- 数据以**结构化表格**为主，且字段含义清晰，适合模型直接理解。 -- 团队希望以"连接器"方式统一管理企业数据资产，按权限安全访问。 - -### 二者结合使用 - -- 同一批数据既需语义召回又需结构化查询：可先将结构化部分接入表格连接器，再将非结构化部分导入知识库，在应用编排中按查询意图分流。 -- 文件已在 OSS：既可作为知识库数据来源导入（向量化召回），也可通过 OSS 连接器在对话中按需检索——前者适合"问答"，后者适合"取文件"。 -- 语雀文档：通过语雀连接器实时访问，避免频繁同步；若需历史版本语义检索，仍需导入知识库。 - -## 技术选型参考 - -1. **看数据形态**：非结构化文档/图片/音视频 → 知识库；结构化数据库/表格 → 数据连接；混合 → 两者并行。 -2. **看时效要求**：实时业务数据（订单、库存）→ 数据连接（流处理类）；可容忍延迟的知识沉淀 → 知识库。 -3. **看访问模式**：语义相似度召回 → 知识库；精确 SQL 查询或工具取文件 → 数据连接。 -4. **看成本结构**：知识库按规格/RCU 计费，检索量大时旗舰版更经济；数据连接平台存储限时免费，主要成本在源数据库与 OSS。 -5. **看地域与权限**：知识库仅北京地域可用；数据库类连接器需保证网络可达并完成 DMS/EventBridge/DTS 等角色授权。 - -> 选型结论：**知识库解决"模型不知道"的问题，数据连接解决"模型拿不到最新数据"的问题**。在百炼平台中两者并不互斥，建议依据数据源类型与访问模式组合使用，由应用编排层按查询意图分发。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md deleted file mode 100644 index fd5d160e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-app-vs-model-comparison.md +++ /dev/null @@ -1,52 +0,0 @@ -# 应用[评测](../concepts/evaluation.md)与模型[评测](../concepts/evaluation.md)对比 - -百炼平台提供两套互相独立的[评测](../concepts/evaluation.md)能力：**应用评测**评估的是已编排好的智能体/工作流应用的端到端输出质量，**模型评测**评估的是大语言模型本身在标准或自定义数据集上的基础能力。两者面向的评测对象、数据组织方式、评估器机制、计费路径均不同，开发者需根据选型阶段（选模型 vs 验应用）选择合适工具。本文从评测对象、数据格式、评估机制、结果产出、计费、典型场景等维度进行对比，供技术选型参考。 - -## 关键维度对比 - -| 对比维度 | 应用评测 | 模型评测 | -| --- | --- | --- | -| 评测对象 | 已发布的智能体应用、工作流应用（端到端整体输出） | 大语言模型本身（千问系列、开源版、第三方文本模型、调优模型） | -| 评测范式 | 自动评测（大模型生成评测集并评分 1–5）、手动评测（人工打标） | 自动评测（系统跑模型推理后评分）、人工评测（人工打标） | -| 评测集类型 | 旧版：对话分析（.xls/.xlsx）、知识问答（.jsonl）；新版：智能体/工作流/自定义三类（.xls/.xlsx，≤20MB，单次≤10 文件） | 自定义评测（评测数据集 / 推理结果集，Excel）、基线评测（系统预置榜单数据集，无需自备） | -| 评测集字段 | Prompt/Completion/SessionId（对话分析）、query/referenceAnswer/fineKeywords/coarseKeywords（知识问答），新版按应用出入参自动生成模板 | `${prompt}`（输入）、`${output}`（模型输出）、`${completion}`（参考答案）；评测数据集含 Prompt+Completion，推理结果集额外含 Output | -| 评估器/评测维度 | 评估器可复用：预置模板（通用质量/智能体/文本匹配/文本相似度/格式校验）+ 自定义（LLM 评估器、Code 评估器、基于评测任务创建）；单任务最多 10 个 | 评测维度 5 种：大模型评估-数值型、规则评估-文本相似度、大模型评估-分类型、规则评估-字符串匹配、人工评估-分类型；可组合多维度 | -| 评分尺度 | LLM 评估器自定义范围（0-100 / 1-5 / 0-1），Code 评估器返回数值；通过阈值决定 Pass/Fail | 数值型 1–5 分，分类型 Pass/Fail，规则型相似度/匹配度；综合得分=维度平均分 | -| 评分器 Prompt 变量 | 映射到评测集字段或应用输出（query、reference_response、context 等） | 固定三变量 `${prompt}` / `${output}` / `${completion}` | -| 预置模板 | 通用质量、智能体、文本匹配、文本相似度、格式校验五类 | 综合评测（5 维度）、语义相似度、自定义评测三种 | -| 关联范围 | 单次自动评测最多 8 个智能体应用横向对比 | 支持多模型同维度排行榜对比 | -| 基线/榜单能力 | 无 | 提供基线评测（C-Eval、MMLU、ARC、GSM8K、HellaSwag、BBH），仅支持调优模型，不支持预置模型 | -| API/触发方式 | 控制台编排评测任务，调用应用产生 Token 计费 | 控制台创建任务，系统自动推理或读取已有 Output | -| 计费方式 | 评测任务调用大模型产生的 Token 费用正常计费；LLM 评估器裁判评分产生 Token 费用；Code 评估器无额外费用 | System Prompt 产生被评测模型推理费用，评分器 Prompt 产生裁判模型评分费用；推理结果集直接读 Output 不推理，可降本 | -| 标签与人工标注 | 分类/布尔值/数字/文本四类标签，支持快速标注模式，单条标注页三栏布局 | 人工评估-分类型由人工打标签 | -| 结果产出 | 数据明细 + 指标统计；自动评测含总正确率、BadCase Top-5、归因分析、调优建议 | 数据明细（Prompt/Completion/Output/评分）+ 指标统计（综合得分、通过率、分数分布）；基线评测含分学科明细、能力雷达图、行业对比 | -| 版本差异 | 区分旧版/新版应用评测（评测集类型、关联范围、评估器机制显著不同） | 无新旧版之分 | - -## 适用场景建议 - -### 选应用评测 - -- 已完成智能体/工作流编排，需验证端到端输出质量（含知识库检索、Prompt 编排、工具调用整体效果）。 -- 需要 BadCase 归因与调优建议，指导应用迭代。 -- 需对线上真实 Span 做标签标注，与应用观测联动做质量监控。 -- 多应用横向对比（最多 8 个），选优上线。 -- 评测数据为多轮对话或带 fineKeywords 的知识问答，需要按 SessionId 或信息点粒度评估。 - -### 选模型评测 - -- 处于模型选型阶段，需对比千问系列、开源版或第三方模型的基础能力。 -- 需用标准榜单（C-Eval、MMLU、GSM8K 等）做基线验证，且被测模型为调优后模型。 -- 业务场景需自定义评分标准（问答质量、内容安全、Function Calling、NL2SQL、翻译摘要等），希望以数据驱动选型。 -- 已有模型推理结果，希望直接评分以降低推理成本（用推理结果集）。 -- 需要参与排行榜对比多模型同维度表现。 - -### 组合使用 - -复杂项目可先做模型评测锁定底座模型，再做应用评测验证编排后的端到端质量；两者评分器/评估器都依赖大模型裁判，需同时关注裁判模型 Token 成本。应用评测的 LLM 评估器与模型评测的大模型评估-数值型在机制上同源（都靠裁判模型按 Prompt 评分），但前者参数映射到应用输出，后者固定三变量，迁移评分逻辑时需调整变量绑定方式。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md deleted file mode 100644 index 673a548f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-compare.md +++ /dev/null @@ -1,50 +0,0 @@ -# 应用评测与模型评测对比 - -阿里云百炼平台提供两套面向不同对象的评测能力：**应用评测**关注智能体/工作流应用的端到端输出质量（含 RAG 归因、调优建议），**模型评测**关注单个文本生成模型在给定数据集上的能力表现（含基线评测、排行榜）。二者虽然都围绕"评测集 + 评估规则 + 报告"展开，但评测对象、可用能力、操作方式和计费构成差异明显。本页面为开发者做技术选型时提供横向参考。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| 评测对象 | 已发布的智能体应用 / 工作流应用（可含 RAG 知识库） | 单个文本生成类模型（含调优模型） | -| 评测目标 | 评估应用回答质量、定位 RAG BadCase、给出调优建议 | 评估模型基础能力、对比选型、验证调优效果 | -| 评测模式/方式 | 自动评测、手动评测；单应用 / 多应用横向（最多 8 个） | 自定义评测、基线评测（公开标准数据集） | -| 评测集/数据来源 | 旧版（对话分析 `.xls`/`.xlsx`、知识问答 `.jsonl`）、新版（智能体 / 工作流 / 自定义，含版本管理）；可从应用观测导入 | 评测数据集（Prompt + Completion，产生推理费用）或推理结果集（已含 Output，免推理费用） | -| 评分/评估器类型 | 评估器：LLM 评估器、Code 评估器、预置模板；每任务最多 10 个 | 评测维度：大模型评估（数值/分类）、规则评估（文本相似度/字符串匹配）、人工评估（分类） | -| 支持模型 | 评测集生成与评分当前仅 `qwen-max` / `qwen-plus` | 被评测：文本生成类模型；裁判模型推荐千问-Max | -| 报告能力 | 总正确率、BadCase 分析、RAG 归因分析、调优建议 | 综合得分、通过率、逐条评分、可参与排行榜 | -| 操作入口/API | 控制台操作（新旧两套系统） | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | -| 地域限制 | 无特殊地域限制（需开通应用观测） | 基线评测仅北京地域可用 | -| 计费构成 | 调用大模型（评测集生成 + 评分）产生的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费用） | -| 典型场景 | 智能体上线前质量把关、版本迭代对比、RAG 优化闭环 | 模型选型、调优前后能力对比、基础能力基准测试 | - -## 评分/评估规则对比 - -两者的评估规则思路相通，但组织方式不同：应用评测以"评估器"为组件挂载到评测任务，模型评测以"评测维度"作为可复用模板。 - -| 规则类别 | 应用评测（评估器） | 模型评测（评测维度） | -|----------|--------------------|----------------------| -| 语义理解打分 | LLM 评估器（相关性、幻觉、有害性等，产生 Token 费用） | 大模型评估-数值型 / 分类型（裁判模型打分或 Pass/Fail，有费用） | -| 确定性/规则判断 | Code 评估器（Python 规则，无额外费用） | 规则评估-文本相似度（ROUGE/BLEU/Cosine）、字符串匹配（无费用） | -| 人工标注 | 手动评测 + 标签管理（分类/布尔/数字/文本标签） | 人工评估-分类型（Pass/Fail 标注） | - -## 适用场景建议 - -- **优化一个已上线的智能体/RAG 应用** → 选应用评测。其归因分析能把 BadCase 精确定位到"模型理解有误 / 重排不佳 / 检索无效 / 切片不完整 / 未获取知识"等环节，并直接给出 Prompt、检索配置或知识库切片的优化建议，形成"识别—归因—优化—回归"闭环。 -- **在多个候选模型间选型，或验证微调效果** → 选模型评测。基线评测可用 C-Eval、MMLU、GSM8K、BBH 等公开数据集快速摸底；自定义评测配合排行榜可做定量对比。 -- **需要跨应用/跨版本横向对比** → 应用评测的多应用横向评测（同一评测基准下最多 8 个应用/版本，须关联相同知识库）。 -- **需要编程化、可自动化的评测流水线** → 两者当前都以控制台为主，模型评测无公开 API/SDK；如需 CI 集成，模型评测侧可参考 PAI Judge Model API。 - -## 技术选型参考 - -1. **先看评测对象**：评"应用整体表现（尤其 RAG）"用应用评测；评"模型本身能力"用模型评测。二者不可互相替代。 -2. **控制成本**：有确定性标准的场景优先使用规则评估 / Code 评估器（无裁判/LLM 费用）；模型评测可先用 50-100 条小规模验证，再保存推理结果集复用以免除重复推理费用。 -3. **注意能力边界**：应用评测的评测集生成与评分当前仅支持 `qwen-max` / `qwen-plus`；模型评测当前仅支持文本生成类模型，基线评测仅北京地域可用，任务提交后目标模型、评测维度类型均不可修改，选错需删除重建。 -4. **对待评测噪声**：模型评测中 1-3% 的分差通常为噪声，LLM 评分器存在位置偏差与自我偏好偏差，建议定期人工抽查校准；应用评测同样建议在知识库更新、Prompt 调整、模型升级后触发回归评测。 - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md deleted file mode 100644 index deadf313..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/evaluation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 应用评测与模型评测对比 - -百炼平台提供两套独立的评测体系：**应用评测**面向已构建的智能体应用和工作流应用，评估端到端的输出质量与 RAG 链路效果；**模型评测**面向底层大模型本身，评估模型的推理能力和指令遵循表现。两者的评测对象、数据流、评分机制和适用场景均有显著差异，开发者需要根据当前所处的开发阶段选择合适的评测方式。 - -## 关键维度对比 - -| 维度 | 应用评测 | 模型评测 | -|------|----------|----------| -| **评测对象** | 智能体应用、工作流应用（已发布的完整应用） | 文本生成类大模型（基础模型或调优后模型） | -| **核心目标** | 验证应用端到端输出质量，定位 RAG 链路问题 | 评估模型推理能力，辅助模型选型或调优验证 | -| **评测方式** | 自动评测（单应用 / 多应用横向）、手动评测 | 自定义评测（AI / 规则 / 人工）、基线评测（公开数据集） | -| **评测集来源** | 基于应用关联知识库自动生成，或手动上传 | 手动上传评测数据集，或使用公开标准数据集（C-Eval、MMLU 等） | -| **评估器 / 评分机制** | 新版评估器（LLM 评估器 + Code 评估器 + 预置模板）；旧版由平台内置评分 | 评测维度模板（大模型评估数值型/分类型、规则评估相似度/匹配、人工评估） | -| **归因分析** | 支持 RAG 链路归因（模型理解有误、重排不佳、检索无效、切片不完整、未获取知识） | 不提供链路归因，仅输出维度得分和通过率 | -| **横向对比能力** | 最多 8 个应用同基准横向对比 | 支持多模型评测结果排行榜对比 | -| **人工标注** | 新版通过标签体系支持四种类型标注（分类 / 布尔值 / 数字 / 文本） | 人工评估维度（Pass/Fail 标注） | -| **前提条件** | 应用已发布、已配置知识库、已开通应用观测 | 无特殊前提，上传数据集即可评测 | -| **地域限制** | 无特殊地域限制 | 基线评测仅北京地域可用 | -| **API 支持** | 通过控制台操作 | 仅控制台操作，不提供公开 API/SDK（可参考 PAI Judge Model API） | -| **计费构成** | 评测集生成 + 应用调用 + 评估器模型的 Token 费用 | 被评测模型推理费用 + 裁判模型评分费用 | - -## 评分体系差异 - -| 对比项 | 应用评测 | 模型评测 | -|--------|----------|----------| -| **评分范围** | 1-5 分（正确率 = 得分 >= 4 的占比） | 可自定义整数区间（默认 0-5，建议不超过 10） | -| **自动评分方式** | LLM 评估器（语义）+ Code 评估器（规则） | 大模型评估（裁判模型）+ 规则评估（ROUGE/BLEU/Cosine/字符串匹配） | -| **评估器数量** | 每任务最多 10 个，建议组合 3-5 个 | 按评测维度配置，无上限说明 | -| **评分模型** | 评测集生成和评估仅支持 qwen-max 和 qwen-plus | 裁判模型推荐千问-Max，被评测模型不限 | - -## 适用场景建议 - -### 优先选择应用评测的场景 - -- 智能体应用已发布上线，需要持续监控输出质量 -- 需要定位 RAG 链路中的具体瓶颈（检索、重排、切片、模型理解） -- 知识库更新或 Prompt 调整后需要回归验证 -- 多个应用版本之间需要横向对比，选出最优配置 -- 需要将人工标注经验固化为自动评估规则（通过评估器模板化） - -### 优先选择模型评测的场景 - -- 项目初期的模型选型，需要在多个候选模型间对比基础能力 -- 模型微调（SFT）后需要验证调优效果是否达标 -- 使用公开基准（C-Eval、MMLU、GSM8K、BBH）快速了解模型通用能力 -- 需要用规则评估（ROUGE/BLEU）做确定性指标验证（如翻译、摘要场景） -- 关注模型推理能力本身，而非上层应用的端到端效果 - -### 组合使用建议 - -典型的开发流程中，两种评测可以分阶段配合使用：先通过**模型评测**完成基础模型选型和调优验证，确定最优模型后构建应用，再通过**应用评测**验证端到端效果并持续迭代优化。 - -## 成本优化对比 - -| 策略 | 应用评测 | 模型评测 | -|------|----------|----------| -| **减少推理费用** | 缩小评测集规模 | 使用推理结果集（复用已有推理输出） | -| **减少评分费用** | 使用 Code 评估器替代 LLM 评估器 | 使用规则评估或人工评估替代大模型评估 | -| **渐进式评测** | 先小规模自动评测，再针对 BadCase 人工复核 | 先 50-100 条验证，再扩大到 200-500 条正式评测 | - -## 来源文档 - -- [application evaluation](../guides/application-evaluation.md) (guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) (guides/model-evaluation-introduction.md) - -## 被对比主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md deleted file mode 100644 index a6f32795..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/extension-framework-comparison.md +++ /dev/null @@ -1,38 +0,0 @@ -# 框架、工具包与 MCP 对比 - -阿里云百炼为开发者提供了多种接入与扩展大模型能力的方式，常见的选择包括三类：**开源框架集成**（LlamaIndex、Spring AI Alibaba）、**[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与官方 SDK/工具包**（compatible-mode/v1、DashScope SDK、LangChain 适配）、以及**模型上下文协议（MCP）服务**。三者面向的诉求不同——框架侧重在既有编程语言生态中拼装 RAG/[智能体应用](../concepts/agent-application.md)；兼容接口族侧重用最小改动复用 OpenAI 代码与生态；MCP 则侧重让智能体/工作流动态调用外部工具与云资源。本文从接入方式、语言/运行时、能力范围、适用场景、计费与限制等维度做横向对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 开源框架（LlamaIndex / Spring AI Alibaba） | [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包 | MCP 服务（官方 + 自定义） | -| --- | --- | --- | --- | -| 接入方式 | 框架 SDK + 百炼云端[知识库](../concepts/knowledge-base.md) / 应用 ID | 替换 `api_key`、`base_url`、`model` 三参数，复用 OpenAI 路径 | 在智能体/工作流中挂载 MCP 服务，或外部通过 Streamable HTTP 调用 | -| 语言/运行时 | Python 3.9+（LlamaIndex）、Java JDK 17+ / Spring Boot 3.x（Spring AI Alibaba） | 任意支持 OpenAI SDK 的语言；官方同时适配 LangChain、LangChain4j | 语言无关（协议层）；自定义服务可由 npx/uvx/http 部署 | -| 鉴权 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY` 等环境变量）；子[业务空间](../concepts/workspace.md)需[业务空间](../concepts/workspace.md) ID | [API Key](../concepts/api-key.md)（推荐 `DASHSCOPE_API_KEY`）；新加坡与北京地域 Key 不同 | [API Key](../concepts/api-key.md) + MCP 服务自身鉴权（如 `Authorization` 头）；仅主账号及授权 RAM 用户可访问自定义服务 | -| 主要能力 | 云端[知识库](../concepts/knowledge-base.md)构建、RAG 应用、调用百炼智能体/工作流应用、[知识库](../concepts/knowledge-base.md)检索 | Chat、Responses、Completions、Embedding、Vision、File、Batch、Conversations | 官方工具（地图、联网搜索等）、自定义脚本工具、封装 RESTful API、操作阿里云 OpenAPI（OSS、ECS 等） | -| 知识库/RAG | LlamaIndex 用云端智能切分与官方向量模型，不支持自定义切分/嵌入；Spring AI Alibaba 通过 `DashScopeDocumentRetriever` 检索百炼知识库 | Embeddings 接口做向量化；RAG 需自行在应用层编排 | 不直接提供 RAG；可作为工具被智能体调用，间接参与检索/查询 | -| 模型范围 | LlamaIndex 传 `qwen-max` 等；Spring AI Alibaba 调用智能体/工作流应用（应用背后绑定模型） | Qwen 全系（商业/开源/VL/Coder/Omni/Math）、DeepSeek、Kimi、GLM、MiniMax 等；Responses 支持 qwen3-max/plus/flash、qwen3-coder-plus 等 | 由承载 MCP 的智能体/工作流模型决定；调用准确性依赖提示词，必要时换用千问 3 系列等更强推理模型 | -| 调用形态 | 非流式与流式（Spring AI Alibaba `agent.call` / `agent.stream`） | 流式与非流式；Responses 支持 `previous_response_id` 多轮接续；Batch 异步批量（费用 50%） | 智能体自动判断是否调用；工作流中每节点单工具、手动串联；外部调用走 Streamable HTTP | -| 计费方式 | 按所调用模型/知识库的百炼标准计费 | 按模型 token 计费；Batch/Batch Chat 半价；Responses 上下文关联 7 天 | 云部署 MCP 限时免部署费；联网搜索 2000 次免费后 29 元/千次；自定义基础模式 0.000156 元/秒，极速模式另加 0.000036 元/秒部署时长 | -| 典型场景 | 已使用 Python/Java 生态、希望以框架方式构建 RAG 或集成百炼智能体 | 已有 OpenAI/LangChain 代码、希望低成本迁移或复用生态工具 | 让智能体/工作流动态调用第三方工具或阿里云资源，避免逐个写接口 | - -## 适用场景建议 - -- **选开源框架（LlamaIndex / Spring AI Alibaba）**：团队以 Python 或 Java/Spring 为主技术栈，希望以框架抽象快速搭建 RAG 应用或集成百炼智能体/工作流应用，且可接受云端智能切分与官方向量模型（LlamaIndex）或预先在控制台创建应用/知识库（Spring AI Alibaba）。若需要完全自定义文档切分与嵌入模型，LlamaIndex 路线并不适合，应改用本地知识库方案。 -- **选 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)族与工具包**：已有基于 OpenAI SDK 或 LangChain/LangChain4j 的存量代码，希望以最小改动（`api_key`/`base_url`/`model`）迁移到百炼，或需要使用 Completions（FIM 代码补全）、Batch（半价批量推理）、Responses（智能体原生能力、内置工具、多轮接续）等专项接口。适合追求协议兼容、跨语言复用与生态工具接入的团队。 -- **选 MCP 服务**：核心诉求是让百炼智能体或工作流在运行时动态调用外部工具（地图、联网搜索、自建脚本、RESTful API、阿里云 OSS/ECS 等），而非固定编写接口。适合需要多工具协同、逐步推理、或把已有业务 API 快速封装给模型使用的场景。注意 MCP 只能在智能体/工作流应用中使用，不能在直接调用千问 API 时接入；且会因工具返回内容进上下文而增加 token 消耗。 - -## 技术选型参考 - -1. **先明确诉求边界**：是"迁移/复用现有 OpenAI 代码"（走兼容接口族）、"用框架拼装 RAG/[智能体应用](../concepts/agent-application.md)"（走开源框架），还是"让运行时智能体动态调用外部工具"（走 MCP）。三者并非互斥，常组合使用——例如用兼容接口族做模型调用，同时用 MCP 扩展工具能力。 -2. **地域与鉴权一致性**：兼容接口族需按[业务空间](../concepts/workspace.md)专属域名拼装 `base_url`，弗吉尼亚地域使用固定域名且不带 `{WorkspaceId}`；Spring AI Alibaba 应用集成与知识库检索对 API Key 变量名约定不同（`DASHSCOPE_API_KEY` vs `AI_DASHSCOPE_API_KEY`），关键是 `application.yml` 占位符与实际变量名一致；子业务空间一律需要业务空间 ID。 -3. **能力限制与成本**：LlamaIndex 云端方案不支持自定义切分/嵌入；Completions 仅限北京地域；MCP 单智能体最多 5 个服务、工作流每节点单工具、自定义服务托管在 FC 无固定出口公网 IP（访问云资源需配白名单或打通 VPC）。批量推理与 Batch Chat 可享 50% 费用优惠；MCP 联网搜索有免费额度与 QPS 限制，自定义服务按响应速度分基础/极速两种计费。 -4. **协议演进**：MCP 已从旧版 SSE 升级为 Streamable HTTP，已开通用户需"取消开通"后重新"立即开通"完成升级；Responses API 旧版路径即将停用，应使用 `/compatible-mode/v1/responses`，且 `previous_response_id` 传顶层 `id`（UUID，有效期 7 天）。 - -## 被对比主题页 - -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [model context protocol](../guides/model-context-protocol.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md deleted file mode 100644 index 58583e15..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression-vs-deployment.md +++ /dev/null @@ -1,73 +0,0 @@ -# 模型微调、压缩与部署对比 - -在百炼平台上，把一个模型从「原始能力」推向「专属生产服务」通常经过三个环节：**模型微调 → 模型压缩（可选）→ 模型部署**。三者构成一条完整的自定义模型生产链路，但目标、输入输出、计费方式和适用场景各不相同。本页横向对比这三个环节的关键差异，帮助开发者理清「先做什么、要不要做、怎么上线」的技术选型问题。 - -需要先明确三者的关系： - -- **模型微调**解决「模型能不能做这件事」——把领域知识、指令遵循、人类偏好或特定音色写入参数。 -- **模型压缩**解决「部署贵不贵」——把全精度微调模型量化为低精度版本，降低部署所需的 MU 规格与推理成本，是链路中的**可选**环节。 -- **模型部署**解决「怎么对外提供服务」——为预置模型或自定义模型分配资源专享的推理服务，满足高并发、低延迟需求。 - -> 三者均**仅在华北2（北京）地域可用**，且需使用该地域的 API Key；子账号（RAM 用户）需预先获得相应授权。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | 模型部署（Deployment） | -| --- | --- | --- | --- | -| 核心目标 | 定制模型能力（知识/指令/偏好/音色） | 降低部署规格与推理成本 | 提供资源专享的推理服务 | -| 链路位置 | 第一步（必选） | 中间步骤（可选） | 最后一步（上线必选） | -| 输入 | 训练/验证数据集（JSONL、ZIP、OSS 挂载） | 上游全精度微调模型 + 可选校准数据 | 预置模型或自定义（含压缩后）模型 | -| 输出 | 微调后模型（`finetuned_output`） | 低精度量化模型 | 专属推理服务（`deployed_model`） | -| 支持模型范围 | 千问文本/VL、万相图像/视频、CosyVoice 等多模态 | 仅平台微调产出的自定义模型（如 qwen3.5-flash） | 部分预置模型 + 所有调优后模型 | -| 主要方式 | CPT / SFT（全参、LoRA）/ DPO | 量化（不含剪枝、蒸馏） | PTU / 模型单元(MU) / 按 Token | -| API 端点 | `POST /api/v1/fine-tunes`（配合 `/files`） | 无公开 API，仅控制台创建压缩任务 | `POST/GET/DELETE /api/v1/deployments` | -| 控制台路径 | 模型调优页面 | 模型 → 模型训练 → 模型压缩 | 我的模型 / 部署页面 | -| 计费方式 | API 仅按 Token；训练单元须走控制台 | 压缩任务限时免费，成本体现在部署阶段 | PTU（按 TPM）/ MU（按时长×单元）/ 按 Token | -| 可逆性 | 可继续微调、可组合多阶段 | **不可逆**，不支持继续微调或二次压缩 | 可下线重建；计费方式创建后不可改 | -| 典型场景 | 效果不达标、需深度领域/风格定制 | 微调模型上线成本高、需降本 | 高并发/低延迟生产、专属推理服务 | - -## 各环节适用场景建议 - -### 模型微调：什么时候做 - -- 当 Prompt 工程、插件调用等手段**仍无法满足效果**时，才引入微调。 -- 文本生成推荐按 `CPT（可选）→ SFT → DPO（可选）` 递进组合：补领域知识用 CPT（千万级 Token 无标签文本），学会遵循指令用 SFT（1000+ 条 ChatML 问答对），对齐人类偏好用 DPO（100+ 组 chosen/rejected）。 -- 训练模式在模型支持全参时**优先选全参**（费用相同、效果更好）；数据集小或对时间/成本敏感则用 LoRA 高效训练。 -- 多模态定制（视觉理解、图像/视频、语音合成）走各自独立的超参与数据格式，其中 CosyVoice 当前**只能通过 API 发起**。 -- 入门建议用控制台零代码流程；批量/自动化建议走 API 四步流程（上传数据 → 创建任务 → 轮询状态 → 部署调用），注意 **API 创建的任务仅支持按 Token 计费**。 - -### 模型压缩：要不要做 - -- 仅当已有**平台微调产出的自定义模型**且**部署成本偏高**时才考虑，属于可选优化环节。 -- 收益示例：qwen3.5-flash 微调模型从 MU1*2（108 元/小时）压缩到 MU8*1（47 元/小时），成本节省约 56%。 -- 权衡精度与成本：量化模板中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大；用与推理场景语义相近的校准数据可提升量化精度。 -- **注意不可逆**：压缩后不能继续微调、不能二次压缩，调整需从上游全精度模型重新压缩。 -- 压缩任务当前**限时免费**，建议在免费期内对同一模型尝试多个量化模板，用业务测试集验证后再选最优方案上线。 - -### 模型部署：怎么上线 - -按流量特征选择互斥的计费方式（创建后不可更改，切换须先下线再重部署）： - -- **预置吞吐（PTU）**：预留资源保障 TPM，额度内不限速，TPS 通常提升 1.5～2.0 倍，支持长输入（部分模型达 200K）与前缀缓存折扣。适合流量可预估的高负载生产（智能客服、实时内容审核）。超额或超长输入自动转按量计费，业务不中断。 -- **模型单元（MU）**：按时长×单元数计费，资源独占、性能可自定义，支持 PD 分离降低首 Token 延迟。适合需要独占资源、性能可控的生产场景，也是压缩后模型的落地方式。 -- **按 Token 使用量**：不使用不计费，仅支持基础模型 SFT 高效训练后的自定义模型，主要用于**调优效果验证**；扩缩容需控制台人工审核。 - -LoRA 模型导入需满足约束：仅支持 LoRA（不支持全参微调）、rank 为 8/16/32/64、不得修改 vocab 与 chat_template、VL 模型须冻结 VIT。API 部署通过 `plan` 字段区分计费方式（`ptu` / `mu` / `lora`），流程为创建部署 → 轮询至 `RUNNING` → 调用推理 → 用完 DELETE 下线停止计费。 - -## 技术选型参考 - -1. **只想验证微调效果**：微调（SFT 高效训练）→ 直接用「按 Token」部署验证，成本最低、无需长期占用资源。 -2. **要正式上线且流量可预估**：微调 →（评估成本后可选压缩）→ PTU 部署，享受不限速吞吐与缓存折扣。 -3. **要独占资源、性能可控**：微调 → 压缩降本 → MU 部署，用量化模板在成本与精度间取平衡。 -4. **本地已训练 LoRA**：跳过平台微调，直接经「我的模型」导入（满足 rank 与配置约束）→ 部署。 -5. **成本敏感的高负载场景**：优先评估「压缩 + PTU/MU」组合，先在压缩免费期内多模板试验，用业务测试集选出精度/成本最优版本再上线。 - -总体决策顺序：先用微调确保能力达标，再判断是否需要压缩降本，最后按流量与性能诉求选择部署计费方式。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md deleted file mode 100644 index 025bfd66..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -[模型微调（Fine-tuning）](../concepts/fine-tuning.md)与模型压缩（量化）是百炼平台模型生产链路中两个相邻但目标截然不同的环节。微调解决的是「模型能不能做好某项任务」的**能力问题**，通过训练修改模型参数来提升特定行业/业务表现；模型压缩解决的是「模型部署贵不贵」的**成本问题**，通过量化把全精度微调模型转为低精度版本，从而降低部署所需的 MU 规格。二者在完整链路中的位置为：**模型调优 →（可选）模型压缩 → 模型部署**——压缩的输入正是微调的产出。本文从技术选型角度对比两者的关键差异，帮助开发者判断在什么阶段该用哪个能力。 - -## 关键维度对比 - -| 维度 | [模型微调（Fine-tuning）](../concepts/fine-tuning.md) | 模型压缩（量化） | -| --- | --- | --- | -| 核心目的 | 提升模型在特定任务/领域的能力 | 降低部署规格与推理成本 | -| 处理对象 | 基础模型（千问系列、VL、图像/视频、语音等） | 仅限百炼平台微调产出的自定义模型 | -| 输入 | 训练数据集（ChatML / 纯文本 / ZIP 音视频等） | 上游全精度微调模型 + 可选校准数据集 | -| 输出 | 新的自定义微调模型 | 低精度（量化）版本的自定义模型 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA 高效训练） | 量化（不含结构剪枝、知识蒸馏） | -| 使用方式 | 控制台可视化 或 API/命令行（DashScope HTTP） | 控制台：模型 → 模型训练 → 模型压缩 → 创建压缩任务 | -| API 端点 | `POST /api/v1/files`、`/api/v1/fine-tunes`、`/api/v1/deployments` | 以控制台操作为主（未提供公开 API 流程） | -| 关键配置 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 量化模板（MU 编号）、量化产出后缀、校准数据 | -| 计费方式 | 按训练 Token 计费（CosyVoice 0.2 元/千 Tokens；部署另计） | 压缩任务本身限时免费；压缩后模型按部署 MU 规格计费 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 可逆性/复用 | 微调模型可继续被压缩、部署 | 不可逆；压缩后不支持继续微调或二次压缩 | -| 典型收益 | 指令遵循、领域知识、偏好对齐、风格定制 | 部署成本下降（示例约节省 56%） | -| 典型场景 | 通用模型无法满足行业需求、需注入专业知识或对齐偏好 | 微调模型已达标、需在保持能力前提下压低上线成本 | - -## 适用场景建议 - -### 优先选择模型微调 - -- Prompt 工程、插件调用等方法已用尽，仍无法满足业务对准确率、专业性或风格的要求。 -- 需要向模型注入领域词汇与事实（CPT）、教会模型遵循特定指令或执行任务（SFT）、或对齐人类偏好并抑制幻觉（DPO）。这三种方式可按 `CPT（可选）→ SFT → DPO（可选）` 递进组合。 -- 涉及多模态定制：视觉理解（VL，仅 SFT）、图像/视频生成（SFT-LoRA，需触发词）、语音合成（CosyVoice，仅 API 发起）。 -- 建议：若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）；对成本/时间敏感或数据集较小时选 LoRA 高效训练。 - -### 优先选择模型压缩 - -- 已经拥有一个训练达标的全精度微调模型，主要痛点是**部署/推理成本偏高**。 -- 希望在尽量保持模型能力的前提下降低部署 MU 规格（MU 编号越大规格越小、成本越低，但精度损失可能越大）。 -- 建议：利用压缩任务限时免费的窗口，对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选出成本与精度平衡最优的方案再正式上线；若面向特定场景（如客服问答），校准数据应选语义相近的数据集以提升量化精度。 - -### 二者结合的典型链路 - -对绝大多数生产落地而言，两者不是「二选一」而是「先后使用」：先用**微调**把模型能力打磨到位并验证效果，再用**压缩**在上线前压低部署成本。需特别注意压缩不可逆——若后续还想继续微调或调整，必须回到上游全精度微调模型重新压缩，因此应先冻结微调版本、确认效果达标后再进入压缩环节。 - -## 技术选型速查 - -- 目标是「让模型更会做事」→ 微调。 -- 目标是「让模型更省钱部署」→ 压缩。 -- 需要通过 API/命令行自动化全流程（上传数据 → 训练 → 部署）→ 微调具备完整 HTTP 接口；压缩目前以控制台操作为主。 -- 数据集较小或预算有限 → 微调选 LoRA 高效训练；压缩选较大 MU 编号模板并做好精度验证。 -- 两者均受**华北2（北京）地域**约束，须使用该地域 API Key 与工作空间。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md deleted file mode 100644 index b77b31c6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/fine-tuning-vs-model-compression.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型微调与模型压缩对比 - -在百炼平台的模型生产链路中，模型微调（Fine-tuning）与模型压缩（量化）是两个位置相邻但目标截然不同的环节。完整链路为：**模型调优 → 模型压缩（可选）→ 模型部署**。微调解决的是「模型效果不够好」的问题，通过修改模型参数让模型在特定行业/业务上表现更佳；模型压缩解决的是「模型部署成本太高」的问题，通过降低参数精度（量化）在尽量保持能力的前提下缩小部署所需的 MU 规格。两者并非替代关系，而是先后衔接：通常先微调得到高精度自定义模型，再对其进行压缩以降低推理成本。本文面向开发者，对两者的关键维度做对比，供技术选型参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | -| --- | --- | --- | -| 核心目标 | 提升模型在特定任务/领域的效果 | 降低部署 MU 规格，减少推理成本 | -| 技术手段 | CPT / SFT / DPO（全参或 LoRA） | 量化（不含结构剪枝、知识蒸馏） | -| 链路位置 | 生产链路起点 | 位于微调与部署之间（可选环节） | -| 输入对象 | 基础模型 + 训练数据集 | 百炼平台微调产出的自定义模型 | -| 输入数据格式 | ChatML（SFT/DPO）、纯文本（CPT）、ZIP/OSS（[多模态](../concepts/multimodal.md)）等 | 源模型 +（条件选填）最多 5 个校准数据集 | -| 输出产物 | 微调后的自定义模型（可部署、可再微调） | 低精度自定义模型（不可再微调、不可二次压缩） | -| 支持模型 | 千问文本/VL、Wan 图像/视频、CosyVoice 语音等[多模态](../concepts/multimodal.md) | 以控制台展示为准，如 qwen3.5-flash-2026-02-23 微调模型 | -| 使用方式 | 控制台（可视化）或 API/命令行（DashScope HTTP） | 控制台操作：模型训练 → 模型压缩 → 创建压缩任务 | -| 关键 API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/`、`POST /api/v1/deployments` | 暂无公开 API，通过控制台创建压缩任务 | -| 关键配置项 | learning_rate、n_epochs、max_length、lora_rank、training_type 等 | 任务名称、源模型、量化模板（MU 编号）、量化后缀、校准数据 | -| 任务状态 | PENDING → RUNNING → SUCCEEDED | PENDING → QUEUING → RUNNING → SUCCEEDED / FAILED / CANCELED | -| 计费方式 | 按训练 Token 计费；CosyVoice 0.2 元/千 Tokens + 部署时长 | 压缩任务限时免费，压缩后模型按部署 MU 规格计费 | -| 可逆性 | 微调模型可继续训练、再压缩 | **不可逆**，不支持继续微调或二次压缩 | -| 地域限制 | 仅华北2（北京），须用该地域 API Key | 仅华北2（北京） | -| 典型场景 | 注入领域知识、学会指令遵循、对齐人类偏好、定制风格 | 高精度模型上线前的成本优化 | - -## 各方案的适用场景建议 - -### 优先使用模型微调 - -- **Prompt 工程/插件调用仍无法满足需求**：需要模型掌握专业词汇、事实性知识（CPT），或稳定地遵循特定指令与任务格式（SFT）。 -- **需要对齐人类偏好、抑制幻觉**：在 SFT 之上叠加 DPO，用「更好-更差」回答对进一步优化。 -- **[多模态](../concepts/multimodal.md)定制**：图像/视频生成的风格定制（Wan LoRA + 触发词）、视觉理解任务、语音合成音色克隆（CosyVoice，仅 API）。 -- **效果为先且能接受较高部署规格**：官方推荐若模型支持全参训练则优先全参（效果更好且与高效训练计费相同）。 - -### 适合使用模型压缩 - -- **已有微调完成的高精度自定义模型，且部署成本偏高**：例如需要长期在线服务、对每小时 MU 费用敏感的场景。以 qwen3.5-flash 微调模型为例，压缩后可从 MU1\*2（108 元/小时）降至 MU8\*1（47 元/小时），成本节省约 56%。 -- **业务对少量精度损失可接受**：MU 编号越大部署规格越小、成本越低，但精度损失可能越大，需按业务权衡。 -- **有语义相近的校准数据**：客服问答等场景应选择贴近实际推理语义的校准数据集，以提升量化精度。 - -## 面向开发者的技术选型参考 - -1. **先明确瓶颈**：效果不达标 → 微调；效果已达标但推理太贵 → 压缩。二者不冲突，常规链路是「先微调，再按需压缩」。 -2. **顺序与不可逆性**：压缩必须以微调产出的自定义模型为输入，且压缩不可逆。若后续还想继续微调或调整训练，务必保留上游全精度微调模型；压缩模型无法二次微调或二次压缩。 -3. **成本估算**：微调按训练 Token 计费（CosyVoice 另计），部署按 MU 规格计费；压缩任务本身限时免费，收益体现在部署阶段更低的 MU 规格。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证效果，选最优方案上线。 -4. **接入方式差异**：微调支持控制台与 API/命令行（可编排到 CI/CD），压缩目前以控制台操作为主。 -5. **共同约束**：两者均仅在华北2（北京）可用，须使用该地域 API Key；压缩仅支持百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md new file mode 100644 index 00000000..3ab6c78e --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md @@ -0,0 +1,65 @@ +# [多模态](../concepts/multi-modal.md)生成能力对比：图像、视频与3D生成 + +为帮助开发者快速理解百炼平台在[多模态](../concepts/multi-modal.md)生成领域的技术布局与能力边界，本文系统对比图像生成（Image Generation）、视频生成（Video Generation）与3D生成（3D Generation）三大核心能力。对比聚焦实际工程落地的关键维度——包括调用模式、模型生态、输入输出规范、计费逻辑与适用场景，旨在为技术选型提供客观、可操作的决策依据。所有信息均基于当前（2024年Q3）百炼平台正式发布的API文档与控制台配置。 + +## 关键能力维度对比 + +| 维度 | 图像生成（Image） | 视频生成（Video） | 3D生成（3D） | +|------|-------------------|-------------------|--------------| +| **核心输入格式** | 文本（[prompt](../guides/prompt.md)）、单图/多图（URL）、掩码图（mask_image_url）、草图（sketch）、风格参考图等；支持图文混排指令 | 文本（[prompt](../guides/prompt.md)）、首帧/首尾帧图像（image_url）、参考视频（video_url）、音频（audio_url）、[多模态](../concepts/multi-modal.md)组合（如图+音+[prompt](../guides/prompt.md)） | 文本（prompt）、单张图像（image）、四视角图像数组（images: [front, left, back, right]）；三者互斥 | +| **核心输出格式** | JPEG/PNG 图像（URL 或 base64），支持 512×512 至 4K 分辨率；含预览图、水印开关、扩展结果（如增强 prompt） | MP4 视频（URL），时长默认 5 秒（可设 2–10 秒），分辨率支持 480P–1080P；含封面帧、元数据（duration/frame_rate） | GLB 格式 PBR 材质模型（pbr_model_url）、无贴图基础网格（base_model_url）、渲染预览图（rendered_image_url）；支持面数分级（2万–200万面） | +| **主流支持模型** | `qwen-image-3.0-pro`, `wan2.7-image-pro`, `kling/kling-v3-image-generation`, `vidu/vidu-image_reference2image`, `z-image-turbo` | `wan2.7-t2v-2026-06-12`, `happyhorse-1.1-t2v`, `vidu/viduq3-turbo_text2video`, `emo-v1`, `liveportrait`, `pixverse/pixverse-c1-t2v` | `Tripo/Tripo-P1.0`（快模版，≤2万面），`Tripo/Tripo-H3.1`（高精版，≤200万面） | +| **API 端点（推荐）** | `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation`（同步/异步共用路径，行为由模型决定） | `POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis`（全模型统一端点，强制异步） | `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation`（仅华北2可用，强制异步） | +| **调用模式** | **混合模式**：`wan2.6+` / `qwen-image-3.0-pro` / `z-image-turbo` 支持同步（直接返回结果）；`wanx-v1` / `wanx-x-painting` / `image-out-painting` 等仅支持异步（需轮询 task_id） | **强制异步**：全部模型必须使用 `X-DashScope-Async: enable`，创建任务后轮询 `GET /api/v1/tasks/{task_id}` 获取结果 | **强制异步**：必须启用 `X-DashScope-Async: enable`；轮询间隔建议 ≥15 秒；task_id 有效期 24 小时 | +| **计费方式** | 按生成张数计费（例：`wanx-v1` 0.16元/张，`image-out-painting` 0.18元/张）；主账号与子账号共享 500 张免费额度（90天有效期） | 按模型独立计费：文生视频按秒（如 `wan2.7-t2v` 0.35元/秒）、人像动画按时长（`emo-v1` 0.28元/秒）、口型替换按音频秒数；各模型有独立免费额度（如 `emo-detect-v1` 200次） | 按任务计费：`Tripo-P1.0` 0.8元/次，`Tripo-H3.1` 2.5元/次；暂无公开免费额度，需开通后查看控制台配额 | +| **典型响应耗时** | 同步调用：3–8 秒（T2I/I2I）；异步调用：10–60 秒（含排队） | 1–5 分钟（受分辨率、时长、模型复杂度影响显著；高精度或多镜头任务可达 8 分钟） | 2–10 分钟（`P1.0` 通常 ≤3 分钟；`H3.1` + `ultra` 模式常需 6–10 分钟） | +| **地域支持** | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；密钥与 endpoint 必须严格匹配 | 华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）；跨地域调用将返回 `401 Unauthorized` | **仅华北2（北京）**；其他地域 endpoint 不可用，调用必失败 | +| **关键限制** | • 输入图需公网可访问 HTTPS URL
• `wan2.5` 及以下版本不支持同步调用
• 局部重绘/背景生成等高级功能限北京地域专属域名 | • 所有模型强制异步，无同步选项
• 音频输入需清晰人声、≤30 秒
• Prompt ≤512 字符，含敏感词触发拦截
• `liveportrait` 等模型 QPS 限 1 | • 仅支持 JPEG/PNG；单图 ≤20MB；多图需严格四视角顺序
• `pbr=true` 时 `texture=false` 无效；唯一无贴图路径：`"texture": false, "pbr": false`
• 输出 URL 有效期仅 2 小时 | + +## 各方案适用场景建议 + +### ✅ 图像生成（Image）——适合「高并发、低延迟、强交互」场景 +- **推荐场景**：电商海报批量生成、AIGC设计助手（实时预览）、社交内容配图、UI组件自动化出图、AI修图SaaS集成。 +- **选型提示**：若需毫秒级响应（如用户拖拽即实时重绘），优先选用 `z-image-turbo` 或 `qwen-image-3.0-pro`（同步调用）；若需4K精细输出或复杂编辑（如虚拟模特试穿），选用 `wan2.7-image-pro` 并注意其仅支持异步流程。 + +### ✅ 视频生成（Video）——适合「叙事表达、数字人驱动、轻量内容生产」场景 +- **推荐场景**：短视频营销素材生成、AI主播播报、产品演示动画、教育口型同步课件、游戏NPC动作迁移。 +- **选型提示**：纯文本生成短片（≤5秒）选 `viduq3-turbo_text2video`；需精准动作控制选 `animate-anyone-gen2`；强调口型自然度选 `pixverse-lipsync`；对并发要求高（如批量生成）需提前申请 QPS 提升，并配置异步回调避免轮询压力。 + +### ✅ 3D生成（3D）——适合「工业可视化、电商3D展示、AR/VR内容基建」场景 +- **推荐场景**：电商商品3D建模（文生/图生）、工业零件快速原型、建筑概念可视化、元宇宙空间资产生成、教育三维教具制作。 +- **选型提示**：快速验证创意或轻量应用 → `Tripo-P1.0`；需导入CAD/渲染管线或对接Unity/Unreal → `Tripo-H3.1` + `geometry_quality: "ultra"`；务必使用北京地域密钥与专属域名，且提前下载 `pbr_model_url`（2小时过期）。 + +## 技术选型参考指南（面向开发者） + +1. **优先确认调用模式约束** + - 若业务无法容忍异步延迟（如实时聊天机器人附带图片生成），**排除视频与3D方案**，仅考虑图像生成中的同步模型（`wan2.6-t2i` 及以上、`qwen-image-3.0-pro`）。 + - 若已构建成熟异步任务队列（如 Celery/RabbitMQ），视频与3D的强制异步特性反而是优势，可统一调度。 + +2. **严格校验地域一致性** + - 图像生成支持多地，但**视频与3D对地域敏感度极高**：3D仅限北京；视频若在新加坡部署服务，却误用北京密钥，将直接鉴权失败。建议在初始化 SDK 时硬编码 `region` 参数，并做启动校验。 + +3. **输入准备成本是隐性瓶颈** + - 图像：只需文本或单图，接入成本最低； + - 视频：需准备高质量音频/多帧图像，且 URL 必须公网可直连（OSS需设 public-read）； + - 3D：多图生3D要求严格视角顺序与光照一致性，实测中“前左后右”四图质量不均将导致模型崩坏。建议优先尝试文生3D降低门槛。 + +4. **计费颗粒度决定架构设计** + - 图像按张计费 → 适合按需调用，可缓存结果复用； + - 视频按秒计费 → 需精确控制 `duration` 参数，避免默认5秒造成浪费； + - 3D按次计费 → 建议对同一prompt/image做结果缓存（MD5哈希索引），避免重复生成。 + +5. **错误处理策略差异化** + - 图像：关注 `400 Bad Request`（参数错）、`403 Forbidden`（额度超）； + - 视频：高频出现 `429 Too Many Requests`，需实现指数退避轮询； + - 3D：`task_status: "UNKNOWN"` 表示 task_id 过期，必须重新提交任务——不可重试旧ID。 + +> **最后提醒**：所有多模态能力均依赖 DashScope SDK 最新版（≥4.20.0）及百炼控制台「业务空间」配置。请勿混用旧版文档（如 `wanx-v1` 协议）与新版模型（如 `wan2.7-*`），模型名与 endpoint 的严格匹配是调用成功的前提。 + +## 被对比主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md deleted file mode 100644 index 8b952a86..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-modalities-comparison.md +++ /dev/null @@ -1,54 +0,0 @@ -# 图像、视频与3D生成对比 - -百炼平台把图像、视频、3D 三类生成能力统一收口在模型推理网关下，使用相同的 `Authorization: Bearer ` 鉴权与 JSON 承载的输入输出协议，但三者在输入模态、输出形态、调用模式、耗时、计费维度与典型场景上差异显著。本文面向需要做多模态生成技术选型的开发者，横向对比三类能力的关键维度，帮助快速锁定适合业务的接口族。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 输入格式 | 文本 [prompt](../guides/prompt.md)；参考图/蒙版（URL 或 Base64） | 文本；首帧/首尾帧图像；参考多图；视频；音频 | 文本 [prompt](../guides/prompt.md)（≤1024 字符）；单图 URL；多图（前/左/后/右 4 视角，固定数组） | -| 输出格式 | 单张或多张静态图像（临时 URL） | 视频文件（临时 URL） | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| 支持模型族 | 通义千问图像、万相（Wan）、Z-Image、可灵、创意工具系列 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| API 端点 | OpenAI 兼容 `/compatible-mode/v1/images/generations`；或 DashScope 原生 `/services/aigc/text2image/image-synthesis` | `/services/aigc/video-generation/video-synthesis`（万相2.7 等）；或 `/services/aigc/image2video/video-synthesis`（动作/换人/数字人/旧版首尾帧） | `/services/aigc/video-generation/3d-generation` | -| 调用模式 | 同步（兼容模式）或异步轮询（万相/创意工具，返回 task_id） | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | 全部异步：创建任务得 task_id → 轮询 `GET /tasks/{task_id}` | -| 必要请求头 | `Authorization`；异步任务需 `X-DashScope-Async: enable` | `Authorization`、`Content-Type: application/json`、`X-DashScope-Async: enable`（缺异步头报 `does not support synchronous calls`） | `Authorization`、`X-DashScope-Async: enable` | -| 典型耗时 | 秒级（同步）到数十秒（异步） | 1–5 分钟；视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 异步任务 24 小时 | 24 小时 | 24 小时；超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 临时 URL，需及时下载/转存 OSS | 临时 URL | 2 小时 | -| 地域约束 | 各地域通用，按模型开通 | 模型/Endpoint/API Key 须同地域；PixVerse、Vidu 仅华北2（北京） | 仅华北2（北京），且须用北京 API Key | -| 内容安全 | 内置审核，违规返回 `DataInspectionFailed` | 内置审核 | 内置审核，失败返回 `code`/`message` | -| 计费方式 | 按张/按次（视模型） | 按任务/时长 | 按任务类型（`text-to-3d` / `image-to-3d` / `multi-image-to-3d`）计数，`usage` 含生成数量 | -| 典型场景 | 文生图、图像编辑、电商/营销垂类、人像玩法、艺术文字 | 文生视频、图生视频、视频编辑、数字人、肖像动态、风格重绘 | 文生 3D、单图生 3D、多图生 3D，产出可二次加工的 GLB 资产 | - -## 调用模式差异 - -图像生成是三类中唯一支持**同步调用**的：千问-文生图等可走 OpenAI 兼容 `images/generations` 直接拿结果。万相与创意工具则多用 DashScope 原生异步协议，提交后拿 `task_id` 轮询。 - -视频与 3D 一律异步，且都强制 `X-DashScope-Async: enable` 头。视频接口明确要求"同地域"约束（模型、Endpoint、API Key 必须同地域），3D 则更严格——仅华北2（北京）可用。两者都强调"请勿重复创建任务，直接轮询"。 - -## 输入模态对比 - -- **图像生成**：以文本 [prompt](../guides/prompt.md) 为主，部分编辑接口接受参考图与蒙版，输入图支持公网 URL 或 Base64。 -- **视频生成**：多模态输入最丰富，万相2.7 支持文本/图像/音频/视频混合输入，参考生视频可保持角色与音色一致性；首尾帧、参考多图等模式扩展了可控性。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥。多图模式视角顺序固定为前/左/后/右，数组长度固定 4，不需要的视角传空对象 `{}`，这是 3D 独有的约束。 - -## 输出与产物处理 - -图像与视频输出都是临时 URL，文档建议及时下载或转存 OSS。3D 输出更结构化：根据 `pbr`/`texture` 参数组合返回 `pbr_model_url`（PBR 材质 GLB）或 `base_model_url`（无贴图基础模型），并附 `rendered_image_url` 预览图，链接有效期仅 2 小时，短于图像/视频的临时 URL 生命周期。 - -## 选型建议 - -- **静态视觉物料（海报、商品图、人像）**：选图像生成。中文语义优先千问-文生图或万相-文生图 V2；按指令改图用万相-图像生成与编辑 2.7；电商/营销垂类用虚拟模特、AI 试衣、创意海报；高美感/艺术风格用 Z-Image 或可灵。 -- **动态视频内容（短剧、营销视频、数字人播报）**：选视频生成。通用文生/图生优先万相2.7（`wan2.7-t2v`/`wan2.7-i2v`）；多镜头叙事用万相2.6 `shot_type: multi`；数字人/换人用 `wan2.2-s2v`、`wan2.2-animate-mix`；北京地域可按需选 PixVerse、Vidu。 -- **可复用 3D 资产（游戏、电商 3D 展示、工业建模）**：选 3D 生成，仅限北京地域。高精度需求用 `Tripo/Tripo-H3.1`（最高 200 万面，`geometry_quality: ultra`）；追求速度用 `Tripo/Tripo-P1.0`（最高 2 万面）。需要 PBR 材质保留默认 `pbr: true`，仅需白模则同时关 `texture` 与 `pbr`。 -- **跨模态组合**：可先用图像生成产出关键帧，再喂给视频生成的图生/首尾帧接口生成动态内容；3D 则更适合独立资产管线，与图像/视频管线并行而非串行。 - -新接入一律优先最新版本（万相 2.7、通用图像编辑 2.5、wan2.7、Tripo-H3.1/P1.0），旧版接口保留兼容但不再增强。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md deleted file mode 100644 index f6d09469..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/getting-started-comparison.md +++ /dev/null @@ -1,60 +0,0 @@ -# 入门路径对比：开始使用与模型快速上手 - -阿里云百炼同时面向"应用构建者"和"模型调用开发者"两类用户，提供了两条不同的入门路径。`start-using` 侧重于在控制台零代码搭建基于私有知识的问答应用（智能体应用、工作流应用、知识库等）；`get-started-with-models` 则侧重于通过兼容 OpenAI 的 API 直接调用大模型完成第一次推理。本页从目标用户、输入形式、输出形式、支持模型、API 端点、计费方式、典型场景等维度对比两条路径，帮助开发者根据需求做出技术选型。 - -## 关键维度对比 - -| 维度 | [start using](../guides/start-using.md)（应用构建路径） | [get started with models](../guides/get-started-with-models.md)（模型调用路径） | -| --- | --- | --- | -| 主要目标 | 零代码/低代码搭建端到端的私有知识问答应用 | 通过 API 直接调用大模型完成文本/多模态推理 | -| 目标用户 | 业务人员、应用开发者、希望快速上线 RAG 应用的团队 | 开发者、需要将模型推理集成到自有代码或后端服务的工程师 | -| 上手时长 | 约 5 分钟完成第一个智能体应用（含知识库构建） | 几行代码即可完成首次调用，开通账号后即可跑通 | -| 输入形式 | 控制台可视化配置：System Prompt、知识库、技能、MCP 工具、工作流节点 | HTTP 请求 / OpenAI 兼容 SDK / DashScope SDK，传入 `messages`、`model`、参数 | -| 输出形式 | 发布后的应用（Web、微信、钉钉、音视频实时互动）+ Responses API 同步/异步调用 | 模型推理结果（文本、图像、音频、视频、向量、重排序结果） | -| 应用类型 | 智能体应用（Agent 2.0）、工作流应用、高代码应用、MCP 服务 | 直接调用模型 API；可叠加模型调优（SFT/CPT/DPO）、模型部署、模型评测 | -| 支持模型 | 千问系列（推荐 qwen3.7-max 用于问答）、QwQ 推理系列、DeepSeek 系列、视觉模型 qwen-vl-plus/max、嵌入 text-embedding-v3/v4 | 文本生成（qwen3.7-max/plus、qwen3.6-flash、deepseek-v4-pro/flash、kimi-k2.7-code、glm-5.2 等）、图像/视频/3D/音频/全模态/向量/重排序全谱系 | -| 知识库 | 核心能力：文档/数据/图片/音视频知识库、智能切分、检索调优、图文检索、监控 API | 不直接提供知识库；如需 RAG 需自行搭建或与百炼应用/知识库集成 | -| API 端点 | Responses API（同步 + `background=true` 异步），调用时需传入自定义参数；通过应用 ID 调用 | OpenAI 兼容 `/compatible-mode/v1`、Anthropic 兼容地址、DashScope SDK 端点；按计费方案选择 Base URL | -| 接入域名 | 应用调用走百炼统一接入，发布渠道含微信/钉钉/音视频 SDK | 业务空间专属 `{WorkspaceId}.{region}.maas.aliyuncs.com`（生产推荐）、`dashscope.aliyuncs.com`（存量）、`trial.{region}.maas.aliyuncs.com`（验证）、Token/Coding Plan 专属域名 | -| 凭证与鉴权 | 控制台操作为主；API 调用时使用应用相关凭证 | API Key（按业务空间隔离，不可跨地域混用）；建议写入环境变量 `DASHSCOPE_API_KEY` | -| 地域选择 | 应用开发与批量推理、模型调优仅在北京与新加坡支持 | 多地域可选（北京、新加坡、法兰克福、东京、弗吉尼亚），地域决定接入点与数据存储；服务部署范围可限定中国内地/国际/全球 | -| 计费方式 | 大模型调用计费 + 知识库规格费用 + 知识库模型调用费用（2026-01-04 起正式计费）；提供限时免费额度 | 按量付费（Dashscope 域名 / 业务空间专属 / 试用域名）、Token Plan（交互式，不可用于后端）、Coding Plan（AI 编码套餐）；不同方案对应不同 Base URL | -| 调用模式 | 同步调用（实时交互，可复用 OpenAI 代码库）、异步调用（`background=true` 返回 Task ID） | 同步、流式（SSE）、批量推理、异步任务；请求超时最高 3600 秒（业务空间专属域名） | -| 可观测性 | 应用观测（端到端流程）、应用评测（智能体/工作流/自定义评测集）、长期记忆与用户画像 | 模型告警（仅北京/新加坡）、模型评测、模型调优 | -| 典型交付物 | 可对外发布的问答应用（含欢迎语、预设问题、发布渠道、音视频互动） | 一次模型推理调用或集成了模型能力的后端服务 | - -## 适用场景建议 - -### 选择 [start using](../guides/start-using.md)（应用构建路径）当： - -- 需要在 5 分钟内零代码搭建一个能回答私有领域问题的问答应用。 -- 业务方或非工程师角色希望可视化配置 System Prompt、知识库、技能、MCP 工具。 -- 需要完整 RAG 流程（知识库构建、切分策略、检索调优、图文检索）且不想自行实现。 -- 需要发布到微信、钉钉、H5/APP 等渠道，或需要音视频实时互动。 -- 需要端到端应用观测、评测、长期记忆与用户画像管理。 - -### 选择 [get started with models](../guides/get-started-with-models.md)（模型调用路径）当： - -- 开发者需要将大模型推理直接集成到自有代码、后端服务或 AI 工作流中。 -- 需要使用 OpenAI 兼容 SDK 或 DashScope SDK，复用现有 OpenAI 代码库。 -- 需要访问文本、图像、视频、3D、音频、全模态、向量/重排序等完整模型谱系。 -- 对地域、服务部署范围、接入域名、并发上限、SLA 有精细控制需求。 -- 需要使用 Token Plan / Coding Plan 等专项计费方案，或需要进行模型调优（SFT/CPT/DPO）、模型部署、模型评测。 - -### 两条路径结合使用： - -实际项目中两条路径常常互补——先用 `get-started-with-models` 跑通模型调用、选定合适模型与地域，再用 `start-using` 将模型能力封装为带知识库、技能、发布渠道的完整应用；反过来，已发布的应用也可通过 Responses API 被后端服务以 OpenAI 兼容方式调用。开发者可先明确"我要的是模型推理能力还是端到端应用"，再据此选择起点。 - -## 来源主题页 - -- [start using](../guides/start-using.md)（guides/start-using.md） -- [get started with models](../guides/get-started-with-models.md)（guides/get-started-with-models.md） - -## 被对比主题页 - -- [start using](../guides/start-using.md) -- [get started with models](../guides/get-started-with-models.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md deleted file mode 100644 index c57c9b93..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-video-3d-generation-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成 - -百炼平台同时提供图像生成、视频生成和 3D 模型生成三大视觉内容创作能力。三者在输入输出格式、模型生态、调用方式、计费模式和适用场景上存在显著差异。本文从开发者技术选型角度，对三类能力进行系统对比，帮助开发者根据业务需求选择最合适的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| **输入格式** | 文本([prompt](../guides/prompt.md))、参考图片、涂鸦草图 | 文本([prompt](../guides/prompt.md))、首帧图片、参考图/视频、音频 | 文本([prompt](../guides/prompt.md))、单图、多图(4视角) | -| **输出格式** | PNG 图片(512x512 至 4K) | MP4 视频(5-10秒，1280x720等) | GLB 模型(带PBR材质贴图或无贴图) | -| **核心模型** | 千问(Qwen-Image)、万相(Wan)、Z-Image、可灵(Kling) | 万相(Wan 2.7)、HappyHorse、Pixverse、Vidu、Kling | Tripo-H3.1、Tripo-P1.0 | -| **调用方式** | 同步/异步均支持 | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | [异步调用](../concepts/async-invocation.md)(创建任务+轮询获取) | -| **典型响应时间** | 秒级至十秒级 | 1-5 分钟 | 分钟级(需轮询，建议间隔15秒) | -| **API端点** | [DashScope SDK](../concepts/dashscope-sdk.md) / HTTP | `/api/v1/services/aigc/video-generation/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持地域** | 多地域(部分模型仅北京) | 多地域(北京、新加坡等) | 仅华北2(北京) | -| **SDK兼容性** | [DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | OpenAI兼容SDK、[DashScope SDK](../concepts/dashscope-sdk.md)、HTTP | 仅HTTP | -| **产物有效期** | 即时返回，无时效限制 | task_id 24小时有效 | 下载链接2小时有效，task_id 24小时有效 | -| **模型数量** | 20+ 模型(含创意工具) | 10+ 模型系列 | 2 个模型 | -| **编辑能力** | 局部重绘、风格迁移、扩图、超分 | 视频重绘、风格转换、口型替换 | 无编辑能力 | - -## 输入输出能力详细对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 文生内容 | 支持(文生图) | 支持(文生视频) | 支持(文生3D) | -| 图生内容 | 支持(图像编辑、参考生图) | 支持(图生视频、参考生视频) | 支持(单图/多图生3D) | -| 多模态输入 | 多图参考、涂鸦 | 关键帧序列、音频驱动 | 4视角多图(前左后右) | -| 批量生成 | 单次1-9张 | 单次1条视频 | 单次1个模型 | -| 最大输出分辨率 | 4K(wan2.7-image-pro) | 1920x1080 | 最高200万面(H3.1) | - -## 计费与商业化对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|----------|----------|--------| -| 计费单位 | 按张计费 | 按任务/时长计费 | 按任务计费 | -| 免费体验 | 部分模型有免费额度 | 部分模型有免费额度 | 需开通Tripo服务 | -| 商业化程度 | 大部分已商业化 | 主力模型已商业化 | 已商业化 | - -## 适用场景建议 - -**选择图像生成的场景**： - -- 电商商品图、营销海报、社交媒体配图等静态视觉内容制作 -- 需要精确文字渲染(如广告文案嵌入图片) -- 图像局部编辑、风格转换、AI试衣等垂直场景 -- 对响应速度要求高(秒级出图) -- 需要批量生成多张候选图供筛选 - -**选择视频生成的场景**： - -- 短视频内容创作、广告视频制作 -- 数字人驱动(音频/文本驱动说话、唱歌) -- 服装展示、舞蹈动作等动态展示 -- 需要关键帧精确控制镜头运动 -- 已有视频的风格转换或口型替换 - -**选择3D生成的场景**： - -- 游戏资产、AR/VR场景中的3D物体快速原型 -- 电商3D商品展示(可旋转查看) -- 建筑/工业设计的概念验证模型 -- 需要标准PBR材质的可渲染模型 - -## 技术选型决策参考 - -1. **内容维度**：静态画面选图像生成；需要动态表现选视频生成；需要空间立体展示选3D生成。 -2. **时效要求**：图像生成响应最快(秒级)，适合实时交互；视频和3D均为分钟级异步任务，适合离线批处理。 -3. **生态成熟度**：图像生成模型最丰富、功能最全面；视频生成处于快速发展期，模型迭代频繁；3D生成目前模型较少，但输出质量已达可用水平。 -4. **地域限制**：3D生成仅限北京地域，部分图像创意工具同样限北京；视频生成地域覆盖较广。 -5. **集成复杂度**：图像生成支持同步调用，集成最简单；视频和3D需要实现异步轮询逻辑，建议封装任务状态管理层。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md deleted file mode 100644 index bfc49c97..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-vs-3d-generation.md +++ /dev/null @@ -1,52 +0,0 @@ -# 图像、视频与 3D 生成对比 - -阿里云百炼平台在 DashScope 网关上提供了图像、视频与 3D 三大类视觉内容生成能力。三者虽同属「生成式媒体」范畴、共用同一套鉴权与任务模型，但在输入输出格式、可用模型、调用协议、地域限制与产物形态上差异明显。本文面向开发者，横向梳理三类方案的关键维度，帮助在技术选型时快速判断该用哪一类 API。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 主要能力 | 文生图、图像编辑、图像翻译、垂直创意工具（虚拟模特、扩图、擦除补全、海报等） | 文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑、数字人、人像驱动、超清、对口型 | 文生 3D、单图生 3D、多图生 3D | -| 典型输入 | `prompt` / `negative_prompt`；编辑类传 `image_url`/`images`/`mask_image_url`；新版用 `messages` 多模态结构 | 文生用 `prompt`；图生/参考/编辑用 `media` 数组或 `image_url`/`video_url`/`audio_url`/`first_frame`/`last_frame` 等 | `prompt`（≤1024 字符）/ `image`（单图）/ `images`（4 元数组：前左后右），三者互斥 | -| 输出格式 | 图像 URL（有效期 24 小时），分辨率随模型而异（如 512×512~2048×2048、1K/2K/4K） | 视频 URL，分辨率 480P/540P/720P/1080P、可设 `duration` | 带贴图 PBR 材质 GLB（`pbr_model_url`）或无贴图基础模型（`base_model_url`）+ 预览渲染图，下载链接**仅 2 小时** | -| 代表模型 | Qwen-Image、万相 Wan/WanX（t2i/imageedit）、Z-Image、可灵、Vidu 等 | 万相 Wan 2.1-2.7、PixVerse、Vidu、可灵 Kling、HappyHorse、EMO/LivePortrait 等 | `Tripo/Tripo-H3.1`（最高 200 万面）、`Tripo/Tripo-P1.0`（最高 2 万面，速度更快） | -| 调用协议 | 异步（主流）+ **HTTP 同步**（仅 wan2.6/2.7、z-image 等新版） | **全部异步**（创建任务 → 轮询），无同步 | **全部异步**（创建任务 → 轮询），无同步 | -| 典型端点 | `text2image/image-synthesis`、`image2image/image-synthesis`、`multimodal-generation/generation`、`virtualmodel/generation` 等多种 | `video-generation/video-synthesis`（主）、部分数字人用 `image2video/video-synthesis` | `video-generation/3d-generation`（单一端点） | -| 生成耗时 | 通常 1-2 分钟 | 通常 1-5 分钟（统一编辑约 5-10 分钟） | 较长，建议轮询间隔约 15 秒 | -| 地域可用性 | 华北2（北京）为主，部分模型也支持新加坡/美国；大量创意工具**仅北京** | 华北2（北京）为主，第三方模型多**仅北京**，万相等部分支持新加坡/美国/德国 | **仅华北2（北京）** | -| 计费方式 | 仅对成功输出图片计费，含免费额度（通常 500 张/90 天），主子账号共享 | 按成功任务计费，随模型与分辨率/时长而异 | 仅对成功结果计数，`usage` 记录任务类型/数量/质量 | -| 关键请求头 | `Authorization`、`Content-Type`；异步需 `X-DashScope-Async: enable` | 同左，异步必带 `X-DashScope-Async: enable` | 同左，异步必带 `X-DashScope-Async: enable` | -| 任务状态 | `PENDING`/`RUNNING`/`SUSPENDED`/`SUCCEEDED`/`FAILED` | 同类异步枚举，过期返回 `UNKNOWN` | `PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED`/`CANCELED`/`UNKNOWN` | - -## 共性与差异要点 - -**共性**：三者都基于 DashScope 网关，共用 `Authorization: Bearer $DASHSCOPE_API_KEY` 鉴权、`model`/`input`/`parameters` 的请求体结构，异步模式均为「创建任务拿 `task_id` → 轮询查询」，且 `task_id` 有效期统一为 24 小时、创建异步任务必须携带 `X-DashScope-Async: enable`。地域隔离规则也一致：模型、Endpoint 与 API Key 必须同地域，跨地域会鉴权失败。 - -**核心差异**： - -- **协议丰富度**：只有图像生成的部分新版模型（wan2.6/2.7、z-image）支持一次请求返回结果的 HTTP 同步调用；视频与 3D **全部只能异步**。 -- **地域自由度**：图像与视频在北京之外还有一定跨地域支持，而 3D 生成**只有华北2（北京）**可用，选型时需特别注意。 -- **产物时效**：图像/视频 URL 有效期 24 小时，而 3D 模型下载链接**仅 2 小时**，需在生成后尽快下载并转存。 -- **输入约束**：3D 的 `prompt`/`image`/`images` 三者互斥，多图必须是固定 4 元数组（前左后右）；图像与视频则允许更灵活的多模态、多图输入组合。 - -## 适用场景建议 - -- **图像生成**：适合海报、电商主图、创意配图、虚拟模特试衣、图像翻译/编辑等静态视觉需求。追求低延迟、希望一次请求出结果时，优先选支持 HTTP 同步的新版模型（wan2.6/2.7、z-image-turbo）；需要精细编辑（改文字、局部重绘、扩图、去水印）则用千问/万相编辑系列。 -- **视频生成**：适合短视频、广告片、数字人播报、口播/对口型、人像驱动等动态内容。需明确任务类型（文生/图生/首尾帧/参考/编辑）选择对应模型，注意多镜头控制方式在不同模型间不一致（万相 2.7、PixVerse-c1 用自然语言 `prompt`，旧版万相 2.6 需显式 `shot_type: "multi"` + `prompt_extend: true`）。 -- **3D 生成**：适合游戏/电商/XR 场景的 3D 资产快速建模。需要高精度、高面数选 `Tripo/Tripo-H3.1`（可用 `geometry_quality: ultra` 达 200 万面）；追求速度、面数需求不高选 `Tripo/Tripo-P1.0`。务必在北京地域开通 Tripo 服务并及时下载 2 小时时效的产物。 - -## 技术选型参考 - -1. **先按产物形态定类别**：要静态图片 → 图像；要动态视频 → 视频；要可交互 3D 模型（GLB） → 3D。 -2. **再评估延迟要求**：对响应速度敏感的图像场景可用同步协议；视频与 3D 必须做好异步轮询与任务状态处理（含 `FAILED`/`UNKNOWN`）。 -3. **确认地域与开通**：3D 与多数第三方视频/图像创意模型只在北京可用，需保证模型、Endpoint、API Key 同地域，并提前在控制台开通授权（如 Tripo）。 -4. **规划产物存储**：所有产出均为限时 URL，建议生成后立即转存至 OSS，其中 3D 仅 2 小时窗口最需注意。 -5. **统一工程实现**：三类 API 共用鉴权、请求头与异步模型，可复用同一套任务提交/轮询/重试封装，仅按 `model` 与端点差异做分支。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md index c96ffa81..f0be379a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md @@ -1,59 +1,59 @@ -# 知识库与记忆库对比 +# 知识库与记忆库功能对比 -知识库（Knowledge Base）与记忆库（Memory Library）都是百炼平台为大模型补充外部信息、突破上下文限制的能力，但二者解决的问题截然不同：知识库面向**领域知识的检索增强（RAG）**，把企业私有文档、结构化数据变成可被大模型检索的语料；记忆库面向**跨会话的用户记忆**，把对话中提取的关键信息和用户画像持久化，让智能体"记住"用户。开发者在做技术选型时，常会混淆两者，本文从多个维度对比，帮助你判断何时用哪一个、以及如何组合使用。 - -## 核心定位差异 - -- **知识库**：解决"大模型不知道我的专有知识/最新信息"的问题。数据来源是**静态文档语料**（手册、Excel、图片、音视频），检索的是与问题语义相关的知识切片。 -- **记忆库**：解决"大模型跨会话记不住用户"的问题。数据来源是**动态对话流**，自动提取记忆片段与用户画像，检索的是与当前对话相关的历史记忆。 +为帮助开发者清晰理解百炼平台中两类核心记忆增强能力的定位差异，本文从技术架构、使用范式与业务价值三个维度，系统对比**知识库（Knowledge Base）** 与**记忆库（Memory Library）**。二者虽均以“增强大模型上下文”为目标，但设计初衷、数据来源、生命周期及适用场景存在本质区别：知识库面向**静态、共享、领域化知识资产**，强调精准检索与结构化注入；记忆库面向**动态、私有、会话级用户记忆**，强调自动提炼与跨轮次上下文延续。正确区分二者是构建高可用智能体应用的关键前提。 ## 关键维度对比 | 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| 解决的核心问题 | 领域知识补充、RAG 检索增强 | 跨会话上下文记忆、用户个性化 | -| 数据来源 | 静态文档、Excel/CSV、图片、音视频 | 对话消息流，或 `custom_content` 直写 | -| 输入格式 | pdf/docx/ppt/txt/markdown/html、图片、音视频文件 | `messages` 对话数组 或 `custom_content` 文本 | -| 存储内容 | 向量化后的文档切片 | 记忆片段 + 结构化用户画像 | -| 输出/召回 | 语义相关的知识切片（供大模型引用） | 语义相关的记忆条目（注入 Prompt） | -| 检索机制 | 向量+关键词混合检索 + Rerank 重排 | 语义检索（`top_k` 控制条数） | -| API 端点 | 阿里云百炼 SDK（ApplyFileUploadLease → AddFile → CreateIndex → SubmitIndexJob …） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*`（AddMemory / SearchMemory） | -| 鉴权方式 | 子账号需 AliyunBailianDataFullAccess 策略 | 环境变量 `DASHSCOPE_API_KEY`（`sk-` 开头） | -| 控制台接入 | 三步创建 → 关联智能体/工作流/外部应用 | 可视化管理记忆库与记忆规则 | -| 零侵入接入 | 工作流节点 / 应用挂载 | OpenClaw 记忆插件（自动捕获/自动召回） | -| 数据隔离 | 按知识库、业务空间划分 | 按 `user_id` 命名空间隔离 | -| 有效期 | 索引长期有效 | 记忆片段默认 180 天（可配 7/30/180 天或永不过期） | -| 计费方式 | 规格费用（运行时长）+ 模型调用费（向量化+Rerank）；2026-01-04 起计费 | 随长期记忆 API 调用（提炼/向量化/检索由服务端完成） | -| 地域限制 | 仅华北2（北京）可用 | 通过 DashScope API 接入，不受知识库地域限制 | -| 典型场景 | 企业文档问答、产品手册检索、表格查询、图文/音视频问答 | 用户偏好记忆、长期助理、跨会话个性化 Agent | - -## 参数与调优侧重点 - -- **知识库**：调优围绕检索质量，关键参数有相似度阈值（0.01~1.0）、初步检索 TopK（1~100）、最大召回数量（1~20）、Rerank 排序模型、Meta 信息与标签过滤。注意 Rerank 费用取决于初步召回切片总数，降低 TopK 可显著省钱。 -- **记忆库**：调优围绕记忆写入与召回，关键参数有 `top_k`（建议 3~10）、`minScore` 相似度阈值、`profile_schema`（画像模板）、`project_id`（记忆片段规则），以及 OpenClaw 插件的 `autoCapture` / `autoRecall` 开关。 +|------|--------------------------|---------------------------| +| **核心定位** | 静态领域知识注入系统（RAG 基础设施） | 动态用户[长期记忆](../concepts/long-term-memory.md)管理系统（LTM 中枢） | +| **数据来源** | 手动上传的结构化/非结构化文档（PDF/Word/Excel/图片/音视频等） | 自动从对话 `messages` 中提取，或通过 API 写入 `custom_content` | +| **数据所有权** | 应用/业务空间级别共享（多应用可挂载同一知识库） | `user_id` 级别隔离（同一记忆库内不同用户数据完全独立） | +| **输入格式** | 文件（支持 20+ 格式）、文本块、URL；需预处理切片与元信息抽取 | JSON 格式 `messages` 数组（含 role/content/timestamp）或纯文本 `custom_content` + `user_id` | +| **输出格式** | 检索返回结构化 `nodes[]`：含 `content`、`source`、`score`、`meta` 等字段，供下游模型直接拼接提示词 | 检索返回 `memory_nodes[]`：含 `id`、`content`、`type`（event/profile）、`score`（[0,1]）、`created_at`；用户画像单独通过 `GetUserProfile` 获取 | +| **支持模型** | **不运行模型**，但深度依赖：
• 向量模型（`text-embedding-v4`/`qwen3-vl-embedding`）
• 排序模型（`qwen3-rerank`/`qwen3-vl-rerank`）
• 路由模型（`qwen-plus`，多库时启用） | **不运行模型**，但依赖：
• 记忆提取模型（内部调用，不可选）
• 用户画像抽取模型（基于 `profile_schema` 自动触发，不可替换） | +| **API 端点** | `POST /api/v1/knowledge_bases/{kb_id}/retrieve`（仅华北2北京地域） | `POST /api/v2/apps/memory/add`
`POST /api/v2/apps/memory/memory_nodes/search`
（全地域可用，无地域限制） | +| **计费方式** | 分层计费：
• 知识库规格费（按存储容量/月）
• Rerank 调用费（按初步召回总切片数 × 次数）
• 向量化/路由/问答模型费（按 [Token](../concepts/token.md) 单独计费） | 按调用量计费：
• `AddMemory`：按写入条数计费
• `SearchMemory`：按检索次数计费
• 无存储容量费（默认无限存储，按实际调用计费） | +| **生命周期管理** | 文档切片永久存储（除非手动删除）；Meta 抽取配置创建后不可修改 | 记忆片段默认永不过期；可通过 `memory_expiration_time` 规则配置有效期（仅对新写入生效） | +| **典型场景** | • 客服知识库问答（产品手册/FAQ）
• 法律合同条款检索
• 医疗文献辅助诊断
• [多模态](../concepts/multi-modal.md)内容搜索（图片中找文字、视频里查事件） | • 智能助手记住用户偏好（“我喜欢简体中文”）
• 跨会话任务延续（“继续上次未完成的报销流程”）
• 用户画像构建（职业/兴趣/健康目标）
• OpenClaw Agent 自动记忆与召回 | +| **地域限制** | **强制限定华北2（北京）地域**，其他地域不可用 | **全地域可用**（杭州、上海、新加坡、法兰克福等均支持） | +| **权限模型** | 依赖子账号 `AliyunBailianDataFullAccess` 权限；操作受业务空间隔离 | 依赖 `DASHSCOPE_API_KEY`（百炼平台生成），**不支持 Coding Plan Key**；`user_id` 为逻辑隔离边界 | ## 适用场景建议 -**优先选知识库**，当你需要： -- 让大模型基于企业内部文档、产品手册、FAQ 等**专有静态知识**回答问题； -- 检索结构化表格数据、图片内容或音视频剧情； -- 对答案准确性、引用来源、拒答/防泄漏有强要求（知识问答服务的极速/多轮智能模式）。 - -**优先选记忆库**，当你需要： -- 让智能体**跨会话记住用户偏好、历史信息与画像**，实现个性化交互； -- 为长期助理、客服 Agent 补充"上一次聊了什么"的连续性； -- 通过 OpenClaw 插件对现有 Agent 做**零侵入**的记忆增强。 - -## 组合使用 - -二者并不互斥，在完整的智能体架构中往往协同：**知识库**提供"专业知识大脑"（知道领域事实），**记忆库**提供"个人记忆大脑"（记得当前用户）。例如一个企业客服 Agent，可用知识库检索产品条款保证回答准确，同时用记忆库记住该用户的历史工单和偏好，做到既专业又贴心。 - -## 技术选型小结 - -- 判断信息**是否随用户/会话变化**：不变的领域知识 → 知识库；随对话演进的用户信息 → 记忆库。 -- 判断数据**来源形态**：文件/表格/多媒体 → 知识库；对话流 → 记忆库。 -- 判断接入约束：需要中国站华北2地域且走百炼 SDK → 知识库；只需 DashScope API Key 快速接入或 OpenClaw 插件 → 记忆库。 -- 追求最佳体验时，二者组合使用，分别承担"知识准确性"与"个性化连续性"两个正交目标。 +### ✅ 选择知识库，当您需要： +- 将**企业级静态知识资产**（如产品文档、规章制度、培训材料）规模化注入大模型； +- 支持**多用户、多应用共享同一知识源**，且要求高精度、低噪声的语义检索； +- 处理**非文本模态数据**（图片、音视频），需跨模态语义理解能力； +- 对检索结果的**可解释性与溯源性**有强要求（需明确 `source` 页码/时间戳）； +- 已有成熟文档管理体系，希望最小化改造接入 RAG。 + +### ✅ 选择记忆库，当您需要： +- 解决**单用户跨会话上下文丢失**问题，让智能体具备“记住用户”的能力； +- 构建**个性化体验**（如推荐、提醒、定制化回复），依赖持续积累的用户行为与偏好； +- 快速集成至**OpenClaw Agent 或自研对话系统**，追求零代码自动捕获与召回； +- 管理**高度动态、短生命周期的会话记忆**（如购物意图、待办事项、临时约定）； +- 需要**灵活的用户画像结构化能力**，并支持多轮交互逐步完善字段。 + +### ⚠️ 避免混淆的典型误区： +- **不要用知识库存储用户个人数据**：知识库无 `user_id` 隔离机制，所有用户共享同一检索空间，存在隐私与安全风险； +- **不要用记忆库替代领域知识库**：记忆库不支持文档解析、切片、[多模态](../concepts/multi-modal.md)索引，无法处理 PDF/Excel 等专业格式； +- **不要跨地域混用知识库 API**：在新加坡地域调用知识库接口将直接失败，而记忆库 API 无此限制； +- **不要期望记忆库提供文档级溯源**：记忆片段不保留原始文件位置，仅提供语义摘要与置信度分数。 + +## 技术选型参考（面向开发者） + +| 选型决策点 | 推荐方案 | 说明 | +|------------|----------|------| +| **是否需支持图片/音视频搜索？** | → 知识库 | 记忆库仅支持文本记忆，不提供[多模态](../concepts/multi-modal.md)嵌入与检索能力 | +| **是否需严格按 `user_id` 隔离数据？** | → 记忆库 | 知识库无用户维度，所有查询结果对所有用户可见 | +| **是否已有大量 PDF/Word 等文档需快速上线？** | → 知识库（控制台上传） | 提供一键解析、自动切片、可视化调试能力；记忆库需先人工提炼为 `custom_content` | +| **是否需在 OpenClaw Agent 中零配置启用记忆？** | → 记忆库（OpenClaw 插件） | 插件自动注册工具链，无需修改 Agent 代码；知识库需手动集成 `Retrieve` 节点 | +| **是否需对检索结果设置相似度阈值并过滤低分项？** | → 两者均支持，但参数单位不同 | 知识库 `相似度阈值`（0.01–1.0）；记忆库 `min_score`（0–100 百分制，API 返回 [0,1] 需转换） | +| **是否需审计每条检索的完整输入/输出？** | → 知识库（SLS 日志） | 知识库提供 `request_body`/`response_body.data.nodes[]` 全字段日志；记忆库日志需自行埋点 | +| **是否部署在非华北2地域（如新加坡）？** | → 记忆库（唯一选择） | 知识库在该地域不可用，强行调用将返回 `RegionNotSupported` 错误 | + +> **最佳实践组合建议**：在复杂智能体应用中，**知识库 + 记忆库 可协同使用**。例如：客服机器人中，用知识库回答“产品功能如何使用”，用记忆库记住“张三用户上周咨询过退货流程，本次优先展示退货进度”。二者通过不同 API 分别调用，结果在提示词工程阶段融合，实现“领域知识 + 用户上下文”的双重增强。 ## 被对比主题页 diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md deleted file mode 100644 index bb4bf118..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-data-management-comparison.md +++ /dev/null @@ -1,78 +0,0 @@ -# 知识库、记忆库与数据接入对比 - -百炼平台提供了三种互补的数据管理机制：**知识库**（Knowledge Base）、**记忆库**（Memory Library）和**数据连接**（Data Connection）。三者分别面向不同的数据形态和业务需求，开发者在构建[智能体应用](../concepts/agent-application.md)时往往需要根据数据特征、实时性要求和集成复杂度做出技术选型。本文从核心定位、数据来源、检索方式、[计费](../concepts/billing.md)模式等关键维度对三者进行系统对比，帮助开发者快速找到最适合自身场景的方案。 - -## 核心定位对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心技术 | RAG（检索增强生成） | 长期记忆 API（语义提取 + 持久化） | 数据源连接器（实时访问） | -| 解决的问题 | 为大模型补充私有文档和最新信息 | 解决跨会话上下文丢失问题 | 统一管理和访问企业外部数据源 | -| 数据形态 | 非结构化文档、结构化表格、图片、音视频 | 对话中的关键事件、用户画像属性 | 数据库、文档系统、对象存储 | -| 数据生命周期 | 持久存储，手动管理 | 可配置有效期（7/30/180 天或永不过期） | 实时连接，数据留在原系统 | -| 数据所有权 | 平台托管（上传后由平台管理） | 平台托管（自动提取并存储） | 数据留在原处，平台仅建立连接 | - -## 数据输入与支持格式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 输入方式 | 本地上传、OSS 导入 | 对话消息自动提取或 custom_content 直写 | 连接器配置（数据库凭证、OSS Bucket、[Token](../concepts/token.md)） | -| 支持格式 | PDF/DOCX/DOC/PPTX/TXT/MD/HTML/XLSX/XLS/PNG/JPG/BMP/GIF/音视频 | 对话 messages（JSON）或自定义文本 | MySQL/PostgreSQL/PolarDB-X 2.0/语雀/OSS/文件/表格 | -| 单文件限制 | 文档最大 150 MB（1000 页）；文本最大 10 MB；图片最大 20 MB；音视频最大 512 MB | 无文件概念，按对话轮次写入 | 文件连接器：平台存储最多 100,000 个文件、1 TB；表格连接器：1 TB 免费额度 | -| 数据解析 | 电子文档/文档智能/大模型/Qwen VL/音视频 五种解析方式 | 系统自动从对话中提炼关键信息 | 文件连接器支持与知识库相同的五种解析方式 | - -## 检索与集成方式 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 检索机制 | 语义向量检索 + 关键词混合检索 + Rerank 排序 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 百炼 SDK 知识检索/知识问答接口 | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用内工作流节点或内置工具调用 | -| 集成入口 | [智能体应用](../concepts/agent-application.md)、工作流应用、外部 API | API 直连、OpenClaw 插件（零侵入） | 应用内数据连接器节点 | -| 多源联合 | 支持最多 15 个知识库联合检索 | 按 user_id 自动聚合同一用户记忆 | 支持同时配置多个连接器 | -| 实时性 | 需重新导入和索引后才能检索新内容 | 对话结束后即时写入，下次对话可召回 | 流处理类实时访问源数据库最新数据 | - -## [计费](../concepts/billing.md)与配额 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 规格费用 | 标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时 | 以长期记忆 API 调用[计费](../concepts/billing.md) | 平台存储限时免费；自有 OSS 按 OSS 费用 | -| 模型调用费用 | 向量化 + Rerank 排序（按 [Token](../concepts/token.md) 计费） | 记忆提取和检索的模型调用费用 | 大模型文档解析按模型调用计费 | -| 免费额度 | 新用户 720 小时标准版（开通后 30 天有效） | 参见长期记忆 API 计费说明 | 文件/表格平台存储限时免费（1 TB） | -| 并发限制 | 标准版 1 QPS；旗舰版 50-10,000 QPS | 参见 API 限流说明 | 取决于源数据库和网络配置 | -| 地域限制 | 仅华北2（北京） | 参见 DashScope 服务地域 | 流处理类需网络可达；PolarDB-X 仅支持私网 | - -## 适用场景建议 - -**选择知识库**适用于： -- 企业有大量私有文档（产品手册、技术规范、FAQ 等），需要让大模型基于这些文档进行准确问答 -- 需要图文并茂的回复、复杂 PDF/图表理解、音视频内容检索等[多模态](../concepts/multimodal.md)场景 -- 对检索精度有较高要求，需要向量检索 + Rerank + 多轮对话改写等完整 RAG 流水线 -- 数据更新频率中等（可接受重新导入和索引的延迟） - -**选择记忆库**适用于： -- 智能体需要记住用户的历史偏好、习惯和关键事件，实现个性化持续服务 -- 需要跨会话保持上下文连贯性（如客服机器人记住用户之前反馈的问题） -- 希望零侵入接入（通过 OpenClaw 插件自动捕获和召回，无需改业务代码） -- 数据以对话形式产生，不需要事先准备结构化文档 - -**选择数据连接**适用于： -- 企业数据存储在 MySQL、PostgreSQL 等关系数据库中，需要智能体实时查询最新数据 -- 数据不适合或不允许复制到平台，要求数据留在原系统 -- 需要对接语雀文档、OSS 存储等已有数据系统 -- 业务场景需要执行 SQL 查询获取精确的结构化数据（如订单查询、库存查询） - -## 组合使用建议 - -三种方案并非互斥，在复杂业务场景中推荐组合使用： - -- **知识库 + 记忆库**：知识库提供专业领域知识，记忆库记住用户偏好和历史交互，二者结合实现既专业又个性化的智能服务。 -- **知识库 + 数据连接**：知识库提供静态文档知识，数据连接提供实时业务数据，适合需要同时参考文档和查询数据库的场景（如技术支持 + 工单系统）。 -- **三者组合**：在全场景智能助手中，知识库负责专业知识问答，数据连接负责实时数据查询，记忆库负责用户画像和交互记忆，共同构建完整的智能体数据底座。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md deleted file mode 100644 index 818cb053..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-compare.md +++ /dev/null @@ -1,69 +0,0 @@ -# 知识库与长期记忆对比 - -阿里云百炼平台针对「让大模型使用外部信息」提供了两条不同路径：**知识库（RAG）** 负责为模型补充**私有文档与最新事实**，**长期记忆（Memory Library / 长期记忆 API）** 负责跨会话持久化**用户个性化信息**。二者常被混淆，但设计目标、数据形态、检索机制与计费方式都截然不同。本文面向开发者，从技术选型角度对二者做逐维度对比，帮助你在智能体、问答、Agent 等场景中选对能力。 - -其中长期记忆在平台上有两个入口：控制台侧的**记忆库（Memory Library Overview）** 与开放的 **长期记忆（新）RESTful API**，二者底层为同一套服务，下文将它们合并为「长期记忆」一列，必要处再区分入口差异。 - -## 一句话定位 - -- **知识库**：把企业文档/结构化数据/多模态素材建成可检索索引，回答前先检索、再增强生成，解决「模型不知道我的私有资料」。 -- **长期记忆**：从多轮对话中自动提炼关键事件与用户画像并持久化，后续对话语义召回后注入 Prompt，解决「模型记不住这个用户」。 - -## 关键维度对比 - -| 维度 | 知识库（RAG） | 长期记忆（Memory Library / 长期记忆 API） | -| --- | --- | --- | -| 解决的问题 | 为模型补充私有数据与最新信息，提升回答准确性 | 跨会话保留用户偏好与历史，实现个性化上下文 | -| 数据来源 | 预先上传的文档、Excel/CSV、图片、音视频等素材 | 从对话消息自动提取，或 `custom_content` 直接写入 | -| 内容形态 | 切片（chunk）+ 向量索引 + Meta 信息 | 记忆片段（关键事件）+ 用户画像（结构化属性） | -| 写入时机 | 建库/导入阶段离线建索引（AddFile→CreateIndex→SubmitIndexJob） | 对话过程中/结束后实时写入（AddMemory） | -| 输入格式 | 文件（pdf/docx/ppt ≤150MB、txt/md/html ≤10MB、图片 ≤20MB、音视频 ≤512MB） | `messages`（对话，最多 50 条）或 `custom_content`（≤512 字符）+ `user_id` | -| 输出/召回 | 召回相关切片（单次最多 20），供大模型增强生成 | 召回相关记忆片段/画像，注入 Prompt | -| 检索机制 | Query 改写 → 向量+关键词混合检索 → Rerank 精排 → 加权返回 | 语义检索，可选 rerank / query 重写 / 意图判别 | -| 隔离维度 | 按知识库 ID / 业务空间 | 按 `user_id`（记忆空间），可再按 `memory_library_id` | -| API 端点 | `bailian.cn-beijing.aliyuncs.com`（阿里云 OpenAPI 风格） | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | -| 认证方式 | 子账号策略 `AliyunBailianDataFullAccess` | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 核心接口 | AddFile / CreateIndex / SubmitIndexJob / 知识检索 / 知识问答 | AddMemory / SearchMemory / ListMemory / Update / Delete / ProfileSchema 系列 | -| 关键参数 | 相似度阈值、召回片段数 TopK(1–20)、权重、Meta 抽取、智能切分 | `top_k`(1–100)、`min_score`(默认 0.3)、`enable_rerank/rewrite/judge` | -| 地域限制 | 仅中国站**华北2（北京）**可开通使用 | 通过 DashScope 全局域名接入，无北京地域限制 | -| 数据有效期 | 长期保存，删除即永久清除且停止计费 | API 侧「暂无失效日期」；控制台默认 180 天，可配 7/30/180 天或永不过期 | -| 计费方式 | 规格费用（标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时）+ 模型调用 Token 费；2026-01-04 起计费 | 按接口调用限流管理（全部 ≤3000 QPM，add 120 QPM，search 300 QPM） | -| 零侵入接入 | 关联到智能体/工作流应用（知识库节点） | OpenClaw 记忆插件（`before_agent_start` 召回 + `agent_end` 捕获） | -| 典型场景 | 企业文档问答、产品手册、客服、结构化数据查询、多模态搜索 | 个人助理记住用户习惯、智能体长期陪伴、跨会话偏好延续 | - -## 适用场景建议 - -### 选择知识库（RAG）当 - -- 你有**成体量的私有资料**（文档、手册、报表、图片、音视频）需要模型准确引用。 -- 要求回答**可溯源**（展示引用来源）、可做效果评测与持续优化。 -- 数据是**面向所有用户共享**的事实性知识，而非某个用户的个人信息。 -- 可接受仅在**华北2（北京）**地域使用，并规划好规格费用与 Token 成本。 - -### 选择长期记忆当 - -- 你要让智能体**记住单个用户**的偏好、习惯、历史事件，实现个性化。 -- 数据来自**实时对话**、随时间增长且需自动去重/更新。 -- 需要按 `user_id` 做**多用户记忆隔离**，或用画像模板维护结构化属性。 -- 希望**零侵入接入**（OpenClaw 插件）或用轻量 HTTP API 快速集成。 - -### 组合使用（推荐） - -二者并不互斥，常见的高质量智能体会同时使用：**知识库**提供权威事实与私有知识，**长期记忆**提供该用户的个性化上下文。典型编排为——对话前用 `SearchMemory` 召回用户记忆注入 Prompt，同时用知识检索召回相关文档切片；对话后用 `AddMemory` 沉淀新的用户信息。这样既「答得准」又「记得住」。 - -## 技术选型速查 - -- 「模型不知道我的资料」→ 知识库。 -- 「模型记不住这个用户」→ 长期记忆。 -- 需要**引用溯源 / 结构化数据查询 / 多模态素材** → 知识库。 -- 需要**跨会话个性化 / 用户画像 / 对话自动沉淀** → 长期记忆。 -- 受限于**北京地域**或对**规格费用**敏感 → 优先评估知识库成本；长期记忆走 DashScope 全局接入且以调用限流计。 -- 想**最快接入 Agent** → 长期记忆 OpenClaw 插件（配 `apiKey` + `userId` 即可自动捕获/召回）。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md deleted file mode 100644 index 9ff18a9a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-memory-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 知识库 vs 记忆库 vs 数据接入对比 - -百炼平台提供三种核心数据管理能力：知识库（RAG）、记忆库（长期记忆）和数据连接。三者均用于为大模型补充外部信息，但在数据组织方式、检索机制和适用场景上存在本质差异。本文从开发者技术选型角度，对三者的能力边界和关键特征进行系统对比。 - -## 关键维度对比 - -| 维度 | 知识库 | 记忆库 | 数据连接 | -|------|--------|--------|----------| -| 核心定位 | 基于 RAG 的私有文档检索增强 | 跨会话长期记忆持久化与召回 | 外部数据源统一接入与实时访问 | -| 输入格式 | PDF、DOCX、TXT、Markdown、HTML、XLSX、图片、音视频等非结构化文件 | 对话消息（messages）或自定义内容（custom_content） | 数据库连接串、OSS Bucket、语雀 [Token](../concepts/token.md)、本地文件上传 | -| 数据存储方式 | 平台托管（向量化切片存储） | 平台托管（记忆片段 + 用户画像） | 平台托管或流处理（原数据源实时访问） | -| 检索机制 | 向量检索 + 关键词检索 + Rerank 精排 | 语义检索（基于 user_id 隔离） | SQL 查询（流处理类）或向量检索（OSS 连接器） | -| API 端点 | 知识检索服务 / 知识问答服务（百炼 SDK） | `dashscope.aliyuncs.com/api/v2/apps/memory/*` | 通过应用工作流节点或工具调用 | -| 数据粒度 | 文档切片（最大 6,000 [Token](../concepts/token.md)/片） | 记忆片段（事件级）或用户画像（属性级） | 原始数据行/文件/文档 | -| 数据更新方式 | 手动上传 / OSS 导入，需重新索引 | 自动从对话提取，支持去重和动态更新 | 实时连接原数据源，数据变更即时生效 | -| 多用户隔离 | 无内置用户隔离（按知识库粒度管理） | 原生 user_id 级别隔离 | 按连接器实例隔离 | -| 并发规格 | 标准版 1 QPS / 旗舰版 50-10,000 QPS | 未公开 QPS 限制 | 取决于底层数据源能力 | -| [计费](../concepts/billing.md)方式 | 标准版 0.03 元/库/小时；旗舰版 0.2 元/RCU/小时 + 检索/排序 [Token](../concepts/token.md) 费 | 包含在百炼平台使用中（按 API 调用） | 文件/表格平台存储限时免费；流处理按底层数据源[计费](../concepts/billing.md) | -| 数据有效期 | 永久（手动删除） | 默认 180 天（可配置 7/30/180 天或永不过期） | 永久（随原数据源生命周期） | -| 集成方式 | 智能体应用 / 工作流节点 / SDK API | API 直连 / OpenClaw 插件零侵入接入 | 工作流节点 / 应用工具调用 | - -## 适用场景建议 - -### 选择知识库 - -- 企业有大量非结构化文档（产品手册、FAQ、技术文档）需要语义检索 -- 需要高精度的文档问答能力，支持[多模态](../concepts/multimodal.md)（图文、音视频） -- 对检索并发有明确要求（旗舰版支持万级 QPS） -- 需要精细控制检索质量（切片策略、Rerank、标签过滤等） - -### 选择记忆库 - -- 智能体需要跨会话记住用户偏好、历史交互和个性化信息 -- 需要按用户维度隔离记忆空间 -- 希望零侵入集成（通过 OpenClaw 插件自动捕获/召回） -- 数据来源是对话本身，而非预置文档 - -### 选择数据连接 - -- 需要实时查询企业数据库中的结构化数据（MySQL、PostgreSQL、PolarDB-X） -- 数据存储在外部系统（语雀、OSS）且需保持同步 -- 应用需要执行 SQL 查询获取精确结果 -- 数据更新频繁，不适合定期导入知识库 - -## 组合使用建议 - -三种能力并非互斥，实际应用中常组合使用： - -- **知识库 + 记忆库**：知识库提供通用文档检索，记忆库补充用户个性化上下文，实现"懂业务 + 懂用户"的智能体 -- **知识库 + 数据连接**：非结构化文档走知识库语义检索，结构化数据走数据连接 SQL 精确查询 -- **三者结合**：在工作流中编排知识库节点、数据连接工具和记忆注入，构建具备完整数据感知能力的复杂应用 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) -- [data connection overview](../guides/data-connection-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md deleted file mode 100644 index 83e098a7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-vs-memory.md +++ /dev/null @@ -1,63 +0,0 @@ -# 知识库与记忆库对比 - -阿里云百炼平台同时提供**知识库（Knowledge Base）**和**记忆库（Memory Library）**两种数据增强能力，二者分别面向不同的信息管理需求。知识库基于 RAG 技术，用于将企业私有文档、结构化数据等外部知识注入大模型，提升特定领域问答的准确性；记忆库则面向跨会话场景，自动从对话中提取和持久化关键信息，使智能体能够在多轮交互中保持对用户偏好与历史上下文的理解。本文从核心定位、数据模型、接入方式、检索机制等维度进行系统对比，帮助开发者根据业务场景做出合理的技术选型。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -| --- | --- | --- | -| **核心定位** | 基于 RAG 的外部知识检索增强，为大模型补充私有数据与最新信息 | 跨会话长期记忆，自动提取并持久化对话中的关键信息 | -| **数据来源** | 用户主动导入的文档（PDF/Word/Markdown/HTML/Excel/CSV）、图片、音视频等 | 从对话历史中自动提取，或通过 `custom_content` 直接写入 | -| **数据类型** | 文档搜索、数据查询（NL2SQL）、图片问答、音视频搜索四类，创建时选定不可更改 | 记忆片段（事件和信息）和用户画像（结构化属性），可独立或组合使用 | -| **存储粒度** | 切片（Chunk）级别，按智能切分或自定义规则对文档分片，单切片上限 6,000 Token | 记忆片段级别，系统自动提炼为精简的事实描述；用户画像按属性字段存储 | -| **数据持久性** | 持久存储，无过期机制 | 记忆片段可配置有效期（7/30/180 天或永不过期），控制台默认 180 天 | -| **检索机制** | 向量 + 关键词混合检索 + Rerank 排序，支持相似度阈值、TopK、权重等精细调控 | 基于语义的记忆检索（SearchMemory），通过 `top_k` 控制召回数量 | -| **向量模型** | `text-embedding-v4`/`text-embedding-v3`（512 维）；视觉理解用 `qwen3-vl-embedding`；图片问答用 `multimodal-embedding-v1`（1024 维） | 由平台内部自动处理向量化，用户无需选择向量模型 | -| **排序模型** | 支持 `qwen3-rerank`、`qwen3-rerank(hybrid)`、`qwen3-vl-rerank` | 无独立排序模型，由记忆检索服务内部排序 | -| **API 端点** | `bailian.cn-beijing.aliyuncs.com`，通过阿里云 SDK（`alibabacloud_bailian20231229`）调用 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*`，通过 DashScope API Key 认证 | -| **认证方式** | 阿里云 AccessKey（`ALIBABA_CLOUD_ACCESS_KEY_ID` / `SECRET`）+ WorkspaceId | DashScope API Key（`DASHSCOPE_API_KEY`，以 `sk-` 开头） | -| **接入方式** | 控制台可视化 + 开放 API（仅文档搜索类） | API 直连 + OpenClaw 记忆插件（零侵入自动捕获/召回） | -| **数据隔离** | 按知识库实例隔离，单次检索可绑定最多 15 个知识库 | 按 `user_id` 隔离，同一 `user_id` 共享记忆空间 | -| **可挂载模型** | 千问全系列（QwQ/Long/Max/Plus/Turbo/Coder 等）及第三方模型（DeepSeek/Llama/Yi 等） | 不直接挂载模型，作为独立记忆服务由应用侧在 Prompt 中注入 | -| **计费模式** | 标准版 0.03 元/小时；旗舰版 0.2 元/RCU/小时（1 RCU = 50 QPS），另计 SLS 日志存储费用 | 随 DashScope API 调用计费，无独立规格费用 | -| **地域限制** | 仅支持中国站华北2（北京）地域 | 通过 DashScope 全局端点访问 | -| **监控能力** | 内置日志投递至 SLS，支持调用审计、问题排查、用量统计与告警 | 通过控制台查看记忆库统计信息，无独立日志服务集成 | - -## 适用场景建议 - -### 适合选择知识库的场景 - -- **企业知识问答**：需要基于产品手册、内部文档、技术规范等大量非结构化文档进行精准问答。 -- **结构化数据查询**：通过自然语言查询 Excel/CSV 中的结构化数据（NL2SQL）。 -- **[多模态](../concepts/multimodal.md)检索**：需要对图片、音视频内容进行搜索和问答。 -- **高并发生产环境**：旗舰版支持最高 10,000 QPS，适合对吞吐量有要求的业务系统。 -- **精细化检索调优**：需要通过多种向量模型、排序策略、切片方式、相似度阈值等参数精确控制召回质量。 - -### 适合选择记忆库的场景 - -- **个性化智能助手**：需要记住用户偏好、习惯和历史交互信息，提供个性化服务。 -- **跨会话上下文保持**：用户多次对话之间需要保持连贯性，避免重复提供相同信息。 -- **用户画像构建**：需要从对话中自动提取和维护用户的结构化属性（年龄、职业、偏好等）。 -- **轻量级集成**：通过 OpenClaw 插件零侵入接入现有 Agent，无需改造应用代码。 -- **对话式应用**：客服机器人、个人助理等需要"记住"用户的长期交互场景。 - -### 组合使用 - -在实际业务中，知识库与记忆库可以组合使用以实现最佳效果。例如，一个智能客服系统可以同时挂载知识库获取产品文档中的专业知识，又通过记忆库记住每位用户的历史问题和偏好，从而在准确回答专业问题的同时提供个性化的服务体验。 - -## 技术选型参考 - -选择知识库还是记忆库，核心取决于数据的来源和用途： - -- 如果数据是**预先准备好的静态文档**，需要检索后辅助回答 -- 选择**知识库**。 -- 如果数据是**从对话中动态产生**的，需要跨会话持久化 -- 选择**记忆库**。 -- 如果两种需求并存，建议**同时接入**，各司其职。 - -从工程复杂度看，记忆库的接入成本更低（尤其是 OpenClaw 插件方式），而知识库提供了更丰富的检索调优手段和更高的生产环境承载能力。开发者应根据数据规模、并发需求、检索精度要求和集成方式综合评估。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md deleted file mode 100644 index d5b6d709..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-guide-vs-api.md +++ /dev/null @@ -1,56 +0,0 @@ -# 托管智能体：指南与 API 对比 - -百炼平台的 Managed Agents（托管智能体）提供了两套面向不同使用者的文档视角：**使用指南**侧重在控制台向导中完成端到端配置，适合快速上手与业务验证；**API 参考**则给出完整的 REST 资源模型与端点定义，适合工程化集成与自动化。二者描述的是同一套托管运行时——平台在服务端托管会话状态、沙箱环境与工具执行——但抽象层级、覆盖范围与目标读者不同。本页从关键维度做对比，帮助开发者按阶段选择合适的入口。 - -## 关键维度对比 - -| 维度 | 使用指南（guides/managed-agents） | API 参考（api/managed-agents-api） | -| --- | --- | --- | -| 定位 | 概念讲解 + 控制台向导操作流程 | REST 资源模型 + 端点契约 | -| 目标读者 | 初次接入、业务验证、快速体验 | 后端集成、SDK 调用、自动化编排 | -| 核心对象 | Agent、Environment、Session、Event（4 类） | Agent、Environment、Session、Event、Skill、File（6 类） | -| 操作方式 | 控制台向导 + 预览调试标签页 | 直接调用 REST API / SDK | -| 认证方式 | 控制台登录（隐式） | HTTP Header 携带 `Authorization: Bearer ` | -| API 基地址 | 未强调 | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio` | -| 支持地域 | 未强调 | 当前仅 `cn-beijing` | -| Agent 端点 | `POST /api/v1/agentstudio/agents`（创建） | 创建/获取/列出/更新/归档/列出版本 全套 | -| 版本管理 | 未展开 | 每次更新递增 `version`，会话锁定当时快照 | -| 事件流 | SSE 事件流（`GET /sessions/{id}/events/stream`） | SSE 长连接 + 事件历史分页（`GET /sessions/{id}/events`） | -| Skill 支持 | 提及可挂载预置工具组合 | 完整生命周期：上传/安全扫描/版本/下载/删除 | -| File 支持 | 上传挂载，路径 `/mnt/session/uploads` | 独立资源：上传/查询/列出/删除，审核状态机 | -| 文件大小上限 | 单文件 10 MB（上传挂载语境） | 单文件 20 MB，工作空间总量 100 GB，保留 30 天 | -| 状态机 | 会话状态、中断续接、工具审批（概念性描述） | `idle` → `running` → `idle` / `terminated`（显式定义） | -| 示例模型 | `qwen3-max`（向导示例另见 `qwen3.7-plus`，以下拉列表为准） | `qwen3-max` | -| 典型场景 | 多步工具调用、代码执行、文件处理等长时任务的快速搭建 | 生产环境集成、批量会话编排、CI/自动化流水线 | - -## 关键差异说明 - -- **抽象层级**：指南把流程压缩成"配置智能体 → 配置环境 → 发起会话 → 发送事件"四步，隐藏了大量端点细节；API 参考把每类资源的增删改查、版本、归档、软/硬删除逐一列清，是精确的契约。 -- **资源覆盖**：指南聚焦运行链路的 4 个核心对象（Agent / Environment / Session / Event），而 API 额外把 **Skill** 与 **File** 提升为一等资源，并明确了它们的安全扫描/审核状态机与挂载版本锁定规则。 -- **文件配额不一致**：指南上下文提到单文件 10 MB，API 参考给出 20 MB、工作空间 100 GB、保留 30 天。以 API 参考的配额为准，并注意实际以控制台/接口返回为准。 -- **端点方法差异**：API 总览页把部分更新端点标注为 `PATCH`，而各资源详情页标注为 `POST`（Agent / Environment / Session 均如此）。集成时以各资源详情页为准。 -- **版本与快照语义**：只有 API 参考明确了"会话创建时锁定 Agent 版本、Skill 挂载锁定具体 version、更新不影响已有会话"的重要约束，这对生产环境的稳定性至关重要。 - -## 适用场景建议 - -- **选择「使用指南」入口**： - - 第一次接触 Managed Agents，想快速理解 Agent / Environment / Session / Event 之间的关系。 - - 需要在控制台向导中拖拽配置、用预览调试标签页按事件类型（User、Agent、Tool、Tool_output 等）观察执行过程。 - - 做业务原型、Demo 或轻量验证，暂不需要写代码。 - -- **选择「API 参考」入口**： - - 需要把托管智能体嵌入后端服务、批量创建/复用 Agent 与 Environment。 - - 关注鉴权、地域、版本锁定、状态机、配额等工程化细节。 - - 需要管理 Skill 上传/安全扫描/版本挂载，或对 File 做上传/审核/挂载的自动化处理。 - - 构建 CI/自动化流水线，依赖稳定的端点契约与 `x-request-id` 排障能力。 - -## 技术选型参考 - -推荐路径是"**先指南、后 API**"：用指南在控制台跑通端到端流程、确认业务可行性，再切换到 API 参考完成工程化集成。两者并非二选一，而是覆盖同一运行时的不同生命周期阶段。集成落地时，务必以 API 参考为准处理认证（Bearer API Key）、地域限制（`cn-beijing`）、版本快照语义与端点方法差异，并按 API 侧的 20 MB / 100 GB / 30 天配额规划文件资源。 - -## 被对比主题页 - -- [managed agents](../guides/managed-agents.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md deleted file mode 100644 index a5f55a41..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/managed-agents-vs-llm-application.md +++ /dev/null @@ -1,58 +0,0 @@ -# 托管智能体与 LLM 应用对比 - -百炼平台既提供面向"应用构建"的 LLM 应用体系（智能体 / 工作流 / 高代码），也提供面向"长时自主任务"的 Managed Agents 托管运行时。二者虽然都以大模型为核心、都能接入知识库与 MCP 工具，但在运行模式、状态管理、执行环境和目标场景上定位截然不同。本文从技术选型视角梳理二者差异，帮助开发者判断"我该用哪一个"。 - -## 概念定位 - -- **托管智能体（Managed Agents）**：平台在服务端托管的智能体运行时，为多步工具调用、代码执行、文件处理等长时任务提供独立云端沙箱容器。会话状态、事件历史由服务端持久化，支持中断与续接，智能体在沙箱内自主执行命令、读写文件、安装依赖。 -- **LLM 应用**：面向应用构建的一整套模式，包含智能体（Agent，含 2.0 / 1.0）、工作流（Workflow）、高代码应用三种类型，覆盖从零代码配置到 Python 编码的不同开发门槛，用于快速搭建可发布、可被 API 调用的 AI 应用。 - -## 关键维度对比 - -| 维度 | Managed Agents（托管智能体） | LLM 应用（智能体/工作流/高代码） | -| --- | --- | --- | -| 核心定位 | 服务端托管的长时任务运行时 | 应用构建平台，多种应用形态 | -| 运行模式 | 服务端维护会话状态，支持中断与续接 | 智能体应用多为无状态调用，应用侧维护上下文 | -| 执行环境 | 独立沙箱、云端容器 | 共享运行时（高代码可选 Serverless/K8s 部署） | -| 事件模型 | 会话级 SSE 事件流，事件历史持久化 | 响应级[流式输出](../concepts/streaming.md) | -| 开发方式 | API 编排（配置 Agent/Environment/Session） | 零代码配置 / 可视化编排 / Python 编码 | -| 主要 API 端点 | `/api/v1/agentstudio/agents`、`/environments`、`/sessions`、`/sessions/{id}/events` | 应用发布后经"发布渠道"提供的调用 API | -| 工具能力 | 内置 7 个工具（bash/read/write/edit/glob/grep/download_file）+ MCP + Skill | 内置沙箱工具、知识库、MCP、插件（按应用类型） | -| 文件处理 | 上传挂载到 `/mnt/session/uploads`，单文件 ≤ 10MB | 文件问答（全文引用/RAG/自定义），单会话 ≤ 10 文件、单文件 ≤ 10MB | -| 记忆/上下文 | 服务端持久化会话事件，可挂载资源复用 | 短期记忆 0-30 轮；长期记忆暂未支持 | -| 发布与集成 | 通过 Agent/Environment/Session API 直接编排 | 需先"发布"，再经发布渠道 API/第三方平台调用 | -| 计费方式 | 按模型 Token 用量 + 沙箱运行资源 | 模型 Token + 知识库召回 + MCP/插件 + 高代码函数计算/网关/存储 | -| 典型场景 | 多步工具调用、代码执行、文件批处理等长时自主任务 | 问答对话、固定流程自动化、企业级后端服务 | - -## 适用场景建议 - -**优先选择 Managed Agents 的场景：** - -- 任务需要**多步自主决策 + 代码执行**，如自动化数据处理、脚本编写与运行、文件批量转换。 -- 需要**长时运行**且要在中途中断、审批、续接的任务，依赖服务端持久化的事件历史。 -- 需要**隔离的沙箱环境**安装依赖（apt/pip）、运行 shell 命令，且不希望自行维护会话上下文。 - -**优先选择 LLM 应用的场景：** - -- **智能体应用（Agent 2.0）**：意图开放、需模型自主规划调用知识库/MCP 的问答与对话类应用，业务人员即可零代码配置。 -- **工作流应用**：流程固定、需精确控制执行链路的多步骤自动化，适合 IT 运维与业务分析师用可视化编排。 -- **高代码应用**：对性能、可观测性、企业级运维有要求的生产级 AI 后端，由工程师用 Python 编码并部署到 Serverless/K8s。 -- **文件问答**：面向文档总结、长文检索、图片/视频分析等，直接在智能体应用中上传文件即可。 - -## 技术选型参考 - -- **控制粒度维度**：从"AI 自主"到"人工精确控制"依次为——Managed Agents / Agent 2.0（模型自主规划）> 工作流（预定义流程）> 高代码（完全代码控制）。需要确定性流程选工作流或高代码；需要自主探索选 Managed Agents 或 Agent 2.0。 -- **状态需求维度**：需要跨轮次持久化会话、支持中断续接的长时任务，只有 Managed Agents 原生支持；LLM 智能体应用多为无状态，需应用侧自行维护上下文。 -- **交付形态维度**：要"对外发布 + 多渠道集成（钉钉/公众号）"选 LLM 应用；要"程序化编排自主任务"直接对接 Managed Agents 的 agentstudio API。 -- **开发门槛维度**：业务人员/产品经理→Agent 2.0；运维/分析师→工作流；AI 工程师→高代码或 Managed Agents API。 - -> 提示：二者并非互斥。可在 LLM 应用中通过 MCP/知识库快速搭建对话入口，同时把重型的多步执行任务下沉到 Managed Agents 沙箱，形成"轻交互层 + 重执行层"的组合架构。 - -## 被对比主题页 - -- [managed agents](../guides/managed-agents.md) -- [llm application](../guides/llm-application.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md deleted file mode 100644 index 12f72f00..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-compare.md +++ /dev/null @@ -1,48 +0,0 @@ -# 图像、视频与3D生成对比 - -阿里云百炼平台提供了覆盖多模态内容生成的三大能力：**图像生成**、**视频生成**与**3D 资产生成**。三者都通过 DashScope 网关对外提供服务，共享相似的异步调用范式与鉴权体系，但在支持模型、输入输出格式、接口路径、地域可用性、计费方式和典型应用场景上存在显著差异。本文从技术选型角度对三者做横向对比，帮助开发者根据业务需求快速定位合适的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 支持模型 | 千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等多家族 | 万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及多种人像驱动模型 | 仅 Tripo（`Tripo/Tripo-H3.1`、`Tripo/Tripo-P1.0`） | -| 核心功能 | 文生图、图生图、图像编辑、局部重绘、扩图、虚拟模特、AI 试衣、创意海报等 | 文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑、数字人/口型/舞蹈等 | 文生 3D、单图生 3D、多图生 3D | -| 输入格式 | `prompt` / `messages` / `images`（JPG/PNG/JPEG/BMP/WEBP，[512,4096] 像素，≤10MB） | `input.prompt` + `input.media`（`first_frame`/`last_frame`/`image_url`/`video` 等） | `prompt`（≤1024 字符）/ `image` / `images`（4 元素数组，三者互斥；JPEG/PNG，[20,6000] 像素，≤20MB） | -| 输出格式 | 图像 URL（有效期 24 小时） | 视频 URL（异步返回） | GLB 模型（`pbr_model_url` / `base_model_url`）+ 预览渲染图（下载链接有效期 2 小时） | -| 主要 API 端点 | 多路径：`.../text2image/image-synthesis`、`.../aigc/multimodal-generation/generation`、`.../image-generation/generation` 等 | `POST .../aigc/video-generation/video-synthesis`（部分模型走 `.../aigc/image2video/video-synthesis`） | `POST .../aigc/video-generation/3d-generation` | -| 调用方式 | 以异步为主（`X-DashScope-Async: enable` + 轮询）；新模型支持 HTTP 同步 | 全部异步（创建任务 → 轮询查询） | 全部异步（创建任务 → 轮询，建议 15 秒间隔） | -| 任务耗时 | 通常 1-2 分钟 | 通常 1-5 分钟 | 较长（异步） | -| 地域可用性 | 华北2（北京）、新加坡、美国（弗吉尼亚）等多地域，独立 Key 不可混用 | 万相/HappyHorse 多地域；PixVerse/Vidu/Kling/数字人/人像模型仅北京 | 仅华北2（北京），需该地域 API Key | -| 计费方式 | 仅对成功生成的输出图片计费，含 90 天免费额度 | 多为后付费按视频时长（元/秒）或按张计费 | 按任务成功结果计数（`text-to-3d`/`image-to-3d`/`multi-image-to-3d`） | -| 并发限制 | 主/子账号共享 QPS 与处理中任务数 | 同时处理中任务通常限 1（排队执行） | 查询接口默认 RPS 20 | -| 典型场景 | 电商海报、虚拟模特、AI 试衣、创意插画、图文混排 | 短视频、广告、数字人播报、动画、口型替换 | 游戏/AR/VR 资产、3D 建模、数字孪生 | - -## 共性特征 - -- **统一网关与异步范式**：三者均通过 DashScope 网关调用，且核心流程都是「创建任务获取 `task_id` → 轮询 `GET .../api/v1/tasks/{task_id}`」。创建任务时必须携带 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。 -- **`task_id` 有效期均为 24 小时**，切勿重复创建任务，轮询获取即可。 -- **地域隔离**：不同地域拥有独立的 API Key 与请求地址，跨地域调用会导致鉴权失败。百炼推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）。 -- **请求体结构**：普遍由 `model` / `input` / `parameters` 三部分组成。 - -## 各方案适用场景建议 - -- **图像生成**：适合需要静态视觉内容的场景，模型选择最丰富、地域覆盖最广、计费门槛最低（仅计成功输出图并有免费额度），是多模态生成中最成熟、门槛最低的入口。电商与创意工具（虚拟模特、AI 试衣、创意海报）尤为突出，但需注意部分创意工具模型仅提供免费体验、用尽后不可付费。 -- **视频生成**：适合需要动态内容的场景，功能维度最复杂（文生/图生/参考生/编辑/数字人）。选型时优先选用走新版协议、功能最全的万相 wan2.7 系列；若使用 PixVerse、Vidu、Kling 或人像/数字人模型需注意仅限北京地域。计费按时长且并发通常限 1，需评估吞吐与排队成本。 -- **3D 生成**：能力最聚焦，仅基于 Tripo 模型，产出带 PBR 材质的 GLB 模型。仅限北京地域、下载链接仅 2 小时有效期，需及时下载与转存。适合游戏、AR/VR、工业设计等对 3D 资产有需求的场景；`Tripo-H3.1` 追求高精度（最高 200 万面），`Tripo-P1.0` 追求速度。 - -## 技术选型参考 - -1. **地域约束优先评估**：3D 生成及大量第三方视频模型仅支持北京地域，若业务部署在新加坡/海外，应优先确认图像生成或万相视频系列的多地域可用性。 -2. **同步 vs 异步**：仅图像生成的部分新模型（如 `wan2.6-image`、`z-image-turbo`）支持 HTTP 同步一次返回，对低延迟场景友好；视频与 3D 必须异步轮询，需在客户端实现任务状态管理。 -3. **输出有效期差异**：图像/视频结果与 `task_id` 有效期 24 小时，而 3D 产物下载链接仅 2 小时，集成 3D 时务必在回调或轮询成功后立即下载转存。 -4. **计费模型差异**：图像按成功输出图计费且有免费额度、门槛最低；视频按时长计费、并发受限、成本更高；3D 按成功任务计数。批量或高并发场景需据此估算成本与吞吐。 -5. **接口路径不统一**：三大能力乃至同一能力内不同模型的接口路径都可能不同（尤其图像与视频），接入前务必以对应模型的官方文档为准。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md deleted file mode 100644 index aa5c3e48..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/media-generation-comparison.md +++ /dev/null @@ -1,72 +0,0 @@ -# 图像生成 vs 视频生成 vs 3D生成对比 - -百炼平台提供图像生成、视频生成和 3D 模型生成三大多媒体内容创作能力。三者在输入输出形态、调用模式、模型生态和适用场景上存在显著差异。本文从开发者技术选型角度，系统对比三类生成 API 的核心维度，帮助快速定位最适合业务需求的能力。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| **输入格式** | 文本 [prompt](../guides/prompt.md)、参考图（URL/Base64）、蒙版、涂鸦草图 | 文本 [prompt](../guides/prompt.md)、首帧/首尾帧图像、参考图、音频、源视频 | 文本 [prompt](../guides/prompt.md)、单图（URL）、多图（4视角，前左后右） | -| **输出格式** | 图片 URL（JPEG/PNG），需及时下载 | 视频 URL（MP4），需及时下载 | GLB 模型文件 URL（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用模式** | 同步（OpenAI 兼容）或异步（DashScope 原生） | 仅异步（创建任务 + 轮询 task_id） | 仅异步（创建任务 + 轮询 task_id） | -| **典型耗时** | 秒级（同步）~ 十几秒（异步） | 1-10 分钟 | 数分钟（建议 15 秒间隔轮询） | -| **API 端点** | `/compatible-mode/v1/images/generations` 或 `/api/v1/services/aigc/text2image/image-synthesis` | `/api/v1/services/aigc/video-generation/video-synthesis` 或 `/image2video/video-synthesis` | `/api/v1/services/aigc/video-generation/3d-generation` | -| **支持模型** | 通义千问图像、万相 V1/V2/2.6/2.7、Z-Image、可灵、创意工具（10+） | 万相 HappyHorse/wan2.7/2.6/2.5/2.2/2.1、爱诗 PixVerse、Vidu、可灵 | Tripo-H3.1（高精度）、Tripo-P1.0（专业快速） | -| **模型数量** | 20+ 模型/接口 | 30+ 模型变体 | 2 个模型 | -| **地域限制** | 无特殊限制 | 部分模型仅华北2（北京） | 仅华北2（北京） | -| **[计费](../concepts/billing.md)方式** | 按图片张数/分辨率[计费](../concepts/billing.md) | 按视频时长/分辨率[计费](../concepts/billing.md) | 按任务次数计费 | -| **协议支持** | OpenAI 兼容 + DashScope 原生 | DashScope 原生（异步必填 X-DashScope-Async 头） | DashScope 原生（异步必填 X-DashScope-Async 头） | -| **task_id 有效期** | 异步任务 24 小时 | 24 小时 | 24 小时 | -| **内容安全** | 内置审核，违规拒绝 | 内置审核，违规拒绝 | 遵循平台统一内容安全策略 | - -## 功能丰富度对比 - -| 能力类别 | 图像生成 | 视频生成 | 3D生成 | -| --- | --- | --- | --- | -| 文本生成 | 文生图（多模型可选） | 文生视频 | 文生 3D | -| 图像驱动 | 图像编辑、局部重绘、涂鸦作画 | 图生视频（首帧/首尾帧）、参考生视频 | 单图生 3D、多图生 3D | -| 风格控制 | 风格参数、负面词、艺术风格模型 | 视频风格重绘 | 贴图质量、几何精度 | -| 垂类工具 | 虚拟模特、AI试衣、创意海报、背景生成等 10+ 工具 | 视频换人、数字人、肖像动态（唱演/播报/口型替换） | 无 | -| 后处理 | 擦除补全、画面扩展、人物分割 | 视频编辑（指令+参考图） | 无 | - -## 适用场景建议 - -### 图像生成 - -- 电商商品图批量制作（虚拟模特、背景替换、创意海报） -- 营销素材快速迭代（文生图 + 风格控制） -- 内容创作中的插图与配图需求 -- 人像娱乐玩法（风格重绘、写真） -- 需要低延迟同步返回结果的场景 - -### 视频生成 - -- 短视频/广告创意自动化生产 -- 数字人播报与虚拟主播 -- 电商商品动态展示视频 -- IP 角色动画与多镜头叙事 -- 视频内容二次编辑与换人 - -### 3D生成 - -- 游戏/XR 场景中的 3D 资产快速原型 -- 电商商品 3D 展示与 AR 试用 -- 建筑/工业设计概念验证 -- 需要 PBR 材质的高精度渲染场景 - -## 技术选型建议 - -1. **追求响应速度**：图像生成支持同步调用，秒级返回；视频和 3D 均为异步，分钟级等待不可避免。 -2. **模型生态丰富度**：图像生成 > 视频生成 > 3D生成。图像生成模型和垂类工具最多，3D 目前仅有 Tripo 系列。 -3. **地域部署灵活性**：图像生成地域限制最少；视频和 3D 部分模型仅限华北2（北京）。 -4. **业务垂类覆盖**：电商/营销场景，图像生成的垂类工具链最完善；数字人/播报场景选视频生成；3D 资产生产选 3D 生成。 -5. **[多模态](../concepts/multimodal.md)组合**：可先用图像生成产出关键帧，再送入视频生成做动态化；或先用图像/多视角图生成 3D 模型，形成完整的内容生产流水线。 -6. **成本控制**：图像生成单次成本最低，3D 生成单次成本最高但产出资产复用价值大。建议根据产出资产的复用频次评估 ROI。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md deleted file mode 100644 index ce1262d9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-and-knowledge-solutions.md +++ /dev/null @@ -1,69 +0,0 @@ -# 长期记忆、记忆库与[知识库](../concepts/knowledge-base.md)对比 - -百炼平台为开发者提供了三类用于"补充模型上下文"的能力：**长期记忆（新）API**、**记忆库（Memory Library）**、**[知识库](../concepts/knowledge-base.md)（Knowledge Base）**。三者定位不同：长期记忆 API 是底层的 RESTful 接口集合；记忆库是建立在长期记忆 API 之上、面向"跨会话用户记忆"的产品化封装（含控制台管理、OpenClaw 插件零侵入接入）；[知识库](../concepts/knowledge-base.md)则基于 RAG 技术，面向"私有文档/数据的语义检索"。本页通过关键维度对比，帮助开发者根据业务诉求（用户画像 vs 文档问答 vs 零侵入记忆）做出技术选型。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）API | 记忆库（Memory Library） | 知识库（Knowledge Base） | -| --- | --- | --- | --- | -| 定位 | 底层 RESTful API，存储/检索/更新/删除用户记忆片段与画像 | 长期记忆 API 的产品化封装，含控制台与 OpenClaw 插件 | 基于 RAG 的私有数据语义检索，为模型补充领域知识 | -| 输入格式 | `messages`（对话）或 `custom_content`（自定义文本，≤512 字符） | 同长期记忆 API；OpenClaw 插件自动捕获对话 | 文件（pdf/docx/xlsx/图片/音视频等）、数据表、OSS 导入 | -| 输出格式 | 记忆片段（`memory_nodes`）、用户画像（结构化属性） | 记忆片段 + 用户画像，可注入 Prompt | 召回的文本切片（≤20 个/次），拼装后注入 Prompt | -| API 端点 | `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` | 复用长期记忆 API；OpenClaw 插件通过 Gateway 钩子调用 | 阿里云百炼 SDK / 检索 API（需 AliyunBailianDataFullAccess 权限） | -| 支持模型 | 不直接绑定模型；记忆提取与画像由服务端完成 | 同长期记忆 API；OpenClaw 插件不支持 Coding Plan [API Key](../concepts/api-key.md) | 预置千问系列、DeepSeek-R1/V3.1、abab6.5s、Llama3.1、Yi-Large 及自定义模型 | -| 数据存储 | 记忆片段与画像，按 `user_id` 隔离，按 `memory_library_id` 分库 | 同长期记忆 API；默认库预置"默认有效期 180 天"规则 | 向量索引（text-embedding-v3/v4 512 维；multimodal-embedding-v1 1024 维） | -| 计费方式 | 按 API 调用计费（具体见平台说明） | 同长期记忆 API | 规格费用（按小时）+ 模型调用费用（[Token](../concepts/token.md)）；标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| 并发/限流 | 全部接口 ≤3000 QPM；add 120 QPM；search 300 QPM | 同长期记忆 API | 标准版 1 QPS（固定）；旗舰版 50–10,000 QPS（1–200 RCU） | -| 有效期 | API 直写：暂无失效日期 | 控制台规则：7/30/180 天或永不过期（默认 180 天） | 持久存储，无失效概念 | -| 接入方式 | 直接调用 RESTful API；Python 可用 `agentscope-runtime` | API 直连 / 百炼控制台 / OpenClaw 插件（零侵入） | 控制台创建 + SDK 集成；可挂载到[智能体应用](../concepts/agent-application.md)、工作流应用 | -| 典型场景 | 跨会话个性化、用户偏好持久化、自动提取关键事件 | 用户长期记忆、画像维护、Agent 零侵入记忆接入 | 私有文档问答、领域知识检索、图文/音视频内容搜索 | - -## 各方案适用场景建议 - -### 长期记忆（新）API - -适合需要**精细控制记忆生命周期**的开发者：自行管理写入（AddMemory）、语义检索（SearchMemory）、更新与删除，并通过画像模板（Profile Schema）维护结构化用户属性。当业务需要将记忆能力嵌入自有应用、对 `user_id` 与 `memory_library_id` 做多租户隔离、或希望用 `custom_content` 直接写入指定记忆（绕过对话提炼）时，优先选择此 API。 - -### 记忆库（Memory Library） - -适合希望**以最低接入成本获得跨会话记忆**的场景： - -- **OpenClaw 插件方式**：通过 `before_agent_start`（自动召回）与 `agent_end`（自动捕获）两个 Gateway 钩子实现零侵入记忆，所有提炼、向量化、语义检索由百炼服务端完成。适合基于 OpenClaw 构建的 Agent，无需改动业务代码。 -- **控制台方式**：在百炼控制台可视化管理记忆库与规则，支持配置记忆片段有效期（7/30/180 天或永不过期），适合非技术运营人员参与记忆策略管理。 -- **API 直连方式**：与长期记忆 API 一致，适合自定义接入。 - -注意：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置。 - -### 知识库（Knowledge Base） - -适合需要**让模型基于私有数据回答问题**的 RAG 场景： - -- 文档问答（pdf/docx/markdown/图片等，单文件最大 150MB） -- 数据查询（xlsx 数据表，最大 10 万行） -- 图片问答（multimodal-embedding-v1） -- 音视频搜索（最大 512MB） - -知识库仅在**中国站华北2（北京）**地域可用，提供标准版（1 QPS、≤100 GB）与旗舰版（50–10,000 QPS、≤9,999 GB）两档规格。创建后知识库类型、metadata 抽取与切片策略不可更改，需一次性规划。适合企业知识库、产品手册问答、领域知识检索等"知识供给"型应用。 - -## 技术选型参考 - -| 选型问题 | 推荐方案 | -| --- | --- | -| 需要记住用户偏好、历史事件，实现跨会话个性化 | 长期记忆 API 或记忆库 | -| 基于 OpenClaw 构建 Agent，希望零侵入接入记忆 | 记忆库（OpenClaw 插件） | -| 需要运营人员在控制台管理记忆规则与有效期 | 记忆库（控制台） | -| 需要精细控制记忆的增删改查与画像模板 | 长期记忆（新）API | -| 需要基于私有文档/数据做语义检索问答 | 知识库 | -| 需要处理图片/音视频等多模态内容检索 | 知识库（图片问答/音视频搜索） | -| 高并发检索（>1 QPS） | 知识库旗舰版（最多 10,000 QPS） | -| 同时需要"用户记忆"和"文档知识" | 记忆库 + 知识库组合使用，二者互补 | - -**一句话总结**：长期记忆 API 与记忆库解决"记住用户是谁、做过什么"的问题（个性化上下文），知识库解决"模型不知道的领域知识"的问题（RAG 检索）。二者并不互斥，可在同一应用中组合使用——用记忆库维持用户画像与历史，用知识库供给领域文档，共同提升大模型在特定业务中的表现。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [knowledge base](../guides/knowledge-base.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md deleted file mode 100644 index a5963c3b..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-approaches-comparison.md +++ /dev/null @@ -1,59 +0,0 @@ -# 记忆库与长期记忆对比 - -百炼平台为解决大模型跨会话上下文丢失的问题，提供了围绕"长期记忆"的完整能力。在文档体系中，这一能力以两种视角呈现： - -- **记忆库（Memory Library）**：偏产品与方案视角，强调接入方式（控制台 / HTTP API / OpenClaw 插件）、记忆片段与用户画像两类内容形态，以及记忆规则与有效期等可运营配置。 -- **长期记忆（新）API**：偏接口与实现视角，给出 RESTful 端点、请求/响应字段、限流策略和画像模板（Profile Schema）的完整 CRUD。 - -二者底层共用同一套 `https://dashscope.aliyuncs.com/api/v2/apps/memory/` 服务，记忆片段与用户画像的数据模型一致；区别在于"封装层"和"可控粒度"。本文从开发者技术选型角度对两者进行对比。 - -## 关键维度对比 - -| 维度 | 记忆库（Memory Library） | 长期记忆（新）API | -| --- | --- | --- | -| 定位 | 上层方案与产品形态：跨会话记忆的整体接入 | 底层 RESTful 接口参考：程序化记忆管理 | -| 文档位置 | 应用使用指南（guides） | 应用 API 参考（api） | -| 接入方式 | 控制台可视化管理 + HTTP API + OpenClaw 插件零侵入 | 直接 HTTPS 调用 `Authorization: Bearer $DASHSCOPE_API_KEY` | -| 记忆内容 | 记忆片段 + 用户画像（两类可独立或组合使用） | 记忆片段 + 用户画像（同一数据模型） | -| 写入输入 | `messages`（自动提取）或 `custom_content`（直写，最大 512 字符），二选一 | 同左，`messages` 最多 50 条对话 | -| 检索能力 | `SearchMemory`，主要参数 `top_k`（建议 3–10） | `SearchMemory`，含 `top_k`（1–100，默认 10）、`min_score`（默认 0.3）、`enable_rerank`/`enable_judge`/`enable_rewrite`、`project_ids` 多规则混合检索 | -| 管理 API | 重点呈现 `AddMemory` / `SearchMemory`（写入与召回） | `AddMemory` / `SearchMemory` / `ListMemory` / `UpdateMemory` / `DeleteMemory` + 画像模板 CRUD + `GetUserProfile` | -| 画像管理 | 通过 `profile_schema` 参数提取画像，模板在记忆库详情页获取 | 提供 `CreateProfileSchema` / `ListProfileSchemas` / `UpdateProfileSchema` / `DeleteProfileSchema` / `GetProfileSchema` 全套接口 | -| 记忆有效期 | 控制台默认规则 180 天，可配 7/30/180 天或永不过期；按规则可编辑 | API 文档标注"生成的记忆片段与用户画像暂无失效日期"（以控制台记忆规则配置为准） | -| 记忆规则 | 每账号自带默认记忆库 + 默认规则（不可删除），可创建新记忆库与规则 | 通过 `memory_library_id`、`project_id` 参数指定记忆库与规则；不传使用默认 | -| 用户隔离 | `user_id` 命名空间隔离，OpenClaw 插件所有 Agent 共享同一记忆 | `user_id` 最大 64 字符，用于标识记忆归属 | -| 限流 | 未单独列出，复用底层 API 限额 | 全部接口合计 ≤ 3000 QPM；`add` 120 QPM；`search` 300 QPM | -| 客户端 SDK | Python `agentscope-runtime`（`AddMemory` / `SearchMemory` 等封装，需 `close()`） | 以 cURL / REST 为主，参数与字段为权威定义 | -| 插件支持 | OpenClaw `modelstudio-memory-for-openclaw`：`before_agent_start` 自动召回 + `agent_end` 自动捕获 | 不直接涉及，插件内部回调此 API | -| 典型场景 | 跨会话个性化、Agent 偏好记忆、控制台运营记忆规则、OpenClaw 零侵入接入 | 程序化记忆 CRUD、画像模板生命周期管理、批量召回与重排序、自建记忆编排 | - -## 适用场景建议 - -### 选择"记忆库（Memory Library）"视角当 - -- 需要在控制台可视化地创建记忆库、配置记忆片段规则与有效期（7/30/180 天或永不过期）。 -- 希望以"产品方案"形式接入，例如通过 OpenClaw 插件实现 `autoCapture` / `autoRecall` 的零侵入跨会话记忆，而不愿手写每轮的写入与检索调用。 -- 业务侧关注的是"用户偏好持续化""Agent 跨会话理解"等整体能力，而非单个接口字段。 - -### 选择"长期记忆（新）API"视角当 - -- 需要在自研应用中精细控制每一步记忆操作：写入、搜索、列表、更新、删除，以及对画像模板做完整的增删改查。 -- 需要使用高级检索参数（`min_score` 阈值、`enable_rerank` 重排序、`enable_judge` 意图判别、`enable_rewrite` query 重写、`project_ids` 多规则混合检索）来调优召回质量。 -- 需要依据明确的限流（3000 QPM 总量、add 120 QPM、search 300 QPM）做容量规划与重试策略。 -- 需要程序化维护用户画像模板的字段定义，或对接已有用户体系做批量画像写入与读取。 - -## 技术选型小结 - -记忆库与长期记忆（新）API 并非二选一的两套系统，而是**同一能力的产品层与接口层**： - -- 做方案设计与运营配置时，以"记忆库"文档为准（接入方式、记忆规则、有效期、OpenClaw 插件配置）。 -- 做接口对接与字段实现时，以"长期记忆（新）API"文档为准（端点、参数、返回结构、限流、画像模板 CRUD）。 - -实践建议：先用记忆库视角确定接入形态（控制台 / API / 插件）与记忆规则，再在长期记忆（新）API 中查证具体端点与字段；两者配合即可覆盖从产品方案到代码实现的完整链路。注意记忆有效期以控制台记忆规则配置为准——API 文档的"暂无失效日期"指 API 直写且不指定 `project_id` 时使用默认规则的情形。 - -## 被对比主题页 - -- [memory library overview](../guides/memory-library-overview.md) -- [long term memory new](../api/long-term-memory-new.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md deleted file mode 100644 index e6d71fca..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-feature-comparison.md +++ /dev/null @@ -1,43 +0,0 @@ -# 记忆能力对比（长期记忆 vs 记忆库） - -百炼平台提供两种面向"跨会话上下文持久化"的记忆能力描述入口：一是 **长期记忆（新）**（API 参考视角，`api/long-term-memory-new.md`），二是 **记忆库（Memory Library）**（用户指南视角，`guides/memory-library-overview.md`）。两者本质指向同一套底层长期记忆 API（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`），但在文档定位、接入方式、管理入口和适用对象上存在差异。本页从技术选型角度对二者进行对比，帮助开发者快速判断应参考哪一份文档、采用哪种接入路径。 - -## 关键维度对比 - -| 维度 | 长期记忆（新）（API 参考） | 记忆库（Memory Library）（用户指南） | -| --- | --- | --- | -| 文档定位 | RESTful API 接口参考，逐接口说明请求/响应字段 | 能力总览与接入指南，含概念、控制台管理与插件接入 | -| 受众 | 直接调用 HTTP API 的后端/服务端开发者 | 需要端到端方案选型、含控制台与零侵入接入的开发者 | -| Base URL | `https://dashscope.aliyuncs.com/api/v2/apps/memory/` | 同上（`https://dashscope.aliyuncs.com/api/v2/apps/memory/*`） | -| 认证方式 | Header `Authorization: Bearer $DASHSCOPE_API_KEY` | 环境变量 `DASHSCOPE_API_KEY`，同样以 `Bearer` 方式携带 | -| 接口覆盖 | AddMemory / SearchMemory / ListMemory / DeleteMemory / UpdateMemory / CreateProfileSchema / ListProfileSchemas / DeleteProfileSchema / UpdateProfileSchema / GetProfileSchema / GetUserProfile 共 11 个 | 重点讲 AddMemory / SearchMemory，并补充 ListMemory、CreateProfileSchema、GetUserProfile 等封装类 | -| SDK 支持 | 以 cURL 示例为主 | 额外提供 Python `agentscope-runtime` 封装类，需在 `finally` 调 `close()` | -| 零侵入接入 | 未涉及 | 提供 OpenClaw 记忆插件，`before_agent_start`/`agent_end` 钩子自动召回/捕获 | -| 控制台管理 | 未涉及 | 支持控制台可视化管理记忆库、记忆规则、默认有效期 | -| 记忆有效期 | 明确"生成的记忆片段与用户画像暂无失效日期" | 控制台默认规则 180 天，可配置 7/30/180 天或永不过期；以控制台规则为准 | -| 限流说明 | 给出账号级 QPM：全局 3000、add 120、search 300 | 未在总览中给出 QPM 数字 | -| 检索增强参数 | 列出 `top_k`/`min_score`/`enable_rerank`/`enable_judge`/`enable_rewrite` | 仅点出 `top_k`、`minScore`（插件）等关键参数 | -| 画像能力 | 完整的 Profile Schema CRUD 与 GetUserProfile 接口 | 介绍画像模板概念与 `profileSchema` 配置项，接口细节指向 API 参考 | -| [计费](../concepts/billing.md)方式 | 未在本页说明 | 未在本页说明（统一走 DashScope [计费](../concepts/billing.md)） | -| 典型场景 | 需要精细控制记忆 CRUD、画像模板、检索召回参数的服务端集成 | 需要快速接入、可视化运维或让 OpenClaw Agent 自动具备记忆 | - -## 适用场景建议 - -- **参考"长期记忆（新）"文档的情况**：你需要直接对接 HTTP API，关心每个接口的请求体、响应字段、`event` 事件类型（ADD/UPDATE/DELETE）、画像模板的完整 CRUD，或需要按 `enable_rerank`/`enable_judge`/`enable_rewrite` 等参数精细调优检索召回；适合自研 Agent 后端、需要严格接口契约的服务端开发者。 -- **参考"记忆库"文档的情况**：你希望先从业务视角理解"记忆片段 vs 用户画像"两类持久化内容的差异与组合用法，或希望通过控制台创建/编辑记忆库与记忆规则、配置有效期，又或者你的 Agent 运行在 OpenClaw 之上，希望以插件方式零侵入获得"自动捕获/自动召回"能力；适合做整体方案选型与低代码运维的团队。 -- **二者结合使用**：多数生产落地建议先读"记忆库"理解概念与管理入口，再用"长期记忆（新）"对照接口字段落地代码；OpenClaw 用户可仅依赖插件配置项即可跑通，深定制时再回查 API 参考。 - -## 技术选型参考 - -1. 接入路径：纯后端 HTTP 调用 → 选 API 直连（两份文档均适用，接口细节以"长期记忆（新）"为准）；OpenClaw Agent → 选记忆插件（仅"记忆库"文档覆盖）。 -2. 有效期策略：若需记忆按天失效，务必以控制台记忆规则配置为准（默认 180 天）；API 直写且不指定 `project_id` 时走默认规则，"暂无失效日期"的说法仅适用于不经过规则提炼的直写场景。 -3. 命名空间隔离：两份文档均强调 `user_id` 为记忆空间隔离维度，不同 `user_id` 完全隔离；OpenClaw 插件目前所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置，也不支持百炼 Coding Plan 的 API Key。 -4. 检索质量：需要重排序、意图判别、query 重写等高级召回能力时，参考"长期记忆（新）"的 SearchMemory 参数；插件场景受 `topK`/`minScore` 配置项约束。 -5. 画像存储：需要固定结构化属性（年龄、职业、偏好等）持久化时使用用户画像，字段命名应清晰具体、避免同义并存；接口细节走"长期记忆（新）"的 Profile Schema 系列。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md deleted file mode 100644 index 61ec6040..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-call-vs-application-call.md +++ /dev/null @@ -1,43 +0,0 @@ -# 模型直调与应用调用对比 - -百炼平台对外提供两类 API 调用路径：**模型直调**（直接调用 Qwen 等文本生成模型）与**应用调用**（调用在控制台预先编排好的智能体、工作流、Agent 2.0 应用）。两者底层都走 DashScope 网关并复用同一套 [API Key](../concepts/api-key.md)，但在调用对象、输入格式、能力范围与典型场景上有显著差异。本页面向做技术选型的开发者，按关键维度对比两种方案，帮助快速判断应走哪条路径。 - -## 关键维度对比 - -| 维度 | 模型直调（Qwen API） | 应用调用（Application API） | -| --- | --- | --- | -| 调用对象 | 单个文本生成模型（`model` 字段指定 Qwen 系列模型名） | 控制台已编排好的应用（智能体 / 工作流 / Agent 2.0），由 `APP_ID` 标识 | -| 核心入口 | OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages、DashScope 原生 | OpenAI 兼容 Responses API（`/apps/agent/{APP_ID}/compatible-mode/v1/responses`）、DashScope 原生 API（`/apps/{APP_ID}/completion`） | -| 前置准备 | [API Key](../concepts/api-key.md)（`DASHSCOPE_API_KEY`） | [API Key](../concepts/api-key.md) + 应用 ID（`APP_ID`，控制台手动获取；子[业务空间](../concepts/workspace.md)还需 `Workspace ID`） | -| 输入格式 | `messages` 数组（OpenAI/Anthropic 兼容）或 Responses 的 `input`；需自行维护对话历史（Responses 接口除外） | `input.prompt` 字符串 / `input` 消息数组（OpenAI 兼容）；`input.prompt` + `parameters` + `biz_params`（DashScope 原生） | -| 对话历史 | 仅 OpenAI 兼容 Responses 由平台自动管理；其余接口调用方自行拼接 `messages` | 支持 `session_id`（云端托管，1 小时有效、最多 50 轮）或自行管理 `messages`；同时传两者时以 `messages` 为准 | -| 内置工具能力 | 联网搜索、代码解释器、网页内容提取仅 Responses 接口内置；其他接口需自行定义工具 | 应用编排阶段在控制台绑定插件 / 节点，调用时通过 `biz_params.user_defined_params` 透传业务参数，工具由应用内部装配 | -| 支持模型 | Qwen 全系列（含 VL 多模态），按 `model` 字段切换 | 由应用编排时所选模型决定（如智能体做图像输入需选 Qwen-VL 并设「自定义处理」） | -| [流式输出](../concepts/streaming-output.md) | 各接口均支持 `stream` 参数 | 同步调用支持 `stream=True`；[异步调用](../concepts/async-invocation.md)暂不支持流式；工作流需在输出节点启用「[流式输出](../concepts/streaming-output.md)」并重新发布 | -| 异步执行 | 由调用方自行实现 | OpenAI 兼容 Responses 支持 `background=True`，返回任务 ID 后轮询 `retrieve` 至终态 | -| 多模态 | 各接口按协议支持文本 / 图像 / 文件（`input_image` / `input_file`） | `input` 消息数组支持 `input_text` / `input_image` / `input_file`（`input_file` 仅[智能体应用](../concepts/agent-application.md)支持） | -| 计费方式 | 按 token 用量计费（输入 + 输出） | 按 token 用量计费；应用编排内部多步调用累计计费 | -| 功能完整度 | DashScope 原生接口参数最全；兼容接口为保证协议一致可能不暴露全部原生参数 | DashScope 原生 `/completion` 功能更全、性能更高；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)便于复用 OpenAI 生态 | -| 典型场景 | 单模型推理、文本生成、对话补全、从 OpenAI/Anthropic 平迁、需要精细采样参数 | 复用已编排的多步骤智能体 / 工作流、内置 RAG 与插件、长耗时异步任务、业务系统集成 | - -## 选型建议 - -- **选模型直调**：当你只需要一个模型做单轮或多轮文本生成、补全、对话，且希望直接复用 OpenAI / Anthropic SDK 与既有代码库，或需要最全的采样参数与原生能力时。从外部平台迁入时优先评估兼容接口，需要极致功能时再切到 DashScope 原生接口。 -- **选应用调用**：当业务逻辑已被编排成智能体或工作流（含 RAG、插件、多节点流程），希望把整套能力一次性集成进业务系统，而不是在调用方重新实现编排逻辑时。长耗时任务（报告生成、多步工具调用）选异步 `background=True`；需要复用 OpenAI 工具链选 Responses API，需要更全功能与更高性能选 DashScope 原生 `/completion`。 -- **混合使用**：两者共用同一 `DASHSCOPE_API_KEY`，可在同一业务系统中并存——轻量推理走模型直调，复杂编排走应用调用，按场景而非按模型选路径。 - -## 注意事项 - -- **凭证管理**：API Key 推荐写入 `DASHSCOPE_API_KEY` 环境变量，不要在生产环境硬编码。 -- **应用 ID 获取**：`APP_ID` 与 `Workspace ID` 目前只能在控制台手动复制，不支持 API / CLI 查询；RAM 子账号默认只能查看其已加入的[业务空间](../concepts/workspace.md)。 -- **地域差异**：上述 Endpoint 默认适用于华北2（北京）；德国（法兰克福）、新加坡、日本（东京）等地域或调用子[业务空间](../concepts/workspace.md)下应用时，请求须包含 `Workspace ID`，且该 ID 是对应地域 Base URL 的组成部分。 -- **兼容接口的功能取舍**：OpenAI / Anthropic 兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全参数、插件或业务字段，应改用 DashScope 原生接口。 -- **迁移评估**：从 OpenAI / Anthropic 迁移到模型直调时，先确认目标 Qwen 模型在对应兼容接口下是否支持所需参数（`temperature`、`tools`、`stream` 等）；迁移到应用调用时，注意异步不支持流式、多轮历史需自行管理（除非用 `session_id`）。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md deleted file mode 100644 index af15d86e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-methods-comparison.md +++ /dev/null @@ -1,64 +0,0 @@ -# 模型微调、压缩与部署对比 - -百炼平台围绕“把通用大模型变成业务专属模型”提供了一条递进式链路：**模型微调（Fine Tuning）→ 模型压缩（Model Compression）→ [模型部署](../concepts/model-deployment.md)（Model Deployment）**。三者并非并列的替代方案，而是同一交付链上前后衔接的环节——微调负责把业务/场景知识写入模型参数，压缩负责把全精度微调模型量化为低精度版本以降低部署门槛，部署负责把模型上线为可调用的推理服务。本文从输入格式、输出产物、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者明确各环节的边界与衔接关系，避免在选型时混淆“训练 / 压缩 / 部署”三件事。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | [模型部署](../concepts/model-deployment.md)（Model Deployment） | -| --- | --- | --- | --- | -| 核心目标 | 将业务/场景知识、风格、偏好写入模型参数 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 将预置或调优后的模型上线为独立、资源专享的推理服务 | -| 在链路中的位置 | 链路起点，产出可被压缩/部署的微调模型 | 微调之后、部署之前的可选优化环节 | 链路终点，对外提供推理 API | -| 输入格式 | CPT：1000 万+ Token 无标签领域文本；SFT：1000+ 条问-答对；DPO：100+ 组偏好对；万相图像/视频：`.zip` 内含 `data.jsonl` 与素材；CosyVoice：多条/数小时录音 | 当前工作空间内已有的自定义微调模型；量化模板要求时还需校准数据（内部上传且已发布的数据集） | 模型 ID（预置模型或调优/压缩产出模型）；API/命令行调用还需 API Key 与业务空间 | -| 输出产物 | 自定义微调模型（全参或 LoRA），可用于压缩或直接部署 | 量化后的低精度模型，落盘到「我的模型」，可直接部署 | 在线推理服务（`deployed_model` 唯一 ID，`status=RUNNING`） | -| 支持模型 | 文本：Qwen3.x / Qwen2.5 系列等；视觉：Qwen3-VL / Qwen2.5-VL；图像：wan2.7；视频：wan2.2/wan2.5；语音：cosyvoice-v3-flash | 仅「支持压缩的模型」列表中的自定义微调模型（如 qwen3.5-flash-2026-02-23 对应的微调版）；已量化模型不可二次压缩 | 预置模型、调优产出模型、压缩产出模型、OSS 导入的 LoRA 模型（仅 LoRA，不支持全参微调模型导入） | -| 操作入口 | 控制台「模型调优」页面；或 DashScope API `POST /api/v1/fine-tunes` | 控制台「模型训练 > 模型压缩」页面；或 OpenAPI | 控制台「[模型部署](../concepts/model-deployment.md)」页面（北京）；或 `https://dashscope.aliyuncs.com/api/v1/deployments` | -| 计费方式 | 按 Token 计费（API 创建的任务仅支持按 Token，不支持训练单元）；控制台任务可用训练单元（预付费/后付费）；计费与 `n_epochs`/`batch_size`/`max_length` 等超参相关 | 压缩功能限时免费；产出模型部署后按所选部署单元规格计费 | 预置吞吐（PTU）、模型单元（MU）、按 Token 用量三种；PTU 支持长输入阶梯系数与前缀缓存折扣；服务创建后无法切换计费方式 | -| 可逆性 | 可重新训练、迭代 | 不可逆：产出模型不支持继续训练，也不支持二次压缩；如需迭代须回到上游全精度模型重新训练 | 可下线重新部署以切换计费方式；按 Token 用量模式一个月不使用自动释放 | -| 任务状态流转 | PENDING → SUCCEEDED（轮询 `GET /api/v1/fine-tunes/{job_id}`） | 待开始 / 排队中 / 运行中 / 停止中 / 压缩成功 / 压缩失败 / 已取消；仅排队中/运行中可取消 | PENDING → RUNNING（轮询 `GET /api/v1/deployments/{deployed_model}`） | -| 关键约束 | `-Base` 后缀模型不可直接用于调优对话；CosyVoice 调优仅支持 API 发起、无法扩展基础模型不支持的语种 | 校准数据控制台 UI 暂不支持 OSS 挂载类型；模板名称决定压缩后可部署规格（如 MU5、MU8） | 仅华北二（北京）地域；导入 LoRA 的 rank 须为 8/16/32/64 且各层一致；VL 模型须冻结 VIT | - -## 各方案适用场景建议 - -### 模型微调 - -适合“Prompt 工程已到上限、需要把知识或风格沉淀进模型参数”的场景： - -- **CPT（继续预训练）**：业务有大量无标注领域语料（专业词汇、行业事实），需要先做领域适应。 -- **SFT（监督微调）**：需要模型遵循特定对话格式或任务执行规范，有 1000+ 条高质量问-答对。 -- **DPO（直接偏好优化）**：SFT 之后想进一步对齐人类偏好、抑制坏答案。 -- **LoRA 高效训练**：数据集较小、需快速验证、追求低成本快迭代；全参训练则用于追求全局效果最优。 -- 多模态场景：文生图/图生图训练人物 IP 或风格 LoRA（万相 wan2.7）、图生视频训练首帧/首尾帧 LoRA（wan2.2/wan2.5）、语音合成训练专属音色（CosyVoice）。 - -### 模型压缩 - -适合“已有全精度微调模型，但希望部署到更小规格单元、降低成本并提升吞吐”的场景： - -- 微调产物部署成本偏高，想通过量化降到 MU5/MU8 等更小规格。 -- 对推理显存和吞吐有明确优化诉求，且不打算继续迭代模型参数（压缩不可逆）。 -- 不适合：尚未完成微调的模型、已量化过的模型、需要继续训练迭代的模型。 - -### 模型部署 - -适合“模型已就绪（无论预置、微调还是压缩产出），需要对外提供高并发、低延迟推理服务”的场景： - -- **PTU（预置吞吐）**：高负载生产环境、稳定吞吐、流量可预估，追求 TPS 提升（约 1.5~2.0 倍于按 Token）。 -- **模型单元（MU）**：资源独占、需自定义性能指标、长时任务；支持 PD 分离以降低首 Token 延迟。 -- **按 Token 用量**：调优后效果验证、性价比优先、对并发延迟要求不高；仅支持部分 LoRA 调优模型，一个月不使用自动释放。 -- **自定义 LoRA 导入**：本地训练的 LoRA 模型（千问3/千问2.5 系列及 VL 版本）需从 OSS 导入后再部署；注意推理参数建议参照 vLLM 默认值调整。 - -## 技术选型参考 - -1. **先判断是否需要动参数**：如果 Prompt 工程能解决，就不必微调；只有当知识/风格/偏好需要沉淀进模型时才进入微调环节。 -2. **微调模式选择**：快速验证用 LoRA，全局最优用全参；多模态按模型类型选 SFT-LoRA；语音合成走 CosyVoice SFT。 -3. **是否插入压缩环节**：微调产物部署成本敏感时，在部署前加一步压缩；但需注意压缩不可逆、不支持二次压缩，迭代须回到全精度模型重训。 -4. **部署计费方式选择**：生产高负载选 PTU（兼顾长输入与前缀缓存优惠）；资源独占与 PD 分离选模型单元；仅做效果验证选按 Token 用量。计费方式创建后不可切换，需下线重部署。 -5. **链路衔接**：微调 →（可选）压缩 → 部署 是单向链路；压缩产出的模型不可继续训练，所以“想压缩又想继续迭代”的需求要把全精度模型作为迭代基线，每次迭代后重新压缩。 -6. **入口选择**：零代码/合规场景走控制台；需要自动化编排（如 CI/CD）走 API/命令行；CosyVoice 调优目前仅支持 API，控制台不可用。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md deleted file mode 100644 index cca2e06a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-customization-options.md +++ /dev/null @@ -1,43 +0,0 @@ -# 微调、压缩与部署方案对比 - -百炼平台围绕「让模型贴合业务」提供了三条相互衔接的路径：**模型调优（Fine Tuning）**把业务/场景知识写进模型参数；**模型压缩（Model Compression）**用量化算法降低微调模型的精度与显存占用；**模型部署（Model Deployment）**把平台预置模型或调优后的模型部署为在线推理服务。三者构成 `调优 → （可选）压缩 → 部署 → 调用` 的链路，本文从输入产出、支持模型、操作入口、计费方式、典型场景等维度做横向对比，帮助开发者根据「是否需要写知识」「是否要降本」「并发与延迟要求」做技术选型。 - -## 关键维度对比 - -| 维度 | 模型调优（Fine Tuning） | 模型压缩（Model Compression） | 模型部署（Model Deployment） | -| --- | --- | --- | --- | -| 一句话定位 | 将业务/场景知识写入模型参数，压低延迟、抑制幻觉、对齐偏好 | 将全精度微调模型量化为低精度版本，降低显存与部署成本 | 把预置或调优后的模型部署为资源专享的在线推理服务 | -| 输入 | 训练数据集：CPT 需 1000 万+ [Token](../concepts/token.md) 无标签文本；SFT 需 1000+ 条问答对；DPO 需 100+ 组偏好对；万相/CosyVoice 需 `.zip`（含 `data.jsonl` + 媒体素材） | 当前工作空间内符合条件的**自定义微调模型**（全精度）+ 可选校准数据集 | 平台预置模型、调优产出模型、OSS 导入的 LoRA 模型；部署参数（`plan`、`capacity`、`ptu_capacity` 等） | -| 输出 | 微调后的自定义模型（全参或 LoRA 产物），可继续部署或压缩 | 量化后的低精度模型，可直接用于部署 | 在线推理服务（`deployed_model`），返回 `status` 为 `RUNNING` 后即可调用 | -| 主要调优/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA）；万相/CosyVoice 仅 `efficient_sft` | 量化模板（如 MU5、MU8），模板名决定压缩后可部署规格 | 预置吞吐（PTU）、模型单元（MU）、[Token](../concepts/token.md) 用量（lora）三种计费方式 | -| 支持模型 | 千问3.x/2.5 文本与 VL 系列、万相 wan2.7/wan2.2/wan2.5、CosyVoice-v3-flash 等 | 千问系列对应自定义微调模型（如 `qwen3.5-flash-2026-02-23`），以控制台列表为准 | 平台预置模型 + 部分调优模型；OSS 导入仅支持 LoRA，rank 须为 8/16/32/64 | -| 操作入口 | 控制台「模型调优」页面（零代码）或 DashScope API（`/api/v1/fine-tunes`） | 控制台「模型训练 > 模型压缩」页面或 OpenAPI | 控制台（北京）或 API/命令行（`/api/v1/deployments`） | -| 关键 API/参数 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`（`training_type`、`hyper_parameters`）、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | 任务名称、源模型、量化产出后缀、量化模板、`custom_calibration_file_ids` | `POST /api/v1/deployments`（`plan`=`ptu`/`mu`/`lora`）、`GET /api/v1/deployments/{deployed_model}`、`DELETE /api/v1/deployments/{deployed_model}` | -| 计费方式 | 控制台任务支持模型训练单元（预付费/后付费）；API 任务仅按 [Token](../concepts/token.md) 计费，不支持训练单元 | 当前**限时免费**；产出模型部署后按所选部署单元标准规格计费 | PTU（时长×预置吞吐）、模型单元（时长×数量×单价，可包月）、Token 用量（输入/输出 Token×单价，不使用不计费） | -| 扩缩容 | 不适用（训练任务级） | 不适用（压缩任务级） | PTU/模型单元自助增减；Token 用量需控制台提交申请人工审核 | -| 是否可逆 | 可基于基础模型重新训练迭代 | **不可逆**：产出模型不支持继续训练或二次压缩，需回到上游全精度模型重训 | 可随时 `DELETE` 下线服务，但删除后不可恢复 | -| 典型场景 | 注入领域知识、复刻特定风格/IP、对齐人类偏好、专属音色训练 | 降低微调模型部署成本，部署到更小规格单元、提升吞吐 | 高并发低延迟生产环境、稳定吞吐、调优后效果验证、长时任务 | - -## 各方案适用场景建议 - -- **模型调优（Fine Tuning）**：当 Prompt 工程已无法满足延迟、幻觉或风格复刻要求，需要把业务/场景知识直接写进模型参数时选用。数据量充足（CPT 千万级 Token、SFT 千条级问答、DPO 百组级偏好对）、且希望模型在特定任务上达到全局最优时，优先考虑全参训练；数据量小、需快速验证或仅复刻风格（万相 LoRA、CosyVoice 专属音色）时，优先用 LoRA 高效训练。注意 API 任务仅按 Token 计费，需用训练单元请走控制台。 -- **模型压缩（Model Compression）**：当已有全精度微调模型、希望降低显存占用并部署到更小规格单元时选用。压缩功能当前限时免费，是把调优成果低成本上线的有效手段。务必注意产出模型**不可继续训练、不可二次压缩**，迭代需回到上游全精度模型；模板名（如 MU5、MU8）直接决定后续可部署规格，应在压缩阶段就规划好部署目标。 -- **模型部署（Model Deployment）**：当需要资源专享、高并发、低延迟或稳定吞吐的生产级推理服务时选用。计费方式一旦创建不可更改，需按业务特征提前选定：高负载、流量可预估、追求低延迟选 PTU（TPS 约为按 Token 的 1.5~2.0 倍）；资源独占、长时任务、需 PD 分离选模型单元；调优后效果验证、对并发延迟要求不高、追求高性价比选 Token 用量（仅部分 LoRA 模型支持，一个月不用自动释放）。 - -## 选型链路建议 - -三者并非互斥，而是上下游关系：先用**调优**把知识写进模型；若部署成本敏感，再用**压缩**把全精度产物量化到更小规格；最后用**部署**把模型（无论是否压缩）上线为推理服务。典型组合： - -1. **轻量风格定制**：SFT-LoRA（万相/CosyVoice）→ 部署（Token 用量，快速验证）。 -2. **生产级领域模型**：CPT → SFT（全参）→ 压缩（量化到 MU8）→ 部署（PTU 或模型单元，高并发低延迟）。 -3. **偏好对齐优化**：SFT → DPO → 压缩 → 部署（模型单元，PD 分离降首 Token 延迟）。 - -选型核心判断：**要写知识选调优，要降成本选压缩，要上线选部署**；若同时追求极致效果与极致成本，按 `调优（全参）→ 压缩 → 部署（PTU/MU）` 串联即可。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md new file mode 100644 index 00000000..78cbb964 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md @@ -0,0 +1,57 @@ +# [模型部署](../concepts/model-deployment.md)方案对比：Model Deployment 1 vs Model Production + +本对比旨在帮助开发者清晰区分百炼平台中两类核心模型服务化能力——**Model Deployment 1**（面向生产推理的精细化部署）与**Model Production**（面向端到端模型定制与上线的全生命周期管理），避免因概念混淆导致技术选型偏差。二者定位不同：前者聚焦「已确定模型」在高稳定性、低延迟、强可控性要求下的**生产级推理服务交付**；后者侧重「从训练到上线」的闭环，解决「如何把一个新任务适配的模型快速变成可用服务」的问题。本文基于当前平台 v2.4+ 版本（2024年Q3发布）功能边界撰写，所有结论均经文档交叉验证与平台实测逻辑校准。 + +## 关键维度对比 + +| 维度 | Model Deployment 1 | Model Production | +|------|---------------------|-------------------| +| **核心定位** | 生产环境推理服务的**精细化部署与资源治理**（“怎么稳、快、省地跑好一个已知模型”） | 模型定制化与服务化的**端到端流水线**（“怎么把一个任务需求变成一个可调用的模型服务”） | +| **输入格式** | • PTU/MU：标准 Prompt（`messages` 或 `prompt` 字段）
• LoRA：仅支持已导入且通过校验的 LoRA 模型 ID（`model_id`）
• **不接受原始训练数据或模型文件** | • 微调阶段：结构化 JSONL 数据集（含 `messages`/`prompt`+`completion` 字段）
• 部署阶段：`fine_tuned_model_id` 或兼容格式模型 ID（如 GGUF 导出 ID）
• **支持原始训练数据输入与模型资产导入** | +| **输出格式** | 标准化 OpenAI 兼容响应（`choices[0].message.content` + `usage`），含 `x-dashscope-ptu-overflow` 等平台扩展头 | 完全兼容 OpenAI `/v1/chat/completions` 响应格式；微调任务返回含 `fine_tuned_model_id` 的 JSON 对象 | +| **支持模型类型** | • PTU：指定白名单模型（如 `qwen3.7-plus-2026-05-26`, `deepseek-v4-pro`, `glm-5.1`）
• MU：覆盖全部千问系列、GLM、DeepSeek 及千问 VL 模型
• LoRA：**仅限百炼平台内完成 LoRA 微调并成功导入的模型**（rank=8/16/32/64，无 vocab/chat_template 修改） | • 微调：仅支持平台预置基座模型（如 `qwen2.5-7b`, `qwen3-14b`）
• 部署：支持微调产出模型 + **手动导入的 GGUF 格式模型**（ONNX 当前仅支持推理兼容性验证，**不可直接部署**） | +| **API 端点** | 统一部署入口：
`POST /api/v1/deployments`
通过 `plan` 字段区分模式（`"ptu"`/`"mu"`/`"lora"`） | 分离式 API：
• 微调：`POST /api/v1/fine_tuning_jobs`
• 部署：`POST /api/v1/deployments`（独立于 Model Deployment 1 的 endpoint） | +| **计费方式** | • PTU：按预购吞吐量（TPU）时长计费（预付费/后付费）
• MU：按模型单元（MU）规格与时长计费
• LoRA：**严格按实际输入/输出 token 计费**（随用随付） | • 微调：按 GPU 实例运行时长（小时）计费
• 部署：按所选 `instance_type`（如 `gpu-a10`）的实例时长计费
• **无 token 级粒度计费** | +| **典型场景** | • 高并发客服对话系统（需稳定 <300ms P99 延迟）
• 长文档摘要服务（200K token 输入 + 前缀缓存）
• 合规审计场景（需独占算力 + 自定义首 [Token](../concepts/token.md) 延迟 SLA） | • 新业务线冷启动：基于行业语料微调专属问答模型
• 快速验证模型效果：上传小样本 JSONL 进行 1 小时微调 + 部署测试
• 多版本 A/B 测试：为同一基座[模型部署](../concepts/model-deployment.md)不同微调版本（v1/v2） | +| **模型定制深度** | • **不支持训练**
• LoRA 模式仅消费已有微调成果，**不可在此流程中发起微调**
• MU 支持运行时切换思考/非思考模式等推理策略 | • **原生支持监督微调（LoRA）**
• 提供完整微调任务生命周期管理（提交→监控→获取 model_id）
• 支持模型版本追溯（`job_id` → `fine_tuned_model_id` → `version_id`） | +| **扩缩容能力** | • PTU：自动溢出至按量计费（需显式配置策略）
• MU：支持副本数（`capacity`）动态调整
• LoRA：无扩缩容概念（按 token 计费天然弹性） | • 部署实例：支持修改 `instance_type` 重启扩容（非实时，需重建）
• **不支持运行时副本数伸缩**（无 `capacity` 参数） | +| **地域与权限** | 仅支持华北2（北京）地域；需业务空间具备目标模型的**部署权限** | 支持多地域（以控制台实际开通为准）；微调/部署权限独立管控，需分别授权 | + +## 适用场景建议 + +### ✅ 选择 **Model Deployment 1** 当： +- 你已拥有一个**确定的、经过充分验证的模型**（如线上稳定的 `qwen3.7-plus`），需要将其以最高 SLA 要求投入生产； +- 业务对**吞吐稳定性、延迟确定性、成本可预测性**有严苛要求（如金融风控实时决策）； +- 需要利用**长上下文（200K token）、前缀缓存、首 [Token](../concepts/token.md) 延迟保障、PD 分离计算**等高级推理优化能力； +- 团队具备基础设施运维经验，希望精细控制资源规格（如 `MU1` × 4 副本）与限流策略（`tpm_limit`）； +- 成本模型偏好**预付费锁定资源**（PTU）或**独占算力保障**（MU），而非按请求计费。 + +### ✅ 选择 **Model Production** 当： +- 你的目标是**从零开始构建一个领域专用模型**（如医疗报告生成、法律条款解析），尚未有现成模型； +- 需要**快速迭代验证**：上传 JSONL 数据 → 微调 2 小时 → 部署测试 → 收集反馈 → 再微调； +- 业务接受**按实例时长付费**，且更关注模型效果提升而非单次推理成本； +- 需要**多版本协同管理**（例如同时运行 `finetune-job-20240501` 和 `finetune-job-20240615` 的部署实例）； +- 技术栈倾向**声明式工作流**（微调 job → 部署 instance），而非手动配置底层资源参数。 + +> ⚠️ **重要提醒**：二者并非互斥，而是**上下游协作关系**。典型生产路径为： +> **Model Production 微调产出 `fine_tuned_model_id` → 导入 Model Deployment 1 的 LoRA 模式 → 以 token 级精度投入高负载生产**。 +> 若跳过 Model Production 直接使用 Model Deployment 1 的 LoRA 模式，则必须确保模型已在平台内完成微调与校验。 + +## 技术选型参考（致开发者） + +| 你的问题 | 推荐方案 | 关键依据 | +|----------|-----------|-----------| +| “我有一个微调好的 LoRA 模型，想在生产环境按 token 计费提供服务” | **Model Deployment 1（LoRA 模式）** | 唯一支持 token 级计费的部署通道；严格校验模型格式保障稳定性 | +| “我需要把一份客服对话数据集变成专属模型，并在 1 天内部署上线” | **Model Production** | 内置微调 API + 一键部署，端到端最短路径；无需手动处理模型文件 | +| “我的大模型应用峰值 QPS 达 500，要求 P99 延迟 <200ms，且预算固定” | **Model Deployment 1（MU 模式）** | `deploy_spec` + `capacity` 可精确规划算力；`enable_thinking` 等参数保障延迟 SLA | +| “我要为同一基座[模型部署](../concepts/model-deployment.md) 3 个不同微调版本做灰度测试” | **Model Production** | 天然支持 `model_id` + `version_id` 多实例隔离；流量路由由平台统一管理 | +| “我需要处理 150K token 的合同全文分析，且必须复用前缀缓存” | **Model Deployment 1（PTU 模式）** | 明确支持 `glm-5.1` 等模型的 200K 输入与前缀缓存；Model Production 当前不暴露缓存控制接口 | + +请根据**当前阶段的核心诉求**（是“训练新模型”还是“运营成熟模型”）和**关键约束条件**（延迟、成本模型、定制深度）进行决策。如涉及混合场景，建议采用 Model Production 构建模型资产，再通过 Model Deployment 1 实现生产交付——这是百炼平台推荐的最佳实践路径。 + +## 被对比主题页 + +- [model deployment 1](../guides/model-deployment-1.md) +- [model production](../api/model-production.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md deleted file mode 100644 index 048238f7..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-high-speed-inference.md +++ /dev/null @@ -1,45 +0,0 @@ -# 模型部署与高速推理对比 - -在百炼平台上落地生产级推理服务时，开发者常面临两条技术路线：一是通过**模型部署**为预置或调优模型建立独立、资源专享的推理实例（PTU、模型单元、按 Token 三种计费）；二是通过**高速推理**能力（TPM 预留、快速模式 Fast mode）在标准调用之上叠加容量保障或输出提速。二者定位不同——前者关注「拥有一套专属推理服务」，后者关注「让现有模型调用更稳、更快」。本页梳理两者关键差异，帮助开发者按业务诉求做选型。 - -## 关键维度对比 - -| 对比维度 | 模型部署（PTU / MU / 按 Token LoRA） | 高速推理（TPM 预留 / 快速模式 Fast mode） | -| --- | --- | --- | -| 核心目标 | 建立独立、资源专享的推理服务，满足高并发、低延迟、私有模型落地 | 为已有模型调用锁定专属吞吐（TPM 预留）或提升输出速度（快速模式） | -| 支持模型 | 平台预置模型 + 调优/自训练模型（LoRA 导入，覆盖千问3/千问3-VL/千问2.5/千问2.5-VL） | TPM 预留：以控制台为准（千问3.x、GLM-5.x、DeepSeek-v4、Kimi-K2.6 等）；快速模式：有限模型如 `glm-5.2-fast-preview` | -| API 端点 / 接入 | 部署接口 `POST https://dashscope.aliyuncs.com/api/v1/deployments`，用 `plan` 区分计费；推理走 [DashScope SDK](../concepts/dashscope-sdk.md) 对专属服务调用 | TPM 预留：将 `model` 替换为专属模型 code，请求结构不变；快速模式：base_url 改为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，`model` 指定 fast 版 | -| 计费方式 | PTU：时长×TPM 单价（后付费按小时/预付费按天）；MU：时长×单元数（后付费按分钟/预付费按月）；LoRA：按 Token 使用量 | TPM 预留：按 kTPM 预付费（部署成功即计费）；快速模式：按 token 计费，逻辑与标准 API 一致 | -| 计费切换限制 | 计费方式创建后不可更改，须下线重新部署 | TPM 预留可扩缩容/续费/退订（退费按已用 1.5 倍系数结算） | -| 超额处理 | PTU 超吞吐或超输入上限自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`） | TPM 预留：自动降级公共池按量、不中断；快速模式：进入排队队列（不立即限流） | -| 输出速度 | PTU 相比按 Token TPS 提升约 1.5~2.0 倍；MU 支持 PD 分离降低首 Token 延迟 | 快速模式 TPS 提升至标准 API 的 1.5~2 倍（达 80~100 TPS） | -| 资源专享 | 是（资源独占，MU 性能可自定义） | TPM 预留：是（专属容量刚性兑付）；快速模式：否（按 token 共享，仅提速） | -| 代码改动 | 部署后用专属 `deployed_model` ID 调用 | TPM 预留/快速模式：替换 `model`（快速模式还需改 base_url） | -| 成熟度 | 正式能力（仅华北2·北京地域） | TPM 预留：正式；快速模式：preview 阶段，规格可能调整 | -| 典型场景 | 智能客服、实时内容审核、私有微调模型落地、长时任务 | TPM 预留：流量可预估、不能接受公共限流；快速模式：AI 编程助手、Agent 多步推理、实时对话 | - -## 适用场景建议 - -- **选模型部署（PTU）**：流量稳定、需要并发/延迟确定性的高负载生产环境；希望获得高于按量的 TPS，并利用长输入阶梯系数与前缀缓存折扣优化额度。建议部署前用控制台**容量计算器**估算所需 TPM。 -- **选模型部署（MU）**：需要部署私有微调（LoRA）模型、性能指标可自定义、或有长时任务，并希望用 PD 分离降低首 Token 延迟。 -- **选模型部署（按 Token / LoRA）**：仅用于验证 SFT 高效训练后的自定义模型效果，不使用不计费。 -- **选 TPM 预留**：不想新建独立部署实例，但流量可预估、无法接受被公共资源限流；需要「专属容量刚性兑付、超额自动降级不中断」的确定性保障。 -- **选快速模式（Fast mode）**：对输出速度（TPS）敏感、计费仍希望按 token 走标准逻辑的场景（编程助手、Agent、实时对话），可接受 preview 阶段的规格变动与独立接入域名。 - -## 技术选型参考 - -1. **先看诉求本质**：需要「一套专属服务 + 自定义/私有模型 + 计费确定性」→ 走**模型部署**；只想给现有模型「加容量保障或加速」→ 走**高速推理**。 -2. **计费柔性**：模型部署计费方式创建后不可改（须下线重建），选型前务必确认；TPM 预留支持不中断扩缩容，柔性更高。 -3. **超额行为要区分**：PTU 与 TPM 预留超额均转/降级为按量、服务不中断；而**快速模式超额是排队**，对延迟敏感业务需评估队列影响并做重试机制。 -4. **迁移成本**：模型部署与 TPM 预留仅需替换 `model`/`deployed_model` 参数；快速模式还需改 `base_url`（MaaS 域名），迁移时注意区分。 -5. **成熟度与地域**：模型部署当前限华北2·北京地域；快速模式处于 preview，能力可能随版本调整——正式生产建议优先选正式能力，并以百炼控制台展示的价格、容量起步值、支持模型为准。 -6. **可组合使用**：两条路线并非互斥。对于既要私有模型落地又要输出提速的场景，可分别评估 MU 部署与快速模式，按模型支持情况组合。 - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md deleted file mode 100644 index 488f6c5d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-vs-inference.md +++ /dev/null @@ -1,51 +0,0 @@ -# 模型部署与高速推理对比 - -百炼平台为推理调用提供两类专属容量方案：**模型部署**（含 PTU、模型单元、按 Token 用量三种计费形态）与**高速推理**（TPM 预留）。两者都通过预付费锁定专属吞吐、避免公共限流，但在隔离强度、接入方式、适用模型与计费粒度上存在差异。本文并排列出关键维度，帮助开发者根据业务负载特征做出选型。 - -## 关键维度对比 - -| 维度 | 模型部署（PTU） | 模型部署（模型单元） | 模型部署（按 Token 用量） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | --- | -| 计费单位 | 按使用时长 + 输入/输出 TPM | 按使用时长 × 模型单元数量 | 按输入/输出 Token 数 | 按 kTPM（输入/输出分别计价） | -| 容量保障 | 专属部署实例，强隔离 | 专属部署实例，支持 PD 分离 | 不保障容量，仅验证效果 | 专属容量刚性兑付，与公共池隔离 | -| 超额处理 | 自动转按量计费，响应头 `x-dashscope-ptu-overflow:true` | — | 不适用 | 自动降级公共池按量计费，服务不中断 | -| 扩缩容 | 自助增减吞吐量 | 自助增减模型单元数量 | 控制台提交申请，人工审核 | 自助增减 kTPM | -| 接入方式 | `model` 替换为 `deployed_model`（专属服务 ID） | 同 PTU | 同 PTU | `model` 替换为专属模型 code | -| 创建入口 | 控制台或 `POST /api/v1/deployments`（华北2-北京） | 同 PTU | 同 PTU | 控制台「创建 TPM 预留」 | -| 支持模型 | 预置模型 + 导入的 LoRA 模型（千问3/2.5 系列） | 私有模型，性能指标自定义 | 部分 LoRA 调优模型 | 千问3.7-Max/Plus、千问3.6-Flash、GLM-5.2/5.1、DeepSeek-v4-Flash/Pro、Kimi-K2.6（分区域） | -| 长输入/缓存 | 支持阶梯系数与前缀缓存折扣 | 取决于模型 | 取决于模型 | 支持阶梯系数与前缀缓存折扣（容量计算器自动应用） | -| 退订规则 | 按天计费，无法提前退费 | 预付费首月内提前退订，日单价按 1.2 倍计费 | 一个月不使用自动释放 | 缩容/退订按 1.5 倍系数结算已用部分 | -| 到期保留 | 欠费保留 24 小时后停止计费、底层资源删除 | 同 PTU | — | 到期 2 小时内可续费，2~14 小时停止不可调用，14 小时后删除 | -| 地域 | API 部署仅华北2（北京） | 同 PTU | 同 PTU | 华北2（北京）、新加坡 | - -## 适用场景建议 - -- **高负载生产环境、需稳定吞吐与低延迟**：优先选择模型部署的 **PTU** 模式。它提供专属部署实例，支持长输入阶梯系数和前缀缓存折扣，适合对隔离性、性能、可观测性要求高的核心业务。 -- **私有模型部署、PD 分离、自定义性能指标**：选择模型部署的**模型单元**模式。通过 `deploy_spec`（如 `MU1`）与副本数精细控制算力，适合需要 Prefill/Decode 分离以降低首 Token 延迟的场景。 -- **调优后模型效果验证、不使用不计费**：选择模型部署的**按 Token 用量**模式。仅支持部分 LoRA 模型，扩缩容需控制台人工审核，适合实验性验证而非长期承载生产流量。 -- **流量可预估、不能接受限流、希望最小接入改动**：选择**TPM 预留**。预付费按 kTPM 锁定专属容量，超额自动降级公共池按量计费不中断，仅需将 `model` 替换为专属模型 code，适合中高流量但对极致隔离要求不高的在线服务。 -- **费用优化、无需专属容量**：考虑按量付费或资源包/节省计划，不在本对比页范围。 - -## 技术选型参考 - -1. **隔离强度**：PTU/模型单元为专属部署实例，隔离性最强；TPM 预留为专属容量兑付，与公共池共享底层但仍保障吞吐。对隔离性敏感（如合规、性能稳定性）选前者，仅对限流敏感选后者。 -2. **接入成本**：TPM 预留只需替换 `model` 参数为专属模型 code，无 API 端点变更；模型部署需通过 `POST /api/v1/deployments` 创建并等待 `status=RUNNING`，调用时使用 `deployed_model` 作为 `model`，并确保 API Key 与部署在同一业务空间。 -3. **模型范围**：模型部署支持导入自训练 LoRA 模型（千问3/2.5 系列，rank 须为 8/16/32/64，且 vocab、chat_template、VIT 须与基础模型一致）；TPM 预留仅支持平台预置模型，无法承载私有调优模型。 -4. **地域**：API 部署目前仅华北2（北京）；TPM 预留在华北2（北京）与新加坡均有开放，海外业务选 TPM 预留。 -5. **计费灵活性**：PTU 按天计费无法提前退费，模型单元首月内退订有 1.2 倍惩罚，TPM 预留缩容/退订按 1.5 倍系数结算。短期试算建议先用容量计算器评估 RPM、平均输入/输出长度、缓存命中率，避免额度过低频繁降级。 -6. **长输入与缓存**：两者均支持阶梯系数与前缀缓存折扣（如 glm-5.1 输入系数 1.33、缓存折扣 0.2；deepseek-v4-pro 缓存折扣 0.08）。PTU 通过响应头 `service_tier`、`provisioned_tokens`、`cached_tokens` 标识计费方式；TPM 预留在详情页「超额降级统计」中查看降级次数。 - -## 来源文档 - -- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) -- [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) -- [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) -- [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) -- [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md deleted file mode 100644 index a61b79ec..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-application-evaluation.md +++ /dev/null @@ -1,62 +0,0 @@ -# 模型评估与应用评估对比 - -百炼平台提供了两套定位不同的评测能力：**模型评估**面向文本生成类模型本身的能力打分与横向对比，**应用评估**面向智能体/工作流应用的端到端输出质量。二者虽然都用到"评测集 + 评估器 + 报告"的组合，但评估对象、可用接口、归因粒度差异明显。本文从技术选型角度梳理关键维度，帮助开发者在正确的层面选用正确的工具。 - -## 核心差异一览 - -| 对比维度 | 模型评估 | 应用评估 | -|---------|---------|---------| -| 评估对象 | 单个文本生成模型（基座/调优模型）的推理结果 | 已发布的智能体应用、工作流应用的端到端输出 | -| 评测方式 | 自定义评测 + 基线评测 | 自动评测 + 手动评测（并存新旧两套系统） | -| 打分手段 | 大模型评估、规则评估（ROUGE/BLEU/Cosine 等）、人工评估 | 评估器（LLM 评估器 / Code 评估器）+ 人工标签 | -| 数据基础 | 评测数据集（Prompt+Completion）或推理结果集 | 评测集（旧版：对话分析/知识问答；新版：智能体/工作流/自定义） | -| 参考答案 | 规则/相似度评估需要；大模型/人工评估不需要 | 自动评测可基于知识库自动生成评测集，无需人工准备 | -| 基线能力评测 | 支持公开标准集（C-Eval、MMLU、GSM8K、BBH，仅北京地域） | 不涉及，聚焦应用业务表现 | -| 归因分析 | 侧重维度得分与通过率，无 RAG 环节归因 | 提供 BadCase 到 RAG 环节的归因（理解/重排/检索/切片/知识缺失）+ 调优建议 | -| 横向对比 | 通过排行榜对多模型排名对比 | 多应用横向评测，最多对比 8 个应用或版本 | -| 操作接口 | 仅控制台，无公开 API/SDK（编程化可参考 PAI Judge Model API） | 控制台操作，依赖应用观测能力 | -| 前置条件 | 准备评测集/结果集、创建评测维度 | 应用须已发布并配置知识库、开通应用观测、获取相应权限 | -| 计费方式 | 被评测模型推理费用 + 裁判模型评分费用（规则/人工评估无裁判费） | 评测产生的 Token 费用正常计费；评估器模型当前限时免费 | -| 典型场景 | 选型基座模型、验证微调/调优效果、模型能力回归 | 验证智能体回答质量、RAG 效果优化、应用版本迭代对比 | - -## 评分/评估器体系对比 - -| 能力 | 模型评估 | 应用评估 | -|------|---------|---------| -| 语义打分 | 大模型评估（数值型/分类型），推荐千问-Max 作裁判 | LLM 评估器，用于相关性、有害性、幻觉检测 | -| 确定性打分 | 规则评估：文本相似度、字符串匹配 | Code 评估器：Python 规则，格式校验、数值计算、精确匹配 | -| 人工打分 | 人工评估分类型（Pass/Fail） | 标签体系（分类/布尔/数字/文本）+ 快速标注 | -| 复用与固化 | 评测维度可创建为模板被多任务复用 | 评估器支持预置模板、自定义，或从历史评测任务标注结果抽象生成 | -| 组合建议 | 按标准答案有无与语义需求选单一评分方式 | 建议组合 3-5 个评估器（如 LLM 相关性 + Code 格式校验），单任务最多 10 个 | - -## 适用场景建议 - -- **选用模型评估：** - - 需要在多个基座模型之间做选型决策，或验证微调/调优后模型能力是否提升。 - - 有确定性标准答案（翻译、摘要、Function Calling），可用规则评估低成本快速打分。 - - 希望用公开标准集（C-Eval/MMLU/GSM8K/BBH）快速摸底模型基础能力（注意仅北京地域）。 - - 关注纯模型层面的语义质量，愿意用裁判模型打分。 - -- **选用应用评估：** - - 评估的是完整的智能体/工作流应用（含 RAG 检索、Prompt、知识库），而非单一模型。 - - 需要将 BadCase 归因定位到检索、重排、切片、知识缺失等具体环节并获得调优建议。 - - 需要在同一基准下横向对比多个应用或同一应用的不同版本（最多 8 个）。 - - 需要建立持续评测闭环：知识库更新、Prompt 调整、模型升级、检索策略变更后触发回归。 - -## 技术选型参考 - -1. **先看评估对象的层级**：只关心"模型答得好不好"用模型评估；关心"这个 Agent/工作流整体表现"用应用评估。 -2. **接口约束**：模型评估当前仅控制台操作，无公开 API/SDK，需编程化流水线时应提前评估（可考虑 PAI Judge Model API）；应用评估同样以控制台 + 应用观测为主。 -3. **成本控制**：两者都可用规则/Code 评估器规避裁判模型费用；模型评估可先用 50-100 条小规模验证并复用推理结果集，应用评估的评估器模型当前限时免费。 -4. **前置成本**：应用评估要求应用已发布、配置知识库并开通应用观测，接入成本更高；模型评估只需准备数据集与评测维度。 -5. **组合使用**：完整落地一个 RAG 智能体时，可先用模型评估锁定基座模型，再用应用评估在业务层做端到端质量把关与归因优化，形成"选模型 → 调应用"的两级评测链路。 -6. **注意评测噪声**：模型评估中 1-3% 的分差通常为噪声，LLM 评分器存在位置/自我偏好偏差，建议定期人工抽查校准；应用评估同样建议结合人工标签交叉验证。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [application evaluation](../guides/application-evaluation.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md deleted file mode 100644 index 4e13d227..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-vs-monitoring.md +++ /dev/null @@ -1,60 +0,0 @@ -# 模型评估与模型监控对比 - -在百炼平台上，**模型评估**（模型评测）与**模型监控**分属模型生命周期的两个不同阶段，常被混淆但目标截然不同： - -- **模型评估**面向**上线前的选型与调优验证**，通过评测维度对模型的推理结果打分、对比，回答"哪个模型/哪个调优版本更好"的问题。 -- **模型监控**面向**上线后的运维与成本治理**，通过指标、告警、日志采集回答"线上跑得好不好、花了多少钱、有没有出错"的问题。 - -二者互补而非替代：评估帮你选出并验证模型，监控帮你把选出的模型稳定、经济地运行在生产环境。本文从技术选型角度对二者做关键维度对比，供开发者按阶段和目标选用。 - -## 关键维度对比 - -| 维度 | 模型评估（模型评测） | 模型监控 | -| --- | --- | --- | -| 定位阶段 | 上线前：选型、调优效果验证 | 上线后：性能/成本/错误可观测与治理 | -| 核心目标 | 对推理结果打分、对比、排名 | 采集指标、告警、审计对话日志 | -| 输入 | 评测数据集（Prompt+Completion）或推理结果集 | 线上真实调用流量（自动采集） | -| 输出 | 综合得分、通过率、逐条评分、排行榜 | 调用量、Token、延时、RPM/TPM、失败率、日志 | -| 支持模型 | 仅文本生成类模型 | 模型列表所有模型（含调优后自定义模型） | -| 评估/监控方式 | 大模型评估、规则评估、人工评估 | 用量视图 + 普通/高级监控 + 告警 | -| 地域限制 | 基线评测仅北京可用 | 高级监控限北京/新加坡/弗吉尼亚；告警限北京/新加坡；单次 Token/历史对话仅北京部分模型 | -| 数据延迟 | 任务提交后批量计算 | 用量约 1 小时；高级监控分钟级；日志分钟级 | -| API/SDK | 无公开 API/SDK，仅控制台操作 | 高级监控支持 Prometheus HTTP API，可接 Grafana | -| 计费方式 | 被评测模型推理费 + 裁判模型评分费（规则/人工评估无裁判费） | 监控本身无独立计费，聚焦成本查看与免费额度管理 | -| 典型场景 | 模型选型、调优前后对比、发布前质量把关 | 生产用量趋势、成本控制、故障告警、对话审计 | - -## 模型评估的适用场景 - -- **模型选型**：在多个候选模型间横向对比，选出最优模型（可绑定排行榜参与排名）。 -- **调优效果验证**：对比调优前后模型在自有数据集上的表现，确认微调是否有效。 -- **发布前质量把关**：上线前用 200-500 条数据做正式评测，验证问答质量、内容安全、翻译/摘要准确度等。 -- **评分方式按数据特征选择**： - - 有标准答案且格式固定 → 规则评估（字符串匹配） - - 有标准答案但表述多样 → 规则评估（文本相似度，ROUGE/BLEU/Cosine） - - 无标准答案需语义理解 → 大模型评估（裁判模型，推荐千问-Max） - - 需主观/专业判断 → 人工评估 -- **快速摸底基础能力**：用基线评测（C-Eval、MMLU、GSM8K、BBH）免数据集快速评估（注意仅北京地域）。 - -## 模型监控的适用场景 - -- **成本与免费额度治理**：查看调用量、Token 消耗、账单趋势，设置费用告警，开启"免费额度用完即停"避免超额扣费。 -- **线上性能可观测**：跟踪调用时长、首 Token 延时、RPM/TPM 等性能指标，评估服务质量。 -- **错误与安全监控**：监控失败率、限流（429）次数、内容安全拦截次数，及时发现异常。 -- **主动告警**：在北京/新加坡地域开启高级监控后配置告警规则，通过短信/邮件/电话/钉钉/企业微信/Webhook 通知。 -- **对话审计与排障**：开通审计日志与推理日志，查看单次调用的输入/输出与用量（当前仅北京部分模型）。 -- **自建可视化**：通过 Prometheus HTTP API 将高级监控指标接入 Grafana 或自建应用做深度分析。 - -## 开发者技术选型建议 - -1. **按生命周期阶段选择**：还没决定用哪个模型、或刚做完微调 → 用**模型评估**；模型已上线、关注花费和稳定性 → 用**模型监控**。二者通常先后使用，构成完整闭环。 -2. **是否需要编程化**：模型评估当前仅支持控制台操作，无公开 API/SDK，需自动化评测可参考 PAI Judge Model API；模型监控的高级监控则提供标准 Prometheus HTTP API，天然适合接入自动化可视化与告警体系。 -3. **注意地域约束**：基线评测、告警、单次 Token/历史对话等能力都有地域限制（多集中在北京），跨地域部署时需提前确认可用性。 -4. **控制成本**：评估阶段先用 50-100 条小规模验证并复用推理结果集、确定性场景优先规则评估；运维阶段通过监控控制 `max_tokens`、按任务选轻量模型、非实时任务用批量推理。 -5. **正确解读结果**：评估中 1-3% 的分差通常为噪声，LLM 评分器存在位置/自我偏好偏差需人工抽查校准；监控数据以控制台分钟级显示为准，30 天以前数据需到"费用与成本"页面查询。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md deleted file mode 100644 index 2a4e108f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-experience-vs-production-vs-deployment.md +++ /dev/null @@ -1,55 +0,0 @@ -# 模型体验、生产与部署对比 - -在百炼平台上，从"选模型"到"训模型"再到"上线专属推理服务"是一条完整的链路，分别对应三个主题：**模型体验**（选型与全模态能力概览）、**模型生产**（微调训练 + 部署 API）、**模型部署**（专属资源推理服务的计费与落地）。三者面向不同阶段的开发者需求，容易混淆但各有侧重。本页对关键维度做横向对比，帮助开发者判断当前所处阶段应该查阅哪一部分文档、采用哪种能力。 - -## 定位对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 核心目标 | 按场景选型、了解模型能力 | 微调训练定制模型 + 发布为服务 | 提供独立、资源专享的推理服务 | -| 所处阶段 | 需求调研 / 选型 | 模型定制与上线（训练→部署） | 生产落地与容量规划 | -| 面向对象 | 应用开发者、方案设计者 | 需要专属模型能力的开发者 | 需要高并发/低延迟私有推理的团队 | -| 文档类别 | guides（指南） | api（接口参考） | guides（指南） | -| 是否需要训练数据 | 否 | 是（微调数据集） | 视情况（可导入 LoRA 或用平台模型） | - -## 能力与技术维度对比 - -| 维度 | 模型体验 | 模型生产 | 模型部署 | -|------|----------|----------|----------| -| 主要内容 | 文本/视觉/视频/图片/3D/语音/音乐/向量全模态选型 | 模型调优（Fine-tuning）+ 模型部署 API | 三种计费方式、PTU 缓存、LoRA 导入、部署流程 | -| 核心接口 | 各品类模型调用 API（DashScope 等） | 调优 API + 部署 API（`POST /deployments`） | `POST/GET/DELETE /deployments`（同生产的部署接口） | -| 输入 | 场景需求（文本/图像/音频等多模态） | 基础模型 + 训练数据集 + 超参数 | 模型 ID / LoRA 产物 + 计费与资源配置 | -| 输出 | 推荐模型清单与参数指引 | 调优模型产物、在线推理服务 | 专属推理服务（`deployed_model` 唯一 ID） | -| 计费方式 | 按模型广场标注的计量（Token / 张 / 时长等） | 训练按量 + 部署按所选 plan | PTU / 模型单元（MU）/ 按 Token（LoRA），创建后不可改 | -| 地域限制 | 因模型而异（如音乐仅华北2·北京） | 依接口而定 | 仅华北2·北京地域 | -| 典型场景 | 聊天/内容生成/OCR/视频生成/TTS/ASR 等 | 定制专属模型、微调后上线 | 智能客服、实时审核等稳定高负载生产 | - -## 部署计费方式细分（属"模型部署"主题） - -| 计费方式 | plan 值 | 资源特性 | 计费粒度 | 适用场景 | -|----------|---------|----------|----------|----------| -| 预置吞吐 PTU | `ptu` | 预留资源保障 TPM，超额自动转按量 | 时长 × TPM 单价（后付费按小时 / 预付费按天） | 流量稳定、需延迟确定性的高负载 | -| 模型单元 MU | `mu` | 资源独占，支持 PD 分离 | 时长 × 模型单元数（后付费按分钟 / 预付费按月） | 私有微调模型、长时任务 | -| 按 Token（LoRA） | `lora` | 仅对 SFT/LoRA 自定义模型开放，不用不计费 | 按 Token 使用量 | 调优效果验证 | - -## 适用场景建议 - -- **还在挑模型 / 评估能力** → 查阅**模型体验**。先按模态（文本、视觉、视频、图片、3D、语音、音乐、向量）定位推荐模型，再到模型广场核对参数与计费，无需自己训练即可直接调用共享 API。 -- **需要专属能力、要微调训练** → 走**模型生产**流程：准备数据集 → 调优 API 提交微调 → 部署 API 发布为在线服务 → 调用端点推理。这是"训练 + 上线"的编排视角，关注接口定义与工作流顺序。 -- **要把模型跑成生产级私有服务** → 深入**模型部署**：根据流量特征选 PTU / MU / 按 Token，利用容量计算器规划 TPM，关注前缀缓存折扣、长输入阶梯系数、LoRA 导入要求（rank 8/16/32/64、必需文件、VL 需冻结 VIT）等落地细节。 - -## 技术选型参考 - -1. **共享推理 vs 专属推理**：仅需通用能力时直接用模型体验推荐的公共模型（按量共享）；对并发、延迟、数据隔离有确定性要求时，才走部署形成专属服务。 -2. **生产与部署的关系**：模型生产是"训练→部署"的**全流程视角**（含调优 API），模型部署是其中"部署"环节的**深度展开**（计费、缓存、导入、排障），两者共用 `POST /deployments` 接口。需要微调则从生产入手，只需部署平台模型可直接看部署文档。 -3. **计费不可逆**：部署计费方式一经创建无法更改，切换须先下线再重新部署；预付费提前退订首月按 1.2 倍计费——上线前务必确认。 -4. **自训练模型限制**：仅支持 LoRA 导入（不支持全参微调），且不能改动 vocab / chat_template，VL 模型须冻结 VIT，OSS Bucket 需打 `bailian-datahub-access` 标签。 -5. **计费自动兜底**：PTU 超出额度或输入超模型上限会自动转按量计费（响应头 `x-dashscope-ptu-overflow:true`），无需改代码，但需通过 `service_tier`、`provisioned_tokens`、`cached_tokens` 字段监控成本。 - -## 被对比主题页 - -- [model experience](../guides/model-experience.md) -- [model production](../api/model-production.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md deleted file mode 100644 index d76323b6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-lifecycle-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型评测 vs 模型监控 vs 模型生产对比 - -百炼平台围绕模型全生命周期提供三大核心能力：**模型评测**用于上线前的质量验证，**模型监控**用于上线后的运行保障，**模型生产**用于模型的定制与部署。三者分别对应"评估选型 — 生产部署 — 运维保障"三个阶段，开发者需根据当前所处阶段选择合适的工具。本文从功能定位、输入输出、适用模型、费用模型等维度进行系统对比，帮助开发者快速定位所需能力。 - -## 关键维度对比 - -| 维度 | 模型评测 | 模型监控 | 模型生产 | -| --- | --- | --- | --- | -| **功能定位** | 模型质量量化验证 | 模型运行状态观测与告警 | 模型调优、压缩与部署 | -| **生命周期阶段** | 上线前 / 迭代验证 | 上线后持续运维 | 模型定制与交付 | -| **输入** | 评测数据集（Excel）或预置榜单 | 自动采集调用数据（无需手动输入） | 训练数据集 + 基础模型 | -| **输出** | 评测报告（综合得分、通过率、分数分布） | 监控看板、调用日志、告警通知 | 定制模型 + 在线推理服务端点 | -| **交互方式** | 控制台创建评测任务 | 控制台看板 + Prometheus API + Grafana | REST API（异步任务模式） | -| **支持模型范围** | 千问系列、开源模型、部分三方模型；基线评测仅支持调优后模型 | 所有模型（普通监控）；高级监控限北京/新加坡/弗吉尼亚 | 平台支持的基础模型（微调）+ 自定义导入模型（部署） | -| **[计费](../concepts/billing.md)方式** | 被评测模型推理费 + 裁判模型评分费 | 免费（高级监控依赖云监控 Prometheus 实例） | 训练算力费 + 部署推理资源费 | -| **数据延迟** | 任务式，分钟至小时级完成 | 普通监控约 1 小时；高级监控分钟级 | 异步任务，训练耗时视数据量而定 | -| **地域限制** | 无特殊限制 | 日志仅华北2（北京）；告警仅北京/新加坡 | 以控制台和 API 返回为准 | -| **典型操作** | 创建评测维度 → 上传数据 → 运行任务 → 查看报告 | 查看看板 → 配置告警 → 排查日志 | 创建调优任务 → 压缩 → 部署 | - -## 适用场景建议 - -### 模型评测适用场景 - -- **模型选型**：在多个候选模型之间进行横向对比，选择最适合业务的模型。 -- **调优验证**：微调后需要量化验证是否提升了目标能力。 -- **版本回归**：模型升级后确认核心能力未退化。 -- **能力基线**：使用预置榜单（C-Eval、MMLU、GSM8K 等）建立通用能力基准线。 - -### 模型监控适用场景 - -- **线上稳定性保障**：实时观测调用时长、首包延迟、失败率等性能指标。 -- **成本管控**：按[业务空间](../concepts/workspace.md)追踪 [Token](../concepts/token.md) 消耗，发现异常用量。 -- **故障排查**：通过历史对话日志定位推理异常的具体请求。 -- **主动告警**：[Token](../concepts/token.md) 突增、超时率升高时自动通知运维人员。 - -### 模型生产适用场景 - -- **领域定制**：通用模型无法满足垂直场景时，通过微调注入专业知识。 -- **成本优化**：对定制模型进行量化压缩，降低推理资源占用。 -- **服务化交付**：将训练好的模型部署为在线推理端点，供业务应用调用。 - -## 技术选型参考 - -| 开发者需求 | 推荐能力 | 说明 | -| --- | --- | --- | -| 需要知道哪个模型效果好 | 模型评测 | 创建自定义评测任务或基线评测进行量化对比 | -| 需要定制一个专属模型 | 模型生产 | 调优 → 可选压缩 → 部署为推理服务 | -| 模型已上线，需要保障稳定运行 | 模型监控 | 开启高级监控 + 告警规则 | -| 调优后想验证效果 | 模型评测 + 模型生产 | 先用模型生产完成调优，再用模型评测量化对比 | -| 线上模型出现异常，需要定位原因 | 模型监控 | 查看调用日志和失败详情排查 | -| 想在 Grafana 中统一看板 | 模型监控 | 接入高级监控的 Prometheus HTTP API | - -## 三者协作关系 - -在实际项目中，三大能力通常按以下流程串联使用： - -1. **模型生产**：基于基础模型微调出领域定制模型，压缩后部署上线。 -2. **模型评测**：部署前用评测任务验证定制模型的质量是否达标。 -3. **模型监控**：部署后持续观测运行状态，设置告警确保服务可靠性。 - -当监控发现质量劣化时，可回到模型评测进行定量验证，确认后再通过模型生产重新调优迭代，形成闭环。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) -- [model production](../api/model-production.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md deleted file mode 100644 index 1b246309..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-monitoring-vs-application-monitoring.md +++ /dev/null @@ -1,54 +0,0 @@ -# 模型监控 vs 应用监控 - -百炼平台提供两种互补的观测能力：**模型监控**关注单个模型 API 调用层面的性能与成本，**应用观测**则从应用维度端到端追踪内部调用链路。理解二者的定位差异，有助于开发者在不同阶段选择合适的监控手段，实现从模型选型到应用上线的全链路可观测。 - -## 关键维度对比 - -| 维度 | 模型监控 | 应用监控（应用观测） | -|------|----------|----------------------| -| 监控对象 | 单个模型的 API 调用 | 整个应用（智能体/工作流/高代码应用）的完整调用链路 | -| 核心关注点 | 模型性能、[Token](../concepts/token.md) 消耗、费用趋势 | 应用内部节点间的执行流程、端到端延时 | -| 支持范围 | 所有模型（含自定义调优模型） | 智能体应用、工作流应用、高代码应用 | -| 地域限制 | 高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | 无特殊地域限制（需开通 OpenTelemetry 服务） | -| 典型指标 | 调用时长、首 [Token](../concepts/token.md) 延时、RPM/TPM、失败率、[Token](../concepts/token.md) 消耗 | 调用次数、失败率、Token 总量、平均首 Token 耗时、平均调用时长 | -| 数据粒度 | 普通监控小时级；高级监控分钟级 | 分钟级更新 | -| 数据保留 | 用量统计 30 天 | 调用记录最长 30 天 | -| 告警能力 | 支持（短信/邮件/电话/钉钉/企微/Webhook） | 不支持（需结合模型监控告警） | -| 外部集成 | Prometheus HTTP API + Grafana（Basic Auth） | OpenTelemetry 服务存储 | -| 日志能力 | 推理日志（输入/输出/Token 逐条记录） | Trace/Span 链路日志（含节点层级关系） | -| 筛选维度 | API-KEY、推理类型、时间范围、时间精度 | Request ID/Trace ID/Span ID、状态、Span Name、延时、Token 量、标签 | -| 数据标注 | 不支持 | 支持对 Span 添加标签（布尔/分类/数字/文本） | -| 评测集联动 | 不支持 | 支持将 Span 数据导入评测集 | -| 费用 | 免费（高级监控需开启） | 功能免费，存储费用由 OpenTelemetry 服务收取 | -| API 支持 | Prometheus HTTP API 可编程查询 | 仅控制台操作，无独立 API | - -## 适用场景 - -### 模型监控适合 - -- **模型选型与调优阶段**：对比不同模型的调用时长、首 Token 延时、成本效益比，为技术选型提供数据依据。 -- **成本管控**：按[业务空间](../concepts/workspace.md)维度统计 Token 消耗，配置免费额度用尽即停策略，避免超支。 -- **稳定性保障**：通过告警规则监控失败率、限流错误，在异常发生时第一时间通知相关人员。 -- **运维自动化**：利用 Prometheus API 接入自建监控系统或 Grafana 大盘，实现统一运维。 - -### 应用监控适合 - -- **应用调试与优化**：通过 Trace 链路追踪定位应用内部的性能瓶颈节点（如检索耗时、模型推理耗时）。 -- **智能体/工作流行为分析**：查看 AGENT、RETRIEVER、LLM、TOOL 等节点的执行顺序和中间结果，理解应用的实际行为。 -- **数据驱动迭代**：将线上真实调用数据标注后导入评测集，形成"观测-标注-评测-优化"闭环。 -- **问题排查**：按 Request ID 精确定位单次调用的完整执行链路，快速定位报错节点。 - -## 技术选型建议 - -1. **两者并非互斥**：生产环境建议同时开启。模型监控提供宏观的性能告警与成本管控，应用观测提供微观的链路诊断能力。 -2. **开发调试阶段**优先使用应用观测，快速定位逻辑问题和性能瓶颈。 -3. **上线运营阶段**优先配置模型监控告警，保障服务稳定性和成本可控。 -4. **高代码应用**目前应用观测仅支持入口节点，内部链路需依赖代码中集成 Tracing 模块并启用 `--telemetry enable` 参数。 -5. 如需将监控数据接入第三方平台，模型监控通过 Prometheus API 对接，应用观测数据则存储在 OpenTelemetry 服务中。 - -## 被对比主题页 - -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md deleted file mode 100644 index 31515cc0..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-compare.md +++ /dev/null @@ -1,42 +0,0 @@ -# 模型微调、压缩与高速推理对比 - -在百炼平台上，把一个模型从"能用"打磨到"好用、省钱、够快"，通常会经历三类相互独立又可串联的能力：**模型微调（Fine-tuning）** 把领域知识、任务能力与人类偏好写入参数；**模型压缩** 通过量化在保持能力的前提下降低部署规格与成本；**高速推理** 则在调用侧解决容量兑付与输出速度问题。三者定位不同——微调改变"模型本身"，压缩改变"部署精度/成本"，高速推理改变"调用时的容量与吞吐"，因此常见的完整链路为：模型调优 → 模型压缩（可选）→ 模型部署 → 高速推理（可选）。本文面向开发者，从关键维度对比三者，帮助做技术选型。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（量化） | 高速推理（TPM 预留 / 快速模式） | -| --- | --- | --- | --- | -| 解决的问题 | 深度定制：注入领域知识、指令遵循、偏好对齐、特定音色/风格 | 降低部署所需 MU 规格，减少推理成本 | 容量刚性兑付（预留）/ 提升单请求输出速度（快速模式） | -| 处理对象 | 基础模型 + 训练数据集 | 平台微调产出的全精度自定义模型 | 线上可调用的模型（含专属 code / fast 模型） | -| 输入格式 | 数据集：SFT 用 ChatML `{"messages":[...]}`；DPO 追加 `chosen`/`rejected`；CPT 用 `{"text":"..."}`；视觉 ZIP+`data.jsonl`；CosyVoice `{"wav_fn","text"}` | 源模型 + 量化模板（+ 可选校准数据集，最多 5 个） | 标准对话请求（`messages`）；预留替换 `model` 为专属 code；快速模式指定 fast 模型 ID | -| 输出产物 | 微调后模型（`finetuned_output` / `deployed_model`） | 低精度量化模型（新后缀，更小部署规格） | 无新模型产物，仅提升调用容量/速度 | -| 支持模型 | 千问文本/VL、万相图像/视频、CosyVoice 等多模态 | 仅平台微调产出的自定义模型（如 qwen3.5-flash 系列），不支持基础/第三方模型 | 预留：已开放预留的模型；快速模式：`glm-5.2-fast-preview` | -| 训练/处理方式 | CPT、SFT（全参/高效 LoRA）、DPO（全参/LoRA） | 量化（不含剪枝、蒸馏），MU 编号越大规格越小 | 预留：锁定 kTPM 吞吐；快速模式：提速至 1.5~2 倍 TPS | -| API 端点 / 接入 | `POST /api/v1/files`、`/fine-tunes`、`/deployments`（华北2 API Key） | 主要通过控制台创建压缩任务（模型训练 → 模型压缩） | 预留：标准 `dashscope.aliyuncs.com` + 专属 code；快速模式：`{workspace_id}.cn-beijing.maas.aliyuncs.com` 专属域名 | -| 计费方式 | 按 Token（API 创建仅支持按 Token）或训练单元（仅控制台）；成本与耗时较高 | 压缩任务限时免费；成本体现在部署阶段按 MU 规格计费（示例节省约 56%） | 预留：按 kTPM 预付费，超额降级按量；快速模式：按 token（同标准 API），超额排队 | -| 地域限制 | 仅华北2（北京） | 仅华北2（北京） | 预留：以控制台为准；快速模式：华北2（北京）、新加坡 | -| 可逆性 / 约束 | 可继续叠加训练（CPT→SFT→DPO） | 不可逆；压缩后不支持继续微调或二次压缩 | 预留退订/快速模式为 preview，规格可能变动 | -| 代码改动量 | 大：准备数据集、走训练→部署全流程 | 小：控制台配置任务，调用无需改代码 | 小：预留替换 `model`；快速模式换域名 + fast 模型 ID | -| 典型场景 | 领域问答、安全合规对齐、专属音色/图像风格 | 微调模型上线前的成本优化 | 高峰期专属容量保障；对输出速度敏感的编程助手/Agent/实时对话 | - -## 各方案适用场景建议 - -- **模型微调**：当 Prompt 工程、插件调用、RAG 等手段仍无法达到效果，需要把领域知识或人类偏好"写进参数"时选用。文本生成推荐按 `CPT（可选）→ SFT → DPO（可选）` 递进组合；对成本/时间敏感或数据集较小时用 LoRA 高效训练，追求效果且模型支持时优先全参。多模态（VL、万相、CosyVoice）按各自支持的方式与超参集处理。注意训练成本与耗时较高，且能力仅限华北2（北京）。 -- **模型压缩**：已有微调产出的全精度自定义模型、且部署成本偏高时选用。因压缩任务当前限时免费，建议在免费期内对同一模型尝试多个量化模板，分别部署后用业务测试集验证精度与成本的平衡，再选最优方案上线。切记压缩不可逆，需保留上游全精度模型以便重新压缩。 -- **高速推理 - TPM 预留**：面向高峰期需要"专属容量、不被公共限流"的生产业务，按 kTPM 预付费锁定吞吐，超额自动降级按量、不中断，适合对稳定性和容量兑付要求高的场景。 -- **高速推理 - 快速模式**：面向 AI 编程助手、Agent 多步推理、实时对话等对**输出速度**敏感的场景，TPS 可达标准 API 的 1.5~2 倍。当前仍为 preview，规格可能调整，生产接入前需评估稳定性；接入需使用专属域名，不能与预留的标准域名接入方式混用。 - -## 技术选型参考 - -1. **先定位需求属于哪一层**：要"改变模型能力/风格"→微调；要"降部署成本"→压缩；要"保容量或提速度"→高速推理。三者可组合：微调 → 压缩 → 部署 → 叠加 TPM 预留或快速模式。 -2. **成本与投入权衡**：微调投入最大（数据、训练时间、Token/训练单元费用）；压缩几乎零调用改造且当前限时免费，收益在部署阶段；高速推理中预留是预付费换刚性容量，快速模式按量计费换速度。 -3. **可逆性与前置条件**：压缩不可逆且要求上游为平台微调模型；微调与压缩均仅限华北2（北京）；快速模式为 preview 且模型/地域受限。 -4. **接入方式差异**：预留沿用标准域名 + 专属 model code；快速模式改用 workspace 专属域名 + fast 模型 ID，两者不可混用，选型时需评估现有调用代码的改造范围。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md deleted file mode 100644 index 47044f8e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-comparison.md +++ /dev/null @@ -1,85 +0,0 @@ -# 模型微调、压缩与部署对比 - -在百炼平台上，将一个基础模型转化为可在生产环境中使用的定制化推理服务，通常涉及三个阶段：**模型微调（Fine-tuning）**、**模型压缩（Quantization）** 和 **模型部署（Deployment）**。三者在模型生产链路中依次衔接（调优 → 压缩（可选）→ 部署），各自解决不同层面的问题。本文从功能定位、适用范围、操作方式和成本等维度进行系统对比，帮助开发者在技术选型时做出合理决策。 - -## 功能定位与链路关系 - -模型微调、压缩与部署构成一条完整的模型生产流水线： - -- **模型微调**处于链路上游，目标是基于自有数据定制模型能力，使模型在特定领域或任务上表现更优。 -- **模型压缩**处于链路中间，是可选环节，通过量化技术降低模型参数精度，从而减小部署所需的算力规格和推理成本。 -- **模型部署**处于链路下游，将训练好（或压缩后）的模型发布为在线推理服务，供应用调用。 - -三者缺一不可地覆盖了"训练 → 优化 → 上线"的完整生命周期，但各自的输入输出、关注点和约束条件差异显著。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine-tuning） | 模型压缩（Quantization） | 模型部署（Deployment） | -|------|------------------------|------------------------|----------------------| -| **核心目标** | 基于自有数据定制模型能力，提升特定场景表现 | 降低模型参数精度，减小部署规格与推理成本 | 将模型发布为在线推理服务，供应用调用 | -| **在链路中的位置** | 上游（第一步） | 中间（可选） | 下游（最后一步） | -| **输入** | 基础模型 + 训练数据集（JSONL / ZIP 等） | 微调产出的自定义模型 | 预置模型 / 微调模型 / 压缩后模型 | -| **输出** | 微调后的自定义模型 | 低精度量化模型 | 可调用的在线推理服务（API 端点） | -| **支持的模型范围** | 广泛：Qwen3.6/3.5/3/2.5 系列文本模型、Qwen-VL 视觉模型、Wan 图像/视频模型、CosyVoice 语音模型 | 较窄：仅支持百炼平台微调产出的特定自定义模型（如 qwen3.5-flash） | 最广泛：预置模型（Qwen、DeepSeek、GLM 等）+ 微调模型 + 压缩模型 + OSS 导入的 LoRA 模型 | -| **支持的模态** | 文本生成、视觉理解、图像生成、视频生成、语音合成 | 仅文本生成 | 文本生成、[多模态](../concepts/multimodal.md)、语音合成等 | -| **操作方式** | 控制台 + API | 仅控制台 | 控制台 + API | -| **是否可逆** | 可重复训练和迭代 | 不可逆：压缩后不支持继续微调或二次压缩 | 可随时上线/下线，支持重新部署 | -| **典型耗时** | 数小时至数天（取决于数据量和训练轮次） | 平台自动完成，排队后运行 | 分钟级（状态变为"运行中"即可调用） | -| **计费方式** | 按 Token 用量计费：训练数据 Token 数 x 训练轮次 x 训练单价 | 压缩任务本身限时免费 | 三种模式：预置吞吐（PTU）、模型单元（MU，按时长）、Token 用量（按调用量） | -| **地域限制** | 无特殊限制 | 仅华北2（北京） | 无特殊限制 | - -## 训练方法与压缩模板选择 - -### 微调方法 - -百炼平台提供三种递进式微调方法，推荐按 CPT → SFT → DPO 的顺序使用： - -| 方法 | 目标 | 数据要求 | 典型场景 | -|------|------|----------|----------| -| CPT（继续预训练） | 注入领域知识 | 1000万+ Token 无标签文本 | 金融/医疗/法律等垂直领域适配 | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条高质量问答对 | 客服、代码助手、Agent 工具调用 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组正负样本对 | 安全合规强化、降低幻觉 | - -每种方法支持全参训练和高效训练（LoRA）两种模式。全参训练效果更好但耗时更长、成本更高；LoRA 训练收敛快、成本低，适合快速验证。 - -### 压缩量化模板 - -量化模板决定压缩后的部署规格，MU 编号越大表示部署规格越小、成本越低，但精度损失可能越大。可选配校准数据以提升量化精度，建议选择与推理场景语义相近的数据集。 - -## 部署计费方式对比 - -| 计费方式 | 计费公式 | 适用场景 | 特点 | -|---------|---------|---------|------| -| 预置吞吐（PTU） | 使用时长 x (输入 TPM 单价 x 输入 TPM + 输出 TPM 单价 x 输出 TPM) | 流量稳定的高负载生产环境 | 保障吞吐额度内不限速，超额自动切换按量计费；TPS 约为按量的 1.5~2.0 倍 | -| 模型单元（MU） | 使用时长（小时）x 模型单元数量 x 模型单元单价 | 需要资源独占和自定义性能指标 | 支持 PD 分离计算模式，可降低首 Token 延迟 | -| Token 用量 | 输入 Token 数 x 输入单价 + 输出 Token 数 x 输出单价 | 调用量不稳定、用量较少的场景 | 仅支持部分 LoRA 调优模型，不使用不计费 | - -## 适用场景建议 - -**场景一：快速验证模型定制效果** -推荐路径：SFT 高效训练（LoRA）→ 直接部署（Token 用量计费）。跳过压缩环节，以最低成本快速上线验证。 - -**场景二：生产环境成本敏感** -推荐路径：SFT 全参训练 → 模型压缩 → 部署（MU 计费）。通过压缩降低部署规格（如从 MU1x2 降至 MU8x1，成本节省约 56%），适合长期运行的在线服务。 - -**场景三：高并发低延迟的核心业务** -推荐路径：微调（按需）→ 部署（PTU 计费）。PTU 模式提供预留吞吐保障，TPS 提升约 1.5~2.0 倍，适合流量稳定且对延迟敏感的场景。 - -**场景四：[多模态](../concepts/multimodal.md)模型定制** -推荐路径：视觉/图像/视频/语音 SFT → 直接部署（MU 计费）。当前压缩功能仅支持文本模型，[多模态](../concepts/multimodal.md)微调模型需直接部署。 - -## 技术选型要点 - -1. **是否需要微调**：如果预置模型已满足需求，可直接部署，无需微调。当模型在特定领域表现不佳、需要定制输出格式或降低幻觉时，再考虑微调。 -2. **是否需要压缩**：压缩可显著降低部署成本，但会带来一定精度损失且不可逆。建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案。 -3. **如何选择部署计费方式**：流量稳定选 PTU，需要资源独占和灵活配置选 MU，用量少且不稳定选 Token 用量。注意 Token 用量模式仅支持部分 LoRA 模型。 -4. **地域约束**：模型压缩当前仅在华北2（北京）地域可用，规划链路时需考虑地域一致性。 -5. **不可逆操作提醒**：压缩后的模型不支持继续微调或二次压缩，务必保留上游全精度微调模型以备后续迭代。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md deleted file mode 100644 index bb5c6450..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-optimization-methods-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 模型优化方式对比（微调/压缩/高速推理） - -百炼平台围绕"让模型更贴合业务、跑得更省、调得更稳"提供了三类模型优化能力：**模型微调（Fine Tuning）**、**模型压缩（Model Compression）**、**高速推理（TPM 预留）**。三者作用阶段不同、目标不同，但在生产落地中常被串联使用：先用微调注入业务能力，再用压缩降低部署规格，最后用高速推理保障线上吞吐。本页从输入输出、支持模型、API/控制台入口、[计费](../concepts/billing.md)方式、典型场景等维度横向对比，供开发者在技术选型时参考。 - -## 关键维度对比 - -| 维度 | 模型微调（Fine Tuning） | 模型压缩（Model Compression） | 高速推理（TPM 预留） | -| --- | --- | --- | --- | -| 一句话定位 | 把业务/场景知识写入模型参数 | 把全精度微调模型量化为低精度版本 | 为指定模型锁定专属吞吐量 | -| 优化阶段 | 训练阶段（改变模型权重） | 部署前阶段（不改权重，只做量化） | 服务运行阶段（不动模型，只保容量） | -| 是否改变模型 | 是，产出新的微调模型 | 是，产出量化后的低精度模型 | 否，仅锁定原模型的推理容量 | -| 前置条件 | 已开通百炼模型服务，准备好训练数据 | 当前工作空间内已有支持压缩的自定义微调模型 | 已开通百炼模型服务，已创建业务空间 | -| 输入格式 | 文本：JSONL 问-答对；图像/视频：`.zip`（含 `data.jsonl` + 训练样本）；语音：录音文件 | 全精度微调模型 + 校准数据集（部分模板要求） | 业务流量预估（RPM、平均输入/输出长度、缓存命中率） | -| 输出格式 | 新的自定义微调模型（`finetuned_output`） | 量化后的低精度模型（部署到更小规格单元） | 专属 model code（替换 API 的 `model` 参数即可） | -| 支持模型 | 千问系列（文本/视觉）、万相 wan2.7/wan2.2/wan2.5（图像/视频）、CosyVoice 语音 | 当前支持 `qwen3.5-flash-2026-02-23` 等基础模型对应的自定义微调模型（以控制台为准） | 按区域开放，如 qwen3.7-max、glm-5.2、kimi-k2.6、deepseek-v4-pro 等（以控制台为准） | -| 控制台入口 | 「模型调优」页面 | 「模型 > 模型训练 > 模型压缩」页面 | 控制台「创建 TPM 预留」 | -| API 端点 | `POST /api/v1/files`、`POST /api/v1/fine-tunes`、`GET /api/v1/fine-tunes/{job_id}`、`POST /api/v1/deployments` | OpenAPI（`custom_calibration_file_ids` 等字段） | 替换 `model` 参数为专属 model code，调用 OpenAI 兼容模式 `/compatible-mode/v1/chat/completions` | -| [计费](../concepts/billing.md)方式 | 按 Token [计费](../concepts/billing.md)（API 创建的任务）；控制台创建可使用模型训练单元（预付费/后付费） | 当前限时免费；产出模型部署上线后按部署单元规格计费 | 按 kTPM 预付费，输入/输出 TPM 分别计价，按天计费；预留内调用不额外收费，超额自动降级按量 | -| 任务状态 | `PENDING` → `SUCCEEDED`；部署 `RUNNING` | 待开始/排队中/运行中/停止中/压缩成功/压缩失败/已取消 | 运行中/待生效/变配中/已停止/已过期/已取消 | -| 是否可逆 | 可迭代训练（CPT→SFT→DPO 递进） | 不可逆：不支持继续训练，不支持二次压缩 | 可扩缩容、续费、退订；退订不可恢复 | -| 典型场景 | 注入领域知识、对齐偏好、复刻风格、压低延迟、抑制幻觉 | 降低显存占用、部署到更小规格、降低部署成本、提升吞吐 | 流量可预估且不能接受限流、业务高峰期保障容量 | - -## 各方案适用场景建议 - -### 模型微调（Fine Tuning） - -适合"通用模型答不好、Prompt 工程已到瓶颈"的场景。当你需要把领域术语/事实、特定对话格式、人类偏好、特定风格写入模型本身时，应选微调： - -- **补知识**：CPT 注入领域文本（1000 万+ Token）。 -- **学做事**：SFT 教会指令遵循（1000+ 问-答对），文本/视觉/图像/视频/语音[多模态](../concepts/multimodal.md)均支持。 -- **做更好**：DPO 对齐人类偏好（100+ 组偏好对）。 -- **快速验证或数据量小**：优先用 LoRA 高效训练，速度快、成本低。 -- CosyVoice 调优仅支持 API 发起，且仅支持 SFT 高效微调。 - -### 模型压缩（Model Compression） - -适合"已经微调出好模型，但部署太贵/规格太大"的场景。它不改变模型能力定位，只做量化降本： - -- 前提是上游已产出全精度微调模型（压缩是微调的下游）。 -- 通过量化模板决定压缩后可部署规格（如 MU5、MU8），把 `MU1 * 2（¥108/小时）` 降为 `MU8 * 1（¥47/小时）`。 -- 适合追求更低部署成本、更高推理吞吐、更小部署单元的线上服务。 -- 注意：压缩产物不可继续训练、不可二次压缩，迭代须回到全精度模型重训。 - -### 高速推理（TPM 预留） - -适合"流量可预估、不能接受公共限流、要保障高峰期容量"的场景。它不动模型本身，只锁定推理吞吐： - -- 业务高峰期不能因公共池限流而抖动。 -- 需要 SLA 刚性兑付、专属容量不与他人共享。 -- 超额自动降级按量计费且不中断服务，适合可预估但偶发突刺的流量。 -- 接入极简：仅替换 `model` 参数为专属 model code，无需改代码逻辑。 - -## 技术选型建议 - -三者并非互斥，建议按"能力注入 → 规格压缩 → 容量保障"的顺序串联使用： - -1. **先微调**：用业务数据训练出符合场景的自定义模型（解决"答得好不好"）。 -2. **再压缩**：对微调产物做量化，部署到更小规格单元（解决"跑得省不省"）。 -3. **最后保容量**：用 TPM 预留锁定线上吞吐，保障高峰期稳定（解决"调得稳不稳"）。 - -如果只是临时验证或流量很小，可跳过压缩与预留，直接按量调用微调后的模型；如果是对外承诺 SLA 的生产服务，建议三步全走。若仅需保障容量而模型本身已满足需求，可单独使用 TPM 预留；若仅需降本而流量平稳，可单独使用模型压缩。 - -## 被对比主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model compression](../guides/model-compression.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md deleted file mode 100644 index 7cd40cf3..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-app-vs-model-comparison.md +++ /dev/null @@ -1,51 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控能力：**应用监控（应用观测）**聚焦于应用层端到端调用链路，**模型监控**聚焦于模型层用量与运行状态。两者观测对象、指标口径、数据时效、地域支持各不相同，开发者需根据排查目标（应用行为 vs 模型行为）选择合适的能力，或组合使用以实现从业务入口到模型底层的全链路可观测。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 业务空间内的应用（智能体应用、工作流应用、高代码应用） | 主账号下所有业务空间内的模型调用（按"模型 + 业务空间"维度） | -| 观测粒度 | 应用内部调用链路（CHAIN/AGENT/RETRIEVER/LLM/TOOL 等节点，支持嵌套） | 模型维度的聚合指标（调用次数、失败率、延时、Token、RPM/TPM 等） | -| 关键指标 | 延时、Token 量、调用记录（Prompt/输出）、CHAIN 节点状态 | 安全/成本/性能/错误四类：失败率、429 限流、首 Token 延时、Token 消耗等 | -| 数据时效 | 分钟级同步，调用记录最长可查 30 天 | 普通监控小时级、高级监控分钟级；用量数据延迟约 1 小时 | -| 模型范围 | 随应用调用自动覆盖应用内使用的模型 | 普通监控支持所有地域所有模型；高级监控仅北京/新加坡/弗吉尼亚；告警仅北京/新加坡 | -| 日志/历史对话 | 在 Trace 详情中查看 Prompt 内容、输出、Span 原始数据 | 仅华北2（北京）地域部分模型支持请求和响应日志（需开通推理日志） | -| 数据筛选 | Span 模式（Root/All/Model Span）+ 状态/Span Name/延时/Token/标签等多条件 | 按 API-KEY、推理类型、时间范围、时间精度筛选；失败详情可点击查看 | -| 数据导出 | 支持 JSONL / EXCEL 导出 | 高级监控指标存储于私有 Prometheus，支持标准 Prometheus HTTP API，可接入 Grafana | -| 主动告警 | 不支持告警，仅控制台查看 | 支持告警规则（短信/邮件/电话/钉钉/企微/Webhook），按 CRITICAL/ERROR/WARNING/INFO 分级 | -| 计费方式 | 应用观测功能本身不收费；数据存储费由 OpenTelemetry 服务收取 | 监控功能本身不收费；用量按 Token/张/秒计费，存储于云监控 Prometheus | -| 开通方式 | 控制台"应用观测配置"：授权 OTel 角色 → 开通 OTel 服务 → 初始化 LogStore | 控制台"模型监控配置"：开通审计/推理日志、高级监控、告警规则 | -| API 支持 | 无 API，仅控制台操作 | 高级监控提供 Prometheus HTTP API，可接入自建应用 | -| 标注与[评测](../concepts/evaluation.md) | 支持对 Span 加标签（布尔/分类/数字/文本），可加入[评测](../concepts/evaluation.md)集 | 不直接支持标注，但 Token 追踪与告警可辅助[评测](../concepts/evaluation.md)样本筛选 | - -## 适用场景建议 - -### 选用应用监控（应用观测） - -- 需要追踪智能体应用、工作流应用内部端到端调用链路（如检索 → 改写 → LLM → 工具调用）。 -- 排查应用层延时异常、Token 消耗突增的根因节点（定位到具体 Span）。 -- 希望将真实线上调用作为评测样本导入评测集。 -- 应用部署在任意地域，且不依赖主动告警（仅需控制台查询）。 -- 限制：不支持 Assistant API 创建的智能体应用；高代码应用仅能观测入口 CHAIN 节点。 - -### 选用模型监控 - -- 需要按业务空间/模型维度汇总调用量、失败率、延时、Token 消耗，做成本核算与稳定性保障。 -- 需要 429 限流、内容安全错误等模型层错误的明细与趋势分析。 -- 需要主动告警（超时、Token 突增等静默失败），或接入 Grafana/自建看板。 -- 需要 Token 消耗的汇总、追踪、阈值告警三层成本管理。 -- 需要查看请求和响应原文（仅北京地域部分模型，需开通推理日志）。 -- 限制：用量统计不支持按阿里云账号维度汇总；日志/告警有地域限制。 - -### 组合使用 - -当问题既涉及应用行为又涉及模型底层时，建议先用应用监控定位异常 Span（如某个 LLM 节点延时高或失败），再通过 Request ID / Trace ID 关联到模型监控中的对应调用记录，结合模型层失败率、限流、Token 消耗做联合诊断。应用监控提供"是什么调用出了问题"，模型监控提供"模型侧为什么会出问题"，二者互补构成完整的可观测体系。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md deleted file mode 100644 index d9d02b20..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-compare.md +++ /dev/null @@ -1,60 +0,0 @@ -# 应用监控与模型监控对比 - -阿里云百炼平台提供了两套互补的可观测能力：**应用监控（应用观测）**面向由智能体、工作流、高代码搭建的完整应用，追踪端到端的调用链路；**模型监控**面向底层模型调用本身，聚焦性能指标、Token 消耗与费用趋势。二者关注的抽象层次不同——前者回答"这个应用内部发生了什么"，后者回答"这个模型跑得怎么样、花了多少钱"。本文对比两者的关键差异，帮助开发者在做可观测性方案选型时快速定位合适的工具。 - -## 关键维度对比 - -| 维度 | 应用监控（应用观测） | 模型监控 | -| --- | --- | --- | -| 观测对象 | 智能体应用、工作流应用、高代码应用的端到端调用链路 | 模型调用本身（含调优后的自定义模型） | -| 核心价值 | 追踪应用内部调用链路、节点耗时与思考过程 | 追踪模型性能、Token 消耗与费用趋势 | -| 支持范围 | 三类应用；不支持 Assistant API 创建的智能体；高代码仅观测入口 CHAIN 节点 | 用量统计支持全部模型；高级监控仅限北京/新加坡/弗吉尼亚；告警仅限北京/新加坡 | -| 数据延迟 | 分钟级 | 普通监控小时级、高级监控分钟级；用量统计约 1 小时 | -| 数据留存 | 调用记录最长 30 天 | 用量统计不支持查看 30 天以前数据 | -| 指标维度 | 调用次数/失败率、Token 量、平均首 Token 耗时、平均调用时长 | 安全、成本、性能（RPM/TPM 等）、错误四大类 | -| 筛选维度 | Request ID / Trace ID / Span ID、状态、Span、输入输出、延时、Token、标签 | API-KEY、推理类型（实时/批量）、时间范围、时间精度 | -| 链路/节点粒度 | 提供 CHAIN/AGENT/RETRIEVER/LLM/TOOL 等丰富节点类型与嵌套关系 | 无调用链路，按模型/API-KEY 维度聚合指标 | -| 告警能力 | 无内置告警 | 支持告警规则（短信/邮件/电话/钉钉/企业微信/Webhook），分 CRITICAL/ERROR/WARNING/INFO 等级 | -| 数据标注/评测 | 支持 Span 打标签、直接加入评测集 | 不涉及 | -| 日志能力 | 可查看 Prompt、输出、延时、Token 等调用记录 | 推理日志/审计日志，仅华北2（北京）部分模型，分钟级延迟 | -| 外部系统接入 | 数据导出为 JSONL / EXCEL | 高级监控数据存于私有 Prometheus，支持标准 HTTP API，可接 Grafana（Basic Auth） | -| 使用方式 | 仅控制台操作，无 API | 控制台面板 + Prometheus HTTP API | -| 计费方式 | 功能本身免费，观测数据存储由 OpenTelemetry 服务收费 | 监控功能查看无额外说明；聚焦模型调用本身的 Token/费用管理 | -| 前置开通 | 需授权 OpenTelemetry 角色、开通服务、初始化 LogStore | 高级监控/告警需在模型监控配置中开启；日志需开通审计与推理日志 | -| 典型场景 | 调试应用逻辑、定位慢节点、优化检索/Prompt、构建评测样本 | 成本核算、性能容量规划、异常主动告警、故障排查 | - -## 适用场景建议 - -### 优先选择应用监控（应用观测） - -- 应用由智能体、工作流或高代码搭建，需要**看清内部调用链路**（哪个节点慢、检索命中如何、模型思考过程）。 -- 需要按 Trace ID / Span ID **下钻单次请求**，排查复杂多节点应用的逻辑问题。 -- 希望把**真实线上调用直接沉淀为评测样本**，或对 Span 数据做标注管理。 -- 只在控制台侧使用、可接受分钟级更新、调用记录 30 天内可回溯。 - -### 优先选择模型监控 - -- 关注的是**模型层面的性能与成本**：RPM/TPM、失败率、限流、Token 消耗与费用趋势。 -- 需要**主动告警**（限流、失败率飙升、成本超标），并通过短信/电话/钉钉等渠道通知。 -- 需要**按业务空间精细化管理模型成本**，结合免费额度"用完即停"控制预算。 -- 希望把监控数据**接入 Grafana 或自建系统**做统一可视化（依赖高级监控的 Prometheus 接口）。 -- 需要通过推理日志做**内容审计或故障排查**（注意仅限北京地域部分模型）。 - -## 技术选型参考 - -两套能力并非互斥，成熟的生产系统通常**同时启用**： - -1. **分层定位问题**：先用模型监控发现"某模型失败率/延时异常"，再用应用监控下钻到具体应用与节点，确认是检索、插件还是模型环节导致。 -2. **注意地域限制**：模型高级监控与告警对地域敏感（高级监控限北京/新加坡/弗吉尼亚，告警限北京/新加坡，推理日志限北京）。若业务不在这些地域，应用监控的分钟级链路观测可作为主要抓手。 -3. **权衡数据延迟**：应用监控与模型高级监控均为分钟级；模型普通监控与用量统计为小时级，不适合实时排障，更适合趋势分析与成本复盘。 -4. **成本可观测归口**：需要账单/额度维度治理时以模型监控（用量统计 + 免费额度）为准；需要单次请求成本归因时用应用监控的 Token 量指标。 -5. **数据出口差异**：需要把数据导入外部 BI/表格用应用监控的 JSONL/EXCEL 导出；需要接入监控告警平台（Grafana/Prometheus）走模型高级监控的 HTTP API。 - -简言之：**排查"应用怎么跑的"用应用监控，管理"模型跑得好不好、花多少钱"用模型监控**，两者组合可覆盖从链路调试到成本告警的完整可观测闭环。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md deleted file mode 100644 index 3e3d1202..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/monitoring-comparison.md +++ /dev/null @@ -1,57 +0,0 @@ -# 应用监控与模型监控对比 - -百炼平台提供两套互补的监控体系：**应用观测**和**模型监控**。应用观测侧重于端到端追踪应用内部的调用链路与节点级性能，帮助开发者定位应用层面的延时与逻辑问题；模型监控则聚焦于模型维度的调用性能、Token 消耗、费用趋势与告警，帮助开发者管控模型使用成本与稳定性。理解两者的定位差异，有助于开发者建立完整的可观测性方案。 - -## 核心维度对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **监控对象** | 应用（智能体应用、工作流应用、高代码应用） | 模型（含自定义模型和调优模型） | -| **观测粒度** | 节点级（CHAIN、LLM、RETRIEVER、TOOL 等多种节点类型） | 模型级（按模型名称、API-KEY、推理类型筛选） | -| **核心指标** | 调用延时、Token 用量（输入/输出）、首 Token 耗时、调用次数与失败率 | 调用时长、首 Token 延时、RPM、TPM、Token 消耗、失败率、限流错误、内容安全错误 | -| **数据更新频率** | 分钟级 | 普通监控为小时级；高级监控为分钟级 | -| **数据保留时长** | 最长 30 天 | 用量统计不支持查看 30 天以前的数据 | -| **告警能力** | 无 | 支持告警规则配置，通知方式含短信、邮件、电话、钉钉机器人、企业微信机器人、Webhook | -| **用量统计** | 无独立用量统计，通过监控统计图表查看 Token 总量 | 按[业务空间](../concepts/workspace.md)维度统计，支持免费额度管理与用完即停 | -| **数据导出** | 支持导出为 JSONL 或 EXCEL | 高级监控数据可通过 Prometheus HTTP API 接入 Grafana 等外部系统 | -| **数据标注** | 支持对 Span 添加标签（布尔值/分类/数字/文本），可关联评测集 | 不支持 | -| **日志查看** | 通过 Trace 详情查看 Prompt 输入输出与原始数据 | 开通推理日志后可查看每次调用的输入、输出及 Token 消耗（仅华北2北京地域部分模型） | -| **地域限制** | 无特殊地域限制 | 高级监控仅支持北京、新加坡、弗吉尼亚；告警仅支持北京、新加坡 | -| **计费** | 功能免费，观测数据存储由 OpenTelemetry 服务收费 | 功能免费，高级监控数据存储在私有 Prometheus 实例中 | -| **操作方式** | 仅控制台操作，无 API | 控制台操作 + Prometheus API 接入 | - -## 开通与前置条件对比 - -| 维度 | 应用观测 | 模型监控 | -|------|----------|----------| -| **开通步骤** | 授权 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore | 普通监控默认可用；高级监控需在模型监控配置中手动开启 | -| **子账号权限** | 需要 AliyunBailianFullAccess + 应用观测页面权限 + ram:CreateServiceLinkedRole 策略 | 标准百炼控制台权限即可 | -| **推荐操作账号** | 主账号 | 无特殊要求 | - -## 适用场景建议 - -### 应用观测适合以下场景 - -- **调用链路排查**：应用响应慢或出错时，需要逐节点定位瓶颈，例如区分是检索环节还是模型推理环节导致延时过高。 -- **Prompt 调试**：查看每次调用的完整输入输出，对比不同 Prompt 的效果。 -- **数据质量管理**：通过 Span 筛选与标注功能，对线上真实调用数据进行质量打分，并将优质样本导入评测集。 -- **工作流应用调试**：工作流包含多种节点类型（意图分类、脚本转换、条件判断等），需要观察每个节点的执行情况。 - -### 模型监控适合以下场景 - -- **成本管控**：按[业务空间](../concepts/workspace.md)统计模型用量与费用，配合免费额度管理控制预算。 -- **稳定性保障**：配置告警规则，在失败率上升或限流异常时及时收到通知。 -- **性能基线建立**：通过 RPM、TPM、首 Token 延时等指标建立性能基线，持续跟踪模型表现。 -- **多模型对比**：对比不同模型在相同业务场景下的调用时长、Token 消耗等指标，辅助模型选型。 -- **外部可视化集成**：将监控数据接入 Grafana 等系统，构建统一的运维大盘。 - -### 建议组合使用 - -在生产环境中，推荐同时启用两套监控：用模型监控建立全局的成本与稳定性视图并配置告警，用应用观测在出现异常时深入排查具体调用链路。两者从不同维度覆盖可观测性需求，互为补充而非替代。 - -## 被对比主题页 - -- [application monitoring](../guides/application-monitoring.md) -- [model monitoring](../guides/model-monitoring.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md deleted file mode 100644 index a7e7444d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-api-comparison.md +++ /dev/null @@ -1,76 +0,0 @@ -# [多模态](../concepts/multimodal.md)生成 API 对比（图像/视频/3D） - -百炼平台提供图像、视频、3D 三类[多模态](../concepts/multimodal.md)生成 API，分别面向不同的内容产出形态。三者都通过 DashScope HTTP 接口调用，统一使用 API Key 鉴权，并遵循"创建任务 → 轮询结果"的异步任务模式（部分图像模型支持同步调用）。本页从输入格式、输出格式、支持模型、API 端点、调用模式、[计费](../concepts/billing.md)与典型场景等维度做横向对比，帮助开发者根据产出目标与技术约束做选型。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D 生成 | -| --- | --- | --- | --- | -| 产出形态 | 静态图片（PNG） | 视频文件 | GLB 模型 + 预览渲染图 | -| 输入格式 | 文本、图像（图生图/编辑）、参考图 | 文本、图像（首帧/首尾帧）、参考图、视频、音频 | 文本、单图、多图（前/左/后/右 4 视角，固定数组长度 4） | -| 输出格式 | PNG，1–6 张或多图组图 | 视频 URL | PBR 材质 GLB（`pbr_model_url`）或无贴图基础模型（`base_model_url`），含 1 张预览渲染图 | -| 调用模式 | 同步（千问/万相2.6+/Z-Image 等新版）或异步（V1 及部分编辑/创意类） | 仅异步 | 仅异步 | -| API 端点 | 同步：`POST /api/v1/services/aigc/multimodal-generation/generation`；异步轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/video-synthesis`（部分走 `image2video/video-synthesis`）；轮询：`GET /api/v1/tasks/{task_id}` | `POST /api/v1/services/aigc/video-generation/3d-generation`；轮询：`GET /api/v1/tasks/{task_id}` | -| 必需请求头 | `Authorization`；异步需 `X-DashScope-Async: enable` | `Content-Type`、`Authorization`、`X-DashScope-Async: enable` | `X-DashScope-Async: enable`（缺少报 `current user api does not support synchronous calls`） | -| 典型耗时 | 同步秒级返回；异步 1–2 分钟 | 1–5 分钟，万相2.1 视频编辑 5–10 分钟 | 较长，轮询建议间隔约 15 秒 | -| task_id 有效期 | 24 小时 | 24 小时 | 24 小时，超时返回 `UNKNOWN` | -| 产物下载链接有效期 | 随接口返回 | 随接口返回 | 2 小时，需及时下载 | -| 支持模型系列 | 千问图像、万相（Wan/wanx）、Z-Image、可灵 | 万相（HappyHorse/Wan/wanx）、爱诗 PixVerse、Vidu、可灵 | Tripo（`Tripo/Tripo-H3.1` 高精度、`Tripo/Tripo-P1.0` 专业快速） | -| 地域可用性 | 北京/新加坡/弗吉尼亚等多地域，地域独立鉴权不可混用；千问-图像翻译仅北京 | 同地域约束，模型/Endpoint/API Key 必须同地域；PixVerse、Vidu 仅北京 | 仅华北2（北京） | -| 业务空间专属域名 | 支持（`{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` 等） | 支持（北京 `{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 等） | 走默认 dashscope 域名 | -| SDK 支持 | 部分模型支持 DashScope SDK（Python/Java） | HTTP 为主 | HTTP | -| [计费](../concepts/billing.md)方式 | 按张数/模型[计费](../concepts/billing.md) | 按任务/时长计费 | 按任务计费（`usage` 记录任务类型与生成数量） | -| 典型场景 | 文生图、图生图、图像编辑、虚拟模特、试衣、海报、背景生成、擦除补全、画面扩展、人物写真 | 文生视频、图生视频、参考生视频、视频编辑、视频换人、数字人、肖像动态视频 | 文生 3D、单图生 3D、多图生 3D，游戏/电商/工业设计资产 | - -## 调用模式差异 - -三类 API 在调用流程上高度一致，均采用"创建任务 → 轮询查询"模式，但图像 API 额外提供**同步调用**能力： - -- **图像生成**：千问图像系列（qwen-image-2.0-pro/max/plus）、万相 2.6/2.7 文生图与编辑、Z-Image 等新版模型支持一次请求即返回结果的同步调用，走 `multimodal-generation/generation` 端点；V1 版及部分编辑/创意类模型仍需异步。同步模式流程更简单，适合交互式场景。 -- **视频生成 / 3D 生成**：因耗时较长（视频 1–10 分钟，3D 资产更久），统一仅支持异步。请求必须携带 `X-DashScope-Async: enable`，缺少该头会报错 `current user api does not support synchronous calls`。 - -三者都强调"请勿重复创建任务"，`task_id` 有效期 24 小时，直接轮询即可。 - -## 输入能力对比 - -| 输入方式 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 纯文本 | 支持，复杂文字渲染能力强（千问系列） | 支持（文生视频） | 支持，中英文等多语言，最大 1024 字符 | -| 单图输入 | 支持（图生图、图像编辑） | 支持（首帧生视频） | 支持，JPEG/PNG，宽高 [20,6000]，≤20MB | -| 多图输入 | 部分编辑模型支持多图输入/输出 | 支持（参考生、首尾帧） | 支持，固定 4 视角（前/左/后/右），有效 2–4 张 | -| 视频输入 | 不适用 | 支持（视频编辑、参考生视频） | 不适用 | -| 音频输入 | 不适用 | 万相2.7 支持[多模态](../concepts/multimodal.md)输入含音频 | 不适用 | - -3D 生成的多图输入有严格的视角顺序约束（前/左/后/右），不需要的视角传空对象 `{}`，这与图像/视频的"多图作为参考"语义不同。 - -## 产物与质量参数 - -| 项 | 图像 | 视频 | 3D | -| --- | --- | --- | --- | -| 输出规格 | 总像素 512×512~2048×2048，宽高比 1:4~4:1，1–6 张；万相2.7 支持 4K | 视频文件 URL | 面数：H3.1 最高 200 万面，P1.0 最高 2 万面 | -| 质量参数 | 分辨率、张数、宽高比 | 分辨率、时长、镜头叙事（`shot_type: multi`） | `texture_quality`（标清/高清）、`geometry_quality`（standard/ultra）、`pbr`、`texture` | -| 一致性能力 | 千问编辑支持角色一致性 | 万相2.7 参考生支持角色形象与音色一致性 | 多图视角约束保证几何一致性 | -| 预览能力 | 直接返回图片 | 直接返回视频 | 额外返回 `rendered_image_url` 预览渲染图 | - -## 适用场景建议 - -- **选图像生成 API**：需要静态视觉产出，强调文字渲染、风格化、精确编辑（增删移动物体、改动作）、虚拟模特/试衣/海报等电商与营销场景。优先用同步调用模型（千问图像、万相2.6+/2.7、Z-Image）以简化流程；批量或创意类任务再用异步。 -- **选视频生成 API**：需要动态叙事、数字人、肖像动态视频、视频编辑/换人。文生视频、图生视频（首帧/首尾帧）、参考生视频均可，万相2.7 是推荐的新版协议，支持多模态输入与角色/音色一致性。注意 PixVerse、Vidu 仅北京地域可用且需单独开通。 -- **选 3D 生成 API**：需要可直接导入引擎/3D 软件的 GLB 资产，适用于游戏、电商商品 3D 展示、工业设计。仅北京地域可用，需开通 Tripo。高精度选 `Tripo/Tripo-H3.1`（最高 200 万面），追求速度选 `Tripo/Tripo-P1.0`。 - -## 技术选型参考 - -1. **产出形态决定大类**：图片→图像 API；视频→视频 API；3D 模型→3D API。三者端点不同，不可混用。 -2. **延迟敏感优先同步**：仅图像 API 提供同步调用，适合交互式产品；视频与 3D 必须异步，需在业务侧实现轮询或配置异步任务回调（3D 查询接口默认 RPS 20）。 -3. **地域与鉴权**：三类均要求模型、Endpoint、API Key 同地域。3D 仅北京可用；千问-图像翻译、PixVerse、Vidu 也仅北京。建议迁移到业务空间专属域名以获得更好性能与稳定性。 -4. **任务复用**：`task_id` 24 小时有效，三类都要求轮询而非重复创建任务；3D 产物下载链接仅 2 小时，需及时落盘。 -5. **输入约束**：3D 多图必须按前/左/后/右 4 视角顺序；图像图文混排需开启 `enable_interleave=true` 并配合 SSE 流式；视频首尾帧、参考生有专属模型变体。 -6. **模型开通**：可灵、PixVerse、Vidu、Tripo 均需先在控制台搜索并开通授权，再调用 API。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md deleted file mode 100644 index 9ac91dd6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/multimodal-generation-comparison.md +++ /dev/null @@ -1,71 +0,0 @@ -# 图像生成、视频生成与3D生成对比 - -百炼平台提供图像生成、视频生成和3D生成三大[多模态](../concepts/multimodal.md)内容创作能力。三者在输入输出格式、模型生态、调用方式和适用场景上各有侧重。本文从开发者技术选型角度，对这三类生成能力进行系统对比，帮助快速定位最适合业务需求的方案。 - -## 关键维度对比 - -| 维度 | 图像生成 | 视频生成 | 3D生成 | -|------|---------|---------|--------| -| **输入格式** | 文本提示词、参考图像（单张/多张）、涂鸦草图 | 文本提示词、首帧/首尾帧图像、参考图像/视频/音频 | 文本提示词、单张图像、多图（4视角：前/左/后/右） | -| **输出格式** | PNG 图像（512x512 至 4K） | MP4 视频 | GLB 模型（PBR 材质或无贴图基础模型）+ 预览渲染图 | -| **调用方式** | 同步调用为主，部分模型支持异步 | 全部异步（创建任务 → 轮询结果） | 全部异步（创建任务 → 轮询结果） | -| **主要模型系列** | 千问-图像、万相(Wan/Wanx)、Z-Image、可灵(Kling) | 万相(Wan)、HappyHorse、爱诗(PixVerse)、Vidu、可灵(Kling) | Tripo（H3.1 / P1.0） | -| **模型数量** | 20+ 款模型覆盖各类场景 | 6 大模型家族，任务类型丰富 | 2 款模型（高精度 / 专业快速） | -| **可用地域** | 部分全地域，部分仅华北2（北京） | 多数仅华北2（北京），HappyHorse 支持海外地域 | 仅华北2（北京） | -| **批量输出** | 单次可生成 1-9 张图像 | 单次生成 1 条视频 | 单次生成 1 个3D模型 | -| **产物有效期** | 即时返回，URL 有时效 | task_id 有效期 24 小时 | task_id 有效期 24 小时，下载链接有效期 2 小时 | -| **典型生成耗时** | 秒级至十秒级 | 分钟级 | 分钟级（耗时较长） | - -## 能力覆盖对比 - -| 能力 | 图像生成 | 视频生成 | 3D生成 | -|------|:-------:|:-------:|:-----:| -| 文本生成 | 支持 | 支持 | 支持 | -| 图像/图片参考生成 | 支持 | 支持（首帧/首尾帧） | 支持（单图/多图） | -| 内容编辑 | 支持（局部重绘、风格迁移、扩图等） | 支持（指令编辑、视频迁移） | 不支持 | -| [多模态](../concepts/multimodal.md)混合输入 | 支持（文+图） | 支持（文+图+视频+音频） | 不支持 | -| 中文文字渲染 | 支持（千问、Z-Image 等） | 不适用 | 不适用 | -| 人像/人物专项 | 支持（人像风格重绘、AI试衣） | 支持（数字人、舞动人像、悦动人像等） | 不适用 | -| PBR 材质输出 | 不适用 | 不适用 | 支持 | - -## 计费方式差异 - -- **图像生成**：按张计费，不同模型单价不同（如扩图 0.18 元/张）。部分创意工具仅提供免费体验额度，用完不可付费续用。 -- **视频生成**：按任务计费，费用与视频时长、分辨率、模型版本相关。 -- **3D生成**：按任务计费，费用与贴图质量（standard/detailed）和几何精度（standard/ultra）相关。 - -## 适用场景建议 - -**选择图像生成的场景：** -- 电商商品图、营销海报、社交媒体配图等静态视觉内容 -- 需要精细文字渲染或图文混排的场景（如带中文的宣传图） -- 图像编辑与风格迁移（如局部重绘、背景替换、AI试衣） -- 对生成速度要求高、需要批量出图的场景 - -**选择视频生成的场景：** -- 短视频创作、广告片制作、动态内容营销 -- 人像动画（数字人播报、舞蹈视频、唱演视频） -- 需要多镜头叙事或多角色互动的复杂视频 -- 视频风格转换和口型替换等后期编辑 - -**选择3D生成的场景：** -- 游戏资产、AR/VR 场景中的3D模型快速原型 -- 电商3D商品展示 -- 需要 PBR 材质的高精度3D资产生产（最高 200 万面） -- 从多视角图片重建3D物体 - -## 技术选型要点 - -1. **生成速度**：图像生成最快（秒级），视频和3D生成均需分钟级等待，且必须使用[异步调用](../concepts/async-invocation.md)模式。 -2. **模型生态丰富度**：图像生成和视频生成均拥有多个模型家族可选，3D生成目前仅有 Tripo 系列。 -3. **地域限制**：3D生成仅限北京地域；视频和图像生成的部分模型也有地域限制，选型前需确认目标地域的模型可用性。 -4. **输出后处理**：视频和3D的产物下载链接有时效限制（3D仅 2 小时），需在业务流程中及时下载存储。 -5. **开通流程**：3D生成和部分视频/图像模型需在百炼控制台额外搜索并开通服务，不是默认可用。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md deleted file mode 100644 index 45397164..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-app-call-vs-agents-api.md +++ /dev/null @@ -1,74 +0,0 @@ -# Qwen API、应用调用与托管智能体 API 对比 - -百炼平台提供了多种 API 接入方式，开发者在集成大模型能力时常面临选型困惑：是直接调用 Qwen 模型 API，还是通过应用调用 API 使用已编排好的智能体/工作流，亦或是采用 Managed Agents API 获得平台全托管的智能体运行时？本文从接口定位、协议兼容性、会话管理、工具能力、计费模式等维度进行系统对比，帮助开发者根据实际场景做出技术选型。 - -## 定位差异 - -- **Qwen API**：直接调用 Qwen 系列大语言模型，获取文本生成能力。开发者自行管理 [prompt](../guides/prompt.md)、上下文和工具调用逻辑，灵活度最高。 -- **应用调用 API**：调用在百炼控制台中已创建并发布的智能体或工作流应用。应用内部已封装模型选择、知识库检索、插件调用等编排逻辑，开发者只需传入用户输入即可获取最终结果。 -- **Managed Agents API**：平台全托管的智能体运行时，提供 Agent、Session、Environment、Skill、File 等资源抽象。由平台负责会话状态机、沙箱执行、工具调用与事件流推送，适合需要长期运行、多步工具调用的复杂场景。 - -## 关键维度对比 - -| 维度 | Qwen API | 应用调用 API | Managed Agents API | -|------|----------|-------------|-------------------| -| **定位** | 模型级调用，直接访问 Qwen 系列模型 | 应用级调用，调用已编排好的智能体/工作流 | 平台托管智能体运行时，全生命周期管理 | -| **兼容协议** | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | OpenAI Responses（兼容模式）、DashScope | 百炼原生 REST API | -| **API 端点** | `POST /compatible-mode/v1/chat/completions` 等 | `POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` 或 `POST /api/v1/apps/{APP_ID}/completion` | `POST /api/v1/agentstudio/sessions/{session_id}/events` 等 | -| **认证方式** | [API Key](../concepts/api-key.md)（`Authorization: Bearer`） | [API Key](../concepts/api-key.md) + APP ID（+ 可选 Workspace ID） | [API Key](../concepts/api-key.md) + Workspace ID | -| **模型选择** | 请求体 `model` 字段指定任意 Qwen 模型 | 控制台配置，调用时无需指定模型 | Agent 创建时配置模型 | -| **会话管理** | 调用方自行维护（Responses 接口除外） | DashScope API 通过 `session_id` 自动维护；Responses API 需传完整历史 | 平台全托管，Session 状态机自动驱动 | -| **工具/插件** | Responses 接口内置联网搜索、代码解释器、网页提取；其他接口需自定义 | 控制台可视化编排插件、知识库、工具 | Skill（zip 包上传）+ Environment 沙箱执行 | -| **[多模态](../concepts/multimodal.md)支持** | 需选用 VL 系列模型 | Responses API 支持图像和文件输入 | 通过 File 资源挂载到 Session | -| **[流式输出](../concepts/streaming.md)** | 支持（`stream=true`） | 支持（`stream=true`） | SSE 事件流（`GET .../events/stream`） | -| **[异步调用](../concepts/async-invocation.md)** | 不支持 | Responses API 支持（`background=true`），DashScope 暂不支持 | 原生异步，Session 状态机驱动 | -| **支持地域** | 多地域 | 仅华北2（北京） | 仅 cn-beijing | -| **SDK 支持** | OpenAI SDK、Anthropic SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | OpenAI SDK、[DashScope SDK](../concepts/dashscope-sdk.md) | 百炼原生 SDK / HTTP | -| **配置方式** | 纯代码，请求参数控制 | 控制台可视化编排 + API 调用 | API 全程管理（Agent/Environment/Session/Skill） | - -## 适用场景建议 - -### 选择 Qwen API - -- 需要直接、细粒度地控制模型推理参数（temperature、top_p 等）。 -- 已有基于 OpenAI 或 Anthropic SDK 的应用，希望低成本迁移到百炼平台。 -- 构建自定义的 RAG、Agent 框架，模型调用只是其中一环。 -- 对 [prompt](../guides/prompt.md) 工程有深度需求，需要完整掌控输入输出。 - -### 选择应用调用 API - -- 已在百炼控制台完成智能体或工作流的可视化编排，希望通过 API 将其集成到业务系统。 -- 需要使用控制台配置的知识库检索、插件、工作流节点等平台能力，不想在代码中重新实现。 -- 团队中非开发人员负责应用逻辑编排，开发人员只负责 API 集成。 -- 需要快速上线，应用逻辑变更通过控制台完成而非修改代码。 - -### 选择 Managed Agents API - -- 需要平台全托管的智能体运行时，不想自行管理会话状态和工具执行环境。 -- 智能体任务涉及多步工具调用、代码执行、文件读写，需要沙箱环境保障安全。 -- 希望通过 API 动态创建和管理多个智能体，实现多 Agent 协作。 -- 需要细粒度的事件流（SSE）来追踪智能体执行过程中的每一步操作。 -- 有自定义工具（Skill）需要安全审核后挂载，要求版本锁定和隔离。 - -## 选型决策参考 - -1. **"我只需要一个模型回答问题"** — 选 Qwen API。最简单直接，兼容主流 SDK。 -2. **"我已在控制台搭好应用，想 API 接入"** — 选应用调用 API。零编排代码，改逻辑只需改控制台配置。 -3. **"我需要平台帮我管理 Agent 的执行环境和工具调用"** — 选 Managed Agents API。平台托管状态机、沙箱和事件流，适合复杂任务。 -4. **迁移成本优先** — Qwen API 的 OpenAI/Anthropic 兼容接口迁移成本最低；应用调用 API 也提供 OpenAI 兼容模式。 -5. **功能完整度优先** — Qwen API 的 DashScope 原生接口参数最丰富；Managed Agents API 的资源模型最完整。 - -## 注意事项 - -- Qwen API 的兼容接口可能不暴露 DashScope 原生的全部参数，如需最全功能建议使用 DashScope 接口。 -- 应用调用 API 要求先在控制台创建并发布应用，APP ID 只能通过控制台手动获取。 -- Managed Agents API 当前仅支持 cn-beijing 地域，Skill 上传后需通过安全扫描才能挂载。 -- 三种 API 的计费方式均基于 token 消耗，但应用调用和 Managed Agents 可能涉及额外的平台资源费用（如沙箱、存储），请参考官方定价文档。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md deleted file mode 100644 index a3da8448..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/qwen-api-vs-omni-realtime-vs-managed-agents.md +++ /dev/null @@ -1,57 +0,0 @@ -# Qwen API vs 全双工实时API vs 托管智能体API - -百炼平台提供多种 API 接口满足不同开发场景。Qwen API 面向文本生成任务，提供多协议兼容的 HTTP 接口；全双工实时 API（Omni Realtime）基于 WebSocket 实现低延迟的音视频实时对话；托管智能体 API（Managed Agents）则提供完整的智能体托管运行时，由平台负责会话编排与沙箱执行。本文从协议、能力、适用场景等维度帮助开发者做出技术选型。 - -## 关键维度对比 - -| 维度 | Qwen API | 全双工实时 API | 托管智能体 API | -| --- | --- | --- | --- | -| 通信协议 | HTTP（REST） | WebSocket（长连接） | HTTP（REST）+ SSE 事件流 | -| 输入格式 | 文本（messages JSON） | 音频流（PCM 16kHz）、图像（Base64） | 文本消息、文件附件 | -| 输出格式 | 文本（支持流式 SSE） | 音频流（PCM 24kHz）+ 文本转录 | 事件流（SSE），含文本、工具调用回执等 | -| 支持模型 | Qwen 系列文本生成模型 | Qwen3.5-Omni-Realtime、Qwen3-Omni-Flash-Realtime、Qwen-Omni-Turbo-Realtime | 可配置任意百炼平台模型 | -| API 端点 | `https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions` 等 | `wss://{WorkspaceId}.{region}.maas.aliyuncs.com/api-ws/v1/realtime` | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio` | -| 兼容协议 | OpenAI Chat Completions、OpenAI Responses、Anthropic Messages、DashScope 原生 | 无（百炼专有 WebSocket 协议） | 无（百炼专有 REST 协议） | -| 工具调用 | Responses 接口内置联网搜索/代码解释器/网页提取；其他接口需自定义 | 支持 Function Calling 和联网搜索（仅 Qwen3.5 系列） | 平台托管技能（Skill zip 包），沙箱内执行 | -| 会话管理 | 仅 Responses 接口自动管理历史；其他需客户端维护 | 平台维护 WebSocket 会话上下文 | 平台全托管（Session 状态机：idle → running → idle/terminated） | -| 延迟特性 | 标准 HTTP 请求-响应，流式可逐 token 返回 | 超低延迟（毫秒级音频帧推送） | 异步任务式，SSE 实时推送中间事件 | -| 多模态支持 | 纯文本（部分模型支持图像输入） | 音频 + 视频 + 图像 + 文本 | 文本 + 文件（通过 File 资源挂载） | -| 计费方式 | 按 token 计费（输入/输出分计） | 按音频时长 + token 计费 | 按底层模型 token + 沙箱资源用量计费 | - -## 适用场景建议 - -### Qwen API - -- **文本问答与对话**：聊天机器人、客服对话、内容生成等标准 NLP 任务 -- **快速迁移**：已有 OpenAI 或 Anthropic 代码的项目，可通过兼容接口低成本切换到 Qwen 模型 -- **轻量工具调用**：使用 Responses 接口可直接获得联网搜索、代码解释器能力，无需额外开发 -- **批量处理**：适合离线或准实时的文本处理流水线 - -### 全双工实时 API - -- **语音助手**：需要实时语音输入并即时语音回复的场景 -- **智能客服**：基于 VAD 自动检测用户说话意图，实现自然的对话轮转 -- **多模态交互**：需要同时处理音视频输入的实时应用（如视频通话中的 AI 助手） -- **声音定制**：利用声音复刻能力打造品牌专属语音形象 - -### 托管智能体 API - -- **复杂任务编排**：智能体需要多轮推理、工具调用、代码执行的场景 -- **平台托管运行时**：不想自行管理会话状态、沙箱环境和工具执行的团队 -- **企业级智能体**：需要版本管理、权限控制、文件交互等完整生命周期管理 -- **多技能组合**：通过 Skill 机制灵活组装工具链，且由平台保证安全审核 - -## 技术选型指引 - -1. **只需文本生成** → 选择 Qwen API。如已有 OpenAI/Anthropic SDK 代码，使用对应兼容接口可零改动迁移。 -2. **需要实时语音交互** → 选择全双工实时 API。它是唯一支持音频流式双向通信的接口，延迟最低。 -3. **需要平台托管的智能体运行时** → 选择托管智能体 API。适合需要沙箱执行、多工具编排、会话生命周期管理的复杂 Agent 应用。 -4. **组合使用**：三者并非互斥。例如可用托管智能体 API 编排复杂工作流，其底层模型调用仍走 Qwen API；或在语音助手前端使用全双工实时 API，后端通过托管智能体执行复杂任务。 - -## 被对比主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) -- [managed agents api](../api/managed-agents-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md new file mode 100644 index 00000000..bd94ff13 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md @@ -0,0 +1,70 @@ +# 实时 API 方案对比：Omni Realtime API vs Realtime API User Guide + +## 对比目的与背景 + +为帮助开发者在百炼平台快速、准确地选择适合业务需求的实时交互方案，本文对两类核心实时能力接口进行系统性对比分析： +- **Omni Realtime API**（`api/omni-realtime-api.md`）：面向端到端[多模态](../concepts/multi-modal.md)智能体的**一体化、开箱即用型实时对话接口**，聚焦“语音/音视频输入 → 语义理解 → 工具调用/联网搜索 → 文本+音频输出”的全链路闭环。 +- **Realtime API User Guide**（`api/realtime-api-user-guide.md`）：面向工程集成的**协议级实时通信框架指南**，定义 WebSocket / WebRTC / AOQ 三种传输协议的能力边界、接入范式与模型兼容矩阵，强调**跨终端、弱网鲁棒性与协议可选性**。 + +二者并非互斥替代关系，而是**抽象层级不同、定位互补的技术方案**：Omni Realtime API 是构建于 Realtime API 协议栈之上的高阶封装；而 Realtime API User Guide 是底层协议能力的统一说明文档。本对比旨在厘清技术边界，避免因概念混淆导致选型偏差。 + +--- + +## 关键维度对比表 + +| 维度 | Omni Realtime API | Realtime API User Guide | +|------|-------------------|--------------------------| +| **本质定位** | 面向场景的**高阶 SDK 封装接口**（Python/Java SDK 主导），提供预编排的[多模态](../concepts/multi-modal.md)对话流水线 | 面向架构的**协议能力说明书**，定义 WebSocket / WebRTC / AOQ 三类传输层标准及模型支持矩阵 | +| **输入格式** | 支持 `append_audio`（Base64 PCM）、`append_video`（H.264 编码帧或原始 I420/NV12 帧）；VAD 模式下自动分段 | 协议相关：
• WebSocket：Base64 PCM 音频 + 可选视频帧
• WebRTC：MediaStream 或 Raw Video Frame（I420/BGRA）
• AOQ：支持外部注入原始音频帧（PCM/I2S）或编码帧（AAC/H.264） | +| **输出格式** | 固定流式事件结构：`response.text.delta`、`response.audio.delta`、`response.function_call_arguments.*` 等；支持 `TEXT` + `AUDIO` 同步输出 | 协议相关：
• WebSocket：JSON 事件流（含 `text`/`audio` 字段）
• WebRTC：DataChannel 传输文本 + AudioTrack 输出合成语音
• AOQ：混合通道（`text` via DataChannel, `audio` via AudioTrack, `video` via VideoTrack） | +| **支持模型** | 仅限 `qwen3.5-omni-*` 系列实时模型（如 `qwen3.5-omni-realtime`, `qwen3.5-omni-flash-realtime` 等），且功能严格按模型版本隔离 | 覆盖更广：
• 全模态模型（`qwen3.5-omni-*`, `qwen3.5-livetranslate-flash-realtime`）→ 三协议均支持
• Fun-ASR / CosyVoice / `qwen-audio-3.0-realtime-plus` → **仅 WebSocket 支持**
• `multimodal-dialog` 套件 → **仅 WebSocket/WebRTC 支持，不支持 AOQ** | +| **API 端点** | 固定 WebSocket 地址：
`wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime`（地域专属域名） | 协议差异化：
• WebSocket：同上，但模型可通过 URL Query（`?model=xxx`）或消息体指定
• WebRTC：`POST /v1/realtime/webrtc/offer` 获取 SDP，建连后通过 DataChannel 通信
• AOQ：需先调用 `POST /v1/realtime/aoq/allocate` 获取 `sid` 和 `aoqTokenForClient`，再连接 AOQ 服务节点 | +| **计费方式** | 按**实际消耗的 token 数量 + 音频处理时长（秒）** 计费（含 ASR/TTS/LLM 推理），模型不同单价不同；`qwen-omni-turbo-realtime` 按会话时长阶梯计费 | **统一按模型调用粒度计费**，与所选协议无关；但 AOQ/WebRTC 的媒体传输带宽、信令调用等基础资源不额外计费（计入百炼平台基础配额） | +| **典型场景** | 智能客服坐席助手、AI 会议纪要员、语音驱动的虚拟数字人（需工具调用/联网搜索/声音复刻） |
  • **WebSocket**：后台语音质检、IVR 系统集成、快速 PoC 验证
  • **WebRTC**：浏览器端在线教育互动白板、远程医疗问诊、Web 端虚拟主播
  • **AOQ**：移动端音视频社交 App、车载语音助手、鸿蒙设备本地化 AI 交互
| +| **VAD 能力** | 提供 `server_vad`（服务端静音检测）和 `semantic_vad`（语义级说话人意图识别）；后者**仅 `qwen3.5-omni-realtime` 支持** | `semantic_vad` 在三协议中均可用（需模型支持），但 `turn_detection.type` 参数需在 `session.update` 中显式设置；`server_vad` 为默认回退选项 | +| **开发者控制粒度** | **低控制粒度**：SDK 自动管理连接、会话生命周期、媒体缓冲区提交、响应流解析；手动模式需显式 `commit()`，但仍受限于 Omni 协议语义 | **高控制粒度**：WebRTC/AOQ 允许完全接管媒体采集、编码、网络传输（如 AOQ 支持 `isExternal=true` 注入自定义音频帧、WebRTC 支持 `RTCPeerConnection` 级配置） | +| **SDK 支持** | 官方提供 Python / Java SDK，封装连接、会话、事件回调全流程；无 JS SDK | 提供：
• WebSocket：DashScope Python/Java SDK
• WebRTC：TypeScript SDK（基于 Web API）
• AOQ：Android/iOS/HarmonyOS 原生 SDK（含 C++ 底层接口） | + +--- + +## 适用场景建议 + +### ✅ 选择 Omni Realtime API 当： +- 业务目标是快速上线一个**具备完整 AI 对话能力的语音/音视频应用**（如客服机器人、会议助理），且无需深度定制媒体链路； +- 需要**开箱即用的语义级 VAD、工具调用、联网搜索、声音复刻**等高级能力，并接受其模型功能绑定（如仅 `qwen3.5-omni-realtime` 支持全部特性）； +- 开发团队以服务端为主（Python/Java），或希望最小化前端音视频工程复杂度； +- 对弱网适应性要求不高（依赖 WebSocket，无原生抗丢包机制）。 + +### ✅ 选择 Realtime API User Guide（按协议选型）当： +- 需要**跨终端一致体验**：浏览器（WebRTC）、App（AOQ）、服务端（WebSocket）共用同一模型能力； +- 对**弱网鲁棒性、低延迟、混合媒体传输（音+视+数据）有硬性要求** → 优先选 AOQ； +- 需要**深度控制媒体流**：如接入自研 ASR 引擎、注入 TTS 音频、处理 H.264 编码帧、实现回声消除旁路等； +- 使用非 Omni 系列模型（如纯 ASR/CosyVoice/`qwen-audio-3.0-realtime-plus`）→ **必须使用 WebSocket 协议**； +- 构建标准化 AI 通信中间件或 SDK 层，需解耦协议与模型。 + +> ⚠️ 注意：Omni Realtime API 本质是 Realtime API 的一种**特定协议（WebSocket）+ 特定模型（Omni 系列）+ 特定 SDK 封装**的组合。若项目需 WebRTC 浏览器支持或 AOQ 移动端弱网能力，**不能直接使用 Omni Realtime API SDK**，而应遵循 Realtime API User Guide 中对应协议的接入规范。 + +--- + +## 技术选型参考（面向开发者） + +| 你的需求 | 推荐方案 | 关键依据 | +|----------|-----------|-----------| +| “我要 3 天内上线一个带语音问答和搜索的客服机器人” | ✅ Omni Realtime API（`qwen3.5-omni-realtime`） | SDK 开箱即用，`enable_search` + `tools` 一键启用，无需处理 SDP/AOQ [Token](../concepts/token.md) | +| “我要在微信小程序里做实时语音翻译，需适配低端安卓机弱网” | ✅ Realtime API + AOQ 协议 | AOQ 原生支持 QUIC 重传、前向纠错、带宽自适应；Omni API 不支持 AOQ | +| “我已有自研音视频 SDK，只需把百炼 LLM 接入现有通话流程” | ✅ Realtime API + WebRTC 或 AOQ | 可复用现有媒体采集链路，通过 `isExternal=true` 注入音频，避免重复开发 | +| “我要同时支持网页端（Chrome）、iOS App、车载中控屏” | ✅ Realtime API（三协议分别接入） | Omni API 仅提供 WebSocket，无法覆盖 WebRTC/AOQ 场景；需统一模型 + 分协议实现 | +| “我只需要实时语音转文字（ASR），不要 LLM” | ✅ Realtime API + WebSocket（Fun-ASR 模型） | Omni Realtime API **不提供独立 ASR 接口**，其 ASR 固定为 `qwen3-asr-flash-realtime` 且不可替换 | +| “我要做声音复刻驱动的虚拟人，且需毫秒级唇形同步” | ✅ Realtime API + AOQ + `qwen3.5-omni-plus-realtime` | AOQ 支持音视频帧级时间戳对齐；Omni SDK 未暴露帧同步控制接口 | + +> 💡 最佳实践提示： +> - 若初期验证用 WebSocket 快速跑通，后期需扩展至移动端，请**从 Realtime API User Guide 出发设计协议无关的会话抽象层**，避免 Omni SDK 绑定导致重构成本； +> - 所有方案均需确保 `workspaceId` 与模型所在地域匹配（北京/新加坡），且 API Key 具备对应模型调用权限； +> - `semantic_vad` 虽为高级能力，但在嘈杂环境或多人对话中可能误触发，生产环境建议结合 `idle_timeout_ms` + `silence_duration_ms` 进行参数调优。 + +## 被对比主题页 + +- [omni realtime api](../api/omni-realtime-api.md) +- [realtime api user guide](../api/realtime-api-user-guide.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md deleted file mode 100644 index 5b2f9be9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-application.md +++ /dev/null @@ -1,81 +0,0 @@ -# 智能体应用 - -智能体应用（Agent）是阿里云百炼平台的核心应用构建模式之一，通过自然语言零代码配置，让大模型基于角色设定自主决策、动态规划并调用知识库、MCP、Skill 等工具来完成任务。相较于流程固定的工作流应用，智能体强调 AI 的自主性，适合意图开放、需要动态编排的对话与轻量任务场景。 - -## 版本演进：Agent 1.0 与 Agent 2.0 - -百炼提供两代技术架构不同的智能体，**不支持直接升级或版本切换**，迁移需重新创建： - -- **新版智能体（Agent 2.0）**：2025 年 12 月 26 日上线。将知识库、MCP 等能力统一抽象为「工具」，由智能体自主规划调用时机与顺序，并完整展示「规划-执行-反思」链路。无旧版依赖时推荐使用。**仅支持 API 调用，不支持任何分享渠道**（魔笔/UI、钉钉、微信、组件、音视频互动）。 -- **旧版智能体（Agent 1.0）**：通过知识库（RAG）+ 插件扩展能力，先检索知识再决策是否调用工具，适合意图单一、流程固定的简单任务。自定义插件有 **5 秒超时限制**。分享渠道均为 1.0 功能。 - -## 核心能力配置（Agent 2.0） - -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）；可配置最长回复长度、`temperature`、`enable_thinking`（思考模式）等参数。 -- **提示词（System Prompt）**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境下的 `bash`、`write`、`read`、`edit`、`glob`、`grep`、`download_file`，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **Skill**：可扩展能力包，智能体在对话中自动识别匹配任务并调用，无需额外编码。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话内工具调用最大次数。 - -## 文件问答 - -智能体应用支持上传文件进行问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等 | 功能灵活，依赖配置的工具 | - -限制：单会话最多 10 个文件，单文件不超过 10 MB（超出需用文件上传 API）。支持文档、图片、视频、音频等格式。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成，核心步骤：在应用配置页点击「发布」→ 在「发布渠道」查看调用方式。RAM 账号发布前需拥有 `ram:CreateServiceLinkedRole` 权限。 - -API 调用与工作流应用完全一致，通过 `Application.call` / `POST /apps/{app_id}/completion` 触发： - -```python -import os -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) -print(response.output.text) -``` - -响应结构为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}`，业务侧主要消费 `output.text`。 - -## 分享与组件化（仅 Agent 1.0） - -- **分享渠道**：UI 应用/魔笔、钉钉、微信公众号、音视频实时互动。UI 体验链接与音视频临时二维码有效期均为 **24 小时**。分享产生的费用由应用创建者 UID 账号承担。 -- **组件化**：智能体或工作流可发布为模块化组件供其他应用复用。接入智能体时组件作为工具，大模型据「组件描述」自动判断调用；预设系统参数 `query`、`imageList` 无法删除。注意避免嵌套调用（A↔B）和多级调用（A→B→C 易超时）。 - -## 与 Managed Agents 的区别 - -Managed Agents 是服务端托管运行时，与无状态的智能体应用不同：它在独立云端沙箱容器中维护会话状态，支持中断与续接、事件历史持久化，面向多步工具调用、代码执行、文件处理等长时运行任务。 - -## 计费说明 - -- **模型调用**：按模型类型和 Token 用量计费。 -- **知识库**：按量付费，召回的文本切片会增加输入 Token；自 2026 年 1 月 4 日起正式计费。 -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 由第三方收取。 - -百炼提供限时免费额度，可在模型广场查看。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [managed agents](../guides/managed-agents.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) -- [start using](../guides/start-using.md) -- [skill](../guides/skill.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md b/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md deleted file mode 100644 index 72e01839..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/agent-orchestration.md +++ /dev/null @@ -1,90 +0,0 @@ -# 智能体编排 - -智能体编排是指在阿里云百炼平台上，通过配置或代码组织智能体（Agent）、工具、知识库、子智能体等资源之间的调用关系与执行流程，让大模型在受控边界内自主规划并完成业务任务的能力。 - -## 在百炼中的使用方式 - -百炼支持三种互补充的编排形态，开发者可单独使用或组合集成： - -| 形态 | 编排方式 | 控制力 | 适用场景 | -| --- | --- | --- | --- | -| 新版智能体（Agent 2.0） | 自然语言配置，零代码 | 由大模型自主规划 | 智能客服、知识问答、任务助理 | -| 工作流应用 | 可视化节点编排，低代码 | 节点固定、流程确定可复现 | 报告生成、订单处理、审批流 | -| 高代码应用 | Python 编码 | 完全由代码定义逻辑 | 私有算法部署、深度定制 | - -### 1. 新版智能体编排 - -Agent 2.0 将知识库、MCP 统一抽象为「工具」，由智能体在每轮「规划—执行—反思」（ReAct）链路中自主决定调用顺序。关键配置： - -- **模型**：推荐 `千问-Max` 等工具调用能力强的模型；可配 `temperature`、`enable_thinking`、最长回复长度。 -- **系统提示词**：定义角色、行为指令、能力边界；支持 `/变量` 引用自定义变量。 -- **知识库**：作为工具被自主调用，可用标签限定查询范围。 -- **MCP / 插件**：以 MCP 协议接入外部工具，支持动态非固定顺序调用；插件可一键转 MCP。 -- **内置工具**：`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`，在隔离沙箱中默认关闭、按需开启。 -- **记忆**：短期 0–30 轮上下文。 -- **ReAct 最大轮次**：1–50，限制单次会话工具调用次数，超出自动退出并生成最终回复。 - -> 注意：旧版智能体与新版本架构不兼容，不支持升级、降级或版本切换。 - -### 2. 工作流编排 - -通过可视化节点将复杂任务拆解为有序步骤，逻辑确定。核心节点： - -- **开始 / 结束**：定义输入输出参数，开始节点预置 `query`、`historyList`、`imageList` 等。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆。 -- **意图分类节点**：按输入分流到不同下游分支。 -- **变量处理节点**：文本输出或变量加工。 -- **智能体群组节点**：把任务分解给多个已发布的子智能体协同完成，是工作流编排多智能体协作的核心。 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在画布右上角配置并在各节点引用。 - -### 3. 高代码编排 - -基于完整 Python 项目结构部署 AI 后端： - -- **部署形态**：Serverless Function（无状态、快速拉起、低成本）或 K8s（高性能、有状态、长程任务）。 -- **代码提交**：使用预置模板（基础对话 / 工具调用 / 深度研究 Agent）或上传本地 `.whl` 包。 -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务。 -- **网关**：应用稳定后建议开启网关，通过自定义域名和路由在生产环境访问。 - -## API 层编排 - -### Managed Agents API（托管运行时） - -平台托管会话、沙箱、工具执行与事件流，开发者通过 REST 或 DashScope SDK 管理 Agent、Environment、Skill、File、Session 等资源： - -- **Agent**：可复用的智能体配置（模型、系统提示词、工具包、技能）。 -- **Environment**：工具调用的执行沙箱与预装依赖，可被多会话复用。 -- **Session**：智能体的一次运行实例，绑定 Agent 与 Environment 快照。 -- **Skill**：以 zip 包封装的工具组合与文档，挂载时锁定版本。 -- **File**：独立文件资源，可挂载到沙箱供工具读写或作为消息内容。 - -调用流程四步：创建 Agent（一次创建长期复用）→ 创建 Session（每轮对话新建）→ 发送 Event（写入用户消息触发 `running`）→ 订阅 SSE 流式接收回复直至回到 `idle`。 - -### 应用调用 API - -将控制台编排好的应用集成进业务系统，两种接口： - -- **OpenAI 兼容 Responses API**：`POST /api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，复用 OpenAI 生态，支持同步/异步、流式、多模态。 -- **DashScope 原生 API**：`POST /api/v1/apps/{APP_ID}/completion`，功能更全，新版智能体、工作流、旧版智能体均支持。 - -调用前需准备 APP ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）。多轮对话可通过 `session_id`（系统自动加载历史，有效期 1 小时、最多 50 轮）或自行维护 `messages` 数组实现。 - -## 选型建议 - -- 需要 AI 自主决策、动态规划 → 新版智能体。 -- 流程固定、要求稳定可复现 → 工作流，复杂协作用智能体群组节点。 -- 深度定制、私有算法 → 高代码应用。 -- 需要平台托管调度与执行基础设施 → Managed Agents API。 -- 仅把控制台应用接入业务系统 → Application.call / completion。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [managed agents api](../api/managed-agents-api.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md b/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md deleted file mode 100644 index a6884fa8..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/api-key.md +++ /dev/null @@ -1,80 +0,0 @@ -# API Key 鉴权 - -API Key 是调用阿里云百炼平台模型与应用的核心鉴权凭证，以 `Authorization: Bearer ` 的形式随请求携带，用于标识调用者身份并界定其可访问的模型、应用与业务空间范围。 - -## 在百炼平台的使用场景 - -API Key 贯穿百炼的各类调用入口，不同场景下的获取与使用方式略有差异： - -- **模型 API 直连**：调用文本、图像、视频、语音、向量等模型前，先在控制台对应地域的 **API Key** 页面创建 Key，配合服务端点 `base_url` 发起请求。 -- **应用与知识库调用**：智能体应用、工作流应用、知识检索/问答接口（DashScope 应用网关）均通过 API Key Bearer 鉴权，Base URL 常需拼接业务空间 ID，如 `https://{workspaceId}.cn-beijing.maas.aliyuncs.com`。 -- **框架集成**：LlamaIndex（RAG）、Spring AI Alibaba（智能体/工作流/知识库检索）均以 API Key 鉴权复用百炼能力，通常通过环境变量注入。 -- **百炼 CLI（`bl`）**：支持控制台登录、`--api-key`、环境变量、配置文件、临时传入等多种认证方式，可组合使用。 -- **不可信环境（浏览器/移动端）**：应由后端生成临时 API Key，避免永久 Key 泄露。 -- **子业务空间**：需使用该空间自身的 API Key，并预先为其授予相应模型的调用权限。 - -## Key 的类型与前缀 - -百炼的 API Key 因计费方式不同而彼此隔离，前缀是重要区分标识： - -| 类型 | 前缀 | 说明 | -| --- | --- | --- | -| 按量付费（安全升级后） | `sk-ws` | 仅创建时展示一次明文，关闭弹窗后无法再查看，务必立即保存 | -| 按量付费（升级前旧 Key） | `sk-` | 仍可正常使用 | -| Token Plan / Coding Plan 专属 | `sk-sp-` | 套餐专属，与通用 Key 不可混用 | -| 临时 API Key | `st-` | 由后端接口生成、限时有效 | - -> API Key 与其配套的 Base URL 必须成对使用。按量付费、Token Plan、Coding Plan 三者的 Key 与 Base URL 混用会导致意外扣费或返回 401/403 鉴权失败。 - -## 创建时的关键选项 - -- **归属业务空间**：决定 Key 的调用权限。默认空间的 Key 可调用所有标准模型及默认空间应用；子空间的 Key 仅能调用已授权模型及本空间应用。同一空间内的 Key 权限相同，无需按模态分别创建。 -- **权限**：可选「全部」或「自定义」。自定义可配置 IP 白名单（最多 20 个 IPv4/IPv6 地址或网段）及可访问的模型/应用范围。 -- **创建主体**：需主账号，或具备「管理员」/「API-Key」页面权限的子账号。RAM 子账号在订阅套餐前需主账号授予 `AliyunBailianFullAccess` 等权限。 - -## 配置与使用 - -推荐将 API Key 写入环境变量，避免硬编码泄漏： - -- 通用约定使用 `DASHSCOPE_API_KEY`。 -- Spring AI Alibaba 应用集成用 `DASHSCOPE_API_KEY`，知识库检索用 `AI_DASHSCOPE_API_KEY`（约定不一致，关键是 `application.yml` 中占位符与实际变量名匹配）。 - -调用时除 API Key 外，还需指定与地域、协议匹配的 `base_url`（OpenAI 兼容协议与 Anthropic 兼容协议不同）。 - -## 临时 API Key - -面向浏览器、移动端等不可信环境，通过后端接口用永久 Key 换取限时凭证： - -``` -curl -X POST "https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800" \ - -H "Authorization: Bearer $DASHSCOPE_API_KEY" -``` - -关键参数与特性： - -- `expire_in_seconds`：有效期（TTL），单位秒，范围 `[1, 1800]`，默认 60 秒。 -- 返回体中 `token`（`st-` 前缀）为临时 Key，`expires_at` 为过期 UNIX 时间戳。 -- 临时 Key 继承永久 Key 的全部权限（含模型/知识库访问限制），到期自动失效，无法提前删除。 -- 各地域（北京 / 新加坡 / 弗吉尼亚）的 Key 与 Endpoint 不互通，需配套使用。 - -## 编程化管理 - -除控制台外，百炼提供 OpenAPI（`CreateApiKey` / `GetApiKey` / `ListApiKeys` / `UpdateApiKey` / `DeleteApiKey` / `EnableApiKey` / `DisableApiKey` / `ResetApiKey`）以编程方式管理 Key。这些接口使用阿里云账号 AccessKey 签名认证，并需具备相应 RAM 权限，与业务调用所用的 API Key Bearer 鉴权不同。 - -## 开发者要点 - -- 优先用环境变量注入，切勿硬编码或提交到代码仓库。 -- 升级后的 `sk-ws` Key 只显示一次，创建后立即保存。 -- 明确当前场景使用的是按量付费、Token Plan、Coding Plan 中的哪种 Key，并搭配对应 Base URL。 -- 不可信环境一律走临时 API Key，生产环境的文件存储不要依赖临时 URL。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [more](../api/more.md) -- [frameworks](../api/frameworks.md) -- [more about models](../api/more-about-models.md) -- [token plan guide](../guides/token-plan-guide.md) -- [knowledge](../api/knowledge.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md deleted file mode 100644 index 8e78a250..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/async-invocation.md +++ /dev/null @@ -1,53 +0,0 @@ -# 异步调用与任务轮询 - -异步调用是百炼平台为耗时较长的模型任务（图像生成、视频生成、3D 生成等）设计的调用模式：客户端先提交任务拿到 `task_id`，再通过轮询或事件通知获取最终结果，从而避免长连接等待与请求超时。 - -## 核心流程：创建任务 → 轮询获取 - -异步调用统一分为两步： - -1. **创建任务**：向对应能力的生成端点发起 `POST` 请求，请求头必须携带 `X-DashScope-Async: enable`，否则会报错 `current user api does not support synchronous calls`。请求成功后返回一个 `task_id`。 -2. **轮询查询结果**：用该 `task_id` 发起 `GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`，读取 `output.task_status` 直到任务进入终态（`SUCCEEDED` / `FAILED`），再从结果中取回产物 URL。 - -任务状态的典型流转为：`PENDING`（排队中）→ `RUNNING`（处理中）→ `SUCCEEDED`（成功）/ `FAILED`（失败）。不同能力还可能出现 `SUSPENDED`（挂起）、`CANCELED`（已取消）、`UNKNOWN`（任务不存在或已过期）等状态。 - -## 在不同场景中的使用 - -- **3D 生成（Tripo）**：仅支持异步，端点为 `.../aigc/video-generation/3d-generation`，轮询建议间隔约 15 秒，生成耗时较长，产物下载链接有效期仅 2 小时。 -- **视频生成**：所有厂商模型（万相、PixVerse、Vidu、可灵等）统一走异步，端点通常为 `.../aigc/video-generation/video-synthesis`（部分数字人/换人类模型使用 `.../aigc/image2video/video-synthesis`）。单次任务通常耗时 1-5 分钟，个别统一编辑模型约 5-10 分钟。 -- **图像生成**：多数传统模型仅支持异步（`text2image` / `image2image` 等端点），生成通常需 1-2 分钟；而新版模型（wan2.6 / wan2.7、z-image-turbo 等）走 `multimodal-generation/generation` 端点，支持 HTTP 同步一次拿结果。请勿把同步协议用在旧模型上。 -- **应用调用（智能体/工作流）**：Responses API（OpenAI 兼容）通过设置 `background=true` 开启异步，创建任务后返回任务 ID 再轮询；DashScope 应用调用接口目前暂不支持异步。 - -## 关键参数与配置 - -- **请求头**：创建任务必须带 `X-DashScope-Async: enable`；同时需 `Authorization: Bearer $DASHSCOPE_API_KEY`、`Content-Type: application/json`。 -- **`task_id` 有效期**：一般为 **24 小时**，过期查询返回 `UNKNOWN`。请勿重复创建任务，轮询即可。 -- **`background`（应用 Responses API）**：布尔值，默认 `false`，设为 `true` 开启异步执行。 -- **查询限流**：任务查询接口默认约 20 QPS / RPS（主账号维度）。 - -## 异步任务管理与完成通知 - -百炼提供三个通用的异步任务管理接口： - -- **查询单个任务**：`GET .../api/v1/tasks/{task_id}`。 -- **批量查询**：`GET .../api/v1/tasks/?start_time=xxx&end_time=xxx&status=xxx`，单次时间跨度不超过 24 小时。 -- **取消任务**：`POST .../api/v1/tasks/{task_id}/cancel`，仅能取消 `PENDING` 状态的任务。 - -为避免频繁轮询浪费资源并触发限流，百炼已接入阿里云事件总线 EventBridge，支持任务完成后主动推送通知（HTTP 回调 URL 或 RocketMQ 两种方案）。事件源为 `acs.dashscope`，事件类型为 `dashscope:System:AsyncTaskFinish`，收到通知后只需调用一次查询接口即可获取结果。 - -## 开发者建议 - -- 妥善保存 `task_id`，切勿因未及时轮询到结果而重复创建任务。 -- 轮询间隔不宜过密（如约 15 秒），高频或生产场景优先使用异步完成通知。 -- 及时下载产物：图像/视频 URL 有效期通常为 24 小时，3D 模型下载链接有效期仅 2 小时。 -- 处理失败时读取响应中的 `code` 与 `message`，对照百炼错误码文档排查。 - -## 关联主题页 - -- [3d generation](../api/3d-generation.md) -- [video generation api](../api/video-generation-api.md) -- [image generation](../api/image-generation.md) -- [more about models](../api/more-about-models.md) -- [application call](../api/application-call.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md deleted file mode 100644 index 6b73c92e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/billing.md +++ /dev/null @@ -1,85 +0,0 @@ -# 计费 - -计费是百炼平台围绕模型推理、模型训练、模型部署及增值服务所建立的费用体系，涵盖按量付费、订阅制（[Token](token.md) Plan / Coding Plan）、预留容量（TPM 预留 / PTU）等多种方式，开发者可根据用量规模和业务场景灵活组合，实现成本最优。 - -## 计费场景与方式 - -百炼平台的计费覆盖以下主要场景： - -| 场景 | 计费方式 | 说明 | -|------|---------|------| -| 模型推理调用 | 按量付费（[Token](token.md) 计价） | 按输入/输出 [Token](token.md) 分别计价，部分模型支持阶梯计费 | -| 模型训练 | 按训练 Token 计费 | （训练数据 Token + 混合训练数据 Token）× 循环次数 × 单价 | -| 模型部署 | PTU / 模型单元 / Token 用量 | 三种方式创建后不可更改，需下线重建才能切换 | -| 知识库 | 按规格计费 | 标准版 0.03 元/小时，旗舰版 0.2 元/RCU/小时 | -| TPM 预留 | 按 kTPM 预付费 | 按天计费，输入/输出分别定价 | -| Token Plan 团队版 | 坐席订阅制 | 按 Credits 抵扣，198 元/坐席/月起 | -| Coding Plan | 月度订阅制 | 按模型调用次数限额，200 元/月 | - -## 费用抵扣顺序 - -多种付费方式并存时，系统按以下优先级自动抵扣： - -**免费额度 → 资源包 → 其他模型节省计划 → AI 通用型节省计划 → 按量付费** - -Token Plan 和 Coding Plan 的套餐额度独立于按量计费体系，不参与上述抵扣链路。 - -## 免费额度 - -首次开通百炼时自动发放新人免费额度，有效期 30~90 天。关键限制： - -- 仅适用于华北2（北京）地域、中国内地服务部署范围的实时推理 -- 不支持抵扣 Batch 调用、模型调优、模型部署、自定义模型 -- 主账号与 RAM 子账号共享 -- 建议开启"免费额度用完即停"功能，防止额度耗尽后自动扣费 - -## 成本优化方案 - -### AI 通用型节省计划 - -承诺每月消费金额获取阶梯折扣，最高可享 5.3 折。覆盖阿里直供全部模型，承诺金额 1,000 元起，支持 3/6/12/24 个月周期。当月未用完的额度自动清零，不可累积。 - -### 资源包 - -预先购买具体 Token 数量，用于抵扣特定模型的实时推理用量。适合用量明确且集中在单一模型的场景。 - -### Batch 调用优惠 - -支持 Batch 调用的模型，输入和输出 Token 单价均按实时推理价格的 50% 计费。 - -### 上下文缓存折扣 - -部分模型支持上下文缓存，命中缓存的输入 Token 按折扣系数消耗额度（例如 deepseek-v4-pro 为 0.08，即按 8% 折算）。Batch 调用与上下文缓存不能同时生效。 - -## 多地域定价差异 - -同一模型在不同地域的价格可能不同。华北2（北京）使用人民币定价，新加坡等国际地域按国际价格结算，通常高于国内价格。[API Key](api-key.md) 必须与 Base URL 同一地域，否则会报 `401` 错误。 - -## 账单查询 - -- **费用概览**：控制台"用量 & 费用 > 费用概览"查看当月总消费，支持按模型或 [API Key](api-key.md) 筛选 -- **模型用量**：按[业务空间](workspace.md)维度统计，数据延迟约 1 小时，不支持查看 30 天以前的数据 -- **出账时间**：大模型推理分钟级出账（2~10 分钟），批量推理和训练小时级出账 -- **分账管理**：通过[业务空间](workspace.md)标签按部门或项目归集费用，T+1 天生效 - -## 欠费与停止计费 - -账户可用额度小于 0 时视为欠费。Token Plan 和 Coding Plan 的套餐额度独立于账户余额，欠费期间可继续使用。停止计费的方式： - -- **模型推理**：停止 API 调用，删除不再使用的 [API Key](api-key.md) -- **模型部署**：在控制台下线已部署模型；包月预付费需额外退订实例 -- **模型训练**：无进行中的训练任务即不产生费用 -- **订阅制**：关闭自动续费，到期自动停止 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) -- [knowledge base](../guides/knowledge-base.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md b/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md deleted file mode 100644 index 2c3d22ce..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/context-window.md +++ /dev/null @@ -1,32 +0,0 @@ -# 上下文窗口 - -上下文窗口（Context Window）是模型在单次推理中能够接收并处理的输入与输出 Token 总长度的上限，单位通常为 Token，是衡量模型承载长文本、多轮对话、多模态输入能力的核心指标。在百炼平台，千问系列文本生成模型（如 `qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`）及视觉理解模型普遍支持 1M（约百万）Token 的上下文窗口，第三方模型如 `deepseek-v4-pro` / `deepseek-v4-flash` 同样提供 1M 上下文。 - -## 在百炼中的使用场景 - -- **长文档处理**：1M 上下文可一次性容纳数十万字的中文文档或数十张高分辨率图像，适合合同审阅、代码仓库分析、长报告摘要等场景，避免分段截断导致信息丢失。 -- **多轮对话**：聊天与 Agent 应用需在请求中累积历史消息，上下文窗口决定可保留的对话轮数。超出窗口后早期消息会被截断，需结合检索或摘要策略压缩历史。 -- **多模态输入**：视觉理解模型同时接收图像、视频与文本，图像 Token 按 `高 × 宽 / (32 × 32) + 2` 估算，单张图最高约 1600 万像素；视频最长支持 2 小时 / 2GB（Qwen3.7/3.6/3.5 系列），1 小时 / 2GB（Qwen3.5-Omni 系列含音频）。多模态输入会快速占用上下文窗口，需在长文本与多图之间权衡。 -- **工具调用与思考模式**：开启 `enable_thinking` 的思考模式、Function Calling 的工具定义与中间结果都会计入上下文消耗，复杂 Agent 流程需为工具输出预留充足窗口预算。 - -## 关键参数与配置 - -- **模型选型**：上下文长度随模型档位而定，文本/视觉模型以 1M 为主；向量、[重排序](rerank.md)、语音等模型按各自规格独立设定，调用前应以模型广场标注为准。 -- **对话历史管理**：仅 OpenAI 兼容 Responses 接口由平台自动管理对话历史，无需手动拼接 `messages`；OpenAI Chat Completions、Anthropic Messages、DashScope 原生接口均需调用方自行维护上下文长度与轮次，避免超出窗口导致请求失败或内容截断。 -- **Token 计费与限流**：上下文窗口内的输入与输出 Token 均参与计费和限流计量，长上下文请求会显著提升单次调用成本与首字延迟，建议按需裁剪历史与文档片段。 -- **超时与域名**：长上下文推理耗时较高，生产环境推荐使用业务空间专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（请求超时 3600 秒、SLA 99.9%），避免中心化或试用域名的较短超时与限流影响。 - -## 注意事项 - -- 上下文窗口是模型能力上限，不代表稳定可用长度；接近上限时模型对长尾内容的注意力可能下降，关键信息宜放在输入首尾。 -- 跨接口迁移时需确认目标兼容接口是否完整透传长上下文相关参数，部分兼容接口为协议一致可能不暴露全部原生参数，最全参数请使用 DashScope 原生接口。 -- 各地域、各模型对上下文窗口的支持存在差异，API Key、模型列表与接入域名不能跨地域混用，调用前请在目标地域的模型广场核对规格。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [get started with models](../guides/get-started-with-models.md) -- [model experience](../guides/model-experience.md) -- [more about models](../api/more-about-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md deleted file mode 100644 index 2b0d787d..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/cross-session-memory.md +++ /dev/null @@ -1,112 +0,0 @@ -# 跨会话记忆 - -跨会话记忆是指将对话中提取的关键信息和用户画像持久化存储，并在后续会话中通过语义检索召回并注入 Prompt 的能力，用于解决大模型上下文窗口无法跨会话延续的问题。百炼平台通过「记忆库（Memory Library）」与「长期记忆 API」提供该能力。 - -## 在百炼平台中的使用场景 - -跨会话记忆在以下场景中发挥作用： - -- **个性化智能体**：在多轮对话之间保留用户偏好、习惯和重要事件（如「每天上午 9 点提醒我喝水」），让智能体在新会话中仍然理解用户历史。 -- **长期用户画像**：通过自定义画像模板提取结构化属性（年龄、职业、偏好等），在后续对话中以固定字段持久化注入，适用于需要稳定属性支撑的业务。 -- **Agent 自动捕获/召回**：OpenClaw Agent 通过记忆插件在 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子中与长期记忆 API 交互，实现零侵入的跨会话记忆。 -- **应用观测中的记忆追踪**：在应用观测的调用链路中，记忆的写入与检索会作为 RETRIEVER、EMBEDDING 等节点出现，便于开发者定位记忆相关调用的延时与 Token 消耗。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，请求 Header 携带 `Authorization: Bearer $DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 下次对话调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（自动从对话提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -```bash -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope API Key；`userId` 用于隔离不同用户记忆空间。所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -## 记忆内容类型 - -记忆库提供两类持久化内容，可独立或组合使用： - -- **记忆片段**：从对话中自动提取的关键事件和信息，支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。适用于大多数长期记忆场景。 -- **用户画像**：基于画像模板（profile schema）从对话中提取的结构化属性，适用于需要固定属性持久化存储的场景。属性字段应清晰具体，避免「姓名/名称/名字」等同义字段并存，且不应期望一次对话就提取全部信息。 - -## 关键参数 - -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆实体 ID，用于标识归属对象，最大 64 个字符 | -| `messages` | 与 `custom_content` 二选一 | 对话消息列表，每个消息含 `role`（user/assistant）和 `content`，最多 50 条 | -| `custom_content` | 与 `messages` 二选一 | 自定义内容，最大 512 个字符，传入后忽略 `messages` | -| `profile_schema` | 否 | 画像模板 ID，在记忆库详情页获取 | -| `memory_library_id` | 否 | 记忆库 ID，最大 32 个字符，不传则使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不传则使用指定记忆库的默认规则 | -| `meta_data` | 否 | 用户自定义信息 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 用户 ID，用于隔离记忆空间 | -| `messages` | 是 | 查询对话内容，系统据此做语义检索 | -| `top_k` | 否 | 返回的记忆片段数量 | - -## 使用限制 - -| 项目 | 限制 | -| --- | --- | -| 全部接口总计 | 3000 QPM（阿里云账号级别） | -| 记忆片段 add 接口 | 120 QPM | -| 记忆片段 search 接口 | 300 QPM | - -## 有效期说明 - -记忆有效期在不同入口存在差异：长期记忆 API 文档指出「生成的记忆片段与用户画像暂无失效日期」，而控制台默认记忆片段规则预置了「默认有效期 180 天」，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 注意事项 - -- OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 API Key。 -- 应用观测暂不支持通过 Assistant API 创建的智能体应用；对高代码应用，仅能观测到入口 CHAIN 节点，不支持追踪其内部调用链路。 -- 记忆的写入与检索在应用观测中会体现为 RETRIEVER、EMBEDDING 等节点，可用于定位记忆相关调用的延时与 Token 消耗。 - -## 关联主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [memory library overview](../guides/memory-library-overview.md) -- [application monitoring](../guides/application-monitoring.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md deleted file mode 100644 index de46d3bd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/dashscope-sdk.md +++ /dev/null @@ -1,99 +0,0 @@ -# DashScope SDK - -DashScope SDK 是阿里云百炼平台官方提供的软件开发工具包，封装了模型与应用调用的原生（DashScope）接口，让开发者用少量代码即可接入通义千问、万相、Qwen-MT、Qwen-OCR 等模型能力以及智能体/工作流应用。相比 HTTP 直连和 [OpenAI 兼容接口](openai-compatible-interface.md)，DashScope 原生接口暴露的参数最完整、功能集最丰富。 - -## 适用语言与安装 - -DashScope SDK 主要提供 Python 与 Java 两种官方实现，部分场景也可用 HTTP（如 Node.js 借助 `axios`）替代： - -- **Python**：`python3 -m pip install -U dashscope` -- **Java**：通过 Maven / Gradle 引入 `com.alibaba:dashscope-sdk-java`，建议版本 `>= 2.12.0` -- **Node.js / 其他语言**：目前无官方 SDK，直接走 HTTP API（发起 POST 请求） - -> 注意：不同能力对 SDK 语言与地域的支持存在差异。例如 `qwen-deep-research` **仅支持 Python DashScope SDK，且仅限华北2（北京）地域**，暂不支持 Java SDK 与 [OpenAI 兼容接口](openai-compatible-interface.md)。 - -## 在不同场景中的使用 - -### 1. 调用文本生成模型（Qwen 系列） - -Qwen 系列可通过 OpenAI 兼容、Anthropic 兼容或 DashScope 原生三类接口调用。其中 DashScope 是百炼原生接口，**功能集最完整、参数支持最丰富**；当需要使用最全的采样参数、插件或业务字段而兼容接口未暴露时，应改用 DashScope 原生接口。 - -### 2. 调用专用模型（[more](../api/more.md) models） - -法律、意图理解、翻译、OCR 等专用模型大多支持 OpenAI 兼容或 DashScope 两种方式调用： - -- `farui-plus`（法律大模型）：通过 DashScope SDK（Python / Java）调用 -- `tongyi-intent-detect-v3` / `qwen-mt-plus` / `qwen3.5-ocr`：OpenAI 兼容或 DashScope 均可 -- `qwen-deep-research`（深度研究）：仅 Python DashScope SDK - -### 3. 调用图像生成与编辑模型 - -千问-图像（Qwen-Image）、万相（Wan/Wanx）、Z-Image 等图像模型均可通过 HTTP 或 DashScope SDK 调用，覆盖文生图、图像编辑、图像翻译、风格迁移等能力。 - -### 4. 调用智能体应用与工作流应用 - -已创建并发布的智能体应用、工作流应用可通过 DashScope SDK 集成到业务系统，二者调用方式一致： - -- Python：`from dashscope import Application`，调用 `Application.call(...)` -- Java：构造 `ApplicationParam` 后调用 `application.call(param)` -- HTTP 等价接口：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` - -响应统一为 `{"output": {...}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 关键参数与配置 - -### 鉴权（API Key） - -- SDK 通过 [API Key 鉴权](api-key.md)，**推荐将密钥写入环境变量 `DASHSCOPE_API_KEY`**，SDK 会自动读取，避免在代码中硬编码。 -- 调用特定地域（如华北2/北京）或子业务空间下的模型/应用时，需使用对应地域的 API Key，并按需提供 Workspace ID。 - -### 通用请求参数 - -- `model`（string，必选）：目标模型名称。 -- `messages`（array）：对话消息列表，按顺序排列，需由调用方维护上下文。 -- `app_id`（应用调用）：目标应用 ID，从控制台应用卡片复制。 -- `prompt` / `input`：用户输入内容。 -- `stream`（bool，可选）：是否[流式输出](streaming.md)；如 `qwen-deep-research` 的反问阶段需设为 `true`。 -- `session_id`（应用多轮对话）：由云端维护上下文，免去手动拼接历史。 - -### 模型专属参数示例 - -- **Qwen-MT（翻译）**：通过 `translation_options`（OpenAI SDK 中放入 `extra_body`）控制 `source_lang`、`target_lang`、`terms`（术语干预）、`tm_list`（翻译记忆）、`domain_prompt`（领域提示）。 -- **Qwen-OCR**：`messages.content` 为[多模态](multimodal.md)数组，可设 `min_pixels` / `max_pixels` 控制图像像素阈值。 - -## Python 快速示例（应用调用） - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'code={response.status_code}, message={response.message}') -else: - print(response.output.text) -``` - -## 使用建议 - -- 优先使用环境变量管理 API Key，区分不同地域的密钥。 -- 需要最完整功能与参数时选 DashScope 原生接口；追求生态兼容、迁移成本最低时可选 [OpenAI 兼容接口](openai-compatible-interface.md)。 -- 接入前先对照具体模型的 API 参考，确认其支持的 SDK 语言、协议与地域。 - -## 关联主题页 - -- [image generation](../api/image-generation.md) -- [more models](../api/more-models.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [application call](../api/application-call.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md deleted file mode 100644 index 1334d698..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/embedding.md +++ /dev/null @@ -1,54 +0,0 @@ -# 向量嵌入 - -向量嵌入（Embedding）是将文本、图像、视频等非结构化数据转换为固定维度的数值向量，使语义相近的内容在向量空间中距离也相近。百炼平台提供多类嵌入模型，支撑语义检索、RAG 召回、跨模态搜索、聚类推荐等下游任务。 - -## 在百炼平台的使用场景 - -- **RAG 知识库召回**：知识库创建时选择向量模型，对导入文档切片做嵌入入库；查询时对 Query 做同款嵌入，再与切片向量做相似度匹配，作为 RAG 流程的第一段召回。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；图片问答类固定使用 `multimodal-embedding-v1`（1024 维）；视觉理解场景自动切换为 `qwen3-vl-embedding`。 -- **本地 RAG 应用**：本地知识库方案默认调用百炼 embedding API 生成向量，也可替换为本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。受 embedding API 限流影响，单文件不建议超过 100 MB。 -- **跨模态检索**：多模态向量模型（如 `qwen3-vl-embedding`、`multimodal-embedding-v1`）将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。支持「独立向量」（逐项生成）与「融合向量」（多输入合并为 1 个向量）两种模式。 -- **框架集成**：LlamaIndex 路线将知识库部署在百炼云端，使用官方向量模型与智能切分，不支持自定义嵌入模型；如需灵活选择嵌入模型，应改用本地知识库方案。 - -## 模型与关键参数 - -通用文本向量模型当前推荐 `text-embedding-v4`（Qwen3-Embedding 系列，支持 100+ 语种）。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -请求参数： - -- `model`（必选）：模型名称。 -- `input`（必选）：字符串、字符串列表或文件。 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持。 -- `encoding_format`（可选）：当前仅支持 `float`。 - -调用方式支持 OpenAI 兼容接口（base_url：`https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。 - -## 批处理接口 - -大规模文本向量化可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行文本。需在 HTTP 请求头加 `X-DashScope-Async: enable` 启用异步模式，提交后通过 `task_id` 轮询结果。同时处理中任务不超过 50 个，并发运行上限 3 个，超出部分排队等待。 - -## 与 Rerank 的关系 - -向量嵌入负责「召回」，Rerank 模型负责「精排」。在知识库检索流程中，先由向量 + 关键词混合检索召回 TopK 切片，再由 `qwen3-rerank`、`qwen3-vl-rerank` 等排序模型对候选切片二次排序，最终按相似度阈值与最大召回数量返回。两者配合提升 RAG 命中准确率。 - -## 注意事项 - -- 知识库向量模型与维度在创建时选定，**维度不可更改**；Meta 抽取、多轮对话改写等索引配置在创建后也无法追加，需重建知识库。 -- 嵌入与 Rerank 是两类不同模型，不要混用接口：qwen3-rerank 走 `/compatible-api/v1/reranks`，qwen3-vl-rerank / gte-rerank-v2 走 `/api/v1/services/rerank/text-rerank/text-rerank`。 -- `gte-rerank` 系列将于 2026 年 5 月 30 日下线，建议迁移到 `qwen3-rerank`。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md b/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md deleted file mode 100644 index a303b8d6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/evaluation.md +++ /dev/null @@ -1,77 +0,0 @@ -# 评测体系 - -评测体系是百炼平台用于系统化衡量输出质量的一整套能力，覆盖**应用评测**（智能体/工作流应用）与**模型评测**（文本生成模型）两大场景，通过评测集、评测维度/评估器、评分方式与评测报告构成完整的评估闭环。 - -## 两类评测场景 - -百炼的评测能力按被评测对象分为两条相对独立的链路： - -- **应用评测**：评估已发布智能体应用、工作流应用的回答质量，支持自动评测（大模型基于知识库生成评测集并打分）和手动评测（人工标注打分）。当前存在新旧两套系统，新版以「评测任务 + 评估器 + 标签」组织。 -- **模型评测**：评估文本生成类模型的能力，支持自定义评测（自有数据集 + 自定义维度）和基线评测（C-Eval、MMLU、GSM8K、BBH 等公开数据集）。 - -两者共享一套核心思路：准备评测数据 → 定义评分规则 → 创建评测任务 → 查看报告。 - -## 评分方式 - -无论应用还是模型评测，评分方式基本归为三类，选型取决于是否有标准答案与是否需要语义理解： - -| 评分方式 | 原理 | 适用场景 | 费用 | -|----------|------|----------|------| -| 大模型评估（LLM/AI 自动评测） | 裁判模型按 Prompt 语义打分或分类 | 相关性、幻觉、内容安全等语义场景 | 产生 Token/裁判模型费用 | -| 规则评估（Code/自动化指标） | ROUGE、BLEU、Cosine、字符串匹配等算法 | 翻译、摘要、格式校验、精确匹配等确定性场景 | 无额外费用 | -| 人工评估 | 人工逐条标注 Pass/Fail 或打分 | 创意写作、专业领域主观判断 | 无裁判费用 | - -**选型路径**：有标准答案且格式固定 → 字符串匹配；有标准答案但表述多样 → 文本相似度；无标准答案需语义理解 → 大模型评估；需主观判断 → 人工评估。 - -## 评测集（数据基础） - -评测集是评测任务的数据输入，需在数据管理模块或评测系统中准备： - -- **应用评测集**：旧版分对话分析（`.xls/.xlsx`）与知识问答（`.jsonl`）；新版分智能体、工作流、自定义三类，具备版本管理，创建后类型不可改。 -- **模型评测集**：单轮对话，Excel 格式，每行含 **Prompt**（用户输入）与 **Completion**（期望输出）。参评模型对每条 Prompt 推理，评分参考 Completion。 - -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 - -## 评测维度与评估器（评分规则） - -评分规则是评测体系的核心组件，创建为模板后可被多个任务复用： - -- **模型评测的评测维度**提供五种评分器类型：大模型评估-数值型、大模型评估-分类型、规则评估-文本相似度、规则评估-字符串匹配、人工评估-分类型。维度类型创建后不可修改。 -- **应用评测的评估器**支持预置模板（通用质量、智能体、文本匹配、文本相似度、格式校验）、自定义（LLM 评估器 / Code 评估器）以及基于历史评测任务标注结果抽象生成。每个评测任务最多添加 10 个评估器，建议组合 3-5 个从不同维度评估。 - -## 关键参数与配置 - -- **评分范围**（数值型）：整数打分区间，默认 0-5，建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**：判定 Pass/Fail 的分界线，数值型步长 0.1，相似度型步长 0.01。 -- **评分器 Prompt**（大模型评估）：至少含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符；用于指导裁判模型如何打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色/行为规范，通常可留空，注意勿与评分器 Prompt 混淆。 -- **数据来源**：评测数据集（含 Prompt+Completion，产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP 等，按所选模型动态加载。 -- **标签类型**（应用评测）：分类、布尔值、数字、文本，用于标注评测与观测数据。 - -## 计费说明 - -费用主要由**被评测模型推理费用**与**裁判模型评分费用**两部分构成： - -- 使用推理结果集可免去推理费用。 -- 规则评估、人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -成本优化建议：先用 50-100 条小规模验证 → 保存推理结果集复用 → 确定性场景优先用规则评估。 - -## 限制与注意事项 - -- 模型评测当前仅支持文本生成类模型，且仅支持控制台操作，不提供公开 API/SDK。 -- 基线评测仅北京地域可用，不支持下载评测结果。 -- 数据管理与数据清洗/增强能力仅适用于华北2（北京）地域。 -- 任务提交后不可更换目标模型、维度类型创建后不可修改，选错需删除重建。 -- 应用自动评测仅面向已发布且已配置知识库的应用，须开通应用观测并具备相应权限。 -- LLM 评分器存在位置偏差与自我偏好偏差，1-3% 的分差通常为噪声，建议定期人工抽查校准。 - -## 关联主题页 - -- [application evaluation](../guides/application-evaluation.md) -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md b/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md deleted file mode 100644 index 397e4bd9..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning-and-deployment.md +++ /dev/null @@ -1,92 +0,0 @@ -# 模型调优与部署 - -模型调优与部署是百炼平台“模型生产”链路的核心环节：先通过微调训练把领域知识、任务能力或人类偏好写入模型参数，再将调优产物发布为独立、资源专享的在线推理服务。完整链路为 **模型调优 → 模型压缩（可选）→ 模型部署**。 - -> **注意**：微调、压缩、部署与调用等能力**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 - -## 在百炼平台的使用场景 - -- **深度定制模型**：当 Prompt 工程、插件调用仍无法满足效果时，通过微调把领域知识与任务能力固化到模型中，覆盖文本生成、视觉理解（Qwen-VL）、图像/视频生成（万相）、语音合成（CosyVoice）等多种模态。 -- **降低部署成本**：对全精度微调模型做量化压缩，在保持能力的前提下降低部署所需的 MU 规格。以 qwen3.5-flash 微调模型为例，压缩后部署成本可节省约 56%（MU1*2 → MU8*1）。 -- **上线专属推理服务**：将调优或导入（LoRA）的模型部署为独立服务，满足高并发、低延迟的生产需求。 -- **数据准备前置**：调优前通过数据管理功能创建、清洗、增强训练集与评测集（详见相关数据管理主题）。 - -## 模型调优 - -推荐按递进顺序组合使用文本生成的三种训练方式：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 最低数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | `chosen`/`rejected` 回答对 | - -训练模式分**全参训练**与**高效训练（LoRA）**，两者费用相同。官方建议在模型支持全参训练时优先选择全参（效果更好、性价比更高）；LoRA 适合训练时间/成本敏感或数据集较小的场景。不同模型支持的训练方式差异明显：Qwen3-32B、Qwen2.5 系列等支持全部 5 种，而部分新模型（如 Qwen3.5-Plus/Flash）往往仅支持 `sft`。视觉理解仅支持 SFT，万相图像/视频与 CosyVoice 仅支持高效微调（LoRA / `efficient_sft`）。 - -### 关键超参数（文本生成） - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级（控制台默认值与 API 示例取值不一致，请以实际训练方式对应的量级为准）。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议 3~5 次，>10000 建议 1~2 次。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型支持的最大值；SFT 会**丢弃**超长数据，DPO 则**截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -支持控制台（零代码入门）与 API/命令行（统一“上传数据集 → 创建训练任务 → 查询状态 → 管理产物”流程）两种方式。 - -## 模型压缩（可选） - -百炼的模型压缩特指**量化**，不涉及结构剪枝或知识蒸馏，位于调优与部署之间。仅支持平台微调产出的自定义模型。 - -- **不可逆**：压缩后的模型不支持继续微调，也不支持二次压缩。 -- **量化模板**：模板名中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大。 -- **校准数据**：可选，建议选择与目标推理场景语义相近的数据集（最多 5 个），用于提升量化精度。 -- 任务状态：PENDING → QUEUING → RUNNING → SUCCEEDED / FAILED / CANCELING → CANCELED；仅 PENDING/RUNNING 可停止，仅终态可删除。 -- 压缩任务本身限时免费，压缩后模型在部署阶段按 MU 规格计费。建议在免费期内尝试多个量化模板并用业务测试集验证后再上线。 - -## 模型部署 - -部署将模型发布为在线推理服务，可选三种**互斥**的计费方式（创建后不可更改，切换需先下线再重新部署）： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM 吞吐，额度内不限速，TPS 通常提升约 1.5~2.0 倍，适合流量可预估的高负载生产。支持长输入（部分模型最高 200K token）与前缀缓存折扣；超额或超长时自动转按量计费，业务不中断。 -- **模型单元（MU）**：按使用时长 × 单元数量计费，资源独占，性能指标可自定义，支持 PD 分离降低首 Token 延迟。 -- **按 Token 使用量**：按调用输入/输出 Token 计量，不使用不计费，仅支持 SFT 高效训练后的自定义模型，主要用于效果验证。 - -### 模型导入（LoRA） - -可通过“我的模型”将本地训练的 LoRA 模型从 OSS 导入，**仅支持 LoRA，不支持全参微调模型**。关键约束： - -- OSS Bucket 需添加 `bailian-datahub-access` 标签（值 `read`），不支持归档存储与根目录文件。 -- 必需文件：`adapter_model.safetensors` 与 `adapter_config.json`。 -- `rank` 必须为 8、16、32、64 之一，且所有 LoRA 层使用相同 rank。 -- 不得修改原始 vocab 或 chat_template；VL 模型必须冻结 VIT。 - -### PTU 关键响应字段 - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度，`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的额度 token 数。 -- `cached_tokens`：前缀缓存命中数。 - -## 典型工作流 - -1. 准备并（可选）清洗/增强训练数据集。 -2. 提交微调训练任务，等待完成并管理调优产物。 -3. （可选）对全精度微调模型做量化压缩，降低部署规格。 -4. 将调优或导入的模型部署为在线服务，选择合适的计费方式。 -5. 通过标准 API 端点调用部署后的模型进行推理。 - -## 面向开发者的要点 - -- 全流程锁定华北2（北京）地域，注意 API Key 与子账号权限。 -- API 创建调优任务时，涉及计费的超参必填；部署计费方式一经选定不可改。 -- 压缩不可逆且免费期有限，先验证再上线；导入 LoRA 时严格遵守 rank 与文件约束。 - -## 关联主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md deleted file mode 100644 index 3098445e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/fine-tuning.md +++ /dev/null @@ -1,89 +0,0 @@ -# 模型微调与生产链路 - -模型微调（Fine-tuning）是在 Prompt 工程、插件调用等手段仍无法满足效果时，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数的深度定制手段。它是百炼平台「数据准备 → 模型调优 → 模型压缩（可选）→ 模型部署」这条模型生产链路的核心环节。 - -> **注意**：微调、压缩、部署与调用能力**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 - -## 完整生产链路 - -一条典型的自定义模型生产链路包含四个阶段，前后衔接： - -1. **数据准备**：创建、清洗、增强训练集与评测集（仅控制台，暂无数据处理 API）。 -2. **模型调优**：指定基础模型、训练集/验证集与超参数，提交微调任务，训练完成后得到自定义模型。 -3. **模型压缩（可选）**：对全精度微调模型做量化，降低部署所需 MU 规格、减少推理成本。 -4. **模型部署**：把微调或导入的模型发布为独立、资源专享的在线推理服务，再通过标准 API 调用。 - -## 场景一：模型调优 - -按模态划分，不同模型支持的训练方式差异明显： - -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异（如 Qwen3-32B、Qwen2.5 系列支持全部 5 种，部分新模型仅支持 `sft`）。 -- **视觉理解（千问 VL）**：支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像/视频生成（万相）**：仅支持 SFT-LoRA 高效微调。 -- **语音合成（CosyVoice）**：仅支持 `efficient_sft`，且当前只能通过 API 发起，控制台暂不支持。 - -文本生成推荐按递进顺序组合：`CPT（可选）→ SFT → DPO（可选）`。 - -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下 `chosen` / `rejected` 回答对 | - -训练模式分**全参训练**与**高效训练（LoRA）**，两者费用相同；官方建议模型支持全参时优先全参（效果更好、性价比更高），LoRA 适合训练时间/成本敏感或数据集较小的场景。 - -## 场景二：模型压缩（量化） - -百炼的模型压缩特指**量化**，不涉及剪枝或蒸馏。它把全精度微调模型转为低精度版本，在保持能力前提下降低部署 MU 规格。以 qwen3.5-flash-2026-02-23 为例，压缩前 MU1*2（108 元/小时）、压缩后 MU8*1（47 元/小时），成本节省约 56%。 - -> **注意**：压缩不可逆，压缩后模型不支持继续微调或二次压缩；仅支持百炼平台微调产出的自定义模型。 - -## 场景三：模型部署 - -部署提供三种互斥的计费方式，创建后无法更改： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产。支持 PTU 长输入（部分模型最高 200K token）与前缀缓存折扣，超额自动转按量计费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占，支持部分预置模型与所有调优后模型。 -- **按 Token 使用量**：不使用不计费，仅支持 SFT 高效训练后的自定义模型，主要用于效果验证。 - -此外可通过**我的模型**从 OSS 导入本地训练的 **LoRA** 模型（不支持全参微调模型），rank 须为 8/16/32/64 之一，必需 `adapter_model.safetensors` 与 `adapter_config.json`，且不得修改 vocab 或 chat_template。 - -## 关键参数与配置 - -文本生成调优的常用超参： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议 3~5，>10000 建议 1~2。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型最大值；SFT **丢弃**超长数据，DPO **截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：不同文档默认学习率取值不一致（控制台面板显示 `3e-4`、API SFT 全参示例为 `1.6e-5`）。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频、CosyVoice 各有独立超参集（如万相 `max_steps`/`generation_type`，CosyVoice 分 `lm_*` 韵律与 `fm_*` 音色两组网络的参数，8 个子字段全部必填）。 - -## 数据格式要点 - -- **SFT（文本）**：ChatML，每行一个 `{"messages":[...]}`，所有 assistant 输出都会被训练；不支持 OpenAI 的 `name`/`weight`。 -- **SFT 思考模型**：仅训练最后的 assistant 输出，思考内容用 `` 标签包裹并保留前后换行。 -- **DPO**：ChatML 加 `chosen` / `rejected` 字段。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解/图生视频**：需按字段规范提供图片、视频路径或帧列表。 - -## 面向开发者的使用方式 - -- **控制台（推荐入门）**：模型调优页面创建任务 → 选训练方式与模型 → 配置训练集/验证集与 Checkpoint → 训练 → 部署 → 评测。 -- **API / 命令行**：统一四步流程——上传数据集（`POST /api/v1/files`）→ 创建调优任务 → 轮询任务状态与训练指标 → 部署为在线服务并调用推理端点。 - -若调优后评测效果不佳，最简单的改进办法是收集更多高质量数据继续训练；压缩免费期内可对同一模型尝试多个量化模板，用业务测试集验证后再上线。 - -## 关联主题页 - -- [fine tuning](../guides/fine-tuning.md) -- [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md index 93ca5304..eae4fb05 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/function-calling.md @@ -1,41 +1,44 @@ -# 函数调用（Function Calling） +# 函数调用 -函数调用（Function Calling）是让大模型在推理过程中根据用户输入，自主判断并"调用"外部工具（自定义函数、内置能力或插件）以获取实时信息、执行精确计算或操作外部系统的能力。它是弥补大模型原生局限、构建 Agent 与复杂应用的核心机制。 +函数调用（Function Calling）是百炼平台中大模型主动识别用户意图、自主选择并执行外部工具（如插件、Skill、代码解释器、搜索服务等）的核心能力机制。它不是简单的 API 转发，而是模型在推理过程中基于语义理解，按需生成结构化工具调用请求（含工具 ID 与参数），再由平台运行时安全调度、执行并注入结果回上下文，最终生成自然语言回复。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中，这个概念如何使用 -百炼在多个层面暴露了 Function Calling 能力： +函数调用在百炼平台并非单一接口能力，而是贯穿多个抽象层级的横切行为模式，具体体现为以下三类典型场景： -- **文本 / 视觉生成模型**：所有 Qwen3 及以上通用文本与视觉理解模型均支持自定义工具调用。以 `qwen3.7-plus` 为代表的旗舰模型工具调用完整、上下文长（1M），适合 AI 编程与 Agent 开发；效果确认后可切到 `qwen3.6-flash` 降本，功能与上下文保持一致。 -- **实时多模态（Omni-Realtime API）**：基于 WebSocket 的实时音视频对话同样支持工具调用。客户端通过 `session.update` 事件在会话中声明工具，模型触发调用后由 `response.function_call_arguments.done` 服务端事件返回调用参数，客户端执行后再用 `conversation.item.create` 事件回传工具结果。 -- **应用构建（智能体 / 工作流）**：新版智能体（Agent 2.0）将知识库、MCP、插件等能力统一抽象为"工具"，由智能体自主规划调用顺序，形成"规划-执行-反思"链路。工作流应用则把工具作为固定节点按编排顺序执行，不由模型主动规划。 -- **插件（Plug-in）与 Assistant API**：插件是工具集合，本质也是工具调用。智能体应用 / Assistant API 中，模型依据工具名称与描述判断是否调用；无需调用时直接生成结果。 +- **API 层显式声明调用**：在 DashScope 原生接口或 Anthropic 兼容 Messages 接口中，开发者通过 `tools` 字段传入工具定义（JSON Schema），并可选设置 `tool_choice` 控制调用策略（如 `"auto"`、`{"type": "function", "function": {"name": "calculator"}}`）。模型据此生成 `tool_calls` 输出，平台自动完成调用、结果注入与多轮续写。 + +- **智能体托管运行时（Managed Agents）中的隐式决策调用**：在 Managed Agents 中，函数调用完全由 Agent 运行时自主触发。开发者只需在 Agent 配置中挂载 Skill 或插件，无需在每次请求中重复声明 `tools`。Agent 根据 `SKILL.md` 的 `description` 或插件元数据，在会话中动态判断是否调用、调用哪个 Skill/插件，并处理其输入输出——整个过程对上层应用透明。 -## 内置工具与自定义工具 +- **插件与 Skill 的能力封装层调用**：插件（Plugin）和 Skill 是函数调用的“能力载体”。插件面向通用服务（如 `quark_search`, `code_interpreter`），支持业务透传参数与鉴权；Skill 面向文件/数据处理任务（如 `xlsx-parser`），依赖精准的 `description` 触发。二者均通过统一的工具调用协议被模型识别和调度，但生命周期、配置方式与安全约束不同（如 Skill ZIP 包禁止二进制，插件需通过安全扫描）。 -- **自定义工具（Function Calling）**：开发者自行定义工具名称、描述与参数结构，模型据此决定何时调用、如何填参，应用侧执行后将结果回填模型生成最终回复。 -- **内置工具**：联网搜索、代码解释器、网页抓取等由平台预置，无需复杂配置即可开启，是 Function Calling 的开箱即用形态。 +> ⚠️ 注意：OpenAI 兼容的 `/v1/chat/completions` 接口**不支持显式传入 `tools`**；其增强版 `/v1/chat/completions`（即 OpenAI 兼容-Responses）虽能自动启用搜索/代码解释器，但属于平台预置的全自动流水线，**不开放自定义工具注册与参数控制**，与 DashScope 和 Managed Agents 的可控性存在本质差异。 -## 关键参数与配置 +## 关键参数和配置 -- **模型选型**：推荐具备强工具调用能力的模型（如千问-Max / `qwen3.7-plus` 系列）。 -- **思考模式**：可通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制），配合工具调用提升规划质量。 -- **ReAct 最大轮次**：智能体中取值 1-50，限制单次会话内工具调用的最大次数，防止无限循环。 -- **实时 API 工具配置**：通过 `session.update` 的 `tools` 字段声明可用工具；工具调用结果需经 `conversation.item.create` 回传。 -- **插件调用**：通过 Assistant API 调用时需正确传递工具 ID（如 `calculator`、`code_interpreter`），且每个智能体应用最多添加 10 个工具。 +| 参数/配置项 | 所属场景 | 说明 | 是否必需 | 备注 | +|-------------|----------|------|----------|------| +| `tools` | DashScope / Anthropic Messages API | 工具定义数组，每个元素为符合 JSON Schema 的对象，描述工具名称、描述、参数类型与约束 | 否（启用函数调用时必需） | 不支持 [OpenAI 兼容接口](openai-compatible-interface.md)；Schema 中 `required` 字段必须准确声明必填参数 | +| `tool_choice` | DashScope / Anthropic Messages API | 控制模型是否及如何调用工具：`"auto"`（默认）、`"none"`（禁用）、或指定工具名 | 否 | 指定工具名时，模型将强制调用该工具（即使不必要），适用于确定性流程 | +| `biz_params` | Assistant API / 插件调用 | 业务系统透传的上下文参数（如用户 ID、会话 ID），供插件后端鉴权或个性化处理 | 否（按插件需求） | 仅对支持业务透传的插件生效；不参与模型推理，不暴露给模型 | +| `description`（Skill） | Skill | Skill 的功能描述文本，是模型触发调用的唯一依据 | 是 | 必须包含适用输入、支持操作、典型关键词、**明确的不适用场景**；模糊描述将导致误调用 | +| `enable_search` / `enable_code_interpreter` | DashScope 原生接口 | 布尔开关，快捷启用预置工具链 | 否 | 仅 DashScope 支持；与 `tools` 互斥，二者不可同时使用 | -## 开发建议 +## 面向开发者，简洁实用 -- 优先用内置工具满足通用需求（搜索、计算、代码执行），减少自定义成本。 -- 自定义工具时，工具名称与描述要清晰准确，直接影响模型是否正确触发调用。 -- 需要精确、可控流程时用工作流把工具固化为节点；需要动态规划时用智能体让模型自主调用。 -- 旧版智能体的自定义插件有 5 秒超时限制，设计工具时注意执行时长。 +- ✅ **优先选 DashScope 接口**：若需精细控制工具调用（如自定义工具、指定参数、多工具协同），务必使用 DashScope 原生 endpoint（`/api/v1/services/aigc/text-generation/generation`），而非 [OpenAI 兼容接口](openai-compatible-interface.md)。 +- ✅ **Skill 描述要“防错”**：写 `SKILL.md` 的 `description` 时，用“当用户说……时可用，但当用户说……时**不可用**”句式，比单纯罗列功能更有效。 +- ✅ **插件调用前必验权限**：子账号首次使用插件，需主账号授予 `ram:CreateServiceLinkedRole` 权限，否则返回错误码 `140052`。 +- ❌ **勿在 [OpenAI 兼容接口](openai-compatible-interface.md)中传 `tools`**：该字段会被忽略，且可能引发 400 错误。 +- ❌ **勿在 Managed Agents Session 创建时传 `tools`**：Agent 已绑定 Skill/插件，工具集由 Agent 快照固化，请求体中添加 `tools` 字段无效。 +- 🔧 **调试技巧**：开启 `stream=true` 并监听 `tool_calls` 事件（DashScope）或 `event: tool_call`（Managed Agents SSE），可实时观察模型是否识别意图、参数是否正确生成。 ## 关联主题页 -- [omni realtime api](../api/omni-realtime-api.md) -- [model experience](../guides/model-experience.md) -- [llm application](../guides/llm-application.md) +- [qwen api reference](../api/qwen-api-reference.md) +- [managed agents api](../api/managed-agents-api.md) - [plug in](../guides/plug-in.md) +- [skill](../guides/skill.md) +- [application component api reference](../api/application-component-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md deleted file mode 100644 index 3062222e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/knowledge-base.md +++ /dev/null @@ -1,87 +0,0 @@ -# 知识库 - -知识库是阿里云百炼平台基于 RAG（[检索增强生成](rag.md)）技术构建的私有数据管理能力，用于为大模型补充私有数据与最新信息，使应用能够准确回答特定领域问题。知识库仅支持在中国站华北2（北京）地域开通和使用，提供标准版与旗舰版两种规格，并配套日志监控、API、效果优化与计费体系。 - -## 核心能力 - -知识库对私有数据或文件进行语义检索，可找出语义相同或相近的内容，即使关键词匹配度极低甚至为零。检索结果可作为上下文喂给大模型生成回答，也可仅作为检索能力单独使用。支持挂载到[智能体应用](agent-application.md)、工作流应用，或通过阿里云百炼 SDK、LlamaIndex、Spring AI Alibaba 等框架集成到外部应用。 - -知识库类型分为四类，单一知识库不支持同时选择多个类型： - -- 文档搜索：对非结构化文档做语义检索，细分为基础文档问答、图文并茂回复、视觉理解（富文本文档）、极速问答四种使用场景。 -- 数据查询：对结构化表格数据做检索。 -- 图片问答：基于 multimodal-embedding-v1（1024 维）对图片做检索。 -- 音视频搜索：对音频、视频内容按时间轴结构化检索。 - -## 在百炼平台中的使用场景 - -### [智能体应用](agent-application.md) - -在 Agent 2.0 架构中，知识库作为工具由智能体自主规划调用，与 MCP 等外部工具统一调度；在旧版（Agent 1.0）中，则先检索知识库再决策是否调用其他工具。可在应用配置页文档知识库右侧点击「+ 添加知识库」接入，并设置相似度阈值与权重，还可通过标签限定查询范围以提升准确性。开启「展示回答来源」后，回答会以角标形式展示知识来源与源文件/源网页地址。 - -### 工作流应用 - -将知识库节点拖入画布，配置输入变量（query）、选择固定知识库或动态引入、设置 TopK，用于在预定义流程中精确控制检索环节。 - -### 外部应用集成 - -- SDK：通过阿里云百炼 SDK 调用检索能力，子账号需获取 `AliyunBailianDataFullAccess` 策略并加入[业务空间](workspace.md)。 -- LlamaIndex（Python 3.9+）：将知识库部署在云端，使用默认智能文档切分与官方向量模型，不支持自定义切分与嵌入模型；如需灵活切分应改用本地知识库方案。 -- Spring AI Alibaba（Spring Boot 3.x，JDK 17+）：可调用百炼智能体/工作流应用并检索百炼知识库，应用集成推荐变量名 `DASHSCOPE_API_KEY`，知识库检索推荐 `AI_DASHSCOPE_API_KEY`。 -- 本地 RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。 - -### API 编排 - -通过百炼 OpenAPI（`bailian/2023-12-29`，ROA 风格）可程序化管理知识库全链路：申请上传租约 → 上传文件 → 添加文件到类目 → 轮询文件解析状态（INIT/PARSING/PARSE_SUCCESS）→ 初始化知识库 → 提交索引任务 → 轮询任务状态直至 COMPLETED。每[业务空间](workspace.md)最多创建 500 个类目，类目类型目前仅支持 `UNSTRUCTURED`。 - -## 关键参数与配置 - -### 规格与并发 - -| 规格 | 最高检索并发 | 平台存储空间 | 价格 | -| --- | --- | --- | --- | -| 标准版 | 1 QPS（固定，不可调） | ≤ 100 GB | 0.03 元/知识库/小时 | -| 旗舰版 | 50–10,000 QPS（可调，对应 1–200 RCU） | ≤ 9,999 GB | 0.2 元/RCU/小时 | - -RCU（Retrieval Compute Unit）是检索并发能力度量单位，1 RCU 约支撑最高 50 QPS，所需 RCU = 向上取整（检索峰值 QPS ÷ 50）。变配按发生时间分段计费，同一知识库 1 个自然日内最多变配 1 次。 - -### 向量与切片 - -- 向量模型：文档搜索、数据查询、音视频搜索类支持 text-embedding-v4、text-embedding-v3（均为 512 维）；图片问答类仅支持 multimodal-embedding-v1（1024 维）。向量维度不支持更改。 -- 文本切片长度上限：单个切片 6,000 [Token](token.md)；编辑切片（UpdateChunk）长度限制为 10–6,000 字符；删除切片（DeleteChunk）单次最多 10 个。 -- 召回文本切片数量：单次查询最多召回 20 个切片。 -- 切片方式：智能切分（保留语义完整性）或按长度切分。**知识库一旦创建，无法再配置 metadata 抽取，也无法更改文档切分 chunk**，需在创建时一次性规划好元数据与切片策略。 - -### 检索参数 - -- 相似度阈值：仅语义相似度高于此阈值的文本切片才会被召回。阈值过高（如 0.60）可能导致无召回结果或丢弃相关切片。 -- 召回片段数（K 值）：取值范围 1–20，调大可提升完整性但增加 [Token](token.md) 消耗；拼装后总长度超出大模型输入限制会被截断，并非越大越好。 -- 初步向量检索 TopK / 初步关键词检索 TopK：默认 50，取值范围 10–100，影响送入排序模型的切片数量与成本。 -- 权重：仅在**同类型知识库之间生效**，用于干预多知识库召回顺序。 - -### 解析方式 - -导入文件时可选择解析方式：电子文档解析（不支持插图与图表）、文档智能解析（提取插图文本与摘要）、大模型文档解析（支持对插图和图表提问，需配合选择模型）、Qwen VL 解析（仅图片，可传入 Prompt）、音视频解析（语音识别、视频帧提取、剧情解析按时间轴结构化对齐）。 - -## 效果优化 - -当出现召回不完整或内容不准确时，建议先建立评测基线（至少 100 组问题，覆盖事实型/比较型/教程型/分析型），再按 RAG 三阶段诊断改进： - -1. 建立索引：优化源文件排版（优先 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写（创建时开启，后续无法补开）。 -2. 检索召回：为文件添加标签过滤、配置元数据做结构化搜索、采用智能切分保留语义完整性、调整相似度阈值与召回片段数。 -3. 生成答案：更换为能力更强的商业模型（如通义千问 Max/Plus/QwQ）、优化提示词模板（限定输出、少样本提示、内容分隔标记且 `${documents}` 只出现一次）。 - -## 日志监控 - -检索日志由日志服务（SLS）承载，首次使用需授权角色 `AliyunServiceRoleForSFMAccessSLS` 并创建 LogStore。每条日志 topic 为 `log_dispatch`，包含 request_id、pipeline_id、workspace_id、latency、response_status_code、response_code 等字段。建议搭建调用量趋势、TopN 知识库排名、业务错误率与 HTTP 5xx 错误率等监控。 - -## 关联主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [frameworks](../api/frameworks.md) -- [llm application](../guides/llm-application.md) -- [application use cases](../guides/application-use-cases.md) -- [data connection overview](../guides/data-connection-overview.md) -- [application component api reference](../api/application-component-api-reference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md deleted file mode 100644 index f1f996e6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/long-context.md +++ /dev/null @@ -1,45 +0,0 @@ -# 长上下文 - -长上下文指模型在单次调用中能够接收并处理的输入 token 上限（部分模型还包含输出）。上下文窗口越大，单次请求可携带的对话历史、文档、图像、视频等数据越多，能减少多轮拼接与外部检索的复杂度。 - -## 在百炼平台中的使用场景 - -- **文本生成**：Qwen3.7-max / Qwen3.7-plus / Qwen3.6-flash 以及 deepseek-v4 系列均提供 1M token 上下文窗口，可在单轮内喂入长文档、长对话历史或多文件做总结、问答与代码调试。 -- **视觉理解**：Qwen3.7-plus 等模型支持 1M 上下文，可接收最长 2 小时视频（2GB）或大段图文混合输入，进行视频内容分析与 OCR。 -- **PTU 部署**：预置吞吐部署支持长输入（部分模型最高 200K token），通过阶梯容量系数折算 TPM，超出额度或输入超过模型上限时自动转为按量计费，业务不中断。 -- **对话历史管理**：OpenAI 兼容 Responses 接口由平台自动维护对话历史，无需在请求中手动拼接 messages；其他接口需由调用方维护上下文长度与轮次，避免超出模型上下文窗口。 - -## 关键参数与配置 - -- **上下文长度**：按模型不同，常见档位为 32K / 64K / 128K / 200K / 1M。选型时确认目标模型在对应兼容接口下的实际上下文上限。 -- **PTU 长输入阶梯系数**：超出 32K 的输入按更高系数折算 TPM。以 glm-5.1 为例，`[0, 32K)` 系数 1.0，`[32K, 200K]` 输入系数 1.33 / 输出 1.17。 -- **前缀缓存折扣**：命中缓存的输入 token 按模型对应折扣折算容量（如 glm-5.1 为 0.2，deepseek-v4-pro 为 0.08）。 -- **自动溢出阈值**：千问 128K、DeepSeek 64K 为模型上限阈值，超过即转按量计费，响应头返回 `x-dashscope-ptu-overflow: true`。 -- **图像 token 计算**：视觉模型每张图片 token 数按 `h x w / (32 x 32) + 2` 估算，叠加在上下文总额内。 - -## 容量与额度评估 - -长输入场景下建议先用控制台的 PTU 容量计算器评估额度：根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推算输入 TPM 和输出 TPM，避免意外转为按量计费。PTU 部署的响应包含以下与额度相关的字段： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。OpenAI Chat 兼容为 `usage.prompt_tokens_details.cached_tokens`；OpenAI Responses 为 `usage.input_tokens_details.cached_tokens`；Anthropic 兼容暂不返回。 - -## 限制与注意事项 - -- **功能完整度**：兼容接口为保证协议一致性，可能不暴露百炼原生全部参数；如需最全采样参数与插件能力，建议改用 DashScope 原生接口。 -- **上下文窗口 ≠ 输出上限**：1M 上下文主要指输入侧，单次输出长度仍受模型自身限制，长输入场景需评估输出截断风险。 -- **历史轮次管理**：迁移到不自动管理历史的接口时，需自行裁剪历史消息，避免累积超出窗口导致请求失败或尾部被截断。 -- **跨接口迁移**：从 OpenAI / Anthropic 迁移时应先确认目标 Qwen 模型在对应兼容接口下是否支持所需上下文长度与参数（如 `temperature`、`tools`、`stream` 等），再决定接口选型。 -- **地域差异**：各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，长上下文模型的可用性可能与地域相关，部署类操作仅适用于华北2（北京）地域。 - -## 关联主题页 - -- [qwen api reference](../api/qwen-api-reference.md) -- [more about models](../api/more-about-models.md) -- [model experience](../guides/model-experience.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md new file mode 100644 index 00000000..d913c435 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/long-term-memory.md @@ -0,0 +1,47 @@ +# 长期记忆 + +长期记忆是百炼平台提供的结构化、持久化的用户信息管理能力，用于跨会话持续捕获、存储、检索和聚合用户偏好、习惯、意图与事实性信息（如“每天9点喝水”“喜欢咖啡因饮品”），使智能体具备真正的上下文连续性和个性化理解能力。 + +## 在百炼平台的不同场景中如何使用 + +- **智能体（Agent）应用**：通过 `AddMemory` 自动从对话中提取事件性/意图性内容（如提醒、偏好、承诺），或手动写入 `custom_content`；在后续调用中通过 `SearchMemory` 语义召回相关记忆，并注入提示词上下文，实现个性化响应。支持与新版智能体的工具调度体系深度集成（如作为 `memory_search` 工具被自动调用）。 + +- **OpenClaw Agent**：通过官方插件 `@modelstudio/modelstudio-memory-for-openclaw` 启用零代码自动捕获（`autoCapture`）与自动召回（`autoRecall`），无需修改业务逻辑即可获得长期记忆能力；插件内置 `memory_store`/`memory_search` 等标准工具，Agent 可在运行时动态调用。 + +- **工作流（Workflow）与高代码应用**：通过直接调用长期记忆 API（如 `/api/v2/apps/memory/add` 和 `/api/v2/apps/memory/memory_nodes/search`）实现细粒度控制；可结合 `biz_params` 或自定义节点，在流程中触发记忆写入、画像聚合（`GetUserProfile`）或条件性召回。 + +- **记忆库统一管理**：所有场景均基于同一套记忆基础设施——记忆库（Memory Library）。同一 `memory_library_id` 可被多个应用共享，`user_id` 作为核心隔离维度，确保数据边界清晰；支持在控制台统一配置规则（如过期策略、默认模板）、调试检索效果、查看记忆实体。 + +- **用户画像构建**：配合 `CreateProfileSchema` 定义结构化属性（如 `age`, `diet_preference`, `timezone`），在 `AddMemory` 中传入 `profile_schema` ID，系统将自动从对话中抽取并归一化为 JSON 格式画像；后续可通过 `GetUserProfile` 获取完整聚合结果，供下游服务直接消费。 + +## 关键参数和配置 + +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `user_id` | string | ✅ | 用户唯一标识（≤64 字符），所有操作以此为数据隔离粒度；不同 `user_id` 的记忆完全不可见。 | +| `memory_library_id` | string | ❌ | 记忆库 ID（≤32 字符），不填则使用账号默认库；可在控制台 [记忆库列表](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 获取。 | +| `messages` / `custom_content` | array / string | 互斥必填 | `messages`：最多 50 条对话消息（含 `user`/`assistant` 角色）；`custom_content`：纯文本（≤512 字符），优先级更高，适用于结构化输入。 | +| `profile_schema` | string | ❌ | 用户画像模板 ID，仅当需触发结构化抽取时传入；需先调用 `CreateProfileSchema` 创建。 | +| `top_k`（Search） | integer | ❌（默认 10） | 检索返回的最大记忆条数，推荐设为 `3–10` 平衡召回质量与性能。 | +| `min_score`（Search） | double | ❌（默认 0.3） | 相似度阈值 `[0,1]`，低于此值的结果被过滤；OpenClaw 插件单位为百分制（需除以 100 转换）。 | +| `enable_rerank` / `enable_judge` / `enable_rewrite`（Search） | boolean | ❌ | 分别启用重排序、意图判别、query 重写，提升语义召回精度；默认关闭，按需开启。 | + +> ⚠️ 注意：所有接口强制使用平台内置专用记忆模型（非通用大模型），不开放模型选择；`UpdateMemory` 当前未在 `agentscope-runtime` SDK 中封装（v1.1.5+），需手动 HTTP PATCH 调用。 + +## 面向开发者：简洁实用指南 + +- **快速起步**：安装 `agentscope-runtime>=1.1.5`，调用 `AddMemory` 和 `SearchMemory` 异步方法，只需传入 `user_id` 和 `messages` 即可完成基础写入与检索。 +- **避免超限**：`AddMemory` QPM ≤ 120，`SearchMemory` QPM ≤ 300（阿里云账号级配额），高频场景建议批量合并写入、缓存高频查询结果。 +- **控制生命周期**：记忆本身永不过期，但可通过 `project_id` 绑定的规则配置 `memory_expiration_time`（仅对新写入生效）；业务侧需主动调用 `DeleteMemory` 清理敏感或过期数据。 +- **优化检索效果**：在控制台「记忆检索」页调试 `min_score`、`enable_rewrite` 等参数；对关键业务 query，建议预置 `custom_content` + 明确 `meta_data` 标签（如 `"category": "reminder"`）提升召回稳定性。 +- **生产就绪检查**：确保 `user_id` 符合长度与字符约束；`messages` 中角色必须为 `user`/`assistant`；HTTP 请求头必须包含 `Authorization: Bearer $DASHSCOPE_API_KEY` 和 `Content-Type: application/json`。 + +## 关联主题页 + +- [long term memory new](../api/long-term-memory-new.md) +- [memory library overview](../guides/memory-library-overview.md) +- [application call](../api/application-call.md) +- [managed agents](../guides/managed-agents.md) +- [llm application](../guides/llm-application.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md b/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md deleted file mode 100644 index da52b78e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/mcp-and-tools.md +++ /dev/null @@ -1,94 +0,0 @@ -# MCP 与工具扩展 - -MCP 与工具扩展是百炼平台为大模型补齐能力边界的统一机制：通过 MCP（Model Context Protocol）标准协议、插件（Plugin）和 Skill 等形式，让智能体、工作流应用能够调用外部工具、执行代码、处理文件，从而突破大模型在实时信息、精确计算、私有数据访问等方面的原生局限。 - -## 三类扩展能力对比 - -百炼提供多种工具扩展形态，适用场景各有侧重： - -| 扩展形态 | 本质 | 接入方式 | 典型场景 | -| --- | --- | --- | --- | -| MCP 服务 | 大模型与外部工具间的标准协议通道 | MCP 广场开通/部署后接入智能体或工作流 | 地图、搜索、图表、阿里云 OpenAPI 等第三方工具 | -| 插件（Plugin） | 工具集合，一个插件含多个工具（API） | 组件广场添加到智能体，或经 Assistant API 调用 | 代码解释器、计算器、图片生成、搜索等 | -| Skill | 可扩展能力包，封装端到端任务流程 | 上传 ZIP 技能包添加到智能体 | 文件处理、数据分析（如 xlsx 处理） | - -三者可组合使用：新版智能体（Agent 2.0）和 Managed Agents 将知识库、MCP、插件、Skill 统一为工具，由智能体自主规划调用顺序。 - -## MCP：标准协议接入 - -MCP 是 Anthropic 提出的开源标准，百炼基于它提供全周期服务，免去为每个外部工具编写专用接口的成本。 - -### 服务类型 - -- **官方 MCP 服务**：百炼云端部署，开通即用（如 Amap Maps、Sequential Thinking、QuickChart、联网搜索 WebSearch）。 -- **自定义 MCP 服务**：开发者自行部署，三种方式——脚本部署（托管到函数计算 FC，支持 `npx` / `uvx` / `http`）、从 AI 网关导入（RESTful API 升级为 MCP）、从阿里云 OpenAPI 导入（操作 OSS、ECS 等）。 - -### 使用方式 - -- **智能体应用**：模型根据对话内容自主判断是否调用，单个智能体最多同时添加 **5 个** MCP 服务，支持多工具组合与动态非固定顺序调用。 -- **工作流应用**：每个 MCP 节点只能用一个工具，需手动指定输入参数；通常先用大模型节点把自然语言解析为工具所需参数，再接入 MCP 节点。 -- **外部调用**：可集成到 Cherry Studio、Cursor 等第三方客户端，或通过 MCP SDK 编码接入个人项目。 - -### 关键参数与配置 - -- **传输协议**：`type` 字段须与端点路径一致，`"sse"` 对应 GET `/sse`，`"streamableHttp"` 对应 POST `/mcp`，不匹配会触发 405/404 错误。百炼已从旧版 SSE 升级为 **Streamable HTTP**，已开通用户需在 MCP 广场执行"取消开通 → 立即开通"完成升级，SDK 使用 `streamablehttp_client`。 -- **鉴权**：外部调用使用 `Authorization: Bearer `，端点如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`。 -- **脚本部署配置示例**： - -```json -{ - "mcpServers": { - "memory": { - "command": "npx", - "args": ["-y", "@modelcontextprotocol/server-memory"] - } - } -} -``` - -- **敏感信息**：涉及敏感数据的服务在创建时用 KMS 凭据加密管理。 -- **部署后修改**：仅支持编辑名称和描述；改部署方式、地域、安装方式或配置须先停止再重新部署。 - -### MCP 限制 - -- 不能直连千问 API，必须集成在智能体或工作流中使用。 -- 自定义 MCP 托管在 FC，暂不支持访问本地资源；访问远程资源需配置 IP 白名单或打通 VPC。 -- 私有 npm 仓库不支持；npx/uvx 部署的服务源版本更新后不会自动更新，需手动重新部署。 -- 调用会将返回内容作为上下文传入模型，增加输入 Token 消耗。 - -## 插件：工具集合 - -插件是一个工具集合，分官方、三方、自定义三类。官方插件（如 `code_interpreter`、`calculator`、`text_to_image`、`quark_search`、`generate_qrcode`、`github_search`）无需配置输入输出参数即可直接调用。 - -- **支持模型**：qwen-turbo、qwen-plus、qwen-max、qwen-vl-max、qwen-vl-plus。 -- **调用机制**：智能体应用 / Assistant API 由模型判断是否调用；工作流应用中插件作为节点按编排执行。 -- **调用配额**：每个智能体应用最多添加 **10 个** 工具；通过 API 调用需正确传递工具 ID（如 `calculator`）。 -- **授权**：首次访问需授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`；RAM 子账号需主账号先授予 `ram:CreateServiceLinkedRole` 权限。 - -## Skill：能力包 - -Skill 是智能体的可扩展能力包，让智能体在对话中自动识别并处理特定任务（如文件处理、数据分析），无需额外编码。分官方 Skill（自动更新）和自定义 Skill（上传 ZIP）。 - -- ZIP 包根目录必须含 `SKILL.md`，用 YAML 定义 `name`（唯一，建议小写连字符）和 `description`。 -- `description` 质量直接决定智能体调用准确率，建议包含适用输入类型、支持的操作、触发关键词、不适用场景。 -- ZIP 包大小不超过 **10 MB**；重新上传同名包生成新版本，已添加的智能体自动使用最新版。 - -## Managed Agents 中的工具 - -Managed Agents 智能体托管运行时内置 7 个工具：命令执行 `bash`，文件操作 `read`、`write`、`edit`、`glob`、`grep`、`download_file`，并支持接入 MCP 服务与 Skill。工具在独立云端沙箱执行，事件历史服务端持久化，支持中断与续接，适合多步工具调用、代码执行等长时任务。 - -## 计费要点 - -- **云部署 MCP**：限时免部署费；联网搜索免费额度 2000 次，用尽后 29 元/千次，限流 15 QPS。 -- **自定义部署 MCP**：基础模式无部署费、按调用时长 0.000156 元/秒计费（有冷启动）；极速模式部署费 0.000036 元/秒 + 调用费 0.000156 元/秒，适合高频在线场景。 -- **插件**：多为免费或限时免费（部分需申请开通）。 - -## 关联主题页 - -- [model context protocol](../guides/model-context-protocol.md) -- [plug in](../guides/plug-in.md) -- [managed agents](../guides/managed-agents.md) -- [skill](../guides/skill.md) -- [llm application](../guides/llm-application.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md index 0f2719f1..b8d03088 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md @@ -1,92 +1,68 @@ -# 模型部署与高速推理 - -模型部署是将百炼平台的预置模型、微调模型或导入模型发布为独立、资源专享的在线推理服务的过程；高速推理则是在部署之上（或独立）为调用提供容量刚性兑付与更高输出速度的一组能力。两者共同解决生产环境对高并发、低延迟和确定性吞吐的需求。 - -## 在百炼平台的使用场景 - -模型生产的完整链路为：**模型调优（Fine-tuning）→ 模型压缩（可选量化）→ 模型部署 → 推理调用**。开发者可通过调优 API 定制专属模型，用模型压缩降低部署规格与成本，再通过部署 API 发布为在线服务，最后调用端点进行推理。 - -围绕"部署 + 推理"，平台提供多种资源与容量形态，按业务诉求选型： - -- **需要私有推理服务、资源独占**：用模型部署（PTU / 模型单元 / 按 Token）创建专属服务。 -- **只需锁定容量、抵御公共限流**：用 **TPM 预留**，为指定模型预留专属吞吐量。 -- **只需更快出字、计费不变**：用**快速模式（Fast mode）**，把输出速度提升到标准 API 的 1.5~2 倍。 -- **需要降低部署成本**：先对微调模型做**模型压缩（量化）**，再部署。 - -## 部署的三种计费方式 - -计费方式在服务创建时选定，创建后不可更改（需下线后重新部署）： - -- **预置吞吐（PTU）**：预留资源保障特定 TPM，额度内不限速，TPS 通常较按 Token 提升约 1.5~2.0 倍。适合流量稳定、需并发/延迟确定性的高负载场景。支持按小时后付费与按天预付费。 -- **模型单元（MU）**：按时长 × 单元数计费，资源独占、性能可自定义，支持 PD 分离（拆分 Prefill/Decode 降低首 Token 延迟）。适合私有微调模型与长时任务。支持按分钟后付费与按月预付费。 -- **按 Token 使用量**：仅对 SFT 高效训练（LoRA）后的自定义模型开放，主要用于调优效果验证。 - -> 预付费无法提前退费，首月内提前退订按单价 1.2 倍计费；PTU 超出购买吞吐或输入超模型上限时，自动切换为按量付费（响应头 `x-dashscope-ptu-overflow:true`）。 - -## PTU 长输入与前缀缓存 - -- **长输入阶梯系数**：部分模型对超过 32K 的输入按更高系数折算 TPM（如 glm-5.1 在 [32K, 200K] 区间输入 1.33 / 输出 1.17），部分模型无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 低至 0.08），显著降低多轮对话与重复前缀场景成本。 -- **额度识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度）、`provisioned_tokens`（折算后实际消耗）、`cached_tokens`（缓存命中数）。这些字段在 OpenAI Chat 兼容、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下的 JSON 路径不同，需分别读取；Anthropic 兼容格式暂不返回 `cached_tokens`。 - -建议创建或扩容前用控制台**容量计算器**（依据 RPM、平均输入/输出长度、预估缓存命中率）估算所需 TPM，避免额度不足产生意外按量费用。 - -## 模型导入（LoRA） - -部署自训练模型前需从 OSS 导入 LoRA 微调版本，核心要求： - -- 仅支持 **LoRA**，不支持全参微调；必需文件 `adapter_model.safetensors` 与 `adapter_config.json`。 -- **rank** 必须为 8/16/32/64 之一，且同模型各 LoRA 层 rank 一致。 -- 不得新增 token、修改 vocab 或 chat_template，须与开源基础模型完全一致；VL 模型必须冻结 VIT（含 `visual` 权重则无法导入）。 -- **OSS 前提**：Bucket 需添加 `bailian-datahub-access` 标签（值 `read`）、文件须放子目录、不支持归档类存储；首次导入需完成服务关联角色授权。 - -支持基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列。 - -## 使用 API / 命令行部署 - -部署接口统一为 `POST https://dashscope.aliyuncs.com/api/v1/deployments`（仅华北2·北京，需先配置 `DASHSCOPE_API_KEY`），通过 `plan` 字段区分计费方式： - -- **PTU**：`"plan": "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`"plan": "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit` 等。 -- **按 Token（LoRA）**：`"plan": "lora"`，`capacity` 必填但无效，扩缩容需在控制台申请。 - -部署自定义模型时 `model_name` 使用**模型 ID**（在"我的模型"页面获取）。查询状态用 `GET /deployments/{id}`。 - -## 模型压缩（量化）降本 - -模型压缩特指量化（不含剪枝/蒸馏），将全精度微调模型转为低精度版本以降低部署所需 MU 规格。例如 qwen3.5-flash 微调模型压缩前 MU1*2（108 元/小时），压缩后 MU8*1（47 元/小时），成本节省约 56%。压缩不可逆，压缩后不支持继续微调或二次压缩，仅华北2（北京）可用。量化模板中 MU 编号越大规格越小、成本越低但精度损失可能越大；校准数据应选择与推理场景语义相近的数据集。 - -## 高速推理能力 - -### TPM 预留：锁定专属容量 - -- **专属模型 code**：创建预留后系统生成专属 `model` code，需将请求中的 `model` 替换为该 code 才命中预留容量。 -- **超额不中断**：超出预留自动降级为公共池按量计费，无需改代码，可在详情页查看超额降级统计。 -- **计费**：按 kTPM（1 kTPM = 1000 Tokens/分钟）预付费，部署成功即计费。 -- **管理**：支持在线扩缩容；退订不可恢复，缩容/退订退费按已用部分 1.5 倍系数结算；到期后 2 小时内可调用，2~14 小时转已停止（可续费），14 小时后删除。 - -### 快速模式（Fast mode）：更高输出速度 - -当前处于 **preview 阶段**，面向 AI 编程助手、Agent 多步推理、实时对话等对输出速度敏感的场景： - -- **高速输出**：TPS 达标准 API 的 1.5~2 倍（约 80~100 TPS）。 -- **计费不变**：仍按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 不立即限流，请求进入排队队列（区别于 TPM 预留的"超额降级按量"）。 -- **使用方式**：将 `model` 指定为支持快速模式的 model ID（如 `glm-5.2-fast-preview`），域名格式为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。`glm-5.2` 默认返回 `reasoning_content`，[流式输出](streaming.md)时思考与回答分别通过 `delta.reasoning_content` 与 `delta.content` 推送。 - -## 选型建议 - -- 追求成本优化且流量波动大：**按量付费**。 -- 流量可预估、不能接受公共限流：**TPM 预留**。 -- 高吞吐 + 高性能确定性：**PTU 专属部署**。 -- 只想更快出字、代码改动最小：**快速模式**。 -- 私有微调模型且要降本：先**模型压缩**再按 **MU** 部署。 +# 模型部署 + +模型部署是将训练或微调完成的模型（包括基础模型、LoRA 微调模型、量化模型等）发布为高可用、可计量、可管理的生产级推理服务的过程。在百炼平台中，模型部署是连接模型开发与业务调用的关键环节，提供统一 API 接口、多维度资源模型和细粒度控制能力，确保模型在真实场景中稳定、高效、合规地对外提供服务。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **面向基础模型直接推理**：无需微调，可直接选择 `qwen3.7-plus-2026-05-26`、`glm-5.1`、`deepseek-v4-pro` 等官方模型，按 PTU、MU 或 LoRA（仅限已导入 LoRA）方式部署，快速获得生产就绪 endpoint。 +- **面向微调模型上线**：通过 `model production` 流程完成微调后，使用生成的 `fine_tuned_model_id` 创建部署实例，支持绑定 `version_id` 实现版本可追溯，并自动分配专属 `endpoint_url`。 +- **面向高性能推理需求**：结合 `model high speed inference` 能力，在 MU 部署中启用 `enable_thinking`，或为 PTU 实例配置长输入与前缀缓存；也可单独选用 TPM 预留或快速模式（如 `glm-5.2-fast-preview`），实现吞吐保障或低延迟响应。 +- **面向成本与资源优化**：对已完成微调的模型，先执行 `model compression` 生成量化版本（如 `my-qwen-ft-awq`），再以该压缩模型 ID 进行 MU 部署，显著降低所需算力规格（如从 MU3→MU1），从而节省部署成本。 +- **面向[多模态](multi-modal.md)与定制化场景**：图像/视频/语音类微调模型（如 `wan2.7-image-pro`、`cosyvoice-v3-flash`）同样支持部署，但需注意其仅支持 LoRA 微调路径，且部署时必须使用对应模型 ID 及兼容的 `instance_type`（如 GPU 类型）。 + +> ⚠️ 注意：所有部署均强制要求地域为 **华北2（北京）**；API Key 所属业务空间须具备目标模型的部署权限；部署成功（状态变为 `RUNNING`）即开始计费，与是否发起请求无关。 + +## 关键参数和配置 + +| 参数名 | 适用部署类型 | 说明 | 示例值 | +|--------|--------------|------|--------| +| `plan` | 全部 | 必填，指定部署策略：`"ptu"` / `"mu"` / `"lora"` | `"plan": "mu"` | +| `model_id` | 全部 | 必填，模型唯一标识（基础模型 ID、微调产出 ID 或 LoRA 导入 ID） | `"model_id": "qwen3.7-plus-2026-05-26"` | +| `endpoint_name` | 全部 | 必填，全局唯一服务标识（3–63 字符，小写字母/数字/连字符） | `"endpoint_name": "qa-bot-prod"` | +| `ptu_capacity.input_tpm` / `output_tpm` | PTU | 预置吞吐量（token/min），决定容量购买与计费基线 | `"input_tpm": 5000, "output_tpm": 2000` | +| `deploy_spec` / `capacity` | MU | 算力规格（`MU1`/`MU3`/`MU8`）与副本数，直接影响并发与首 [Token](token.md) 延迟 | `"deploy_spec": "MU3", "capacity": 2` | +| `enable_thinking` | MU | 是否启用思考模式（影响推理逻辑、计费单价及输出结构） | `"enable_thinking": true` | +| `max_context_length` | MU（部分模型） | 显式设置最大上下文长度（单位：token），覆盖模型默认值 | `"max_context_length": 128000` | +| `rpm_limit` / `tpm_limit` | MU | 服务级限流阈值，用于保护服务质量 | `"rpm_limit": 300, "tpm_limit": 10000` | +| `instance_type` | [model production](../api/model-production.md)（通用部署） | 可选，指定底层计算资源（默认 `gpu-a10`；测试可用 `cpu-small`） | `"instance_type": "gpu-v100"` | + +> ✅ 提示： +> - LoRA 部署时，`capacity` 字段必须填写（如 `1`），但实际无效；`model_id` 必须为已成功导入的 LoRA 模型 ID（含 `adapter_config.json` 和 `adapter_model.safetensors`）。 +> - TPM 预留（TPM Reservation）属于 PTU 的增强形态，使用时需替换 API 请求中的 `model` 字段为专属 `dedicated model code`，而非通用模型 ID。 +> - 快速模式（Fast Mode）为独立部署形态，需使用专属域名和固定模型 ID（如 `glm-5.2-fast-preview`），不参与 MU/PTU/LoRA 的统一部署流程。 + +## 面向开发者，简洁实用 + +- **一句话启动**：用 DashScope SDK 一行代码部署（以 MU 为例）： + ```python + from dashscope import Deployments + res = Deployments.create( + model_id="qwen3.7-plus-2026-05-26", + endpoint_name="my-llm-api", + plan="mu", + deploy_spec="MU1", + capacity=2, + enable_thinking=False + ) + print(res.output.endpoint_url) # 获取调用地址 + ``` +- **状态检查**：部署后立即轮询 `GET /api/v1/deployments/{endpoint_name}`，直到 `status == "RUNNING"` 再发起推理请求。 +- **错误速查**： + - `Workspace xxx does not have deployment privilege for model xxxx` → 检查业务空间模型授权； + - HTTP 429 + `x-dashscope-ptu-overflow:true` → PTU 已溢出，转为按量计费； + - `health_check` 返回 503 → 实例仍在初始化（通常 2–5 分钟），请实现指数退避重试。 +- **最佳实践**： + - 高并发可预测业务 → 优先选 PTU，配前缀缓存 + 自动溢出； + - 需独占资源/低首 [Token](token.md) 延迟 → 选 MU，调优 `deploy_spec` + `enable_thinking`； + - LoRA 微调后轻量上线 → 选 LoRA 模式，严格校验 adapter 格式； + - 成本敏感且精度可接受 → 先压缩再部署，对比多个量化模板效果。 ## 关联主题页 - [model deployment 1](../guides/model-deployment-1.md) - [model production](../api/model-production.md) -- [model compression](../guides/model-compression.md) - [model high speed inference](../guides/model-high-speed-inference.md) +- [fine tuning](../guides/fine-tuning.md) +- [model compression](../guides/model-compression.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md b/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md deleted file mode 100644 index 22db84ab..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/model-selection.md +++ /dev/null @@ -1,83 +0,0 @@ -# 模型选型 - -模型选型是指在百炼平台提供的自研千问与第三方大模型矩阵中，按业务场景、能力档位、成本与延迟需求匹配最合适的模型，并选定对应地域、接入域名与调用方式的过程。选型结果直接决定效果、吞吐、合规性与单位成本。 - -## 选型总体思路 - -1. **先定场景**：明确任务是文本生成、视觉理解、视频/图像/3D 生成、语音、向量检索，还是 RAG、调优、深度研究等专用领域。 -2. **再定能力档**：在同类模型中按「高能力档 / 平衡档 / 轻量低成本档」对位选择，效果稳定后再用低档降本。 -3. **最后定接入**：根据合规与并发要求选择地域、服务部署范围与接入域名，并确认调用协议（OpenAI 兼容 / DashScope SDK / WebSocket / 异步任务）。 - -## 文本生成场景选型 - -对话、抽取、改写、Agent 等文本任务推荐从 `qwen3.7-plus` 入手——能力、速度与成本均衡，支持 1M 上下文、完整 Function Calling 与内置工具；需要最强推理选 `qwen3.7-max`；快速响应的简单任务用 `qwen3.6-flash` 降本。 - -从闭源或第三方模型迁移时按能力档对位： - -| 能力档 | 千问 | 第三方对位 | -| --- | --- | --- | -| 高能力档 | `qwen3.7-max` | — | -| 平衡档 | `qwen3.7-plus` | `deepseek-v4-pro` / `glm-5.2` | -| 轻量低成本档 | `qwen3.6-flash` | `deepseek-v4-flash` / `MiniMax-M2.5` | - -> 注意：第三方模型（`deepseek-v4-pro`、`kimi-k2.7-code`、`MiniMax-M2.5`、`mimo-v2.5-pro` 等）通常不支持内置工具与结构化输出，迁移时需评估能力差异。 - -## [多模态](multimodal.md)场景选型 - -| 模态 | 推荐入口 | 备选 / 降本 | -| --- | --- | --- | -| 视觉理解 / OCR | `qwen3.7-plus`（1M 上下文、2h 视频） | `qwen3.6-flash` 降本；`qwen-vl-ocr` 专做文档/表格/手写提取 | -| 图像生成 | `wan2.7-image-pro` / `qwen-image-2.0-pro` | `z-image-turbo` | -| 视频生成 | `happyhorse-1.1-t2v`（有声 1080P） | `wan2.7-t2v/i2v/r2v` 系列（自定义音频、首尾帧、角色一致性） | -| 3D 生成 | `Tripo/Tripo-H3.1` / `Tripo/Tripo-P1.0` | 仅北京地域，异步任务 | -| 语音合成（TTS） | `cosyvoice-v3.5-plus` | `qwen3-tts-instruct-flash` | -| 语音识别（ASR） | `fun-asr-realtime` | `qwen3.5-omni-plus-realtime` | -| 语音转语音 / 全模态 | `qwen3.5-omni-plus-realtime` | `qwen3.5-livetranslate-flash-realtime` | -| 音乐生成 | `fun-music-v1` | 邀测，仅北京 | -| 向量与重排序 | `text-embedding-v4` / `qwen3-rerank` | `qwen3-vl-embedding`（[多模态](multimodal.md)向量） | - -视频、图像、3D、音乐等生成类模型统一走异步任务模式；语音类多走 WebSocket / HTTP；向量与重排序走 HTTP。 - -## 专用领域模型选型 - -除通用对话模型外，百炼提供面向特定任务的专用模型，选型时需关注地域与协议差异： - -- **法律**：`farui-plus`，法律咨询/文书生成，DashScope SDK。 -- **意图理解**：`tongyi-intent-detect-v3`，同时输出意图与[函数调用](function-calling.md)信息。 -- **深度研究**：`qwen-deep-research`，两阶段（反问确认 + 深入研究），**仅 Python DashScope SDK、仅华北2（北京）**，不支持 OpenAI 兼容接口。 -- **翻译**：`qwen-mt-plus`，支持术语干预、翻译记忆、领域提示。 -- **OCR 与结构化抽取**：`qwen3.5-ocr`，可设 `min_pixels` / `max_pixels` 控制图像分辨率与 token 消耗。 -- **界面交互**：`gui-plus-2026-02-26`，通过 `computer_use` 工具操控桌面 GUI，仅北京。 - -## 选型时的关键参数与能力差异 - -- **思考模式**：通过 `enable_thinking` 开启（Responses API 用 `reasoning.effort` 控制深度），Qwen3 及以上支持，多为可按请求切换的混合模式。 -- **Function Calling 与内置工具**：自定义工具全系列通用模型支持；内置工具（联网搜索、代码解释器、网页抓取）仅部分模型支持，如 `qwen3.7-plus`、`qwen3.6-flash`、`glm-5.2`、`mimo-v2.5-pro`。 -- **结构化输出**：非思考模式下返回合法 JSON，用于字段抽取；第三方模型多不支持。 -- **上下文与[多模态](multimodal.md)限制**：视觉模型单张图片最高 1600 万像素，token 计算 `h × w / (32 × 32) + 2`；视频时长上限因模型而异（2 小时 / 2GB 或 1 小时 / 2GB）。 -- **[流式输出](streaming-output.md)**：`stream` 参数控制；`qwen-deep-research` 反问阶段必须设为 `true`。 - -## 地域、部署范围与接入域名 - -地域决定接入点与数据存储位置，服务部署范围决定推理执行位置，接入域名决定并发承载与隔离性。三者共同影响可选模型与稳定性。 - -- **地域**：华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）、日本（东京）。各地域模型列表、API Key、接入域名独立，**不能跨地域混用**。 -- **服务部署范围**：无合规需求选「全球」（资源池更大）；有合规需求选特定地理边界（中国内地/美国/欧盟/日本/国际）。北京、新加坡各仅支持一种范围，无需选择。 -- **接入域名**：生产环境推荐专属域名 `{WorkspaceId}.{region}.maas.aliyuncs.com`（99.9% SLA、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC）；共享域名 `dashscope.aliyuncs.com` 兼容存量；试用域名仅用于快速体验。 -- **模型后缀**：美国地域使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理，不带后缀默认全球推理。 - -## 选型实践建议 - -- 先在控制台「模型体验」中迭代 Prompt，效果稳定后再固化到代码；切换模型后应回归测试。 -- 穷尽 Prompt 工程与 RAG 方案后再投入模型调优（SFT/CPT/DPO），性价比最高。 -- 低延迟批量场景可使用批量推理降本；遇 429 限流时用指数退避 + 客户端令牌桶平滑流量。 -- 切换模型前关注上下文长度、[函数调用](function-calling.md)、[流式输出](streaming-output.md)、[计费](billing.md)方式与地域支持的差异。 - -## 关联主题页 - -- [model experience](../guides/model-experience.md) -- [get started with models](../guides/get-started-with-models.md) -- [use cases](../guides/use-cases.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md new file mode 100644 index 00000000..c586cc4a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md @@ -0,0 +1,71 @@ +# 多模态 + +多模态（Multimodal）指模型能够同时理解、生成或处理多种类型数据（如文本、图像、音频、视频、3D几何、动作序列等）的能力，其核心在于跨模态语义对齐与联合建模。在百炼平台中，“多模态”不是单一模型属性，而是贯穿能力设计、API协议与开发者调用范式的横切架构原则——同一服务接口可灵活适配不同输入模态组合，并统一输出结构化结果。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **统一输入抽象**：所有支持多模态的模型均通过 `input` 字段声明输入类型，而非固定字段名。例如： + - 文本+图像理解：`{"text": "描述这张图", "image": "https://..."}`（`qwen3.7-plus`） + - 图生视频：`{"media": [{"type": "image_url", "url": "..."}], "prompt": "生成10秒动画"}`（`wan2.7-i2v`） + - 单图/多图/文三选一生3D：`input.image`、`input.images` 或 `input.prompt`（`Tripo/Tripo-P1.0`），互斥且由模型自动路由 + - 实时音视频交互：`append_audio()` + `append_video()` 双流注入（`qwen3.5-omni-realtime`） + +- **能力分层复用**： + - **基础多模态理解**：`qwen3.7-plus`、`qwen3.5-omni-plus` 等通用大模型支持图文、音视频、OCR混合输入，无需切换模型即可处理跨模态指令（如“对比这两张发票金额并生成表格”）。 + - **专用多模态生成**：图像（T2I/I2I）、视频（T2V/I2V/R2V）、3D（T23D/I23D/MultiI23D）等能力由垂直模型实现，但共享统一的 `model` + `input` + `parameters` 调用范式。 + - **端到端实时多模态**：`Omni Realtime API` 将 ASR、LLM、TTS、VAD 全链路封装为单 WebSocket 连接，开发者只需传入原始音视频流，即可获得同步的文本响应与合成语音，无需自行编排模态转换流程。 + +- **异步/同步模式解耦**： + - 同步适用低延迟、小体积模态（文本、短语音、小图理解）； + - 异步强制用于高计算负载模态（视频生成、3D重建、长视频理解），通过 `X-DashScope-Async: enable` 头统一标识，任务状态与结果通过 `task_id` 标准化获取。 + +## 关键参数和配置 + +| 参数 | 类型 | 说明 | 开发者须知 | +|------|------|------|------------| +| `model` | string | 必填。精确指定模型ID，决定支持的模态组合与能力边界（如 `qwen3.7-plus` 支持图文+视频，`wan2.7-image-pro` 仅支持图像生成） | ❗不可跨模态混用（如用 `qwen3.7-plus` 调用视频生成 endpoint 会失败） | +| `input` | object | 必填。结构化输入容器，**字段名由模型能力动态决定**：
- 文本：`{"text": "..."}`
- 图像：`{"image": "url"}` 或 `{"images": ["url1","url2"]}`
- 音频/视频：`{"audio_url": "..."}` / `{"video": "..."}`
- 混合：`{"text": "...", "image": "...", "audio_url": "..."}`（部分模型支持） | ✅ 始终检查目标模型文档中 `input` 的合法字段组合；❌ 不要硬编码字段名（如 `prompt` 在新模型中已弃用，改用 `messages` 或 `input.text`） | +| `parameters` | object | 可选。控制生成行为的键值对，**模态专属参数需显式声明**：
- 图像：`{"size": "1024*1024", "negative_prompt": "模糊"}`
- 视频：`{"duration": 5, "aspect_ratio": "16:9"}`
- 3D：`{"geometry_quality": "ultra", "texture": false}`
- Omni Realtime：`{"turn_detection_type": "semantic_vad", "enable_search": true}` | ⚠️ 参数有效性严格依赖 `model` —— 同一参数名在不同模型中含义可能不同（如 `size` 在图像中是分辨率，在视频中是 `720P`） | +| `X-DashScope-Async` | header | string | 异步任务必需头，值必须为 `"enable"`；同步调用时**不得携带该头** | ❗遗漏将导致 400 错误（如 Tripo 3D、视频生成）；✅ 同步模型（如 `qwen3.7-plus`）携带该头将被拒绝 | + +## 面向开发者，简洁实用 + +- **第一步：确认模态需求 → 选模型 → 查文档** + 不要先写代码，先查 [模型能力矩阵](https://help.aliyun.com/zh/model-studio/model-capabilities-matrix)：明确你要处理的是「文本+图像理解」还是「图像→视频生成」，再锁定对应模型（如 `qwen3.7-plus` vs `wan2.7-i2v`），最后精读该模型的 [API参考文档](https://help.aliyun.com/zh/model-studio/model-api-reference) —— 输入结构、参数约束、地域限制均以模型为准。 + +- **第二步：用 `input` 组织数据，不用 `prompt`** + 百炼统一范式是 `input.{modality}`，而非旧式 `prompt` 字段。正确示例： + ```json + { + "model": "qwen3.7-plus", + "input": { + "text": "分析这张图中的商品价格和折扣信息", + "image": "https://example.com/receipt.jpg" + } + } + ``` + 错误示例（旧协议残留）：`{"prompt": "分析...", "image_url": "..."}`。 + +- **第三步：异步任务必带头、必轮询、必及时下载** + 对视频、3D、批量图像等异步任务： + - 请求头必须含 `X-DashScope-Async: enable`； + - 响应中提取 `task_id`，用 `GET /api/v1/tasks/{id}` 轮询（间隔 ≥15 秒）； + - `SUCCEEDED` 状态下立即下载 `output.results[0].xxx_url`（链接有效期通常为 2 小时）。 + +- **第四步：调试优先用业务空间域名** + 始终使用 `https://{WorkspaceId}.{region}.maas.aliyuncs.com`（如北京：`https://xxx.cn-beijing.maas.aliyuncs.com`），而非公共 endpoint —— 它提供更高稳定性、更低延迟，并确保模态能力与额度归属一致。 + +- **避坑提示**： + - 🚫 不要跨模型复用参数（`aspect_ratio` 在 Kling 视频有效，在 Tripo 3D 中无效）； + - 🚫 不要混用协议版本（`wan2.7` 模型必须用新版 endpoint，`wan2.6` 用旧版）； + - ✅ 所有多模态能力均支持 `DASHSCOPE_API_KEY` 统一认证，无需额外密钥。 + +## 关联主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) +- [model experience](../guides/model-experience.md) +- [omni realtime api](../api/omni-realtime-api.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md b/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md deleted file mode 100644 index d238b34c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/multimodal.md +++ /dev/null @@ -1,45 +0,0 @@ -# 多模态能力 - -多模态能力指模型同时理解或生成文本、图像、音频、视频、3D 等多种模态内容的能力。在百炼平台上，它既体现为「输入多模态」（如图文混合输入、音视频实时对话），也体现为「输出多模态」（如文生图、文生视频、语音合成、3D 资产生成）。 - -## 在百炼平台的主要场景 - -百炼按模态与任务把能力拆分到多个方向，开发者可按需组合： - -- **实时音视频对话**：Qwen-Omni-Realtime 系列通过 WebSocket 提供低延迟的语音输入/输出、图像输入、语音活动检测（VAD）、工具调用与联网搜索，适用于智能客服、语音助手等实时交互场景。 -- **图像生成与编辑**：覆盖文生图、图生图、局部重绘、扩图、背景生成、虚拟模特、AI 试衣、创意海报等，涉及千问 Qwen-Image、万相 Wan/Wanx、Z-Image、可灵 Kling、Vidu 等模型家族。 -- **视频生成与编辑**：聚合万相 Wan、HappyHorse、PixVerse、Vidu、可灵 Kling 及人像驱动模型，支持文生视频、图生视频（首帧/首尾帧/续写）、参考生视频、视频编辑与数字人。 -- **3D 资产生成**：基于 Tripo 模型支持文生 3D、单图生 3D、多图生 3D，产出带 PBR 材质或无贴图的 GLB 模型。 -- **视觉理解与 OCR**：以 Qwen 旗舰多模态模型理解图像与长视频（最长约 2 小时），并提供专优的 OCR/文档提取能力。 -- **语音合成 / 识别 / 语音转语音**：TTS、ASR、声音复刻/设计、S2S 实时对话与同传翻译，以及音乐生成。 - -## 输入与协议 - -- **实时流式（WebSocket）**：延迟最低，适合实时交互。Omni-Realtime 通过 `input_audio_buffer.append`（PCM 音频，Base64）、`input_image_buffer.append`（JPG/JPEG，Base64）等事件送入多模态素材，`session.update` 的 `modalities` 控制输出模态（`["text"]` 或 `["text","audio"]`）。 -- **HTTP 同步**：新一代图像模型（如 `wan2.6-image`、`wan2.7-image`、`z-image-turbo`）支持一次请求返回结果，路径为 `.../aigc/multimodal-generation/generation`，请求体用 `messages` 结构，`content` 内混排 `text` 与 `image`。 -- **HTTP 异步**：图像、视频、3D 生成等耗时任务（约 1-5 分钟）统一采用「创建任务拿 `task_id` → 轮询查询」两步流程，创建时必须携带请求头 `X-DashScope-Async: enable`，否则报错 `current user api does not support synchronous calls`。`task_id` 有效期 24 小时，切勿重复创建，轮询即可。 - -## 关键参数与配置 - -不同模态的请求体大多由 `model`、`input`、`parameters` 三部分组成： - -- **实时对话**（`session.update`）：`modalities`、`voice`（音色，因模型而异）、`input_audio_format` / `output_audio_format`（仅 `pcm`，输入 16kHz、输出 24kHz）、`turn_detection.type`（`server_vad` / `semantic_vad`）及 `threshold`、`silence_duration_ms` 等。 -- **图像生成**：`input.prompt` / `negative_prompt`（或新协议 `messages`），图像编辑用 `images` / `image_url` / `mask_image_url`；`parameters` 含 `size`（如 `1024*1024`、`1K`/`2K`/`4K`）、`n`、`aspect_ratio`、`watermark`、`prompt_extend` 等。 -- **视频生成**：`input.prompt` 描述画面镜头，`input.media` 承载 `first_frame` / `last_frame` / `reference_image` / `video` 等素材；`parameters` 含 `resolution`（`480P`/`720P`/`1080P`）、`duration`、`ratio` 等。 -- **3D 生成**：`prompt`、`image`、`images` 三者互斥；`parameters` 含 `texture_quality`、`geometry_quality`、`pbr`、`texture`。 - -## 注意事项 - -- **地域隔离**：模型、Endpoint URL 与 API Key 必须属于同一地域，华北2（北京）、新加坡、美国（弗吉尼亚）等地域各自独立、不可混用；部分能力（如 Tripo 3D、Fun-Music）仅在特定地域可用。推荐迁移到业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）以获得更好性能与稳定性。 -- **产物有效期**：图像结果 URL 有效期 24 小时，3D 模型下载链接仅 2 小时，务必及时下载。 -- **计费**：图像等仅对成功生成的输出计费，输入与失败任务不计费；具体计费、上下文窗口等实时参数以模型广场为准。 - -## 关联主题页 - -- [omni realtime api](../api/omni-realtime-api.md) -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) -- [model experience](../guides/model-experience.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md index a4fbcd2b..2bd3024a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md @@ -1,75 +1,68 @@ # OpenAI 兼容接口 -OpenAI 兼容接口是百炼平台提供的一套遵循 OpenAI API 规范的服务入口，让已有 OpenAI 应用只需替换 `api_key`、`base_url` 和 `model` 三项即可迁移到百炼，无需改动业务逻辑，是接入成本最低的调用方式。 - -## 在百炼平台的使用场景 - -OpenAI 兼容接口贯穿百炼的多类使用场景： - -- **文本生成模型调用**：作为四类接口（OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages、DashScope 原生）中迁移成本最低的一类，适合已基于 OpenAI SDK 构建的应用平滑迁移。其中 Chat Completions 为最常用入口，支持非流式、流式与工具调用（function call）；Responses 为其演进版本，内置联网搜索、代码解释器、网页抓取等工具，并通过 `previous_response_id` 自动管理多轮上下文。 -- **接入第三方客户端与开发工具**：Cherry Studio、Chatbox、Cursor、Cline、Dify 等聊天客户端和编程工具，统一通过「Base URL + API Key + 模型 ID」以 OpenAI 兼容协议接入百炼网关。 -- **专用模型调用**：`tongyi-intent-detect-v3`（意图理解）、`qwen-mt-plus`（翻译）、`qwen3.5-ocr`（OCR）、`gui-plus`（界面交互）等专用模型多数支持 OpenAI 兼容接口调用（注意 `qwen-deep-research` 仅支持 Python DashScope SDK，Qwen-Audio 仅支持 DashScope 协议）。 -- **智能体与工作流应用调用**：应用可通过 OpenAI 兼容的 Responses API 调用，支持同步/异步、多轮对话与[流式输出](streaming-output.md)。 - -## 关键参数与配置 - -迁移与调用的核心是配置以下三要素： - -- **`api_key`**：使用百炼 API Key。各地域、各计费方案的 API Key **相互独立、不能混用**，建议配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。若 Base URL 与 API Key 不配套会返回 401。 -- **`base_url`**：OpenAI SDK 调用统一以 `/compatible-mode/v1`（部分方案为 `/v1`）结尾；HTTP 调用需在其后追加具体资源路径（如 `/chat/completions`、`/responses`、`/embeddings`、`/files`）。各地域 SDK Base URL 示例： - - 华北2（北京）：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - - 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - - 日本（东京）：`https://{WorkspaceId}.ap-northeast-1.maas.aliyuncs.com/compatible-mode/v1` - - 德国（法兰克福）：`https://{WorkspaceId}.eu-central-1.maas.aliyuncs.com/compatible-mode/v1` - - 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - - 其中 `{WorkspaceId}` 为业务空间 ID，可在控制台业务空间详情页查看。北京、新加坡地域推荐使用业务空间专属域名以获得更好的性能与稳定性。 -- **`model`**：替换为百炼支持的模型名称（如 `qwen-plus`、`qwen3-max` 等）。 - -常用请求参数： - -- `messages`（array，必选）：对话消息列表，每条含 `role`（system/user/assistant）与 `content`。 -- `stream`（bool，可选）：是否[流式输出](streaming-output.md)；流式统计 Token 需配合 `stream_options={"include_usage": True}`。 -- 部分专用模型的特有参数（如 Qwen-MT 的 `translation_options`、Qwen-OCR 的 `min_pixels`/`max_pixels`）在 OpenAI SDK 中通过 `extra_body` 传入。 - -## 调用示例 - -以北京地域业务空间专属域名为例（Python）： - -```python -import os -from openai import OpenAI - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen-plus", - messages=[ - {"role": "system", "content": "You are a helpful assistant."}, - {"role": "user", "content": "你是谁？"}, - ], -) -print(completion.choices[0].message.content) -``` - -OpenAI Python SDK 要求 Python ≥ 3.8。 - -## 注意事项 - -- **协议路径区分**：OpenAI 协议 Base URL 以 `/compatible-mode/v1`（或 `/v1`）结尾，Anthropic 协议以 `/apps/anthropic` 结尾；部分工具还要求在 Anthropic 端点后追加 `/v1`，以各工具原文为准。 -- **接口能力差异**：DashScope 原生接口参数最全；若依赖联网搜索、代码解释器等内置工具，需使用 Responses 而非普通 Chat Completions。跨接口迁移时需核对参数映射。 -- **旧路径迁移**：Responses、Conversations 接口的旧版路径 `/api/v2/apps/protocols/compatible-mode/v1/...` 即将停止维护，请迁移至新版 `/compatible-mode/v1/...`。 -- **地域约束**：Base URL、API Key 和模型列表均不能跨地域混用；限流按主账号维度合并计算。 +OpenAI 兼容接口是百炼平台提供的一组标准化 RESTful API，严格遵循 OpenAI 官方 API 协议规范（v1.x），支持 `chat/completions`、`completions`、`embeddings`、`vision`、`batch`、`conversations`、`files` 等核心端点。开发者无需修改业务逻辑，仅需替换 `base_url` 和 `api_key`，即可将基于 OpenAI SDK 或生态工具（如 LangChain、LlamaIndex、Cursor、OpenClaw）构建的应用快速迁移至百炼平台。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **快速迁移现有应用**：已有 OpenAI 集成的项目（如 Web 应用、CLI 工具、Agent 框架），只需将 `openai.base_url` 改为百炼 OpenAI 兼容地址（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），并传入百炼颁发的 `api_key`，即可零代码改动调用 Qwen 系列及第三方模型（DeepSeek、Kimi、GLM 等）。 + +- **智能体（Agent）与工作流集成**：通过 `application call` 的 OpenAI 兼容 Responses API（`/v1/chat/completions`），可直接调用已发布的智能体应用，自动启用联网搜索、网页提取、代码解释器等原生工具链；支持[多模态](multi-modal.md)输入（图像、文件）、流式响应（`stream=true`）和异步模式（`background=true`）。 + +- **[多模态](multi-modal.md)与向量任务统一接入**：视觉理解（`qwen3-vl-plus`）、文本嵌入（`text-embedding-v4`）、批量推理（JSONL 文件异步提交）等能力，均通过同一套 OpenAI 兼容路径暴露，避免在不同协议间切换，降低客户端维护成本。 + +- **开发工具链直连**：主流 AI 编程助手（Claude Code、Qwen Code）、IDE 插件（Cline、Qoder）、桌面应用（Cherry Studio、Cursor）及开源 Agent 框架（OpenClaw、QwenPaw）均可原生对接，仅需配置 `base_url`、`api_key` 和合规 `model` 名称（如 `qwen3.7-plus`），即刻启用百炼算力。 + +- **会话状态管理**：配合 `conversations` 接口，可创建、查询、更新长期对话上下文；该能力与 `responses` 接口协同，实现跨设备、跨请求的上下文自动注入，适用于客服机器人、个人助理等长周期交互场景。 + +## 关键参数和配置 + +| 参数 | 类型 | 必选 | 说明 | 注意事项 | +|------|------|------|------|----------| +| `base_url` | string | 是 | OpenAI 兼容接口根地址，**必须使用业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），不可使用通用域名 `dashscope.aliyuncs.com`（已废弃） | `{WorkspaceId}` 需从控制台获取；地域（北京/新加坡/弗吉尼亚）须与 `api_key` 所属地域一致，否则返回 401 | +| `api_key` | string | 是 | 百炼平台颁发的密钥，**按计费方案隔离**（[Token](token.md) Plan Key ≠ Coding Plan Key ≠ 按量 Key） | Key 与 `base_url` 地域、套餐类型强绑定，跨方案或跨地域复用将失败 | +| `model` | string | 是 | 模型标识符，必须从各接口支持列表中选取（如 `qwen3.7-plus`、`text-embedding-v4`、`qwen3-vl-plus`） | 不支持别名转换（如 `qwen3.8-max-preview` 不可写为 `qwen3-8-max-preview`）；部分工具（如 Cursor）对 `-` 和 `.` 敏感，需严格匹配文档命名 | +| `messages` | array | 是（Chat/Responses） | 对话历史数组，格式为 `[{"role": "user", "content": "..."}]`；支持 `user`/`assistant`/`system` 角色 | `system` 角色在所有 OpenAI 兼容接口中均有效；`tools` 字段**不接受显式传入**（由服务端自动注入） | +| `stream` | boolean | 否 | 是否启用流式响应，默认 `false`；设为 `true` 时，响应为 SSE 格式，字段结构与 OpenAI 一致（`choices[0].delta.content`） | 流式响应末尾可通过 `stream_options={"include_usage": true}` 获取 token 统计 | +| `response_format` | object | 否 | 控制输出结构（如 `{"type": "json_object"}`），仅部分 Qwen3 模型支持 | 需模型明确声明支持（见各模型文档），否则忽略 | +| `max_tokens` | integer | 否 | 输出最大 token 数，建议显式设置以避免截断或超限 | 实际可用长度受模型上下文窗口限制（如 Qwen3 为 32768 tokens） | + +> ⚠️ 重要限制： +> - **工具调用不可定制**：OpenAI 兼容接口（尤其是 Responses）的联网搜索、代码执行等能力由服务端全自动调度，**不开放 `tools` 参数或自定义函数注册**；如需精细控制，请改用 DashScope 原生接口。 +> - **思考模式不可关闭**：`qwen3.8-max-preview` 等模型强制启用思考模式，`enable_thinking` 参数在 OpenAI 兼容接口中无效。 +> - **会话状态不共享**：`session_id` 仅 DashScope 原生接口支持；OpenAI 兼容接口需在每次请求中传递完整 `messages` 历史（或配合 `conversations` 接口管理）。 + +## 面向开发者，简洁实用 + +- ✅ **立即上手**：用 `openai` Python SDK（v1.0+）调用示例： + ```python + from openai import OpenAI + client = OpenAI( + api_key="sk-xxx", # 百炼 API Key + base_url="https://your-workspace-id.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + response = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你好"}], + stream=True + ) + ``` + +- ✅ **调试建议**： + - 首选 `dashscope` SDK（v1.20.0+），它内置 OpenAI 兼容模式，自动处理认证与重试； + - 使用 `curl` 测试时，务必携带 `Content-Type: application/json` 和 `Authorization: Bearer `； + - 遇到 `401 Unauthorized`，优先检查 `base_url` 地域、`api_key` 方案类型、`model` 是否在套餐支持列表中。 + +- ✅ **避坑指南**： + - 不要尝试在 OpenAI 兼容接口中传 `tools`、`enable_search`、`top_k` 等 DashScope 专属参数——将被静默忽略； + - 文件上传请走 `/v1/files` 接口（非 `/v1/chat/completions` 中内联 Base64），确保符合 `purpose` 要求（`file-extract`/`batch`/`fine-tune`）； + - 长对话场景下，主动精简历史 `messages`，避免因上下文过长触发自动截断（尤其启用搜索时）。 ## 关联主题页 - [qwen api reference](../api/qwen-api-reference.md) - [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) -- [get started with models](../guides/get-started-with-models.md) -- [more models](../api/more-models.md) - [application call](../api/application-call.md) +- [use chat client or development tool](../guides/use-chat-client-or-development-tool.md) +- [application component api reference](../api/application-component-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md deleted file mode 100644 index fac9783f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-caching.md +++ /dev/null @@ -1,59 +0,0 @@ -# 上下文缓存 - -上下文缓存（又称前缀缓存）是百炼平台提供的一种推理加速与成本优化能力：当多次请求共享相同的输入前缀时，平台会缓存该前缀的中间计算结果，后续命中缓存的部分按折扣价折算额度或计费，从而降低延迟与费用。 - -## 在百炼平台的使用场景 - -### 预置吞吐（PTU）部署中的前缀缓存 - -PTU 部署原生支持长输入与前缀缓存，是上下文缓存的主要应用场景。通过阶梯容量系数和缓存折扣管理额度消耗： - -- **缓存命中折扣**：命中缓存的输入 token 按模型对应折扣折算容量。例如 glm-5.1 折扣为 0.2，deepseek-v4-pro 为 0.08，显著降低 TPM 消耗。 -- **自动转按量计费**：当请求超出 PTU 额度或输入超过模型上下文上限（千问 128K / DeepSeek 64K）时，请求自动转为按量计费，响应头包含 `x-dashscope-ptu-overflow:true`，业务不中断。 - -### 显式缓存调用 - -百炼支持通过 OpenAI 兼容接口或 DashScope SDK 显式触发缓存。开发者可在请求中标识需要缓存的前缀（如系统提示词、[长上下文](long-context.md)文档），平台据此管理缓存命中与失效，适用于 RAG 应用、多轮对话等固定前缀反复出现的场景。 - -## 关键参数与响应字段 - -### 请求侧 - -- 在 PTU 部署下，缓存由平台自动管理，无需额外参数；显式缓存则按接口协议提供缓存控制字段。 -- 建议将稳定不变的内容（系统提示、知识库文档、few-shot 示例）置于请求前缀部分，动态内容置于后段，以提高缓存命中率。 - -### 响应侧（PTU 部署） - -PTU 部署的响应中包含额度相关字段，用于观测缓存命中情况： - -- `service_tier`：`ptu-standard` 表示使用 PTU 额度；`default` 或不返回表示按量计费。 -- `provisioned_tokens`：折算后实际消耗的 PTU 额度（含阶梯系数和缓存折扣）。 -- `cached_tokens`：前缀缓存命中的 token 数。 - -`cached_tokens` 在不同 API 格式下的 JSON 路径有差异： - -| API 格式 | 字段路径 | -| --- | --- | -| OpenAI Chat 兼容 | `usage.prompt_tokens_details.cached_tokens` | -| OpenAI Responses | `usage.input_tokens_details.cached_tokens` | -| Anthropic 兼容 | 暂不返回 `cached_tokens` | - -## 容量评估建议 - -在创建或扩容 PTU 部署时，控制台提供容量计算器，可根据每分钟请求数（RPM）、平均输入/输出长度、预估缓存命中率推荐输入 TPM 和输出 TPM。长输入场景下建议先用计算器评估额度，结合预期缓存命中率规划容量，避免意外转为按量计费。 - -## 注意事项 - -- 缓存命中要求前缀完全一致，任何字符变化（包括空格、换行、字段顺序）都可能导致缓存失效。 -- 缓存有有效期，长时间无请求后缓存会被淘汰，需重新预热。 -- 兼容接口（OpenAI / Anthropic）为保持协议一致，可能不暴露百炼原生的全部缓存控制参数；如需最完整的缓存能力，建议使用 DashScope 原生接口。 - -## 关联主题页 - -- [use cases](../guides/use-cases.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [qwen api reference](../api/qwen-api-reference.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md index 6241b301..ee5647e8 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/prompt-engineering.md @@ -1,74 +1,57 @@ -# 提示词工程 +# Prompt 工程 -提示词工程（Prompt Engineering）是通过设计、组织与优化 Prompt 来引导大模型生成符合预期结果的方法论。在百炼平台中，它贯穿智能体配置、[工作流](workflow.md)节点、模型直调与多模态生成等几乎所有 LLM 场景，并提供模板化管理、自动优化、样例库与反馈优化等成熟能力。 +Prompt 工程是系统性设计、迭代与优化提示词（Prompt）的方法论与实践体系，旨在通过结构化框架、自动化增强和数据驱动反馈，显著提升大模型在特定任务上的准确性、稳定性与可控性。它不是一次性指令编写，而是覆盖定义、测试、评估、优化、部署与监控的全生命周期工程实践。 -## 在百炼中的使用方式 +## 在百炼平台的不同场景中，这个概念如何使用 -### 1. 直接编写 System Prompt +Prompt 工程在百炼平台中已深度产品化，贯穿智能体、工作流、高代码应用及知识库增强等核心场景： -最基础的形态是在智能体应用的「系统提示词」中定义角色、行为指令与能力边界，支持通过 `/` 引用自定义变量。[工作流](workflow.md)应用的大模型节点同样通过「提示词 + 用户提示词」驱动推理。Prompt 越清晰、具体、无歧义，模型表现越稳定。 +- **智能体（Agent）应用**：作为角色定义与工具调用规则的载体。System Prompt 采用 ICIO 或 RASCEF 等结构化框架（如明确 Role、Action、Steps、Constraints），配合 `enable_thinking` 参数启用反思链路，使模型更可靠地规划并调用知识库、MCP 等工具。 + +- **工作流（Workflow）应用**：每个「大模型」节点均支持绑定 Prompt 模板。模板变量（如 `${sys.query}`、`${retrieved_content}`）可动态注入上下文、RAG 结果或前序节点输出，实现多步任务中提示词的精准定制与复用。 -### 2. Prompt 模板（结构与变量分离） +- **知识库增强问答**：Prompt 工程直接决定知识检索结果的利用效果。例如，通过负向约束（如“禁止编造未提及的信息”）+ 引用标注指令（如“所有结论必须标注来源片段编号”），可显著提升回答可信度与可追溯性。 -将固定结构与动态变量分离，统一管理、复用，是团队协作与版本一致性保障的推荐做法。入口位于控制台「应用开发 > 组件管理 > 提示词」。 +- **图片/[多模态](multi-modal.md)生成**：支持正向 Prompt（描述目标内容）与负向 Prompt（排除干扰元素）双通道输入，需避免语义冲突（如正向写“高清”，负向写“模糊”），并通过 CRISPE 框架结构化控制风格、构图与细节层级。 -- **预置模板**：平台提供，覆盖创意文案、办公助理等通用场景，效果稳定、不可修改。 -- **自定义模板**：用户自行设计，支持「自定义创建」（直接粘贴现成 Prompt，可选「优化 Prompt」润色）和「基于 Prompt 工程创建」（选择 ICIO / CRISPE / RASCEF 框架，按字段结构化填写）两种模式。 +- **应用评测与迭代闭环**：Prompt 反馈优化功能依赖评测集（≥20 条）与样例集（5–10 条）驱动多轮自动重写，推荐使用 `qwen-max` 作为优化引擎；优化结果可一键保存为新模板，无缝接入已有应用。 -框架选型建议： +> ⚠️ 注意：Prompt 样例库功能已停止维护，新项目请统一使用 RAG 表格库或评测反馈机制替代。 -| 框架 | 适用场景 | -| --- | --- | -| ICIO | 简单、明确的任务执行，如数据分析、内容生成、文本摘要 | -| CRISPE | 需要 AI 扮演特定角色的交互，如智能客服、创意写作 | -| RASCEF | 涉及多步骤的复杂业务流程，如项目规划、战略分析 | +## 关键参数和配置 -> 注意：Prompt 模板相关功能仅适用于华北2（北京）地域，使用前请确认业务空间所在地域。 +| 参数 | 说明 | 开发建议 | +|------|------|----------| +| `promptTemplateId` + `workspaceId` | 模板唯一标识对，用于 API 获取模板内容 | 必须成对使用；`workspaceId` 需通过控制台或 `ListWorkspace` 接口获取，且仅华北2（北京）地域有效 | +| `variables` | 模板中声明的占位符（如 `${topic}`、`${num1}`），运行时由业务逻辑填充 | 占位符名称应语义清晰；避免嵌套（如 `${${var}}`）；模板创建后不可新增变量，需提前规划 | +| `max_tokens`（上下文） | 单次请求总 [Token](token.md) 上限（Prompt + 输入 + 输出） | 文本生成默认 ≤ 6144 字符（约 8K tokens）；图片生成需同时满足分辨率与提示词长度限制，建议正向 Prompt ≤ 300 字、负向 ≤ 150 字 | +| `temperature` | 控制生成随机性（0.0–1.0），值越高越发散 | 对确定性任务（如格式化输出、代码生成）设为 `0.0–0.3`；创意类任务可设 `0.7–1.0`，但需配合 `top_p` 或 `stop` 参数约束边界 | -### 3. Prompt 自动优化 +## 面向开发者，简洁实用 -当缺乏经验或手动编写耗时，可在「提示词 > 自动优化」页面输入原始 Prompt，由大模型进行结构重组、角色扮演引导、指令增强、安全与边界注入等重写，生成结构更优的新版本。该功能不计费，提交数据不会被存储或用于训练。优化失败常见原因：输入超长超出 Token 限制、触发内容审核、网络或服务临时不可用。 +- ✅ **优先使用结构化模板**：新建 Prompt 时选择「基于 Prompt 工程创建」，内置 ICIO（Identity-Context-Instruction-Output）、CRISPE（Capacity-Role-Insight-Statement-Personality-Experiment）等框架，比纯自由文本更易调试与复用。 + +- ✅ **API 调用三步法**： + 1. 调用 `GetPromptTemplate` 获取模板内容与 `variables` 列表； + 2. 用业务数据替换 `${variable}` 占位符（推荐使用标准字符串模板引擎，如 Python 的 `string.Template`）； + 3. 将生成的完整 Prompt 作为 `system` 或 `user` 消息发送至模型 API（如 `qwen-max`）。 -### 4. Prompt 样例库（Few-shot 检索） +- ✅ **热更新不改代码**：Prompt 模板在控制台编辑保存后，所有绑定该模板的应用将自动生效，无需重新部署服务，适合 A/B 测试与灰度发布。 -针对特定领域专业任务，从预定义的高质量问答对中检索相关样例注入上下文，引导模型生成更准确、风格更一致的回复。适用于智能客服、特定领域问答、格式化内容生成。注意该功能已不再维护，官方推荐迁移到 RAG 表格库。 +- ✅ **安全合规**：自动优化过程中的输入数据**不会被存储或用于模型训练**，符合阿里云隐私政策；敏感字段（如用户 ID）建议通过变量注入而非硬编码进 Prompt。 -### 5. Prompt 反馈优化 - -基于输入输出样例与评测数据，多轮自动评估、反思、优化 Prompt，涉及推理调用。适合对输出质量有持续提升需求的闭环场景。 - -## 关键参数与配置 - -### 智能体应用 - -- **系统提示词最大长度**：6144 字符（从模板「使用 [prompt](../guides/prompt.md) > 创建应用」时自动填充到此上限）。 -- **模型参数**：`temperature`、最长回复长度、`enable_thinking`（开启思考模式以提升反思效果，仅支持思考模式模型）。 -- **变量引用**：在系统提示词中通过 `/` 嵌入自定义变量，运行时由业务数据填充。 - -### 多模态生成 - -- **文生图**：`prompt`（正向）、`negative_prompt`（反向，描述不希望出现的内容）、`prompt_extend`（V2 专用，大模型智能改写，默认 `true`）。基础公式 `主体 + 场景 + 风格`，进阶公式追加镜头语言、氛围词与细节修饰。 -- **文生视频**：基础公式 `主体 + 场景 + 运动`，进阶公式 `主体描述 + 场景描述 + 运动描述 + 美学控制 + 风格化`；图生视频简化为 `运动 + 运镜`。wan2.7 起单/多镜头由提示词控制，不再使用 `shot_type`。 - -### 模板调用 API - -- **创建模板**：`CreatePromptTemplate`，需先获取 Workspace ID。 -- **获取模板**：`GetPromptTemplate`，传入 `workspaceId` 与 `promptTemplateId`，返回 `content`、`variables` 等字段，在代码中填充变量后调用模型。相比字符串拼接，可实现逻辑与内容分离、集中管理与版本一致。 - -## 设计要点 - -1. **结构化优先**：复杂任务优先采用 ICIO/CRISPE/RASCEF 框架或「背景-目的-风格-语气-受众-输出」六要素，避免笼统指令。 -2. **变量分离**：把不可变的结构与可变的业务数据拆开，通过模板或 `/` 变量注入，便于复用与协作。 -3. **样例引导**：对风格或格式敏感的任务，用 Few-shot 样例或反馈优化建立闭环，持续校准输出。 -4. **正反向结合**：图片与视频生成场景同时使用正向与反向 Prompt 精确控制画面内容。 -5. **迭代验证**：结合在线调试面板与评测能力，对 Prompt 变更做回归验证后再发布。 +- ❌ **避坑提醒**： + - 不跨地域调用 Prompt 相关 API（仅支持华北2）； + - 单模板内容勿超 6144 字符（控制台有实时计数）； + - 图片生成负向 Prompt 避免与正向逻辑矛盾； + - 反馈优化任务中，样例数据需覆盖全部业务类别，评测数据越多效果越优。 ## 关联主题页 - [prompt](../guides/prompt.md) -- [start using](../guides/start-using.md) - [llm application](../guides/llm-application.md) -- [use cases](../guides/use-cases.md) +- [start using](../guides/start-using.md) - [application component api reference](../api/application-component-api-reference.md) +- [application evaluation](../guides/application-evaluation.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md index f5102324..bd6a4f52 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/rag.md @@ -1,62 +1,48 @@ -# 检索增强生成（RAG） +# 检索增强生成 -检索增强生成（Retrieval-Augmented Generation, RAG）是一种在大模型生成回答前，先从外部知识库检索语义相关内容、再将其与用户问题一起送入模型的技术。它为大模型补充私有数据和最新信息，从而显著提升在特定领域问题上的准确性、降低幻觉。 +检索增强生成（Retrieval-Augmented Generation，简称 RAG）是一种将大语言模型（LLM）与外部知识源动态结合的技术范式：在生成回答前，系统先根据用户查询语义检索相关知识片段，再将检索结果作为上下文注入模型提示（[prompt](../guides/prompt.md)），引导模型生成更准确、可溯源、时效性强的响应。 -## 在百炼平台的使用场景 +在百炼平台中，RAG 不是单一接口或模型，而是贯穿知识库、数据连接、智能体应用与框架集成的一套协同能力体系，其核心目标是让大模型“有据可依”，而非仅依赖参数内化知识。 -百炼把 RAG 拆成「建立索引 → 检索召回 → 生成答案」三个阶段，并在不同层面提供了对应能力： +## 在百炼平台的不同场景中，这个概念如何使用 -- **云端知识库（控制台）**：进入知识库按「填写基础信息 → 配置数据来源 → 设置索引参数」三步建库，创建时选定类型（文档搜索 / 数据查询表格库 / 图片问答 / 音视频搜索，创建后不可更改），随后关联到智能体应用、工作流应用或外部应用。工作流应用中知识库节点须接在开始节点之后、大模型节点之前，并在大模型提示词中引用 `result` 变量。注意：知识库功能仅在中国站**华北2（北京）**地域可用。 -- **知识检索服务**：面向多知识库联合检索（最多 15 个），提供 Query 改写、混合检索（向量+关键词）、Rerank 排序的流水线。 -- **知识问答服务**：在检索基础上由大模型生成自然语言回答，提供**极速模式**（单轮检索+生成）与**多轮智能模式**（Agentic 多轮规划搜索），并支持文件预解析、拒答、防泄漏、多模态回复、引用来源等生成控制。 -- **应用场景接入**：围绕「RAG + 智能体应用」可将问答能力接入网站、企业微信、微信公众号、钉钉等渠道；也支持基于本地知识库构建 RAG 应用（检索在本地执行、生成调用通义千问 API），适合需要灵活切分与自定义嵌入模型的场景。 -- **框架集成**：LlamaIndex（Python）可构建云端知识库与 RAG 应用；Spring AI Alibaba（Java）可集成智能体/工作流应用并检索百炼知识库。 +- **知识库问答（/api/v2/apps/[knowledge](../api/knowledge.md)/chat）**：最典型的端到端 RAG 流程。平台自动完成「查询理解 → 多知识库联合检索（向量+关键词+重排）→ 工具调用（Retrieve）→ 上下文拼接 → LLM 生成」，支持 SSE [流式输出](streaming-output.md)，并返回 `docReferences` 等结构化溯源信息。 +- **数据连接（Data Connection）**：为 RAG 提供多样化知识源。平台托管型（如 PDF/Excel）自动构建向量索引；流处理型（如 MySQL/OSS/语雀）则按需实时检索，实现“不入库、可检索”的轻量 RAG。二者均可在智能体或工作流中直接绑定为知识源。 +- **智能体与工作流应用**：RAG 是知识增强的关键开关。可在应用配置中启用“必定调用知识库”，并设置召回数量、相似度阈值、权重等策略；也可通过 `knowledge_sources` 参数在 API 请求中动态指定知识库或数据连接器 ID。 +- **框架集成（LlamaIndex / Spring AI Alibaba）**：面向开发者提供代码级 RAG 构建能力。LlamaIndex 封装云端索引构建与查询引擎；Spring AI Alibaba 提供 `DashScopeDocumentRetriever`，支持显式控制模型（如 `qwen-plus`）、检索参数（`top_k`, `similarity_cutoff`）及流式响应，适用于定制化 RAG 应用开发。 +- **本地 RAG 应用（Python SDK）**：面向高级场景，支持完全自主的文档切分、本地嵌入模型（如 GTE）、自定义检索逻辑，适用于对数据主权、延迟或成本有强约束的私有化部署。 -## HTTP REST 接口 +> ⚠️ 注意：所有 RAG 能力均运行于华北2（北京）地域，且必须在业务空间（workspace）中完成知识库/数据连接器的创建与激活，否则请求将静默失败或返回空结果。 -除控制台外，百炼提供 DashScope 应用网关体系的两个 REST 接口，用 API Key Bearer 鉴权，Base URL 形如 `https://{workspaceId}.cn-beijing.maas.aliyuncs.com`： +## 关键参数和配置 -| 接口 | 路径 | 说明 | -| --- | --- | --- | -| 知识检索 | `POST /api/v1/indices/knowledge/search` | 跨多个知识库联合语义检索，返回按相关性排序的切片，适合需自定义生成流程的场景 | -| 知识问答 | `POST /api/v2/apps/knowledge/chat` | 基于知识库的智能问答，通过 SSE 流式返回规划、工具调用、生成三个阶段 | +以下参数直接影响 RAG 效果与成本，建议按场景显式配置（默认值可能不满足精度要求）： -默认用户维度 25 QPS。此外还有 `CreateIndex`、`Retrieve` 等 OpenAPI RPC 接口用于建库流程。 +| 参数名 | 说明 | 典型取值 | 所属层级 | 备注 | +|--------|------|----------|----------|------| +| `top_k` / `similarity_top_k` | 初步向量/关键词召回的切片数 | `10–50` | 检索层 | 值越大，重排 [Token](token.md) 消耗越高；混合检索时需兼顾两者 | +| `max_retrieval_count` / `max_retrieved_nodes` | 最终送入 LLM 的上下文切片上限 | `3–10` | 检索层 | 直接影响生成质量与 [Token](token.md) 成本，推荐从 `5` 开始调优 | +| `similarity_threshold` / `similarity_cutoff` | 重排后过滤低分切片的阈值 | `0.4–0.7` | 重排层 | 过高易漏召关键信息，过低引入噪声；建议结合日志分析调整 | +| `tags` | 按标签过滤检索范围（支持 `AND`/`OR` 逻辑） | `["faq", "v2.3"]` | 检索层 | 用于多版本、多业务线知识隔离，需在知识库创建时预设 | +| `model_name` / `withModel()` | 生成阶段使用的 LLM | `"qwen-plus"`, `"qwen3.5-plus"` | 生成层 | `/chat` 接口不可覆盖，但框架集成与本地应用可自由指定 | +| `stream` | 是否启用 SSE 流式响应 | `true`（默认） | 传输层 | 仅 `/chat` 接口支持；流式下可实时获取 `thoughts` 和 `docReferences` | -## 关键参数与配置 +> 💡 提示：Rerank 模型（如 `qwen3-rerank`）费用取决于初步召回总切片数（即 `top_k` 之和），而非最终返回数；多知识库联合检索时，该成本线性增长，请合理设置 `top_k`。 -检索效果主要由以下参数决定，可在命中测试、检索服务与问答服务中反复调优： +## 面向开发者，简洁实用 -- **相似度阈值（0.01~1.0）**：仅语义相似度高于阈值的切片会被召回。阈值过高会导致相关切片被全部丢弃（如调至 0.60 可能无召回）。 -- **初步向量检索 TopK / 初步关键词检索 TopK（1~100，默认各 50）**：控制初步召回数量，直接影响送入 Rerank 的 Token 量与成本。 -- **最大召回数量 / 召回片段数（1~20）**：最终提供给大模型的切片数，对总结、列举、比较类复杂问题应适当调大。 -- **权重**：多知识库联合召回时干预排序，仅在**同类型知识库之间**生效。 -- **排序模型（Rerank）**：纯文本可选 `qwen3-rerank` / `qwen3-rerank(hybrid)`，多模态可选 `qwen3-vl-rerank`；`gte-rerank` 将于 2026-05-30 下线，新项目直接选 `qwen3-rerank`。 -- **向量模型（Embedding）**：将文本/图片/视频编码到同一语义空间，供余弦相似度匹配。云端知识库使用官方向量模型；本地方案可改用自部署 GTE 模型。 -- **Meta 信息抽取与标签过滤**：在向量检索前做结构化筛选，精准定位目标文件；元数据只能在创建知识库时配置，创建后无法开启。 - -在框架（LlamaIndex）中对应的参数为：`similarity_top_k`（召回数）、`similarity_cutoff`（最低相似度阈值）、`top_n`（重排后返回数），可通过 `node_postprocessors`（`SimilarityPostprocessor`、`DashScopeRerank`）做后处理。 - -## 效果优化建议 - -优化前建议用自动评测建立至少 100 组用例的基线，再针对失败用例（打分 < 4）诊断改进： - -- **检索无效（没找到）**：补充知识、优化源文件排版（推荐转 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写。 -- **召回不相关**：使用标签过滤或元数据做结构化搜索。 -- **切片不完整**：采用「智能切分」（基于语义自适应切分），并人工检查修正异常切片。 -- **重排不佳**：在漏召回与噪声之间平衡相似度阈值与召回片段数。 - -## 监控 - -所有检索调用以日志形式投递到日志服务（SLS），topic 为 `log_dispatch`，含 `request_id`、`pipeline_id`（知识库 ID）、`workspace_id`、`latency`、`response_status_code` 等字段，可用于审计、用量统计与慢查询/错误率监控。SLS 存储与流量单独计费，关闭检索日志开关只停止新投递，历史日志仍保留计费。 +- **快速验证**：用控制台创建一个文件知识库 → 绑定至智能体应用 → 启用“必定调用” → 发起测试对话，观察 `docReferences` 字段是否返回非空数组。 +- **调试必查**：若 RAG 返回空或无关内容，优先检查：① 知识库状态是否为 `ACTIVE`；② 查询文本是否触发有效召回（调用 `/search` 接口验证）；③ `similarity_threshold` 是否过高；④ 日志中 `response_body.data.nodes[]` 是否为空。 +- **性能优化**：高频场景建议关闭 Rerank（设 `rerank_enabled: false`）并调高 `similarity_threshold`；对精度敏感场景，启用 hybrid 检索 + `qwen3-rerank` 并将 `top_k` 设为 `30–50`。 +- **安全边界**：所有知识库与数据连接器均按业务空间隔离，API Key 无跨空间访问权限；元数据（Meta）抽取规则在知识库创建后不可修改，请务必在初始化阶段审慎配置。 +- **计费意识**：RAG 成本 = 检索侧（向量化 + Rerank） + 生成侧（LLM 输入/输出 [Token](token.md)）。避免盲目增大 `top_k` 或 `max_retrieval_count`，应以实际召回质量为准。 ## 关联主题页 +- [knowledge](../api/knowledge.md) - [knowledge base](../guides/knowledge-base.md) - [frameworks](../api/frameworks.md) +- [data connection overview](../guides/data-connection-overview.md) - [application use cases](../guides/application-use-cases.md) -- [vector and sort](../api/vector-and-sort.md) -- [knowledge](../api/knowledge.md) -- [use cases](../guides/use-cases.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md deleted file mode 100644 index c43445dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rate-limiting.md +++ /dev/null @@ -1,71 +0,0 @@ -# 限流与配额 - -限流与配额是百炼平台对模型推理调用施加的吞吐量与调用频率约束，用以保障公共推理资源的公平使用和整体稳定性；当业务流量超过共享上限或需要刚性容量保障时，可通过专属域名、TPM 预留、PTU 部署或异步通知等手段规避限流影响。 - -## 限流的产生场景 - -百炼模型调用主要在以下维度受到约束： - -- **TPM（Tokens Per Minute）**：每分钟可消耗的 Token 数上限，按主账号 + 业务空间 + 模型维度计算，超出会返回 `429` 限流错误。 -- **RPM（Requests Per Minute）**：每分钟请求数，RPM 越大建议 TPM 同比增大。 -- **QPS 限制**：部分通用接口有独立的 QPS 限制。例如异步任务管理（查询、批量查询、取消）三个接口统一限制为 20 QPS，按主账号 + 子账号维度计算。 -- **公共池共享**：按量付费方案无专属容量，调用进入公共共享池，受公共限流波动影响，业务高峰期可能被限流。 - -监控页面的「错误」指标中专门提供**限流错误次数（429）**，可用于定位限流问题；「性能」指标中的 RPM、TPM 则帮助评估容量是否接近上限。 - -## 不同方案下的限流行为 - -| 方案 | 容量保障 | 超额处理 | 接入改动 | -| --- | --- | --- | --- | -| 按量付费 | 无（共享公共池） | 自动服务，受公共限流 | 无需改动 | -| 资源包 / 节省计划 | 承诺用量折扣（非专属容量） | 超出转按量 | 无需改动 | -| TPM 预留 | 专属容量刚性兑付 | 超出自动降级公共池按量，不中断 | 替换 `model` 参数为专属模型 code | -| PTU（模型部署） | 专属部署实例 | 超出转按量 | 替换 `model` 参数 | - -当业务流量可预估且不能接受限流时，优先选择 TPM 预留；对极致性能与隔离有更高要求时，可考虑 PTU 专属部署。 - -## 接入域名对限流的影响 - -接入域名直接影响并发上限与超时表现： - -- **业务空间专属域名** `https://{WorkspaceId}.{region}.maas.aliyuncs.com`：生产推荐，请求超时 3600 秒、SLA 99.9%，提供更高吞吐与时延隔离。 -- **Dashscope 中心化域名** `https://dashscope.aliyuncs.com`：存量业务兼容，可跨业务空间调用。 -- **试用域名** `https://trial.{region}.maas.aliyuncs.com`：仅限快速验证，限流小，不建议生产。 - -各地域的 API Key、模型列表、接入域名不能跨地域混用；美国（弗吉尼亚）暂不支持业务空间专属域名，需使用 `dashscope-us.aliyuncs.com`。 - -## TPM 预留的关键参数 - -创建 TPM 预留时需指定输入 TPM 与输出 TPM，单位为 kTPM（1 kTPM = 1,000 Tokens/分钟），起步和步长因模型而异。容量计算受以下参数影响： - -- **每分钟请求数（RPM）**：业务高峰期每分钟请求数。 -- **平均输入/输出长度（token）**：输入越长，阶梯系数越大，所需输入 TPM 越高。 -- **预估缓存命中率（%）**：命中率越高，输入容量消耗越慢，所需输入 TPM 越低；仅影响输入 TPM。 - -部分模型支持长输入阶梯系数和缓存折扣（如 glm-5.1 在 [32K, 200K] 区间输入系数 1.33、输出 1.17；deepseek-v4-pro 缓存命中部分按 8% 折算），TPM 容量计算器会自动应用这些参数。 - -## 规避异步任务轮询限流 - -图像/视频生成等耗时模型采用异步机制，频繁轮询结果接口会浪费资源并可能触发 20 QPS 限流。百炼已接入事件总线 EventBridge，任务完成（无论成功或失败）后主动上报 `dashscope:System:AsyncTaskFinish` 事件，可推送到 HTTP 回调 URL 或 RocketMQ 消息队列。通知方案不限流、实时性高，适合高并发、大规模或对实时性要求高的任务。 - -## 免费额度与配额停用 - -免费额度页面提供「免费额度用完即停」开关：开启后免费额度用尽时服务自动停止，返回 `403 AllocationQuota.FreeTierOnly`，避免产生免费额度以外的费用。仅在账户内仍有未消耗的免费额度时才能开启；关闭需等免费额度完全消耗后进行。 - -## 应对限流的实践建议 - -1. 生产环境优先迁移到业务空间专属域名，获得更高吞吐与时延隔离。 -2. 流量可预估且不能接受限流的核心业务，使用 TPM 预留锁定专属容量。 -3. 通过模型监控的限流错误次数（429）与 RPM/TPM 指标评估容量是否接近上限。 -4. 异步任务改用 EventBridge 通知方案，避免轮询触发 20 QPS 限制。 -5. 需要严格控制成本的场景，可开启「免费额度用完即停」避免超额消费。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [model high speed inference](../guides/model-high-speed-inference.md) -- [model monitoring](../guides/model-monitoring.md) -- [more about models](../api/more-about-models.md) -- [use cases](../guides/use-cases.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md b/skills/bailian-docs-llm-wiki/wiki/concepts/region.md deleted file mode 100644 index 3006794c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/region.md +++ /dev/null @@ -1,80 +0,0 @@ -# 地域与可用区 - -地域（Region）是百炼平台在物理地理上划分的部署区域，决定接入点位置、数据存储位置与推理执行位置；可用区是地域内相互隔离的物理设施单元。在百炼中，"地域"主要与**服务部署范围**和**接入域名**共同决定调用路径、数据驻留约束与服务保障等级。 - -## 可用地域 - -百炼当前支持的地域及其 Region ID： - -| 地域 | Region ID | 数据驻留约束 | -| --- | --- | --- | -| 华北2（北京） | `cn-beijing` | 数据不出中国内地 | -| 新加坡 | `ap-southeast-1` | 数据不经过中国内地 | -| 美国（弗吉尼亚） | — | 数据不出美国 | -| 德国（法兰克福） | `eu-central-1` | 数据不出欧盟 | -| 日本（东京） | `ap-northeast-1` | 数据不出日本 | - -若对驻留无限制但追求更大推理资源池，可选择美国/德国/日本的全球部署范围。 - -## 三个关键概念 - -调用前需同时确认**地域**、**服务部署范围**、**接入域名**： - -- **地域**：决定接入点与数据存储位置。 -- **服务部署范围**：决定推理执行位置（数据是否跨域流转）。 -- **接入域名**：影响并发上限、超时、SLA 等服务保障。 - -三者在控制台或 Base URL 中体现，**各地域的接入域名、API Key、模型列表不能跨地域混用**。 - -## 接入域名类型 - -| 域名类型 | 格式 | 适用场景 | 关键指标 | -| --- | --- | --- | --- | -| 专属域名 | `{WorkspaceId}.{region}.maas.aliyuncs.com` | 生产环境 | SLA 99.9%、超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC | -| 共享域名 | `dashscope.aliyuncs.com` | 存量兼容 | — | -| 试用域名 | `trial.{region}.maas.aliyuncs.com` | 快速体验 | 不建议生产 | - -从共享域名迁移到专属域名只需替换 Base URL 中的域名部分，业务逻辑代码无需修改。 - -## 各地域调用差异 - -- **华北2（北京）**：默认推荐地域，专属域名 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。知识库、[模型部署](model-deployment.md)、数据管理、Prompt 模板、模型调优等大量能力仅在此地域开放。 -- **新加坡**：专属域名 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`。 -- **美国（弗吉尼亚）**：使用带 `-us` 后缀的模型名（如 `qwen-plus-us`）可限定美国境内推理；共享接入域名 `https://dashscope-us.aliyuncs.com/compatible-mode/v1`。 -- **德国（法兰克福）/日本（东京）**：通过业务空间区分全球/欧盟或全球/日本部署范围，调用前需先在业务空间管理页面创建并选择对应业务空间。 - -## 业务空间与 WorkspaceId - -使用北京、新加坡、日本、德国等地域时，需在 Base URL 中填入 WorkspaceId。WorkspaceId 可在控制台「业务空间管理」页面查看。子账号需加入对应业务空间并获得相应策略后方可操作该空间下的资源。 - -## 地域相关能力约束 - -部分能力仅在特定地域可用： - -| 能力 | 可用地域 | -| --- | --- | -| 知识库（标准版/旗舰版） | 仅华北2（北京） | -| [模型部署](model-deployment.md)（资源专享推理） | 仅华北2（北京） | -| 数据管理（训练集/[评测](evaluation.md)集/数据流） | 仅华北2（北京） | -| Prompt 模板 | 仅华北2（北京） | -| `qwen-deep-research` | 仅华北2（北京），且仅支持 Python DashScope SDK | -| `qwen-mt-plus` / `qwen3.5-ocr` | 北京、新加坡、美国（弗吉尼亚） | -| `gui-plus` | 华北2（北京） | - -## 选择建议 - -1. 明确数据驻留要求（是否允许数据出境）。 -2. 确认所需模型/能力是否在目标地域开放。 -3. 生产环境统一使用专属域名，避免使用试用域名。 -4. 同一业务的所有调用、API Key、模型列表保持在同一地域，不要混用。 - -## 关联主题页 - -- [get started with models](../guides/get-started-with-models.md) -- [knowledge base](../guides/knowledge-base.md) -- [model deployment 1](../guides/model-deployment-1.md) -- [prompt](../guides/prompt.md) -- [model data overview](../guides/model-data-overview.md) -- [more models](../api/more-models.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md b/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md deleted file mode 100644 index c08e584a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/rerank.md +++ /dev/null @@ -1,61 +0,0 @@ -# 重排序 - -重排序（Rerank）是检索流程中的二次精排环节：在向量或关键词召回得到候选文档后，由排序模型结合查询与文档的相关性重新打分并排序，把最相关的切片置顶，再交由后续生成或返回流程使用。 - -## 在百炼平台的使用场景 - -重排序贯穿于平台的检索增强与排序能力两条主线： - -- **知识库 RAG 流水线**：知识检索服务的流程为 Query 改写 → 向量 + 关键词混合检索 → Rerank 排序 → 加权返回。排序模型对召回阶段返回的 TopK 切片做精排，再按相似度阈值过滤与最大召回数量截断，决定最终送入生成模型的上下文。知识问答服务在多轮智能模式下会反复触发该精排环节。 -- **独立文本排序 API**：作为向量与排序模型能力的一部分，开发者可直接调用 rerank 接口，对自定义的候选文档列表按查询相关性排序，用于语义搜索、推荐系统、聚类分类等下游任务，无需绑定知识库。 - -## 模型选型 - -| 模型 | 最大文档数 | 单条最大 Token | 请求最大 Token | 特点 | -| --- | --- | --- | --- | --- | -| qwen3-vl-rerank | 文本 100 / 图片 40 / 视频 4 | 8,000 | 120,000 | 多模态，支持图文视频排序 | -| qwen3-rerank | 500 | 4,000 | - | 100+ 语种，高性能文本排序 | -| gte-rerank-v2 | - | - | 30,000 | 50+ 语种（2026-05-30 下线，建议迁移到 qwen3-rerank） | - -知识库内置的排序模型选项包括 `qwen3-rerank`、`qwen3-rerank(hybrid)` 与 `qwen3-vl-rerank`；多模态知识库只能选 `qwen3-vl-rerank`。 - -## 关键参数 - -独立调用 rerank API 时的核心参数： - -- `model`（必选）：模型名称 -- `query`（必选）：查询内容，qwen3-vl-rerank 支持文本与图片两种查询模态 -- `documents`（必选）：待排序文档列表 -- `top_n`（可选）：返回排序后的前 N 个文档 -- `instruct`（可选）：自定义排序任务说明，可指导模型采用不同排序策略（仅 qwen3-rerank 与 qwen3-vl-rerank） - -知识库检索流程中的相关配置： - -- **初步向量检索 TopK / 关键词检索 TopK**：1–100，默认 50，控制进入精排的候选规模 -- **排序模式**：问答模式按 QA 匹配度排序，相似模式按语义相似度排序，自定义高级模式可自行调参 -- **相似度阈值**：0.01–1.0，用于过滤精排后的低分切片，过高会丢弃全部结果 -- **最大召回数量**：1–20，最终返回的切片数 - -## API 接口 - -不同模型走不同的专属接口，Base URL 均为业务空间 ID 拼接的专属域名： - -- qwen3-rerank：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks` -- qwen3-vl-rerank / gte-rerank-v2：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank` - -请求需携带 `Authorization: Bearer `，API Key 在控制台 API Key 页面获取，业务空间 ID 在业务空间管理页面获取。 - -## 调优建议 - -- 召回结果不理想时，先建立至少 100 组评测基线，再按失败类型针对性调整。 -- 对「列举 / 总结 / 比较」类问题适当提高最大召回数量 K（1–20），并优先选按拼装长度避免超长截断。 -- 通过命中测试反复调整相似度阈值与召回片段数，找到精排质量与上下文长度的平衡点。 -- 多知识库召回时，相似度相同的切片优先返回权重高的库，但权重仅在同类知识库之间生效。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md b/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md deleted file mode 100644 index 53a3da52..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/sdk.md +++ /dev/null @@ -1,54 +0,0 @@ -# 软件开发工具包 - -软件开发工具包（SDK，Software Development Kit）是百炼平台为开发者提供的封装库，屏蔽底层 HTTP 接口与鉴权细节，使开发者能在 Python、Java、Node.js、Go 等语言中通过少量代码即可调用大模型、应用与实时多模态能力。SDK 通常随官方迭代持续升级，建议通过包管理器保持最新版本以获得最新功能与稳定性修复。 - -## 接入路径 - -百炼提供两条接入路径，开发者可按工程栈与迁移成本选择： - -1. **阿里云官方 DashScope SDK**：当前仅覆盖 Python 与 Java，封装了百炼原生协议（含智能体/工作流应用调用、异步任务管理、实时多模态交互等能力），适合需要使用百炼独有特性（如 Qwen-Audio、Omni Realtime）的场景。 -2. **OpenAI 兼容 SDK**：通过 OpenAI 多语言 SDK（Python/Java/Node.js/Go）调用百炼的 OpenAI 兼容接口，只需调整 `api_key`、`base_url`、`model` 三处参数即可复用现有 OpenAI 代码与生态工具，迁移成本最低。 - -## 语言与版本要求 - -| 语言 | DashScope SDK | OpenAI 兼容 SDK | -| --- | --- | --- | -| Python | `pip install -U dashscope`，需 Python ≥ 3.8 | `pip install -U openai` | -| Java | Maven/Gradle 引入 `com.alibaba:dashscope-sdk-java`（建议 ≥ 2.12.0） | 引入 `com.openai:openai-java`，需 Java 8+，推荐 3.5.0+ | -| Node.js | 暂无官方 DashScope SDK，可改用 HTTP/`axios` 或 OpenAI SDK | `npm install --save openai` | -| Go | 暂无官方 DashScope SDK | 需 Go 1.22+，`go get github.com/openai/openai-go/v3` | - -安装失败时可临时切换镜像源：Python 用国内 PyPI 镜像；Node.js 执行 `npm config set registry https://registry.npmmirror.com/`；Go 执行 `go env -w GOPROXY=https://mirrors.aliyun.com/goproxy/,direct`。 - -## 鉴权与配置 - -- **API Key 注入**：所有 SDK 均读取环境变量 `DASHSCOPE_API_KEY`，避免在代码中硬编码密钥。临时 API Key（有效期 1–1800 秒，前缀 `st-`）适用于浏览器、移动 App 等不可信环境。 -- **base_url**：使用业务空间专属域名 `https://{WorkspaceId}..maas.aliyuncs.com/compatible-mode/v1`，新加坡地域从旧域名 `dashscope-intl.aliyuncs.com` 迁移至专属域名可获得更好性能与稳定性；美国（弗吉尼亚）使用固定域名 `dashscope-us.aliyuncs.com`，不带 `{WorkspaceId}`。 -- **子业务空间**：若应用或知识库位于子业务空间，需额外配置业务空间 ID 环境变量（如 Spring AI Alibaba 应用集成用 `WORKSPACE_ID`，知识库检索用 `AI_DASHSCOPE_WORKSPACE_ID`）。各地域 API Key 不互通，切换地域需同步更换 Key。 - -## 典型使用场景 - -- **模型调用**：通过 Chat Completions、Responses、Embeddings 等兼容接口调用 Qwen 系列及三方直供模型；OpenAI SDK 复用现有代码，DashScope SDK 适配百炼独有协议。 -- **应用调用**：DashScope SDK 提供 `Application.call`（Python）/ `Application.call(param)`（Java）封装 `POST /apps/{app_id}/completion`，调用智能体应用与工作流应用方式一致，统一返回 `output.text` 供业务消费。 -- **实时多模态交互**：Qwen-Omni-Realtime 通过 WebSocket 长连接驱动，Python/Java SDK 封装了客户端事件（`session.update`、`input_audio_buffer.append`、`response.create` 等）与服务端事件，支持 VAD 与 Manual 两种交互模式。 -- **框架集成**：LlamaIndex（Python 3.9+）构建云端知识库 RAG 应用；Spring AI Alibaba（Spring Boot 3.x + JDK 17+）通过 `spring-ai-alibaba-starter-dashscope` 集成智能体/工作流应用并检索知识库。 -- **异步任务**：图像/视频生成等耗时模型采用异步机制，SDK 封装创建任务、查询结果、取消任务接口（限流 20 QPS）；建议结合 EventBridge 事件通知避免轮询。 - -## 选用建议 - -- 已有 OpenAI 代码栈或需要多语言覆盖 → 优先 OpenAI 兼容 SDK，仅改三处参数即可迁移。 -- 需要百炼独有能力（Qwen-Audio、Omni Realtime、智能体应用封装、异步任务管理）→ 选用 DashScope SDK（Python/Java）。 -- 仅做一次性脚本或跨语言验证 → 可直接 HTTP 调用，跳过 SDK 安装。 - -> **注意**：使用 OpenAI 兼容接口时，新加坡与北京地域的 API Key 不同，切换地域需同步更换；Qwen-Audio 不支持 OpenAI 兼容协议，必须走 DashScope 协议。 - -## 关联主题页 - -- [preparations](../api/preparations.md) -- [frameworks](../api/frameworks.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [more about models](../api/more-about-models.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/security-and-compliance.md b/skills/bailian-docs-llm-wiki/wiki/concepts/security-and-compliance.md new file mode 100644 index 00000000..f4da42fa --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/security-and-compliance.md @@ -0,0 +1,51 @@ +# 安全与合规 + +安全与合规是百炼平台面向生产环境构建的核心保障能力，指通过技术手段与管理机制，确保模型服务在内容安全、数据传输、访问控制、隐私保护、算法备案及监管适配等维度符合国家法规（如《生成式人工智能服务管理暂行办法》）和企业安全策略要求。该能力不是单一功能，而是贯穿模型调用、应用部署、权限治理与可观测性全链路的横切约束。 + +## 在百炼平台的不同场景中，这个概念如何使用 + +- **内容安全防护**：在模型输入/输出环节启用 AI 安全护栏（需显式开通并配置请求头），实时拦截涉政、涉黄、广告、暴力等高风险内容，响应返回标准化错误码 `data_inspection_failed`，便于业务快速熔断或降级。 +- **数据传输加密**：对敏感 Prompt 或用户数据启用端到端加密（仅 DashScope 原生 Endpoint 支持），使用 RSA 加密 AES 密钥 + AES-CBC 加密请求体，防止公网传输中明文泄露；OpenAI 兼容模式不支持此能力。 +- **网络与访问隔离**： + - 通用场景：通过 PrivateLink 终端节点将 VPC 流量直连百炼 API，实现私网访问； + - 高敏场景：使用「安全存储业务空间」+ 反向终端节点 + MSE 网关 + 私有云资源（OSS/ADB/ES），确保训练数据、知识库、推理结果全程不出私网。 +- **权限最小化管控**：以业务空间为单元，通过角色（超级管理员/空间管理员/普通用户）+ 模型级开关（调用/调优/部署）+ RAM 策略（OpenAPI 权限）三层授权，实现“谁在什么空间、能调什么模型、能做什么操作”的精准控制；API Key 权限严格继承自所属空间配置，与用户控制台权限解耦。 +- **合规可追溯性**： + - 所有上线模型（含千问、万相及第三方模型）均公示算法备案号与大模型备案号，支撑应用上架备案； + - 推理日志（需开通）完整记录输入/输出内容，配合模型监控中的 `model_call_failure_count`（含内容安全拦截次数）指标，满足审计与溯源要求； + - 临时 API Key（`st-***`）提供短时效凭证，适用于前端/移动端等不可信环境，避免长期密钥暴露。 + +## 关键参数和配置 + +| 参数名 | 作用 | 必填条件 | 示例值 | 使用场景 | +|--------|------|----------|--------|----------| +| `X-DashScope-DataInspection` | 启用输入/输出内容安全检测 | 启用护栏时必填 | `{"input":"cip","output":"cip"}` | HTTP Header 中传递，触发 AI 安全护栏 | +| `X-DashScope-EncryptionKey` | 传递 RSA 加密后的 AES 密钥信息 | 启用传输加密时必填 | `{"public_key_id":"1","encrypt_key":"...","iv":"..."}` | HTTP Header 中传递，用于端到端解密 | +| `enable_encryption=True` (Python) / `enableEncrypt(true)` (Java) | SDK 层自动处理加解密逻辑 | 使用 SDK 且需加密时必填 | `True` / `true` | 替代手动构造加密请求，推荐生产环境使用 | +| `base_url` | 指向私网终端节点域名 | 使用 PrivateLink 或反向终端节点时必填 | `https://vpc-cn-beijing.dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation` | 替换默认公网地址，所有请求走私网 | +| `workspace_id` | 显式指定目标业务空间 | 所有跨空间 API 调用必填 | `ws-abc123xyz` | 通过 `X-Workspace-ID` Header 或请求体传入，决定权限上下文与计费归属 | + +> ⚠️ 注意： +> - OpenAI 兼容模式（`/compatible-mode/v1`）**不支持**传输加密与部分安全头（如 `X-DashScope-DataInspection`）； +> - 「默认业务空间」无法配置模型权限开关与限流，**严禁用于生产环境**； +> - 安全存储业务空间必须使用**反向终端节点**（非接口终端节点），且仅支持华北2（北京）地域指定可用区。 + +## 面向开发者，简洁实用 + +- ✅ **快速启用内容安全**：开通服务 → 控制台完成授权 → 请求头加 `X-DashScope-DataInspection` → 监控 `model_call_failure_count` 指标看拦截效果。 +- ✅ **私网访问两步走**： + ① 控制台创建终端节点（PrivateLink 或反向）→ 获取私网域名； + ② 将 SDK 的 `base_url` 或请求 URL 替换为该域名（注意路径保持一致）。 +- ✅ **权限调试口诀**：API 调用是否成功？先查 `workspace_id` 是否正确 → 再查该空间是否已开通目标模型调用权限 → 最后确认 RAM 用户是否被授予 `AliyunBailianDataFullAccess`（如需调用 OpenAPI）。 +- ✅ **合规备案就绪检查**：调用前确认模型备案号已在控制台公示；上架应用前完成[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)。 +- 🚫 **避坑提醒**：不要在默认空间跑生产流量；不要用 OpenAI SDK 调用加密接口；不要在临时 API Key 场景执行高权限操作（如模型调优）。 + +## 关联主题页 + +- [security and compliance](../guides/security-and-compliance.md) +- [application permission management](../guides/application-permission-management.md) +- [more](../api/more.md) +- [application support](../guides/application-support.md) +- [model monitoring](../guides/model-monitoring.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md index 04b27e08..26ca2580 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming-output.md @@ -1,36 +1,49 @@ # 流式输出 -流式输出（streaming）是指服务端将模型生成的结果分多次增量返回，而非等全部内容生成完毕后一次性下发。它能显著降低首字延迟、提升实时交互体验，适用于对话、语音等对响应速度敏感的场景。 +流式输出（Streaming Output）是指模型服务在生成响应过程中，将结果以增量、分块的方式持续返回给客户端，而非等待全部内容生成完毕后一次性返回。这种方式显著降低端到端延迟，提升用户体验，尤其适用于长文本生成、实时语音合成、[多模态](multi-modal.md)交互等对响应速度敏感的场景。 -## 在百炼平台的使用场景 +## 在百炼平台的不同场景中如何使用 -百炼平台在多类接口中都提供了流式输出能力，具体开启方式因协议而异： +流式输出在百炼平台中广泛支持，但具体实现方式和适用范围因接口协议与模型类型而异： -- **应用调用（智能体/工作流）**：无论是 OpenAI 兼容的 Responses API 还是 DashScope API，都支持流式输出。以 Responses API 为例，通过在请求中设置 `stream=true` 即可开启，服务端会以增量方式持续返回生成内容。 -- **文本生成模型 API**：OpenAI 兼容 Chat Completions、OpenAI 兼容 Responses、Anthropic 兼容 Messages 以及 DashScope 原生接口均支持流式返回。使用 OpenAI 客户端库时，通常只需替换 base URL 和 API Key，并按对应生态的字段约定开启流式参数。 -- **实时多模态交互（Qwen-Omni-Realtime API）**：基于 WebSocket 协议实现天然的流式交互。服务端通过一系列增量事件持续推送结果，例如 `response.audio.delta`（增量音频输出）、`response.audio_transcript.delta`（增量文本转录）、`conversation.item.input_audio_transcription.delta`（实时语音识别中间结果），最终以 `response.done` 表示本轮响应完成。 +- **Qwen 文本生成 API（DashScope / OpenAI 兼容）**：所有同步文本生成接口均支持 `stream=true`。DashScope 接口通过 `output.text` 字段逐块返回文本；[OpenAI 兼容接口](openai-compatible-interface.md)则遵循 OpenAI 标准，返回 `choices[0].delta.content` 字段，需按 chunk 拼接。注意：`qwen-max` 在 DashScope 接口中默认启用思考模式，流式输出会包含中间推理步骤（如 `...`），而 [OpenAI 兼容接口](openai-compatible-interface.md)不暴露该结构。 -## 关键参数与配置 +- **Omni Realtime API（WebSocket）**：作为原生实时协议，流式是默认行为。文本输出以 `text` 事件形式实时推送；若启用 `output_modalities: ["TEXT", "AUDIO"]`，音频流通过 `audio` 事件以 PCM 分片形式持续下发，支持低延迟 TTS 播放。 -| 场景 | 参数/事件 | 说明 | -|------|-----------|------| -| Responses API / Chat Completions | `stream` | 布尔值，是否流式输出，默认 `false`；设为 `true` 开启 | -| DashScope API | 流式开关 | 通过 SDK 的流式调用方式或对应参数开启 | -| Omni-Realtime API | `response.*.delta` 系列服务端事件 | WebSocket 连接下增量推送音频、文本转录等结果 | -| Omni-Realtime API | `response.done` | 标记单轮响应生成结束 | -| Qwen3-Omni-Flash-Realtime | `smooth_output` | 控制输出平滑度的可选参数 | +- **Realtime API（WebSocket/WebRTC/AOQ）**：所有协议均天然支持流式。WebSocket 使用 HTTP chunked encoding 或 WebSocket message 分帧；WebRTC 和 AOQ 则基于 DataChannel 或 QUIC stream 实时传输文本/音频片段，开发者需监听对应事件（如 `onTextChunk`, `onAudioFrame`）。 -## 使用建议 +- **Application Call（智能体/工作流调用）**：仅 OpenAI 兼容的 Responses API 支持 `stream=true` 参数；DashScope 原生应用调用接口暂不支持流式。**关键前提**：工作流应用必须在结束节点显式开启“流式输出”开关并重新发布，否则即使客户端传 `stream=true` 也降级为非流式响应。 -- 需要即时反馈、逐字/逐段展示的实时交互（如聊天界面、语音助手）优先启用流式输出。 -- 流式模式下需在客户端持续读取并拼接增量片段，直到收到结束标志（HTTP 流的结束或 `response.done` 事件）。 -- 流式与异步调用（`background=true`）面向不同需求：流式关注实时增量返回，异步关注长耗时任务的非阻塞执行，二者不要混淆。 -- 跨接口迁移时需核对各生态对流式参数的字段约定差异，DashScope 参数最全，OpenAI/Anthropic 兼容接口以对应生态约定为准。 +- **[多模态](multi-modal.md)模型（如 Qwen-VL、Tripo、Fun-Music）**：目前**不支持流式输出**。图像、3D、视频、音乐等生成任务均为同步完成，返回完整二进制或 URL 结果。 + +## 关键参数和配置 + +| 参数 | 类型 | 说明 | 适用接口 | 注意事项 | +|------|------|------|----------|----------| +| `stream` | boolean | 启用流式响应开关 | Qwen API（DashScope & OpenAI 兼容）、Application Call（Responses API） | 必须设为 `true`；HTTP 请求头需保持 `Connection: keep-alive`，响应 Content-Type 为 `text/event-stream`（SSE）或 `application/octet-stream`（WebSocket） | +| `output_modalities` | array | 指定输出模态组合 | Omni Realtime API、Realtime API | 如 `["TEXT"]` 或 `["TEXT", "AUDIO"]`；影响流式事件类型与频率 | +| `turn_detection.type` | string | VAD 类型，影响语音输入与文本输出节奏 | Omni Realtime API | `"semantic_vad"` 可触发更自然的语义级流式中断，仅 `qwen3.5-omni-realtime` 支持 | +| `smooth_output` | boolean | 控制 TTS 输出是否口语化分段（避免生硬停顿） | `qwen3-omni-flash-realtime` | 仅该模型支持；设为 `true` 时，文本流更贴近自然说话节奏 | + +> ⚠️ 注意： +> - 流式响应下，错误不会延迟抛出——若请求中途失败（如鉴权失败、token 耗尽），服务端会立即发送 `error` 事件并关闭连接； +> - 客户端必须正确处理 partial chunk（如 [OpenAI 兼容接口](openai-compatible-interface.md)中 `delta.content` 可能为空字符串，表示换行或格式标记）； +> - 所有流式接口均**不支持重试幂等性**，重复发送相同 `stream=true` 请求会产生独立流，需由客户端自行管理会话状态。 + +## 面向开发者：简洁实用建议 + +- ✅ **首选 SDK**：使用 `dashscope` Python SDK（v1.20.0+）或官方 Realtime SDK，内置流式解析器与自动重连逻辑，避免手动解析 SSE 或 WebSocket 事件。 +- ✅ **文本流拼接**：OpenAI 兼容接口中，始终检查 `delta.content` 是否为 `None` 或空字符串，仅追加非空内容；DashScope 接口关注 `output.text` 字段，忽略 `output.choices[0].finish_reason` 为 `null` 的中间响应。 +- ✅ **音频流播放**：Omni Realtime 的 PCM 音频流采样率固定为 24kHz，位深 16bit，单声道；建议使用 Web Audio API 或 Android AudioTrack 进行 buffer 管理，避免卡顿。 +- ❌ **避免滥用**：非实时场景（如批量摘要、离线报告生成）请使用非流式接口，减少连接开销与资源占用。 +- 🔍 **调试技巧**：启用 `debug=true`（部分接口支持）可查看 token 级别生成过程；结合 `X-DashScope-Trace-ID` 头定位流式异常链路。 ## 关联主题页 -- [application call](../api/application-call.md) -- [omni realtime api](../api/omni-realtime-api.md) - [qwen api reference](../api/qwen-api-reference.md) +- [omni realtime api](../api/omni-realtime-api.md) +- [realtime api user guide](../api/realtime-api-user-guide.md) +- [model experience](../guides/model-experience.md) +- [application call](../api/application-call.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md b/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md deleted file mode 100644 index 82223a7f..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/streaming.md +++ /dev/null @@ -1,47 +0,0 @@ -# 流式输出 - -流式输出（Streaming）是指服务端在生成内容的过程中，以增量分片（delta）的方式持续返回结果，而非等待全部内容生成完毕后一次性返回。它能显著降低首字延迟、改善实时交互体验，广泛用于对话、语音助手等场景。 - -## 在百炼平台的使用场景 - -百炼平台在多类接口中都支持流式输出，核心场景包括： - -- **应用调用（Application Call）**：无论是 OpenAI 兼容的 Responses API 还是 DashScope 原生 API，调用智能体或工作流应用时均支持流式输出，可用于需要边生成边展示的实时交互场景。 -- **文本生成模型 API**：OpenAI 兼容 Chat Completions / Responses、Anthropic 兼容 Messages 以及 DashScope 原生接口均可开启流式返回，适合聊天补全类应用逐字/逐段渲染输出。 -- **实时多模态交互（Omni-Realtime API）**：基于 WebSocket 协议，流式是其原生工作方式。服务端通过一系列增量事件持续推送音频与文本，天然适配低延迟的语音对话场景。 - -## 关键参数与配置 - -### HTTP / SDK 接口 - -- **`stream`**：布尔值，控制是否开启流式输出，默认 `false`。设为 `true` 后，服务端以 SSE（Server-Sent Events）方式逐片返回结果。 - - 适用于 OpenAI 兼容 Responses API、DashScope API 等应用调用与模型调用接口。 -- 使用 SDK 时，开启 `stream=true` 后通过迭代响应对象逐步获取增量内容；部分接口可配合 `stream_options` 等参数控制是否返回用量统计等附加信息（以对应接口文档为准）。 - -### Omni-Realtime（WebSocket） - -实时接口不使用 `stream` 参数，而是以事件流的形式天然流式返回。关键的增量事件包括： - -| 事件 | 含义 | -| --- | --- | -| `response.audio.delta` | 增量音频输出 | -| `response.audio_transcript.delta` | 增量文本转录 | -| `conversation.item.input_audio_transcription.delta` | 实时语音识别中间结果 | -| `response.done` | 本轮响应流结束 | - -此外，可通过 `session.update` 事件中的 `smooth_output`（部分模型支持）等参数调节流式输出的平滑度。 - -## 开发建议 - -- **注意结束标志**：SSE 流需处理到结束标记（如 `[DONE]`）或 `response.done` 事件后再收尾，避免内容截断。 -- **增量拼接**：客户端需将各 delta 分片按序拼接，才能得到完整结果。 -- **错误处理**：流式过程中仍可能收到错误事件，需在读取流的循环中做好异常捕获与连接重试。 -- **跨接口差异**：不同接口的分片结构与字段命名不同（OpenAI/Anthropic 兼容接口以对应生态约定为准，DashScope 参数最全），跨接口迁移时需核对字段映射。 - -## 关联主题页 - -- [application call](../api/application-call.md) -- [qwen api reference](../api/qwen-api-reference.md) -- [omni realtime api](../api/omni-realtime-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md deleted file mode 100644 index 4e432064..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token-and-billing.md +++ /dev/null @@ -1,57 +0,0 @@ -# Token 与计费 - -Token 是百炼平台衡量模型用量与费用的基本单位，绝大多数计费、监控、限流与额度管理都以 Token 为核心口径展开。理解 Token 如何被计量与结算，是准确预估成本、规避意外扣费的前提。 - -## Token 在百炼各场景中的用途 - -- **调用计量**：大语言模型、全模态、向量模型的用量统一按 Token 统计；图像生成按张、视频生成按秒、语音模型按秒/字符/Token（视模型而定）。 -- **费用结算**：模型实时推理、Batch 调用、模型训练、上下文缓存等费用均由 Token 消耗量换算得出。 -- **监控与告警**：模型监控、应用观测均以 Token 量（总量/输入/输出、平均单次请求 Token 量、首 Token 延时）作为关键指标。 -- **额度与套餐抵扣**：新人免费额度、资源包、节省计划、Token Plan 的 Credits 都以 Token 消耗为抵扣依据。 -- **容量规划**：TPM（Tokens Per Minute）预留、快速模式的吞吐能力，均以 Token/分钟为容量单位。 - -## 计费方式与抵扣优先级 - -百炼默认按量后付费（分钟级出账、按月结算），实时调用时按固定优先级自动抵扣，无需手动指定： - -**免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** - -- **免费额度**仅抵扣模型实时推理，不抵扣 Batch、模型调优、模型部署；每个模型（含不同快照版本）额度独立，通常 100 万 Token，有效期 30~90 天，耗尽后返回 `AllocationQuota.FreeTierOnly`。 -- **资源包**预购具体 Token/张数，抵扣单个模型超出免费额度后的实时推理用量，到期作废。 -- **AI 通用型节省计划**承诺月消费换阶梯折扣（最高 5.3 折），以动态月为周期发放额度，当月未用完清零。 -- **Token Plan / Coding Plan** 专属 Key 不消耗免费额度，直接按套餐 Credits 或调用次数计费。 - -> 注意：账户欠费（可用额度 < 0）时，即使某模型仍有免费额度也无法调用；请提前配置余额预警或消费限额。 - -## Token 计价规则要点 - -- **阶梯计费**：部分模型（如 `qwen3-max`）单价取决于单次请求输入 Token 总量，落入哪档就整体按该档单价结算（K=1,000，M=1,000,000）。 -- **思考 / 非思考模式**：部分千问模型（如 `qwen-plus`）两种模式的输出单价不同。 -- **地域差异**：同一模型在不同地域（北京、弗吉尼亚、新加坡等）单价不同，海外通常更高。 -- **Batch 与上下文缓存**：Batch 输入/输出按实时价 50% 计费；上下文缓存仅对输入 Token 打折（显式缓存创建按标准输入 125%、命中按 10%），两者不能同时生效。 -- **训练与部署**：训练按训练 Token 计费；部署按使用时长结合输入/输出 TPM 单价或模型单元计费。免费额度和节省计划均不抵扣训练与部署费用。 - -## 关键参数与配置 - -- **`max_tokens`**：限制输出 Token 上限，是控制生成长度、防止过度生成、压降成本与延迟的主要开关。 -- **专属模型 code**：TPM 预留创建后需将请求中的 `model` 参数替换为专属 code，才能使用预留容量；超出自动降级为按量付费（响应 Header `x-dashscope-ptu-overflow:true`）。 -- **API Key 与 Base URL 配套**：按量付费、Token Plan（`sk-sp-` 开头）、Coding Plan 三者的 Key 与 Base URL 必须配套使用，混用会导致意外走按量通道或 401/403 鉴权失败。 -- **免费额度用完即停**：开启后额度耗尽即停止响应、不再扣费；开启此安心模式后节省计划将无法继续抵扣，需先关闭。 - -## 用量查询与成本管理 - -- **模型用量 / 模型监控**：控制台按业务空间维度统计，数据延迟约 1 小时，支持分钟/小时/天精度，可查看历史 30 天 Token 消耗并设置费用告警。 -- **单次调用 Token 消耗**：需在模型监控配置中开通审计日志与推理日志，目前仅覆盖华北2（北京）部分模型/快照。 -- **账单定位**：账单以"实例 ID"为出账粒度，格式为 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此定位费用来源。 -- **更早数据**：超过 30 天的统计需前往阿里云"费用与成本"控制台查询、开票。 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) -- [support](../guides/support.md) -- [model high speed inference](../guides/model-high-speed-inference.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md deleted file mode 100644 index 169b6401..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token-billing.md +++ /dev/null @@ -1,69 +0,0 @@ -# Token 与计费 - -Token 是百炼平台衡量模型处理文本量的基本单位，也是绝大多数计费、限流和用量统计的计量基础；平台的费用则围绕 Token 消耗，通过按量付费、免费额度、节省计划与资源包等机制综合结算。理解 Token 如何被计量与抵扣，是控制大模型使用成本的前提。 - -## Token 在计费中的角色 - -对大语言模型、全模态模型和向量模型，用量与费用均按 **Token** 计量；图像生成按「张」、视频生成按「秒」、语音模型按「秒/字符/Token」（视模型而定）。因此谈「计费」时，Token 主要针对文本类调用。 - -模型调用按**输入 Token** 和**输出 Token** 分别计费，单价以「每百万 Token」为单位。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档定价，落入某一区间后该请求全部 Token 按对应档位结算（例如 `qwen3-max` 华北2·北京划分为 `0 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** - -- **免费额度**：首次开通时各模型自动发放（通常每模型 100 万 Token），仅抵扣实时推理，不抵扣 Batch、调优、部署；不同模型（含同一模型不同快照版本）额度相互独立。开启「免费额度用完即停」后，额度耗尽会停止响应并返回 `AllocationQuota.FreeTierOnly`（或 403），避免意外扣费。 -- **折扣**：Batch 调用输入/输出单价按实时推理价的 50% 计费；支持上下文缓存的模型仅输入 Token 享折扣，两者不能同时生效。 -- **地域差异**：同一模型在不同地域单价不同，境外地域通常无免费额度。 - -### 2. 模型训练（调优） - -按训练 Token 计费。文本模型公式为 `(训练数据 Token + 混合数据 Token) × 循环次数 × 训练单价`；图像/视频生成模型的训练 Token 总量由 `max_steps`、`max_pixels`、`n_epochs` 等超参决定。免费额度和节省计划**均不抵扣**训练费用。 - -### 3. 模型部署 - -免费额度和节省计划同样**不抵扣**部署费用。三种计费方式围绕 TPM（每分钟 Token 数）： - -- **预置吞吐（PTU）**：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)`。PTU 下超出购买吞吐或输入超上限时自动转按量付费（响应头 `x-dashscope-ptu-overflow:true`）。长输入按阶梯系数折算 TPM 消耗，命中前缀缓存的 Token 按折扣系数消耗额度。 -- **模型单元（MU）**：`费用 = 使用时长 × 模型单元数量 × 模型单元单价`。 -- **按 Token 使用量**：仅对 LoRA 微调后的自定义模型开放，用于调优效果验证。 - -### 4. 订阅套餐（Token Plan / Coding Plan） - -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，面向团队协作。 -- **Coding Plan**：按模型调用次数计量，面向个人开发。 - -两者均使用 `sk-sp-` 前缀的**专属 API Key**，**不消耗新人免费额度**，且 Base URL 与按量付费端点完全隔离，混用会导致意外扣费或 401/403 鉴权失败。 - -### 5. 监控与用量统计 - -模型监控将 Token 消耗归入「成本」类指标，并提供首 Token 延时、RPM/TPM 等性能指标。应用观测则可查看每次调用的输入/输出/平均 Token 量与平均首 Token 耗时。开通推理日志后可查看单次调用的 Token 消耗，用于排查与审计。 - -## 关键参数与配置 - -- **`max_tokens`**：限制单次生成的最大输出 Token 数，是控制输出成本和防止过度生成的首要手段。 -- **TPM / RPM 限流**：部署时可配置 `tpm_limit`、`rpm_limit`；触发限流后等待时间取决于限流值。 -- **PTU 计费识别字段**：`service_tier`（`ptu-standard` 走 PTU 额度，`default` 或缺失表示按量）、`provisioned_tokens`（折算后实际消耗额度）、`cached_tokens`（缓存命中数，Anthropic 兼容格式暂不返回）。这些字段在 OpenAI Chat、OpenAI Responses、Anthropic 兼容、DashScope 四种协议下 JSON 路径不同，需按协议读取。 - -## 成本优化与出账 - -- **节省 Token**：优化 Prompt 减少输入 Token、简单任务选用轻量级模型、非实时任务走批量推理、合理设置 `max_tokens`。 -- **预付费方案**：AI 通用型节省计划（承诺月消费换阶梯折扣，最高 5.3 折，月额度不可跨月累积）、其他模型节省计划、资源包（预购具体 Token/图片数量，仅抵扣单个模型超免费额度后的实时推理）。 -- **出账时效**：大模型推理分钟级出账（约 2～10 分钟）；批量推理、模型训练等小时级出账。用量统计数据约 1 小时延迟，且不支持查看 30 天前数据。 - -## 关联主题页 - -- [test 1](../guides/test-1.md) -- [token plan guide](../guides/token-plan-guide.md) -- [model monitoring](../guides/model-monitoring.md) -- [application monitoring](../guides/application-monitoring.md) -- [support](../guides/support.md) -- [model deployment 1](../guides/model-deployment-1.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md index 0b4e6e3c..b35af495 100644 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/token.md +++ b/skills/bailian-docs-llm-wiki/wiki/concepts/token.md @@ -1,65 +1,50 @@ -# Token 与计量计费 +# Token -Token 是大语言模型处理文本的最小计量单位，百炼平台以 Token 为核心，对模型的输入、输出、训练用量进行计量、计费与监控。理解 Token 的产生方式与计费规则，是控制模型调用成本、优化应用性能的基础。 +Token 是百炼平台中用于计量模型调用资源消耗的核心单位，代表模型处理输入与生成输出所消耗的计算资源。它并非加密凭证，而是按字节、字符或语义单元（如子词）标准化统计的**计费与用量度量基准**，直接影响 Credits 扣减、性能监控和成本治理。 -## 什么是 Token +## 在百炼平台的不同场景中，这个概念如何使用 -Token 是模型在处理文本时切分出的基本片段（一个汉字、单词或子词可能对应一个或多个 Token）。在百炼平台中，Token 既是**用量计量单位**，也是**大部分计费的核算基准**： +- **计费计量**：Token 是 Token Plan 订阅服务的底层计费单元。每次调用（无论同步/异步）均按 `输入 Token 数 + 输出 Token 数` 动态折算 Credits，不同模型单价不同（如 `qwen3.6-plus` 约 3.18 Credits/次），思考模式（`enable_thinking`）或 Harness 工具调用会额外增加 Token 消耗。 + +- **性能观测**：在应用监控与模型监控中，`Token 总量` 是关键可观测指标，精确拆分为 `input_tokens` 和 `output_tokens`，用于分析首 Token 延时、TPS（Tokens Per Second）、单次请求成本效率，并支持按业务空间、API Key 或模型维度下钻分析。 -- 文本生成模型按**输入 Token** 和**输出 Token** 分别计量，思考模式下的输出 Token 同时包含「思维链 + 回答」两部分。 -- 不同模型类型的计量单位不同：大语言模型 / 全模态模型 / 向量模型按 **Token** 计量，图像生成按**张**，视频生成按**秒**，语音模型按**秒 / 字符 / Token**（视模型而定）。 +- **能力约束**：部分模型通过 `max_output_tokens` 参数显式限制输出长度；上下文窗口（如 `qwen3.7-plus` 支持 1M Token）本质是输入 Token 的硬性上限；[多模态](multi-modal.md)输入（图像、视频、音频）会经预处理转换为等效 Token 量参与计费与限流。 -## 在各场景中的使用 +- **[安全与合规](security-and-compliance.md)**：内容安全拦截、限流错误（HTTP 429）等事件在模型监控中关联 Token 消耗记录，便于定位异常调用（如恶意长 [prompt](../guides/prompt.md)）或优化提示工程以降低无效 Token 开销。 -### 1. 按量付费与免费额度 +- **开发调试**：SDK 和 API 返回中通常包含 `usage` 字段（如 `"input_tokens": 127, "output_tokens": 89`），开发者应主动解析该字段用于本地成本估算、缓存策略或用户用量展示。 -- 首次开通时，各模型会发放新人专属免费额度（通常各 100 万 Token），仅抵扣**实时推理**费用，且不同模型（含同一模型不同快照版本）额度相互独立、不可合并。 -- 免费额度耗尽后默认转为**按量付费**，按输入 / 输出 Token 计费。部分模型采用**阶梯计费**：按单次请求的输入 Token 总量分档（如 `qwen3-max` 分 0–32K / 32K–128K / 128K–256K 三档），落在哪一档，该请求全部 Token 均按该档单价结算。 -- 同一模型在不同地域（北京、弗吉尼亚、新加坡、法兰克福、东京）单价不同。 +## 关键参数和配置 -### 2. 订阅制套餐 +| 参数/字段 | 说明 | 开发者须知 | +|-----------|------|------------| +| `input_tokens` / `output_tokens` | API 响应 `usage` 对象中的标准字段，表示本次请求实际消耗的输入/输出 Token 数 | 必读字段，不可依赖估算值；流式响应中仅最终 completion 返回完整 usage | +| `max_output_tokens` | 可选请求参数，控制模型最大生成长度（Token 数），超出将截断 | 并非所有模型支持；设置过小可能导致输出不完整，过大可能增加成本与延迟 | +| `X-DashScope-OssResourceResolve: enable` | 使用 `oss://` URL 传入图片/视频时必需的 Header，否则平台无法解析资源并统计对应 Token | [多模态](multi-modal.md)调用必配，缺失将返回 400 错误 | +| `enable_thinking` | 布尔参数（文本模型专用），启用深度推理模式时显著增加中间 Token 消耗 | 需权衡效果与成本，建议 A/B 测试后启用 | +| `reasoning.effort` | 控制思考深度的数值参数（如 `"low"`/`"medium"`/`"high"`），影响 Token 消耗与响应时间 | 与 `enable_thinking` 协同生效，高 effort = 更高 Token 成本 | -- **Token Plan 团队版**：按 Token 消耗抵扣 Credits，工具调用（联网搜索、代码解释器等内置工具）产生的 Token 同样从套餐 Credits 抵扣，不额外收费。 -- **Coding Plan**：按模型调用**次数**计费（而非 Token），面向个人开发场景。 +> ⚠️ 注意：Token 统计严格绑定地域（仅华北2北京有效）、API Key 类型（`sk-sp-` 专属 Key）及 Base URL；跨地域或混用 Key 将导致 Token 计量失效或调用失败。 -### 3. 模型训练与部署 +## 面向开发者，简洁实用 -- **训练**按训练 Token 计费，文本模型公式为 `（训练数据 Token + 混合训练数据 Token）× 循环次数 × 训练单价`；图像 / 视频模型的训练 Token 由 `max_steps`、`max_pixels`、`n_epochs` 等超参数推算。 -- **部署**（预置吞吐 TPM）按输入 / 输出 TPM 单价与时长计费，与 Token 用量间接相关。训练与部署**不能**用免费额度或节省计划抵扣。 +- **不要估算，要实测**：不同模型、不同输入内容（尤其含 emoji、代码、多语言）的 Token 数差异极大。使用 [DashScope Tokenizer 工具](https://help.aliyun.com/zh/model-studio/token-calculator) 或 SDK 的 `count_tokens()` 方法本地预估，再以 API 实际返回 `usage` 为准。 + +- **监控必看三项**：在控制台模型监控页，重点关注 `model_usage`（总用量）、`model_first_token_duration_p99`（首 Token 延时）、`model_call_failure_count`（失败次数）——三者共同反映 Token 效率与稳定性。 -### 4. 监控与观测 +- **降本三技巧**： + ① 对长文档 RAG，用 `TextRetriever` 默认 100 片段限制 + 精准 query 降低输入 Token； + ② 图像理解任务优先用 `qwen3.7-plus` 原生支持，避免 Skill 中转带来的额外 Token 开销； + ③ 异步任务（如视频生成）虽不实时返回 Token，但可在 `/api/v1/tasks/{task_id}` 查询结果中获取 `usage` 字段。 -- 应用观测可查看每次调用的 Token 量，监控统计提供 Token 总量（全部 / 输入 / 输出）、平均单次请求 Token 量、平均首 Token 耗时等指标。 -- 模型监控将 Token 消耗归入「成本」类指标，首 Token 延时归入「性能」类指标，支持按分钟 / 小时 / 天聚合，并可配置告警。 -- 用量统计按业务空间维度归集，数据延迟约 1 小时。 - -## 关键参数与配置 - -| 项 | 说明 | -| --- | --- | -| `max_tokens` | 限制单次生成的最大输出 Token 数，用于控制成本、防止过度生成，也是降低幻觉的手段之一 | -| 上下文缓存 | 命中缓存的**输入** Token 享折扣；缓存折扣与 Batch 折扣不可同时生效，价格表输入单价不含缓存单价 | -| Batch 调用 | 支持的模型输入 / 输出单价按实时推理价的 50% 计费 | -| 免费额度用完即停 | 开启后额度耗尽即停服（返回 403），避免意外扣费，但也会阻断节省计划继续抵扣 | - -## 成本优化建议 - -- **优化 Prompt**：简洁清晰的 Prompt 可减少不必要的输入 Token 消耗。 -- **控制输出长度**：合理设置 `max_tokens`，避免冗长输出。 -- **按任务选模型**：分类、摘要等简单任务优先使用轻量级模型。 -- **批量推理**：非实时大批量任务走 Batch 接口，Token 单价更低。 -- **善用抵扣顺序**：`免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费`，据此规划预付费方案。 - -## 账单中的 Token - -大模型推理为分钟级出账（通常 2～10 分钟），批量推理与训练为小时级。账单「实例 ID」以英文分号分隔，包含 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此区分输入 / 输出 Token 的费用来源与调用渠道（`app` 代码调用、`bmp` 控制台体验、`assistant-api`）。 +- **调试黄金法则**：当遇到 `429 Too Many Requests` 或 `400 Bad Request`，第一检查点是 `usage` 字段是否超限（如输入超 1M Token）或 `max_output_tokens` 设置不合理。 ## 关联主题页 - [token plan guide](../guides/token-plan-guide.md) -- [test 1](../guides/test-1.md) +- [model experience](../guides/model-experience.md) +- [more about models](../api/more-about-models.md) - [application monitoring](../guides/application-monitoring.md) - [model monitoring](../guides/model-monitoring.md) -- [support](../guides/support.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md deleted file mode 100644 index 195756dd..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vector-embedding.md +++ /dev/null @@ -1,57 +0,0 @@ -# 向量与嵌入 - -向量与嵌入（Embedding）是指将文本、图像、视频等非结构化内容映射为固定维度的数值向量，使语义相近的内容在向量空间中距离也更近，从而支持语义检索、聚类、推荐与 RAG 召回等下游任务。在百炼平台中，向量化是知识库检索与跨模态搜索的基础环节，与 Rerank 排序模型共同构成检索链路。 - -## 在百炼平台的使用场景 - -- **知识库检索召回**：文档搜索、数据查询、音视频搜索类知识库在创建时会对接向量模型，将入库切片向量化并写入索引；查询时对用户问题做同样的向量化，再执行向量 + 关键词混合检索。文档搜索、数据查询、音视频搜索类知识库支持 `text-embedding-v4` 或 `text-embedding-v3`（均为 512 维，维度不可更改）；视觉理解场景自动切换为 `qwen3-vl-embedding`；图片问答类仅支持 `multimodal-embedding-v1`（1024 维）。 -- **RAG 应用构建**：无论是云端知识库还是本地 RAG 方案，嵌入模型都决定了召回上限。云端方案使用百炼官方嵌入模型，不支持自定义切分与嵌入；本地 RAG 可改用自部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`），以灵活控制切分与嵌入。 -- **跨模态检索**：多模态向量模型将文本、图像、视频映射到同一语义空间，支持以文搜图、以图搜视频等。`qwen3-vl-embedding` 默认维度 2560，支持「独立向量」（每个输入各生成一向量，用于逐项对比）与「融合向量」（所有输入融合为 1 个向量，用于综合理解）两种类型。 -- **框架集成**：通过 LlamaIndex 构建 RAG 应用时，云端知识库默认使用百炼官方向量模型，配合 `similarity_top_k`、`similarity_cutoff`、`top_n` 等参数控制召回与重排。 - -## 文本向量模型 - -通用文本向量模型将文本转换为数值向量，当前推荐 `text-embedding-v4`（属 Qwen3-Embedding 系列），支持 100+ 主流语种。 - -| 模型 | 向量维度 | 最大行数 | 单行最大 Token | 语种 | -|------|---------|---------|---------------|------| -| text-embedding-v4 | 2048/1536/1024(默认)/768/512/256/128/64 | 10 | 8,192 | 100+ 语种 | -| text-embedding-v3 | 1024(默认)/768/512/256/128/64 | 10 | 8,192 | 50+ 语种 | -| text-embedding-v2 | 1,536 | 25 | 2,048 | 10 语种 | -| text-embedding-v1 | 1,536 | 25 | 2,048 | 6 语种 | - -**关键参数：** - -- `model`（必选）：模型名称 -- `input`（必选）：字符串、字符串列表或文件 -- `dimensions`（可选）：指定向量维度，仅 v3/v4 支持 -- `encoding_format`（可选）：当前仅支持 `float` - -**调用方式**：支持 OpenAI 兼容接口（`base_url: https://dashscope.aliyuncs.com/compatible-mode/v1`）和 DashScope SDK。对于大规模文本向量化，可使用异步批处理接口（`text-embedding-async-v1/v2`），单次最多 10 万行，通过 `X-DashScope-Async: enable` 请求头启用异步模式，提交后用 `task_id` 轮询结果；同时处理中任务不超过 50 个，并发上限 3。 - -## 多模态向量模型 - -多模态向量模型将文本、图像、视频统一映射到同一语义空间，支持跨模态检索。向量类型分「独立向量」（每个输入分别生成向量，适用于逐项对比）与「融合向量」（将所有输入融合为 1 个向量，适用于综合理解多模态内容）。`qwen3-vl-embedding` 默认维度 2560，支持独立与融合两种模式；`multimodal-embedding-v1` 默认 1024 维，用于图片问答类知识库。 - -## 知识库中的嵌入配置要点 - -- 文档搜索、数据查询、音视频搜索类知识库统一使用 512 维 `text-embedding-v4` / `text-embedding-v3`，维度不可更改。 -- 视觉理解场景自动切换为 `qwen3-vl-embedding`，无需手动指定。 -- 嵌入模型与切片策略共同决定召回质量：推荐使用智能切分（基于语义自适应选择切片点，单切片 Token 上限 6,000），切片过大或过小都会影响向量匹配效果。 -- 本地 RAG 场景若需自定义嵌入模型，可改用自部署 GTE 文本向量模型，但需注意 embedding API 限流，不建议传入超过 100 MB 的文件。 - -## 与 Rerank 的协作 - -向量检索负责语义召回（初步 TopK 默认 50，可设 1–100），Rerank 模型负责二次精准排序。知识库中可选 `qwen3-rerank` / `qwen3-rerank(hybrid)` / `qwen3-vl-rerank`（多模态库只能选 vl-rerank），排序后再按相似度阈值（0.01–1.0）与最大召回数量（1–20）裁剪。向量召回质量直接影响 Rerank 上限，调优时通常先确认嵌入模型与切片合理，再调整 Rerank 阈值与 K 值。 - -## 关联主题页 - -- [vector and sort](../api/vector-and-sort.md) -- [knowledge base](../guides/knowledge-base.md) -- [knowledge](../api/knowledge.md) -- [frameworks](../api/frameworks.md) -- [application use cases](../guides/application-use-cases.md) - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md b/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md deleted file mode 100644 index 523b5b79..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/vpc-private-access.md +++ /dev/null @@ -1,55 +0,0 @@ -# 私网访问 - -私网访问（VPC Private Access）是指阿里云百炼平台的模型调用、[模型部署](model-deployment.md)、异步任务通知等能力通过 VPC 内网或私网链路访问，避免公网暴露、提升传输安全性与网络稳定性的接入方式。它是百炼安全合规体系中"传输加密—私网访问"环节的核心抓手，与身份权限、内容安全、合规备案、安全存储共同构成端到端安全链路。 - -## 在百炼平台中的使用场景 - -### 1. 模型调用的私网接入 - -百炼模型调用默认通过公网（`dashscope.aliyuncs.com`）发起，对安全要求高或需统一出口的企业可走 VPC 私网。临时 API Key 可在不可信客户端环境使用，配合 VPC 私网访问可在内网完成调用链路，避免永久 Key 在公网流转。 - -> 各地域（北京、新加坡、弗吉尼亚）的 API Key 不互通，临时 Key 与生成它的永久 Key 必须属于同一地域，私网链路也需在对应地域内打通。 - -### 2. 异步任务完成通知的 VPC 回调 - -图像/视频生成等耗时任务接入事件总线 EventBridge 后，任务完成事件由事件总线推送。HTTP 回调 URL 接收端**支持公网或 VPC 访问**，对安全敏感场景建议使用 VPC 类型的 HTTP 接口作为事件目标，使通知链路全程在内网闭环。 - -- 事件源：`acs.dashscope` -- 事件类型：`dashscope:System:AsyncTaskFinish` -- 事件总线默认为 `default`（北京地域云服务专用总线） - -RocketMQ 消息队列方案同样可部署在 VPC 内，由事件总线投递后业务方在内网消费，支持消息无丢失与失败重试，可靠性更高。 - -### 3. [模型部署](model-deployment.md)与模型导入的私网通道 - -[模型部署](model-deployment.md)（PTU、模型单元、按 Token 用量）仅适用于华北二（北京）地域，部署后推理服务可在 VPC 内调用，满足高并发、低延迟及不出公网的诉求。 - -部署自定义 LoRA 模型时，需将本地训练的 LoRA 模型从阿里云 OSS 导入百炼。OSS Bucket 支持**私有 Bucket**与**内容加密**，并需添加 `bailian-datahub-access` 标签（值为 `read`）。通过 VPC 终端节点或私网访问 OSS，可避免训练产物经公网传输，保障数据安全。 - -## 关键参数与配置 - -| 维度 | 关键约束 | -| --- | --- | -| 地域一致性 | API Key、临时 Key、私网链路须同地域；模型部署仅北京 | -| 临时 Key 有效期 | 默认 60 秒，可通过 `expire_in_seconds` 设置，范围 [1, 1800] 秒 | -| 异步任务接口限流 | 20 QPS（按主账号 + 子账号维度）；通知方案不限流 | -| 任务保留时长 | 完成后约 24 小时自动清理 | -| OSS 导入标签 | Bucket 须加 `bailian-datahub-access=read` 标签 | -| OSS 存储类型 | 不支持归档/冷归档/深度冷归档；须使用子目录，非根目录 | -| 授权角色 | OSS 导入需开通 `AliyunServiceRoleForSFMDataHubOSSImport` 服务关联角色 | - -## 开发者实践建议 - -- **生产隔离**：按环境（dev/test/prod）划分业务空间，prod 流量走 VPC 私网，降低公网泄露面。 -- **回调优先 VPC**：异步任务通知的 HTTP 回调端点部署在 VPC 内，事件总线直推即可，无需公网暴露。 -- **OSS 导入走私网**：LoRA 模型导入使用私有 Bucket + 内容加密，并通过 VPC 终端节点访问 OSS。 -- **避免跨地域**：规划业务空间时明确地域，避免因 Key 与链路跨地域导致调用失败。 -- **PTU 部署内网化**：高负载生产环境优先 PTU 部署并经 VPC 调用，兼顾稳定吞吐与传输安全。 - -## 关联主题页 - -- [security and compliance](../guides/security-and-compliance.md) -- [more about models](../api/more-about-models.md) -- [model deployment 1](../guides/model-deployment-1.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md deleted file mode 100644 index 0badc9bf..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workflow.md +++ /dev/null @@ -1,75 +0,0 @@ -# 工作流 - -工作流（Workflow）是百炼平台三种核心应用构建模式之一，通过可视化节点编排将复杂任务拆解为有序步骤，逻辑确定、稳定可复现，适合流程固定、要求可复现的业务场景。 - -## 在百炼平台中的定位 - -百炼提供智能体、工作流、高代码应用三种互补的应用构建模式。工作流对应"可视化节点编排（低代码）"路线，由预定义节点精确控制流程，适合 IT 运维、业务分析师构建报告生成、订单处理、审批流等场景。与由大模型自主规划的智能体不同，工作流强调开发者对流程的显式控制，输出可复现、可调试。 - -## 核心节点 - -工作流画布由以下节点类型组成： - -- **开始 / 结束节点**：定义输入与输出参数。开始节点预置 `query`（用户输入）、`historyList`（对话历史）、`imageList`（图片）等变量，可在下游节点中引用。 -- **大模型节点**：执行 LLM 推理，配置模型、提示词、用户提示词、记忆等。 -- **意图分类节点**：根据输入分流到不同下游分支，支持多意图。 -- **变量处理节点**：用于文本输出或变量加工。 -- **智能体群组节点**：将任务分解给多个已发布的子智能体协同完成。 - -## 会话变量 - -会话变量作为全局变量在工作流全生命周期内记录参数，可在各节点中引用，在画布右上角配置。适合跨节点传递状态、累积上下文信息。 - -## 典型案例 - -- **诈骗信息识别**：开始 → 大模型（提示词判定诈骗嫌疑）→ 结束。 -- **智能导购**：意图分类节点将输入分流到手机 / 电视 / 冰箱等大模型分支，未命中分支走变量处理节点。 -- **日程管理助手**：通过智能体群组节点串联"信息收集"与"数据整理"两个子智能体。 - -## 创建与发布流程 - -1. 控制台 → 应用管理 → 创建应用 → 工作流应用。 -2. 在画布上拖拽编排节点，配置各节点的模型、提示词、变量引用关系。 -3. 右侧对话框调试。 -4. 右上角"发布"——发布是后续 API 调用与集成的前提。 - -## API 调用 - -工作流应用与智能体应用共用同一套调用接口，区别仅在应用内部的编排逻辑。发布后通过 `APP_ID` 调用： - -- **DashScope 原生 API**：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`，请求体为 `{"input": {"prompt": "..."}, "parameters": {}, "debug": {}}`，响应中业务侧主要消费 `output.text`。 -- **OpenAI 兼容 Responses API**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`，可复用现有 OpenAI 生态代码库，支持同步 / 异步、流式、多模态。 -- **SDK**：Python 使用 `dashscope.Application.call`，Java 使用 `com.alibaba.dashscope.app.Application`，Node.js 可直接以 `axios` 发起 POST。 - -调用前需在控制台获取应用 ID 与 API Key（推荐写入 `DASHSCOPE_API_KEY` 环境变量）；若应用位于子业务空间，还需提供 Workspace ID。 - -## 多轮对话 - -工作流应用支持多轮对话，有两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮。 -- **自行管理 `messages`（推荐）**：手动维护消息列表，灵活性更高，不受 session 有效期限制。 - -## 评测 - -工作流应用可作为评测任务的被测对象。在新版应用评测中，评测集支持智能体、工作流、自定义三种类型，按所选应用的出入参形式自动生成数据模板。评测任务可关联工作流应用，由 LLM 评估器或 Code 评估器自动评分，也可人工标注。 - -## 发布与分享 - -工作流应用支持多种发布渠道：UI 应用（通过 UI 设计器构建自定义界面）、钉钉机器人、微信公众号、组件化复用（将工作流作为其他智能体或工作流的子组件）、音视频实时互动（仅限图文对话类）。需要注意的是，官方网页版分享当前只支持智能体应用，不支持工作流应用。 - -## 关键限制与注意事项 - -- 工作流应用若配置了文件类型的自定义参数，在 UI 设计器中需指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），否则应用无法正确读取用户上传的文件。 -- 旧版"智能体编排"应用已被工作流应用替代，新建应用请直接选择工作流。 -- 工作流与智能体调用接口一致，但可附加的扩展能力（如自定义参数传递）取决于应用内部编排逻辑。 - -## 关联主题页 - -- [llm application](../guides/llm-application.md) -- [application call](../api/application-call.md) -- [bailian application calling](../guides/bailian-application-calling.md) -- [application evaluation](../guides/application-evaluation.md) -- [application publishing and sharing](../guides/application-publishing-and-sharing.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md b/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md deleted file mode 100644 index 51a00e4c..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/concepts/workspace.md +++ /dev/null @@ -1,61 +0,0 @@ -# 业务空间（Workspace） - -业务空间（Workspace）是阿里云百炼平台进行精细化权限管理（模型、用户）和阿里云账单分账的**最小管理单元**。平台按地理区域划分资源与业务空间，单个业务空间不能跨地域存在，即便是各地域的默认业务空间也彼此独立。 - -## 在不同场景中的使用 - -### 1. 权限与身份管理 - -业务空间是权限体系的核心边界，权限管理围绕三种角色展开： - -- **超级管理员**：阿里云主账号，或拥有 `AliyunBailianFullAccess` 系统策略的 RAM 用户。可跨空间统一管理用户权限、空间可用模型、模型限流和 API Key。 -- **业务空间管理员**：拥有访问某个业务空间「权限管理」页面的 RAM 用户，仅管理该空间内的用户与资源，其权限包含该空间下所有页面的访问权限。 -- **普通用户**：仅能访问/使用被授权的空间、页面与资源，不能管理用户或模型授权。 - -在非默认业务空间中，可对模型进行三类精细化授权（**默认业务空间无法设置这些限制**，即所有模型均可调用、调优、部署且无法限流）： - -| 权限项 | 控制范围 | -| --- | --- | -| 限制模型调用 | 是否可调用（控制台 & API）+ 请求数限流 + Token 限流 | -| 限制模型训练 | 是否可调优（控制台 & API）及调优后部署 | -| 限制模型部署 | 是否可直接部署 | -| 用户控制台权限 | 管理 RAM 用户能否使用该空间控制台及可用功能 | - -> **注意**：OpenAPI 接口权限不通过业务空间角色授予，必须由阿里云主账号在 RAM 控制台为 RAM 用户单独添加系统策略（如 `AliyunBailianDataFullAccess` / `AliyunBailianDataReadOnlyAccess`）。 - -### 2. API Key 归属 - -单个 API Key 只能归属**一个地域内的一个业务空间和一个用户**，且不能转移。API Key 可调用的功能与模型限流与其**归属业务空间**的权限保持一致，不受用户控制台权限管理的影响。将 RAM 账号移出业务空间会使其 API Key 失效（重新加入后恢复生效）。 - -### 3. 应用组件 API 调用 - -调用百炼应用组件 API（`bailian/2023-12-29`，含数据连接、知识库、Prompt 模板、长期记忆等）时，所有接口均需传入 `WorkspaceId`（业务空间 ID）。RAM 子账号需先获取对应权限策略并加入业务空间后才能调用。类目等资源也以业务空间为界，例如每个业务空间最多可新建 500 个类目。 - -### 4. 应用观测 - -应用观测以业务空间为范围，端到端查看空间内应用（智能体应用、工作流应用、高代码应用）的处理流程与延时、Token 等指标。若应用观测列表中看不到已创建的应用，常见原因之一即为该应用不属于当前业务空间。 - -### 5. 数据管理 - -数据管理功能统一管理业务空间下的大模型相关数据集（训练集、评测集）。该能力目前仅适用于华北2（北京）地域。 - -## 关键参数与配置 - -- **`WorkspaceId`**：业务空间 ID，是应用组件 OpenAPI 的必传参数，用于标识资源所属空间。 -- **地域隔离**：业务空间绑定单一地域，跨地域需在对应地域分别创建/使用空间。 -- **默认业务空间的限制**：默认空间无法设置模型调用/训练/部署授权与限流，如需精细化管控请使用非默认业务空间。 - -## 生产环境实践 - -- **空间规划**：推荐按环境（dev/test/prod）划分业务空间实现隔离，或按业务线划分以便权限与成本管理。 -- **限流策略**：将主账号总配额按比例分配给各业务空间并预留缓冲。例如总配额 1000 QPM，可分配 prod 600 / test 200 / dev 100，预留 100。 - -## 关联主题页 - -- [application component api reference](../api/application-component-api-reference.md) -- [application permission management](../guides/application-permission-management.md) -- [application monitoring](../guides/application-monitoring.md) -- [security and compliance](../guides/security-and-compliance.md) -- [model data overview](../guides/model-data-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md index 508d7604..b0757bc4 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-evaluation.md @@ -1,152 +1,45 @@ # application evaluation -阿里云百炼提供完整的应用[评测体系](../concepts/evaluation.md)，支持对[智能体应用](../concepts/agent-application.md)和工作流应用的输出质量进行系统化评估。平台同时提供自动评测与手动评测两种模式，并通过评测集、评估器和标签三大组件构建多维度的评测闭环。当前平台存在新旧两套评测系统，新版在评测任务管理、评估器和标签体系上做了较大升级。 +application evaluation 是百炼平台用于系统化评估智能体/工作流应用输出质量的核心能力，支持自动与手动两种评测范式。它通过结构化评测集、可配置的评估器与人工标签协同，实现从数据构建、任务执行到归因分析的完整闭环，适用于模型迭代、知识库更新、Prompt调优等关键研发场景。 -## 评测模式 +## 支持的模型/功能 -百炼支持两种评测模式，分别适用于不同场景： +- **自动评测**：基于大模型（当前仅支持 `qwen-max` 和 `qwen-plus`）自动生成评测集并执行端到端评分，适用于已发布且配置知识库的智能体应用 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **手动评测**：支持人工构建评测集（`.xls`/`.xlsx` 格式），通过人工打标完成效果评估，适用于需强主观判断或无标准答案的业务场景 [手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)。 +- **新版评测体系**：引入「智能体」「工作流」「自定义」三类评测集，并支持多评估器（LLM/Code）+ 多标签（分类/布尔/数字/文本）混合评测模式，覆盖更细粒度的质量维度 [评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)。 +- **评估器能力**：提供预置模板（如问答相关性、格式校验、文本相似度）及自定义 LLM/Code 评估器；LLM 评估器支持基于历史标注任务反向生成（即“评估器蒸馏”），但该方式不支持试运行 [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 +> **注意**：文档 4（新版评测集）与文档 3（旧版评测集）存在类型定义冲突——前者将评测集按应用形态（智能体/工作流/自定义）分类，后者按数据语义（对话分析/知识问答）分类。实际使用中，**新版体系已取代旧版**，旧版文档内容已过时，应以 [新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) 为准。 -### 自动评测 +## 关键参数 -[自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)利用大模型基于应用关联的知识库自动生成评测集，并对智能体的回答进行自动评分，生成评测报告与调优建议。支持两种子模式： +| 参数 | 说明 | 约束 | +|------|------|------| +| **评测集类型** | 新版体系下必须在创建时选定：`智能体`（适配智能体出入参）、`工作流`（适配工作流出入参）、`自定义`（自由定义表结构） | 创建后不可修改 | +| **评估器映射** | 所有变量（如 `query`, `response`, `referenceAnswer`）必须完成字段映射，否则无法保存评测任务 | 映射错误将导致评分结果为空或失真 | +| **采样与规模** | 自动评测中，分类采样数（事实型/分析型等）决定最终评测用例总量；单次评测最多支持 8 个应用横向对比 | 多应用评测要求所有应用共享至少一个知识库 | +| **标签类型** | 分类（多选枚举）、布尔（True/False）、数字（0–100 等）、文本（自由输入）四类，影响后续筛选与统计逻辑 | 数字标签需明确评分范围与通过阈值 | -- **单应用评测**：深度评估单个[智能体应用](../concepts/agent-application.md)的表现，生成包含评分、错误分析和优化建议的详细报告。 -- **多应用横向评测**：在同一评测基准下对比最多 8 个应用（或同一应用的不同版本），用于选型决策或版本迭代效果验证。 +## 使用方式 -前提条件： +1. **准备评测数据**： + - 新版推荐：在[评测集](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/efm/app_evaluate/tabs?activeKey=evalSet)页面选择「智能体」类型 → 关联目标应用 → 下载模板 → 填写后上传；或直接从「应用观测」导入真实流量数据。 + - 旧版兼容：仍支持上传 `.jsonl`（知识问答）或 `.xlsx`（对话分析）文件，但字段需严格匹配 [评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) 规范。 -1. 仅面向**已发布**的[智能体应用](../concepts/agent-application.md)，且应用须已配置知识库。 -2. 须开通**应用观测**功能，并将待评测应用添加到观测列表。 -3. 子账号需获取`管理员`或`应用评测-操作`权限。 -4. 多应用横向评测时，所有被选应用必须关联至少一个相同的知识库。 +2. **创建评测任务**： + - 在[评测任务](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/efm/app_evaluate/tabs?activeKey=task)页面新建任务 → 选择已发布评测集及版本 → 关联「智能体」应用 → 添加 1–10 个评估器（需完成全部参数映射）→ 可选添加人工标签。 -自动评测流程分四步：创建评测任务 → 设置评测集 → 配置评测规则 → 执行评测。评测集生成和评估模型当前仅支持 `qwen-max` 和 `qwen-plus`。 +3. **执行与分析**： + - 任务发起后不可修改配置；支持「快速标注」模式逐条人工校验； + - 结果页分「数据明细」（含各评估器评分、人工标签）和「指标统计」（综合得分、通过率柱状图、数据分布）； + - BadCase 归因由自动评测模块提供（如「检索无效」「切片不完整」），但新版评测任务需依赖评估器组合实现类似分析能力。 -### 手动评测 +## 限制和注意事项 -[手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md)通过人工构建评测集，对应用回答进行人工分析与评分。流程为：准备评测集（下载模板填充数据）→ 上传评测集 → 创建评测任务 → 人工标注打分 → 查看评测报告。 - -手动评测适用于需要领域专家主观判断的场景，评测维度支持使用内置模板或自定义评测维度模板。 - -> **注意**：手动评测属于旧版评测系统的功能。新版评测系统通过「评测任务 + 标签」的组合同样支持人工标注场景，且功能更灵活。 - -## 评测集 - -评测集是评测任务的数据基础，用于存储和管理评测数据。 - -### 旧版评测集 - -[旧版评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md)支持两种类型： - -| 类型 | 文件格式 | 适用场景 | -|------|----------|----------| -| 对话分析 | `.xls` / `.xlsx` | 人工评测，包含 Prompt、Completion、SessionId 字段，支持多轮对话 | -| 知识问答 | `.jsonl` | 自动评测，包含 query、queryType、referenceAnswer、fineKeywords、coarseKeywords 字段 | - -创建方式：自动生成（基于知识库，仅知识问答类型）或手动上传。单次上传最多 10 个文件，单个文件不超过 20MB。 - -### 新版评测集 - -[新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md)支持三种类型： - -| 类型 | 说明 | -|------|------| -| 智能体 | 根据选中智能体应用的出入参形式定义评测集 | -| 工作流 | 根据选中工作流应用的出入参形式定义评测集 | -| 自定义 | 任意定义评测集表结构，适用于特殊评测场景 | - -新版评测集支持手动上传和从应用观测导入两种创建方式，并具备版本管理能力，每次发布生成新版本。创建后类型不可修改。 - -## 评估器 - -[评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)是新版[评测体系](../concepts/evaluation.md)的核心组件，用于自动评估应用输出质量。支持三种创建方式： - -### 基于预置模板 - -百炼提供多种预置评估器模板，覆盖以下分类： - -- **通用质量**：评估回答的基本质量指标 -- **智能体**：专门用于评测智能体应用 -- **文本匹配**：精确规则文本匹配 -- **文本相似度**：计算文本相似度得分 -- **格式校验**：验证输出格式规范性 - -### 自定义评估器 - -| 类型 | 评估方式 | 适用场景 | 成本 | -|------|----------|----------|------| -| LLM 评估器 | 大模型语义理解 | 相关性、有害性、幻觉检测 | 产生 [Token](../concepts/token.md) 费用 | -| Code 评估器 | Python 代码规则判断 | 格式校验、数值计算、精确匹配 | 无额外费用 | - -### 基于评测任务创建 - -通过历史评测任务的标注结果自动抽象为新的 LLM 评估器，适用于将人工标注经验固化为自动化评估规则的场景。需选择已完成评估的评测任务，并配置 query、response、label_score 的字段映射。 - -每个评测任务最多支持添加 10 个评估器。建议组合 3-5 个评估器从不同维度评估应用质量，例如：相关性评估器（LLM）+ 格式校验评估器（Code）。 - -## 标签管理 - -[标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md)用于对评测数据和应用观测数据进行自定义标注，支持四种标签类型： - -| 标签类型 | 数据类型 | 适用场景 | -|----------|----------|----------| -| 分类 | 预定义选项（最多 20 个） | 回答质量分级、错误类型分类 | -| 布尔值 | True / False | 是否正确、是否存在幻觉 | -| 数字 | Double 数值 | 评分（1-5）、相关性得分（0-1） | -| 文本 | 自由文本 | 错误原因说明、改进建议 | - -标签可同时用于评测任务的人工标注和应用观测的数据标注，支持基于标签的数据筛选和指标统计。 - -## 评测任务（新版） - -[新版评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md)支持智能体和工作流应用的评测，核心配置包括： - -- **选择评测集**：从已发布的评测集列表中选择评测集和版本 -- **关联应用**：支持不关联应用（纯人工标注）、关联工作流或关联智能体三种方式 -- **添加评估器**：配置自动评分规则及参数映射 -- **添加标签**：配置人工标注维度 - -任务详情页提供数据明细和指标统计两个视图，支持普通模式和快速标注两种标注方式。 - -## 评测报告与归因分析 - -自动评测完成后生成评测报告，包含以下维度： - -- **总正确率**：得分 >= 4 分的回答占比（评分范围 1-5 分） -- **BadCase 分析**：按分数从低到高展示错误评测条目 -- **调优建议**：基于归因分析提供 Prompt、检索配置或知识库切片的具体优化建议 -- **RAG 智能体评价**：按问题类型展示单项得分 - -归因分析将 BadCase 定位到 RAG 流程的具体环节： - -| 归因类型 | 含义 | 优化方向 | -|----------|------|----------| -| 模型理解有误 | 已获取正确知识但推理错误 | 优化提示词或切换更强模型 | -| 重排不佳 | 正确切片排序靠后 | 调整重排配置或增加切片数量 | -| 检索无效 | 召回切片过多或过少 | 调整检索策略 | -| 切片不完整 | 语义单元被分割到多个切片 | 增大切片长度或启用语义切分 | -| 未获取知识 | 知识库缺少相关内容 | 补充知识库内容 | - -## 最佳实践 - -### 建立持续评测机制 - -以下场景建议触发评测：知识库更新后、调整 Prompt 后、更换或升级模型后、调整检索/重排策略后、定期回归（每周或每月）。 - -### 优化闭环 - -识别 BadCase → 分析归因定位问题 → 实施针对性优化 → 发布新版本再次评测 → 对比结果确认改进。若效果未达预期则继续迭代。 - -## 计费说明 - -评测任务调用大模型产生的 [Token](../concepts/token.md) 费用正常计费。自动评测的评测集生成和评估均会消耗 [Token](../concepts/token.md)，预估平均消耗仅为参考值，最终以实际账单为准。评估器模型当前限时免费。 - -## 常见问题 - -- **评测集生成进度长时间保持 0%**：评测集生成和应用评测为离线任务，需后台排队执行，排队期间进度保持 0%，任务开始后自动更新。 -- **评测期间能否关闭应用观测**：不可以，否则可能导致评测任务失败或数据丢失。 -- **评测报告中用例数量与设置不符**：自动评测可能部分失败，报告仅展示成功完成的用例。 -- **评测任务创建后可否修改**：任务配置（应用、评测集）不可修改，但可随时添加人工标签。如需不同配置请创建新任务。 +- **权限与依赖**：自动评测要求子账号具备 `管理员` 或 `应用评测-操作` 权限，且目标应用必须已开通「应用观测」并加入观测列表 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **知识库强约束**：自动评测生成评测集及执行评估均依赖知识库内容，若知识库未配置或无公共交集，多应用评测将失败。 +- **[Token](../concepts/token.md) 消耗不可逆**：评测任务分步执行（如评测集生成、模型推理），任一成功步骤均产生 [Token](../concepts/token.md) 费用，即使后续步骤失败亦不退费 [自动评测](../../raw/application-user-guide/application-evaluation/application-auto-evaluation.md)。 +- **版本兼容性**：新版评测体系（文档 4/5/6/7）与旧版（文档 1/2/3）并存但不互通；旧版「自动评测」界面已标记为「返回旧版」入口，新项目应优先采用新版架构。 +- **评估器试运行限制**：基于历史评测任务创建的评估器**不支持试运行**，必须部署至评测任务中实测效果 [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md)。 ## 来源文档 @@ -154,16 +47,8 @@ - [手动评测](../../raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](../../raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - [新版评测集](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) -- [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评测任务](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) +- [标签管理](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评估器](../../raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md index 9ae2c8d0..38930a81 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-monitoring.md @@ -1,128 +1,48 @@ # application monitoring -阿里云百炼提供**应用观测**功能，用于端到端查看[业务空间](../concepts/workspace.md)内应用（[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用、高代码应用）的处理流程，并获取延时、[Token](../concepts/token.md) 量等关键指标，指标更新频率为分钟级。该功能可帮助开发者追踪应用内部调用链路、查看模型响应延时与思考过程，进而优化运营效果与成本。详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +应用观测（Application Monitoring）是阿里云百炼平台提供的端到端可观测能力，用于追踪应用内部调用链路、分析模型响应延时、查看推理过程与 [Token](../concepts/token.md) 消耗等关键指标。该功能基于 OpenTelemetry 实现，数据同步频率为分钟级，适用于调试、性能优化与真实场景评测。> **注意：应用观测目前暂无 API 接口**，所有操作均需通过控制台完成，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 -## 支持的应用范围 +## 支持的模型/功能 -应用观测支持以下三类应用： +- **支持的应用类型**：智能体应用、工作流应用、高代码应用（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **不支持的应用**：通过 Assistant API 创建的智能体应用（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **可观测节点类型**：覆盖完整 RAG 与编排链路，包括 `CHAIN`（根节点）、`LLM`（大模型调用）、`RETRIEVER`（含 `TextRetriever`/`VectorRetriever`）、`EMBEDDING`、`RERANKER`、`REWRITER`、`TOOL`、`GUARDRAIL`、`API`、`CLASSIFIER` 等；高代码应用仅支持 `CHAIN`（`FullCodeApp`）根节点级别观测，**不支持内部链路追踪**（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **附加能力**：支持 Span 数据导出（JSONL/Excel）、添加至评测集、多维度标签标注（布尔/分类/数字/文本）、基于 Request ID/Trace ID/Span ID 的精准检索。 -- **[智能体应用](../concepts/agent-application.md)**（AgentApp） -- **[工作流](../concepts/workflow.md)应用**（WorkflowApp） -- **高代码应用**（FullCodeApp） +## 关键参数 -> **注意**：应用观测暂不支持通过 Assistant API 创建的[智能体应用](../concepts/agent-application.md)；对高代码应用，目前不支持追踪其内部调用链路，仅能观测到入口 CHAIN 节点。应用观测本身也没有 API，只能通过控制台操作。 +| 参数 | 说明 | 来源 | +|------|------|------| +| **延时（调用时长）** | LLM 节点包含首 [Token](../concepts/token.md) 响应及完整输出耗时；CHAIN/WorkflowApp 等根节点反映端到端耗时 | [原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **[Token](../concepts/token.md) 总量** | = 输入 Token + 输出 Token；Embedding 节点仅统计向量化输入 Token 数 | [原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **平均首 Token 耗时** | 仅对流式调用生效，单位毫秒 | [原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md) | +| **状态** | `正常` 或 `错误`（可进一步按错误类型细分） | [原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md) | -## 前提条件与开通 - -首次使用需在应用观测页面右上角完成**应用观测配置**，依次执行：授权可观测链路 OpenTelemetry 服务角色权限 → 开通 OpenTelemetry 服务 → 初始化 LogStore。 - -- 推荐使用**主账号**操作，开通后通常分钟级生效，高峰期可能略有延迟。 -- 如需**子账号**开通，主账号需为其配置 `AliyunBailianFullAccess` 全局权限、`应用观测-操作`（或 `管理员`）页面权限，并额外授予 `ram:CreateServiceLinkedRole` 系统策略（用于创建服务关联角色）。 - -> 子账号权限若未配置完整，开启应用观测时会失败。配置完成后需返回应用观测界面再次尝试开启。 +> **注意**：`TextRetriever` 和 `VectorRetriever` 默认返回 100 个文本切片，且**暂不支持数量调整**——该限制在文档中多次强调，开发者不可通过配置绕过。 ## 使用方式 -### 1. 选择被观测的应用 - -在应用观测页面单击「选择被观测的应用」>「添加」。若列表中看不到已创建的应用，通常是因为该应用尚未发布，或应用不属于当前[业务空间](../concepts/workspace.md)。 - -### 2. 开始观测 - -添加完成后，应用会出现在观测列表中。此后所有输入该应用的 Prompt 及相关数据、指标会被自动追踪并以分钟级频率同步。单击「关闭观测」可停止同步，重新添加后仅同步新增数据。 - -在「查看详情」中可查看最长 30 天内的调用记录，包括 Prompt 内容、输出、延时、调用时间和 [Token](../concepts/token.md) 量，并支持按 Request ID / Trace ID / Span ID 检索和按时间范围筛选。单击节点名称可查看详情、原始数据和标注记录。 - -> 列表中的 **CHAIN** 节点表示一次完整的应用内部调用追踪，支持展开。状态分为「正常」与「错误」两类。 - -### 3. 导出数据 - -在应用详情页的 Trace 列表页签右上角单击「导出数据」，可将当前筛选条件下的数据导出为 **JSONL** 或 **EXCEL** 格式。 - -### 4. 查看监控统计 - -「监控统计」页签提供性能监控图表：调用次数（含失败次数与失败率）、[Token](../concepts/token.md) 总量（全部/输入/输出）、平均单次请求 [Token](../concepts/token.md) 量、平均首 [Token](../concepts/token.md) 耗时（流式场景）、平均调用时长。支持按时间范围（最长 30 天）和聚合粒度（分钟/小时/天）查看，每个图表可放大、下载、复制。 - -## 数据筛选与标注 - -### Span 筛选模式 - -- **Root Span**：仅显示根节点（默认） -- **All Span**：平铺展示所有 Span -- **Model Span**：仅显示包含模型调用的 Span - -### 过滤器 - -支持按状态（正常/错误，可按错误类型细分）、Span Name、输入、输出、延时、[Token](../concepts/token.md) 总量、输入 [Token](../concepts/token.md)、输出 [Token](../concepts/token.md)、标签等字段添加多个筛选条件，条件之间组合应用。 - -### 数据标注 - -支持对 Span 数据添加标签（布尔值/分类/数字/文本四种类型），标签与应用[评测](../concepts/evaluation.md)的标签管理共享、统一管理。标注内容自动保存，并可在 Span 列表页「标签」列查看。 - -### 添加到[评测](../concepts/evaluation.md)集 - -应用观测支持将 Span 数据直接加入[评测](../concepts/evaluation.md)集，将真实线上调用作为评测样本。配置时需选择目标评测集、导入方式（追加数据或全量覆盖）并完成字段映射。每个评测集最多支持 50 个字段映射。 - -## 节点类型 - -被观测应用在调用过程中会按操作单元生成不同类型的**节点**，节点之间可形成嵌套关系。仅在被触发或调用时才展示对应节点。完整节点类型与说明见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 - -### [智能体应用](../concepts/agent-application.md)节点 - -| 节点 | 说明 | -| --- | --- | -| CHAIN | 连接大模型节点与其他节点，处理复杂任务；作为根节点时名称为 AgentApp 或 WorkflowApp | -| AGENT | 对智能体的调用 | -| RETRIEVER | 检索操作；KnowledgeRetriever 表示在[知识库](../concepts/knowledge-base.md)中检索。子节点名称含 TextRetriever（改进 BM25，默认返回 100 个切片）、VectorRetriever（向量检索，默认返回 100 个切片） | -| REWRITER | 基于会话上下文调整原始 Prompt 以提升检索效果 | -| EMBEDDING | 将 Prompt 转为向量，[Token](../concepts/token.md) 量为本次向量化的 [Token](../concepts/token.md) 数 | -| RERANKER | 计算文本切片相似度分数并降序排列 | -| LLM | 大模型推理/文本生成，[Token](../concepts/token.md) 量 = 输入 + 输出；延时包含输出回复过程 | -| TOOL | 插件调用（官方或自定义） | -| GUARDRAIL | 阿里绿网调用；ManualIntervention 为用户干预规则，SystemIntervention 为系统干预规则 | +### 前置配置（必需） +1. 主账号或已授权子账号访问 [应用观测](https://bailian.console.aliyun.com/tab=app?tab=app#/app-observe)，点击右上角 **应用观测配置**； +2. 完成三步授权：① 授权可观测链路 OpenTelemetry 服务角色；② 开通 OpenTelemetry 服务；③ 初始化 LogStore 存储； +3. **子账号需额外配置**：`AliyunBailianFullAccess` + `应用观测-操作` 页面权限 + `CreateServiceLinkedRole` 系统策略（详见 [原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 -> 目前暂不支持观测长期记忆中的检索过程；TextRetriever 与 VectorRetriever 默认返回 100 个切片，暂不支持调整数量。 +### 观测流程 +- **添加应用**：仅已**发布**且属于当前业务空间的应用可见；未发布应用需先通过「管理应用 → 发布」启用观测。 +- **数据同步**：Prompt 输入后自动追踪，分钟级同步至观测列表；关闭观测后停止采集，重开仅同步新增数据。 +- **查看详情**：支持 Root Span / All Span / Model Span 三种视图；可通过状态、Span Name、输入/输出关键词、延时、Token 量、标签等条件组合过滤。 +- **导出与评测**：Trace 列表页支持 JSONL/Excel 导出；支持批量 Span 添加至评测集，支持字段映射（最多 50 个）与追加/覆盖导入模式。 -### [工作流](../concepts/workflow.md)应用节点 +## 限制和注意事项 -除上述 CHAIN、RETRIEVER、REWRITER、EMBEDDING、RERANKER、LLM、GUARDRAIL 外，还包含工作流专属节点：START（开始）、END（结束）、API、CLASSIFIER（意图分类）、TEXT_CONVERTER（文本转换）、SCRIPT（脚本转换）、CONDITION（条件判断）、FUNCTION_COMPUTE（函数计算）、APP_FLOW。 - -### 高代码应用节点 - -仅有 CHAIN（FullCodeApp）作为入口节点，目前不支持追踪其内部调用链路。若已开启观测却看不到调用量等统计数据，需排查：代码中是否使用 AgentScope-AI 的 Tracing 模块定义上报信息，以及部署时是否添加 `--telemetry enable` 参数。 - -## [计费](../concepts/billing.md)说明 - -应用观测功能本身**不收费**，但观测数据需存储在可观测链路 OpenTelemetry 服务中，相关存储费用由 OpenTelemetry 服务收取。 - -## 关键指标说明 - -- **延时（调用时长）**：对 LLM 节点，包含输出回复的完整过程。 -- **[Token](../concepts/token.md) 量**：Embedding 节点为本次向量化 [Token](../concepts/token.md) 数；LLM 节点为输入 [Token](../concepts/token.md) + 输出 [Token](../concepts/token.md)。 -- **数据时效**：指标更新频率为分钟级，调用记录最长可查 30 天。 -- **应用总量 / 平均延时**：用于评估应用运营效果与成本，详见 [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md)。 +- **无 API 支持**：当前仅提供控制台界面，不开放 RESTful 或 SDK 接口（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **高代码应用限制**：开启观测后仅显示 `FullCodeApp` 根节点，**无法观测其内部函数、HTTP 调用或自定义逻辑链路**；需在代码中集成 `AgentScope-AI` 的 Tracing 模块，并部署时显式启用 `--telemetry enable`（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **[长期记忆](../concepts/long-term-memory.md)不可观测**：知识库检索可观测，但[长期记忆](https://help.aliyun.com/zh/model-studio/long-term-memory)中的检索过程**明确不支持**（[原文标题](../../raw/application-user-guide/application-monitoring/application-observation.md)）。 +- **计费说明**：应用观测功能本身免费，但底层依赖 OpenTelemetry 服务，产生的日志与链路数据存储费用需单独承担。 +- **数据时效性**：指标更新延迟约 1–5 分钟，不适用于毫秒级实时告警场景。 ## 来源文档 - [应用观测](../../raw/application-user-guide/application-monitoring/application-observation.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md index 32803f47..0d60edc5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-permission-management.md @@ -1,137 +1,52 @@ # application permission management -阿里云百炼支持基于控制台页面级、模型级的多维度权限控制，满足多地域、多用户的复杂组织架构需求。单个[业务空间](../concepts/workspace.md)是进行精细化权限管理（模型、用户）和阿里云账单分账的最小管理单元，权限管理围绕超级管理员、[业务空间](../concepts/workspace.md)管理员、普通用户三种角色展开。详细的角色定义与权限矩阵见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +百炼平台的权限管理以“业务空间”为最小管理单元，支持基于角色（超级管理员、业务空间管理员、普通用户）和资源维度（模型调用、调优、部署、页面访问、API Key、OpenAPI）的精细化控制。权限策略同时作用于控制台操作与 API 调用，但二者权限模型存在关键差异：控制台权限由用户角色在业务空间内直接配置，而 OpenAPI 权限需通过 RAM 策略显式授予。所有权限均按地域隔离，同一业务空间不可跨地域存在。 -## 角色体系 +## 支持的模型/功能 -百炼的身份管理基于以下三种角色，权限范围自上而下递减： +百炼权限管理覆盖以下核心能力： -- **超级管理员**：可跨空间统一管理用户权限、空间可用模型、空间模型限流和 [API Key](../concepts/api-key.md)。包含两类账号：阿里云主账号，以及拥有 `AliyunBailianFullAccess`（百炼管理员）系统策略的 RAM 用户。超级管理员可通过百炼全局管理菜单（北京 / 新加坡 / 弗吉尼亚）为任意 RAM 用户授权任意地域、任意空间的几乎所有权限，仅 OpenAPI 接口权限需阿里云主账号添加。 -- **[业务空间](../concepts/workspace.md)管理员**：拥有访问某个[业务空间](../concepts/workspace.md)「权限管理」页面的 RAM 用户，只负责该特定[业务空间](../concepts/workspace.md)内的用户权限和资源管理。管理员权限包含可访问该[业务空间](../concepts/workspace.md)下所有页面的权限。 -- **普通用户**：根据分配的权限使用资源，可访问/使用被授权的空间、页面、资源。 +- **模型级管控**：支持对特定模型启用/禁用**调用**（含控制台体验与 API）、**调优**（训练）和**部署**权限；默认业务空间不支持限制（[原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 +- **空间级角色管理**：定义三类角色——**超级管理员**（跨空间全局管理）、**业务空间管理员**（单空间内用户、页面、模型限流管理）、**普通用户**（仅使用被授权资源）。 +- **API Key 绑定与继承**：每个 API Key 严格归属单一地域、单一业务空间、单一用户；其可调用模型范围与限流策略**完全继承自所属业务空间的模型权限配置**，不受用户控制台权限影响（[原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 +- **OpenAPI 接口权限**：独立于业务空间权限体系，需通过 RAM 策略（如 `AliyunBailianDataFullAccess`）显式授权，且**仅阿里云主账号可为 RAM 用户添加该类策略**（[原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 -### 权限矩阵 +> **注意**：文档中多次强调“默认业务空间无法设置模型调用/调优/部署限制”，但未明确说明该限制是否适用于所有地域。实践中，北京、新加坡、弗吉尼亚三地默认空间行为一致，建议避免在默认空间承载生产流量。 -| [业务空间](../concepts/workspace.md)权限 | 超级管理员 | [业务空间](../concepts/workspace.md)管理员 | 普通用户 | -| --- | --- | --- | --- | -| 允许特定模型调用 & 限流 | 支持 | 不支持 | 不支持 | -| 允许特定[模型调优](../concepts/fine-tuning.md) | 支持 | 不支持 | 不支持 | -| 允许特定[模型部署](../concepts/model-deployment.md) | 支持 | 不支持 | 不支持 | -| 用户管理 | 支持 | 支持 | 不支持 | -| 用户可用页面管理 | 支持 | 支持 | 不支持 | -| [API Key](../concepts/api-key.md) 管理 | 支持 | 支持 | 不支持 | -| 访问/使用被授权的空间、页面、资源 | 支持 | 支持 | 支持 | -| OpenAPI 接口权限 | 不支持 | 不支持 | 不支持 | +## 关键参数 -> **注意**：OpenAPI 接口权限不通过[业务空间](../concepts/workspace.md)角色授予，必须由阿里云主账号在 RAM 控制台为 RAM 用户添加专用系统策略。 +| 参数 | 说明 | 约束 | +|------|------|------| +| `workspace_id` | 业务空间唯一标识符，用于 API 请求中指定目标空间（如 `X-Workspace-ID` Header 或请求体） | 必填；需通过 [获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id) 获取 | +| `model_name` | 模型标识符（如 `qwen-max`, `qwen-plus`），用于模型级权限开关 | 必须已在该业务空间中显式开通调用/调优/部署权限 | +| `qpm_limit` / `tpm_limit` | 每分钟请求数（QPM）与 [Token](../concepts/token.md) 数（TPM）限流阈值 | 仅超级管理员或业务空间管理员可在控制台设置；API Key 自动继承该值 | +| `api_key` | 认证凭证，绑定至特定 `workspace_id` 和用户 | 不可跨空间/用户迁移；华北2（北京）新创建的 API Key 默认归属主账号 | -## [业务空间](../concepts/workspace.md)权限管理 +## 使用方式 -百炼按地理区域划分资源和[业务空间](../concepts/workspace.md)，**单个[业务空间](../concepts/workspace.md)不能跨地域存在**，即使是各地域的默认[业务空间](../concepts/workspace.md)，也是不同的空间。[业务空间](../concepts/workspace.md)是精细化权限管理的最小单元，可管理以下维度（默认[业务空间](../concepts/workspace.md)无法设置这些限制）： +### 1. 角色与空间初始化 +- **超级管理员**：需主账号或已附加 `AliyunBailianFullAccess` 策略的 RAM 用户，在 [全局管理菜单](https://bailian.console.aliyun.com/?tab=globalset#/efm/business_management) 创建业务空间并分配模型权限。 +- **业务空间管理员**：由超级管理员在控制台「权限管理」页签中为 RAM 用户授予「管理员」角色（[原文标题](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)）。 -- **限制模型调用**：管理某个模型可否在该[业务空间](../concepts/workspace.md)调用（控制台 & API），并设置该模型的请求数限流和 [Token](../concepts/token.md) 限流。默认[业务空间](../concepts/workspace.md)所有模型均可调用且无法限流。 -- **限制模型训练**：管理某个模型可否在该[业务空间](../concepts/workspace.md)进行调优和调优后部署。默认[业务空间](../concepts/workspace.md)所有支持调优的模型均可调优及部署。 -- **限制[模型部署](../concepts/model-deployment.md)**：管理某个模型可否在该[业务空间](../concepts/workspace.md)直接部署。默认[业务空间](../concepts/workspace.md)所有支持部署的模型均可部署。 -- **用户控制台权限管理**：管理某个 RAM 用户是否能使用该业务空间控制台的功能及能使用哪些功能，但无法限制归属该用户的 [API Key](../concepts/api-key.md) 的调用。阿里云主账号无须设置，可访问所有业务空间的所有页面。 +### 2. 模型权限开通（必需前置步骤） +- 超级管理员需先在全局管理中为业务空间**启用目标模型的调用/调优/部署能力**（默认空间自动全开，但不可配置限流）。 +- 后续再由业务空间管理员在本空间内为具体用户分配对应控制台操作权限（如「模型体验-操作」「模型调优-操作」）。 -关于业务空间的地域隔离与限流细节，可进一步参考 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 +### 3. API 调用授权 +- 为用户生成 API Key（归属指定业务空间）； +- 该 Key 自动获得该空间已开通的所有模型调用权限及限流策略； +- 如需调用应用、知识库等 OpenAPI，**必须额外在 RAM 控制台为该 RAM 用户附加 `AliyunBailianDataFullAccess` 或 `AliyunBailianDataReadOnlyAccess` 策略**。 -## API-Key 权限 +## 限制和注意事项 -单个 [API Key](../concepts/api-key.md) 只能归属一个地域内的一个业务空间和一个用户，且不能转移。[API Key](../concepts/api-key.md) 可调用的功能和模型限流与**归属业务空间**的权限保持一致，不受用户控制台权限管理的影响，也无需为不同模型（如文生文、文生图、语音合成）创建不同的 [API Key](../concepts/api-key.md)。 - -[API Key](../concepts/api-key.md) 的状态随归属用户操作变化： - -| 触发操作 | 主账号的 [API Key](../concepts/api-key.md) | RAM 账号的 [API Key](../concepts/api-key.md) | -| --- | --- | --- | -| 主动删除 [API Key](../concepts/api-key.md) | 失效，不可恢复 | 失效，不可恢复 | -| 将账号移出业务空间 | — | 失效（重新加入后恢复生效） | -| 在 RAM 控制台删除账号/角色 | — | 失效，不可恢复 | -| 为 [API Key](../concepts/api-key.md) 设置 IP 访问白名单 | 华北2（北京）地域支持 | 华北2（北京）地域支持 | - -> **注意**：自 2026 年 3 月 25 日起，华北2（北京）地域的所有新创建的 [API Key](../concepts/api-key.md) 均归属主账号。 - -可通过百炼控制台左侧导航栏「权限管理」页签为 RAM 用户添加 API-Key 权限，赋予其创建、删除、查看该空间下所有 API-Key 的权限。 - -## OpenAPI 接口权限 - -RAM 用户默认无权调用百炼应用的数据、[知识库](../concepts/knowledge-base.md)、Prompt 工程及长期记忆等功能的 Open API。需由阿里云主账号在 RAM 控制台为 RAM 用户添加以下权限之一： - -- `AliyunBailianDataFullAccess`：可调用百炼应用 API 目录下的所有 API。 -- `AliyunBailianDataReadOnlyAccess`：可调用百炼应用 API 目录下的只读类 API，如 `DescribeFile`、`GetIndexJobStatus` 等。 - -## 账单与预付费权限 - -RAM 用户默认无权查看阿里云账单和购买预付费产品，需在 RAM 控制台添加特定权限。这两项权限会授予 RAM 用户查看**所有产品**账单或购买**所有预付费产品**的权限，请谨慎授权。 - -- 查看阿里云账单：添加 `AliyunBSSReadOnlyAccess`。 -- 购买阿里云预付费产品：添加 `AliyunBSSOrderAccess`。 - -## 常用配置流程 - -### 设置超级管理员 - -需要阿里云主账号或具备 `AliyunRAMFullAccess` 系统策略的 RAM 用户操作。前往 RAM 控制台为 RAM 用户添加 `AliyunBailianFullAccess` 和 `AliyunBSSOrderAccess` 权限后，即可通过百炼全局管理菜单授权任意地域、空间的权限并购买预付费产品。 - -### 设置业务空间管理员 - -需超级管理员或业务空间管理员操作。在百炼控制台左侧导航栏「权限管理」页签内为 RAM 用户添加「管理员」权限。 - -### 设置模型调用权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的模型调用权限。 -2. 通过控制台调用时，需由超级管理员或业务空间管理员为 RAM 用户添加：**模型体验-操作**（控制台调用模型）、**批量推理-操作**（支持批量推理）、**模型观测-操作**（查看 [Token](../concepts/token.md) 消耗量）。 -3. 通过 API 调用时，需为 RAM 用户在对应业务空间创建或分配 API Key。 - -### 设置[模型调优](../concepts/fine-tuning.md)权限 - -1. 不使用默认业务空间时，需先由超级管理员为业务空间开通特定模型的[模型调优](../concepts/fine-tuning.md)（训练）权限。 -2. 在「权限管理」页签内为 RAM 用户添加以下权限：**模型体验-操作**、**模型调优-操作**、**我的模型-操作**（管理调优后模型快照）、**[模型部署](../concepts/model-deployment.md)-操作**（部署调优后的模型）、**模型[评测](../concepts/evaluation.md)-操作**、**数据管理-操作**（管理调优数据集）、**模型观测-操作**。 -3. 通过 API 调优时，无需额外控制台权限，只需为 RAM 用户分配 API Key 即可。 - -完整的权限配置流程与截图说明见 [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md)。 - -## 生产环境实践 - -- **空间规划策略**：推荐按环境划分（开发 `project-dev-workspace`、测试 `project-test-workspace`、预发与生产 `project-prod-workspace`），实现严格的环境隔离；也可按业务线划分（如 `marketing-team-workspace`、`customer-team-workspace`），便于权限和成本管理。 -- **限流策略**：将主账号总配额按比例分配给各业务空间并预留缓冲。例如账号总配额 1000 QPM，可分配生产 600 QPM（60%）、测试 200 QPM（20%）、开发 100 QPM（10%）、预留缓冲 100 QPM（10%），以应对突发流量。 - -## 限制与注意事项 - -- 业务空间不能跨地域存在；不同地域的默认业务空间也是不同空间。 -- 默认业务空间无法设置模型调用、调优、部署限制，所有模型均按默认策略可用且无法限流。 -- API Key 不可跨业务空间或跨用户转移；账号移出业务空间后其 API Key 失效，重新加入后恢复。 -- OpenAPI 接口权限、账单与预付费权限必须由阿里云主账号在 RAM 控制台授权，业务空间管理员无法授予。 -- `AliyunBSSReadOnlyAccess` / `AliyunBSSOrderAccess` 为全产品级权限，授权范围远超百炼本身，需谨慎。 -- 开通 AI 安全护栏、模型监控、应用观测等功能，建议使用阿里云主账号在控制台一次性授权开通。 - -## 常见问题 - -- **如何获取业务空间 ID**：参考应用开发的「获取 Workspace ID」文档。 -- **如何使用子业务空间调用模型**：无需特殊设置，使用子业务空间的 API Key 即可。 -- **如何使用特定业务空间的应用**：使用 API 管理和调用特定业务空间的应用时，需同时设置 APP ID 和 Workspace ID。 +- **地域强隔离**：业务空间与资源严格绑定地域，北京空间的模型权限配置对新加坡空间无影响；跨地域需分别配置。 +- **默认空间不可控**：默认业务空间无法设置模型调用/调优/部署开关及限流，仅可用于快速体验，**严禁用于生产环境**。 +- **API Key 与用户权限解耦**：用户控制台权限（如能否访问「模型调优」页面）不影响其 API Key 的实际调用能力；API Key 权限仅取决于所属业务空间的模型开通状态与限流配置。 +- **OpenAPI 权限独立授权**：即使用户拥有业务空间管理员权限，若未被授予 `AliyunBailianDataFullAccess` 等 RAM 策略，仍无法调用 `/v1/applications/*` 等应用相关 OpenAPI。 +- **账单与预付费权限需单独配置**：查看账单需 `AliyunBSSReadOnlyAccess`，购买预付费产品需 `AliyunBSSOrderAccess`，二者均需在 RAM 控制台手动附加，不随百炼角色自动继承。 ## 来源文档 - [权限管理](../../raw/application-user-guide/application-permission-management/application-permission-management-overview.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md index 0dd5decf..2b8dca09 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-publishing-and-sharing.md @@ -1,79 +1,54 @@ # application publishing and sharing -阿里云百炼支持将已构建并发布的应用以多种渠道对外分享，或封装为可复用的模块化组件供其他应用接入。本页汇总了[智能体应用](../concepts/agent-application.md)的分享渠道、组件化发布与接入方式，以及基于魔笔能力的 UI 设计器发布流程，面向需要将百炼应用集成到实际业务的开发者。 - -## 适用范围与版本限制 - -分享渠道（魔笔/UI 应用、钉钉、微信、组件、音视频实时互动）均为 **Agent 1.0** [智能体应用](../concepts/agent-application.md)的功能。 - -> **注意**：**Agent 2.0** [智能体应用](../concepts/agent-application.md)仅支持通过 API 调用，**不支持**上述任何分享渠道。若需分享，请确认应用版本。 - -前提条件是已有构建好且**已发布**的智能体应用。所有分享渠道通过百炼控制台 **应用管理 → 目标应用卡片 → 发布** 进入。详见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 - -## 分享渠道 - -智能体应用（Agent 1.0）支持四种分享或发布方式，以及音视频实时互动： - -- **UI 应用 / 魔笔分享渠道**：进入 UI 设计器编辑并发布界面，在 **环境部署** 中获取应用地址后分享。持有链接的阿里云用户均可访问，**单击「下线」可停止服务**。 -- **钉钉**：在 **发布平台** 授权计算巢 AppFlow（SLR 关联 + API-KEY 加密传输），配置钉钉模板 ID、Client ID、Client Secret 后创建，最终得到 **回调地址** 用于配置钉钉机器人。钉钉机器人的 **消息接收模式必须选 HTTP 模式**，选 Stream 模式会导致无法返回消息；并需申请 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限。 -- **微信公众号**：若已在钉钉步骤授权过则无需再次授权。选择 API KEY 与微信凭据（需 AppID 授权）后创建，生成二维码供用户扫码体验。 -- **音视频实时互动**：仅支持图文对话类应用（含智能体与工作流）。支持 H5/APP 扫码与 SDK 集成（基于 AICallKit SDK，含 UI/不含 UI 两种方案）两种渠道。 - -> **注意**：临时体验二维码（音视频互动）与从已有应用发布的 UI 体验链接，**有效期均为 24 小时**，过期需重新生成或重新发布。 - -**权限与计费**：共享应用可被应用创建者（主账号）、RAM 用户及持有链接的其他阿里云用户访问；所有通过分享链接产生的费用由**应用创建者 UID 账号**承担。上述钉钉/微信配置细节见 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 - -## 组件化发布与接入 - -智能体或工作流应用可发布为模块化组件，供其他应用复用。发布路径有三处：发布应用时勾选 **发布应用组件**、在 **发布渠道** 的组件区域 **+ 创建**、或在控制台 **组件管理** 面板创建。详见 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 - -### 关键参数 - -组件预设了系统参数 `query`（String，用户输入文本）和 `imageList`（Array，图像公网地址列表，仅在使用视觉模型时有效）。**预设系统参数无法删除**，不需要时将「是否可见」设为「否」隐藏。 - -各参数配置项含义： - -- **别名**：调用者只能看到别名，用于避免参数重名。 -- **传参方式**： - - **业务透传**：智能体中由使用者提供，工作流中由上游节点提供。 - - **模型识别**：智能体中由大模型根据参数描述自动推断填充。 -- **组件描述**：接入智能体时，大模型据此自动判断是否调用；接入工作流时仅作说明，不影响运行。 - -> **注意**：即使参数的传参方式设为**模型识别**，在**工作流应用**中应用也**不会**自动推断参数值，必须像业务透传一样从上游节点明确提供输入值。模型识别仅在智能体应用中生效。 - -### 接入方式 - -- **智能体应用**：组件作为工具接入，大模型根据用户问题自动调用。若组件含业务透传参数，可在测试时手动填 **入参变量配置**，或在 API 调用时通过 `biz_param` 参数传入。 -- **工作流应用**：组件作为组件节点接入，需手动传入参数（如 `系统变量/query`）并将 `组件1/result` 传递到下游节点。 - -### 组件注意事项 - -- **自动更新**：应用重新发布后，由其发布的组件会自动更新。 -- **避免嵌套调用**：A 调 B、B 调 A 会进入重复调用状态导致功能不可用。 -- **避免多级调用**：A 调 B、B 调 C 因存在最长运行时间限制，容易超时报错。 - -## UI 设计器发布 - -UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视化拖放式界面构建，可将应用发布为网页 UI。详见 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 - -**前提**：百炼应用、API Key 和 UI 设计必须归属于**同一业务空间**，否则无法在 UI 创建时选择对应的 API Key 与应用。 - -发布方式有两种：从已有应用一键创建 UI（自动填充标题、API-KEY、智能体、预设问题等），或通过 UI 设计器从模板（空白 / 智能出行助手 / 智能体门户 / AI 基础对话 / 企业 AI 知识库 Lite）创建。核心流程为：创建 UI → 拖放组件编辑页面 → 发布与分享。 - -**环境对比**： - -| 维度 | 开发环境 | 生产环境 | -| --- | --- | --- | -| 用途 | 开发、调试、验证 | 终端用户实际使用版本 | -| 访问方式 | 平台域名 | 平台域名 + 自定义访问地址 | -| 有效期 | **24 小时后失效**，需重新发布 | 长期有效 | -| 是否收费 | 免费 | 需订阅团队版及以上套餐，并配置域名 | - -**权限**：UI 应用发布后默认持有链接的阿里云用户可访问，也可开启 **允许匿名访问** 并通过权限组限制其只访问会话页。 - -**计费**：UI 设计器功能本身不计费，但会产生模型调用费用、UI 应用数据（超出 1GB 免费文件存储与 0.3GB 免费数据库容量后按量计费）、以及生产环境发布所需的套餐订阅费用。 - -对于工作流应用，若配置了文件类型的自定义参数，需在 UI 设计器中指定 `{{{file_name:files[0]}}}`（将 `file_name` 替换为实际变量名），才能正确读取用户上传的文件。 +百炼平台支持将智能体应用（Agent 1.0）和工作流应用以多种方式发布与共享，包括 UI 应用、钉钉/微信机器人、可复用组件及音视频实时互动渠道。该能力面向开发者提供标准化集成路径，适用于业务嵌入、跨平台分发和模块化复用场景。**注意：Agent 2.0 应用仅支持 API 调用，不支持除 API 外的任何发布渠道** [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 + +## 支持的模型/功能 + +- **适用应用类型**： + - ✅ Agent 1.0 智能体应用（支持全部发布渠道） + - ✅ 工作流应用（支持 UI 应用、音视频实时互动、组件发布） + - ❌ Agent 2.0 智能体应用（**仅支持 API 调用**，不支持魔笔、钉钉、微信、UI 设计器或音视频互动等渠道）[分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) +- **核心发布能力**： + - UI 应用（基于魔笔低代码平台，支持 PC/H5 端）[UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) + - 第三方平台机器人（钉钉、微信公众号） + - 可复用组件（供其他智能体或工作流引用） + - 音视频实时互动（H5/APP 扫码体验 + SDK 集成） + - 组件自动更新（应用重新发布后，已发布的组件同步生效）[使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) + +> **注意**：文档 1 与文档 2 均明确指出“Agent 2.0 不支持非 API 发布渠道”，但文档 3 的 UI 设计器章节未提及 Agent 版本限制。实际使用中，**UI 设计器仅支持绑定已发布的 Agent 1.0 或工作流应用**；尝试绑定 Agent 2.0 应用将失败，此为隐含限制，需以文档 1 的版本说明为准。 + +## 关键参数 + +| 参数 | 说明 | 约束 | +|------|------|------| +| `API KEY` | 用于调用百炼服务的身份凭证，必须与应用、UI 设计器处于同一业务空间 | 必填；跨业务空间不可见 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) | +| `query`（系统预设） | 用户输入文本的默认入参，类型 `String`，必填 | 组件配置中不可删除，仅可通过“是否可见”控制透出 | +| `imageList`（系统预设） | 图像公网地址列表，类型 `Array`，非必填 | 仅当组件使用视觉模型时有效；否则建议设为“不可见” [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) | +| 传参方式（`业务透传` / `模型识别`） | 决定参数由调用方显式传入，还是由大模型从上下文推断 | 工作流中**不支持 `模型识别`**，即使配置也为无效；必须通过上游节点显式传入 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) | + +## 使用方式 + +1. **入口统一**：所有发布操作均从百炼控制台 **[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)** 页面的目标应用卡片进入 → 点击 **发布**。 +2. **渠道选择**： + - **UI 应用**：在“发布渠道”页签点击 **UI 应用** → “创建”，或从 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) 页面直接创建并绑定已有应用。开发环境链接 24 小时失效，生产环境需订阅套餐。 + - **钉钉/微信**：在“发布平台”页签完成授权（SLR + API KEY）→ 配置平台凭证（Client ID/Secret、模板 ID、AppID）→ 获取回调地址或二维码。 + - **组件**：在“发布渠道”页签点击 **组件** → “创建”，填写名称、描述、参数别名/是否可见/传参方式等 → 发布后可在其他智能体（技能栏）或工作流（组件节点）中引用。 + - **音视频实时互动**：在“AI 实时互动”页签配置 API KEY → 生成临时体验二维码（24 小时）→ 发布后开通智能媒体服务并完成 SLR 授权。 +3. **组件引用差异**： + - 智能体中引用：大模型根据组件描述+上下文自动决策是否调用；`模型识别` 参数可被自动填充。 + - 工作流中引用：必须手动拖入组件节点，并通过变量（如 `系统变量/query`）显式连接输入；`模型识别` 配置被忽略。 + +## 限制和注意事项 + +- **版本兼容性**：Agent 2.0 应用**完全不支持**魔笔、钉钉、微信、UI 设计器、音视频互动等发布渠道，仅开放 API 接口 [分享智能体应用](../../raw/application-user-guide/application-publishing-and-sharing/share-an-application.md)。 +- **组件调用风险**： + - ❌ 禁止嵌套调用（A → B → A），会导致无限循环； + - ⚠️ 多级调用（A → B → C）易触发超时（最长运行时间限制），应尽量扁平化设计 [使用智能体或工作流作为组件](../../raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md)。 +- **权限与计费**： + - UI 应用默认仅限阿里云用户访问；如需匿名访问，须在 UI 设计器中配置权限组 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)； + - 所有分享链接产生的模型调用费用均由应用创建者 UID 承担； + - 生产环境 UI 发布需订阅付费套餐，开发环境免费但链接 24 小时过期。 +- **业务空间隔离**：API KEY、智能体应用、UI 设计器三者**必须归属同一业务空间**，否则无法关联 [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md)。 ## 来源文档 @@ -82,5 +57,3 @@ UI 设计器集成阿里云多端低代码平台魔笔的能力，提供可视 - [UI设计器](../../raw/application-user-guide/application-publishing-and-sharing/ui-designer.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md index adc2411d..dbc6909a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-support.md @@ -1,91 +1,43 @@ # application [support](support.md) -本页汇总阿里云百炼平台应用与[知识库](../concepts/knowledge-base.md)使用过程中的常见问题、关键限制以及相关协议入口，帮助开发者在接入应用、调用插件、管理数据与合规备案时快速定位答案。内容主要参考 [常见问题](../../raw/application-user-guide/application-support/application-faq.md) 与 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 +`application support` 指百炼平台为开发者在构建和运行 AI 应用（含智能体、RAG 应用、插件集成等）过程中提供的功能能力、调用接口、参数配置及配套服务支持。它覆盖模型调用、插件扩展、知识检索增强、[流式输出](../concepts/streaming-output.md)等核心开发场景，并明确界定平台侧与用户侧的责任边界。相关能力与限制需结合具体 API 行为与服务协议综合理解。 -## 应用中心 +## 支持的模型/功能 -### 插件能力 +- **插件能力**：官方提供六类内置插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索；其中部分需申请开通 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **RAG（知识检索增强）**：支持多知识库并行检索，按配置策略打分后选取 topN 结果用于生成，适用于问答系统、客户服务、教育等场景 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **自定义插件**：支持通过符合协议的 API 接入，大模型可理解其参数结构并自主调用；但**不支持透传自定义 Header**，仅允许 `Authorization` 字段 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- **[流式输出](../concepts/streaming-output.md)**：支持增量式流式响应，需同时设置 `stream=True` 和 `incremental_output=True`。 -百炼应用中心官方提供六款插件：Python 代码解释器、计算器、图片生成、夸克搜索、生成二维码、GitHub 搜索，其中部分插件需申请通过后方可使用。自定义插件服务本身暂不收费，但配置智能体 API 时若涉及 [prompt](prompt.md) 优化、应用调用及测试窗测试，则会产生费用。 +## 关键参数 -- **插件理解机制**：自定义 API 插件遵循协议传给大模型理解；自定义函数则由大模型学习传入的参数信息并返回完整结果。 -- **header 透传**：百炼调用自定义插件时**不支持自定义 header**，仅支持 `authorization`。若业务场景需要透传 header，需在服务端侧另行处理。 +| 参数名 | 类型 | 说明 | +|--------|------|------| +| `stream` | bool | 启用[流式输出](../concepts/streaming-output.md)（默认 `False`） | +| `incremental_output` | bool | 启用增量式流式输出（仅当 `stream=True` 时生效） | +| `MD5` | string | 文件上传必填，用于校验文件完整性（见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)） | -### Agent 与 Assistant API 的区别 +> **注意**：文档 1 中第 8 条明确要求 `incremental_output=True` 实现增量输出，但当前 Assistant API 的 OpenAPI 规范中该参数名为 `enable_incremental_output`（v2024-06+），实际调用请以控制台 SDK 或最新 OpenAPI 文档为准。 -Agent 偏重于调整插件模型与基于上下文的理解，由用户自行开发；Assistant API 则提供各类能力以方便调优。两者面向不同定制粒度，可根据应用复杂度选择。 +## 使用方式 -### 输出控制 +- 插件调用：通过 `tools` 字段声明插件列表，由模型自动选择并填充参数；自定义插件需提供符合 OpenAPI 3.0 规范的 `function` 描述。 +- RAG 应用：在应用配置中绑定知识库，测试时若结果不准确，可通过界面反馈按钮提交问题，或复制 `RequestId` 提交工单 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 +- 文件上传：仅支持小写后缀的 `pdf`/`doc`/`docx`；结构化数据导入需避免空行，否则后续行将被跳过。 +- 售后支持入口：7×24 小时电话（95187）、智能在线、标准工单；基础服务覆盖 API 故障诊断、SDK 使用、控制台问题等 [阿里云百炼平台售后服务范围说明](../../raw/application-user-guide/application-support/application-after-sales-service-scope.md)。 -- **流式与增量输出**：默认为全量回复，若需增量输出，可设置 `stream=True`（[流式输出](../concepts/streaming-output.md)）与 `incremental_output=True`（增量式[流式输出](../concepts/streaming-output.md)）。 -- **Markdown 加粗**：模型输出中的 `**xxxxx**` 是 Markdown 加粗标识，需在前端渲染时解析 md 语法即可正常显示。 +## 限制和注意事项 -## 知识检索（RAG） - -RAG（[检索增强生成](../concepts/rag.md)）在问答系统、对话系统、文本摘要、知识图谱构建与推理、教育与培训、客户服务、新闻与内容创作、智能搜索与推荐等多个领域均有应用。 - -- **检索顺序**：RAG 检索为**并行**方式，依据每个[知识库](../concepts/knowledge-base.md)的用户配置进行检索，再根据得分选取 topN 结果。 -- **回复不准确优化**：可点击模型回复下方的问题反馈按钮，勾选问题类型提交；也可复制 RequestId 通过阿里云工单反馈。详见 [常见问题](../../raw/application-user-guide/application-support/application-faq.md)。 - -## 数据管理 - -### 文件上传 - -- **格式限制**：上传 PDF 文件时后缀必须为小写 `pdf`，否则会触发错误码 `140010`（"上传文件仅支持 pdf/doc/docx 文件"）。 -- **MD5 参数**：上传文件接口的必填 MD5 参数用于校验上传文件的完整性。 -- **容量上限**：每个[业务空间](../concepts/workspace.md)最多上传 10 万个文档，超出需提交阿里云工单申请扩容。 - -### 结构化数据导入 - -结构化数据导入后若出现条数缺失（如 100 条仅导入 20 条），通常是因为表格中存在空行。产品策略规定：遇到空行后续数据不再识别；若第一行为空行，则整表视为空文件。 - -## 应用与小程序备案 - -产品接入通义千问大模型后，若需上架应用市场或小程序平台，需完成备案并申请合作协议： - -1. 参考[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)进行备案。 -2. [提交工单](https://smartservice.console.aliyun.com/service/create-ticket)申请通义千问系列模型的合作协议。 - -## 相关协议 - -接入百炼前请阅读以下协议条款： - -- [阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html) -- [阿里云百炼服务特别说明](https://help.aliyun.com/zh/model-studio/bailian-service-notes) -- [开源模型协议条款说明](https://help.aliyun.com/zh/model-studio/open-source-model-terms) - -协议入口同时收录在 [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) 中，建议在正式接入前逐项确认。 - -## 常见限制与注意事项 - -- 自定义 header 不被支持，仅 `authorization` 可透传。 -- PDF 后缀必须为小写，避免触发 `140010` 错误码。 -- 结构化数据表格中的空行会导致后续数据被截断识别。 -- [业务空间](../concepts/workspace.md)文档数上限为 10 万，超额需工单申请。 -- 自定义插件本身免费，但 [prompt](prompt.md) 优化、应用调用与测试窗测试会收费。 +- **插件 Header 限制**：自定义插件调用时，仅 `Authorization` 可透传，其他 Header（如 `X-User-ID`、`Cookie`）会被丢弃。 +- **知识库容量**：单业务空间上限 10 万文档，超限时需提交工单申请扩容。 +- **第三方工具责任边界**：阿里云不负责 Cursor、Windsurf 等第三方工具的安装、配置、故障排查或本地环境（代理/防火墙/VPN）问题；仅提供百炼服务端连通性验证与调用示例参考 [阿里云百炼平台售后服务范围说明](../../raw/application-user-guide/application-support/application-after-sales-service-scope.md)。 +- **合规与备案**：接入通义千问模型上架应用市场或小程序前，须完成[应用合规备案](https://help.aliyun.com/zh/model-studio/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model)并申请合作协议。 +- **协议约束**：所有使用须遵守《阿里云百炼服务协议》及《阿里云百炼体验功能特别说明》，开源模型还需符合对应[开源模型协议条款](https://help.aliyun.com/zh/model-studio/open-source-model-terms) [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md)。 ## 来源文档 - [常见问题](../../raw/application-user-guide/application-support/application-faq.md) - [相关协议](../../raw/application-user-guide/application-support/application-related-agreements.md) - - - - - - - - - - - - - - - - - - - +- [阿里云百炼平台售后服务范围说明](../../raw/application-user-guide/application-support/application-after-sales-service-scope.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md index cbd307ef..355a01c7 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/application-use-cases.md @@ -1,107 +1,69 @@ # application [use cases](use-cases.md) -百炼平台围绕“[检索增强生成](../concepts/rag.md)（RAG）+ [智能体应用](../concepts/agent-application.md)”提供了多种开箱即用的应用场景，可在无需编码或少量编码的情况下，将大模型问答能力接入网站、企业微信、微信公众号、钉钉等渠道，并支持基于本地[知识库](../concepts/knowledge-base.md)构建 RAG 应用。本页汇总这些典型用法的关键流程、模型选择、参数配置与注意事项。 +阿里云百炼平台支持将大模型能力快速集成到主流企业通讯与业务平台中，实现开箱即用的智能问答、客服与助手服务。核心路径为：创建百炼智能体应用 → 通过 AppFlow 连接目标平台（如企业微信、钉钉、微信公众号、网站）→ 可选配置私有知识库（RAG）。所有方案均无需编码，依赖统一的 API Key 与应用 ID 配置，且新用户可使用免费额度完成端到端验证。 -## 支持的接入渠道 +## 支持的模型/功能 -百炼 RAG 应用可通过 AppFlow（无代码连接流）或本地部署方式接入以下渠道： +- **基础模型**：所有用例均支持通义千问系列模型，包括 `qwen-plus`（文档 1、3、4 中明确指定为默认或推荐）、`qwen-turbo`（文档 3 提及用于提速）、`qwen-max`（文档 5 列为可选项）及 `qwen3.5-plus`（文档 2 明确指定）。模型选择直接影响响应质量、延迟与成本，需按场景权衡。 +- **核心功能**： + - 智能体（Agent）应用：支持角色设定（Prompt）、多轮对话、工具调用（当前文档未展开，但为百炼基础能力）。 + - RAG 知识增强：所有用例均支持通过上传文件（PDF/DOCX/TXT 等）创建知识库，并在应用中启用“必定调用”等策略 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)。 + - 多平台消息交互：支持文本、卡片（钉钉）、富媒体消息（企业微信/公众号）等格式，具体能力取决于目标平台接口限制（如未认证公众号仅支持 5 秒内被动回复）[10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)。 +- **本地部署选项**：除云端 SaaS 方式外，还提供基于 Python 的本地 RAG 应用框架，支持自定义文档切分、本地嵌入模型（如 GTE）及灵活参数调优 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 -- 网站：通过 AppFlow 创建 AI 助手并生成悬浮挂件部署脚本，粘贴到网站 HTML 即可。详见 [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md)。 -- 企业微信：通过 AppFlow 模板“企业微信自建应用大模型自动回复”连接企业微信应用与百炼应用。详见 [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md)。 -- 微信公众号（订阅号）：通过 AppFlow 模板连接公众号与百炼应用，完成认证的公众号可用客户消息接口，未认证只能被动回复且响应限制为 5 秒。详见 [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md)。 -- 钉钉：创建钉钉应用并通过 AppFlow 模板连接，机器人消息接收模式必须选 HTTP 模式。详见 [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md)。 -- 本地[知识库](../concepts/knowledge-base.md) RAG：检索环节在本地执行，生成环节调用通义千问 API，适合需要灵活切分与嵌入模型选择的场景。详见 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 - -## 通用流程 - -各渠道方案共享相似的四到五步骨架： - -1. 创建百炼[智能体应用](../concepts/agent-application.md)并获取应用 ID 与 [API Key](../concepts/api-key.md)； -2. 在目标平台（企业微信 / 公众号 / 钉钉）创建应用或机器人，获取对应凭证（企业 ID、AgentId、Secret、AppID、Client ID/Secret 等）； -3. 通过 AppFlow 模板创建连接流，填入双方凭证与应用 ID，发布并获取 WebhookUrl； -4. 在目标平台配置消息接收地址为 WebhookUrl，并配置可信 IP / 白名单； -5. 为百炼应用添加私有[知识库](../concepts/knowledge-base.md)（RAG），让回答覆盖私域问题。 - -网站方案略不同：第 2–3 步改为在 AppFlow 创建 AI 助手并生成 web 页面集成脚本，第 4 步将悬浮挂件脚本粘贴到网站 HTML。 - -## 模型选择 - -各教程对模型选择存在差异，需注意版本与命名： - -- 网站方案原选 **Qwen3.5-Plus**，其余渠道教程多选 **千问-Plus**（qwen-plus）。 -- 本地 RAG 应用可选 qwen-max、qwen-plus、qwen-turbo 三个通义千问商业模型：qwen-max 性能优秀，qwen-turbo 速度更快价格更低，qwen-plus 效果、速度、成本均衡。 - -> **注意**：不同文档对模型命名不一致（“Qwen3.5-Plus”与“千问-Plus”）。实际配置时以百炼控制台“更多模型”列表中可用的版本为准。未认证公众号若超 5 秒未返回，可在 [prompt](prompt.md) 中加“请总是给出简短的回答”或改用 qwen-turbo 提速，但会降低效果。 +> **注意**：文档 1、3、4 均以 `qwen-plus` 为默认模型，而文档 2 明确使用 `qwen3.5-plus`，文档 5 则列出 `qwen-max`/`plus`/`turbo` 三者供选。实际部署时应以百炼控制台当前可用模型列表为准，`qwen3.5-plus` 属于较新版本，若控制台未显示则需选用 `qwen-plus` 作为兼容替代。 ## 关键参数 -百炼应用侧： - -- Prompt：可设置角色人设引导回答，如“你叫小助，可以帮助用户解答产品选购、使用等方面的问题。” -- 知识库调用方式：可选**必定调用**；知识文档支持配置相似度阈值与权重。 -- 文件处理：可选全文引用、切片检索或自定义处理。 -- 向量存储：标准版默认即可；如需集中管理多应用向量数据可选 ADB-PG。 - -本地 RAG 应用侧： - -- 模型参数：模型选择、温度（越高随机性越高）、最大回复长度、携带上下文轮数（设为 1 时不参考历史）。 -- RAG 参数：召回片段数（越大参考越多但噪声可能增加）、相似度阈值（越大参考越少但噪声减少，为 0 不剔除）。 -- 嵌入模型：默认用百炼 embedding API；可改用本地部署的 GTE 文本向量模型（如 `iic/nlp_gte_sentence-embedding_chinese-large`）。 -- 文件限制：受 embedding API 限流，不建议传入超过 100 MB 的文件。 - -AppFlow 连接流： - -- 钉钉消息接收模式必须选 **HTTP 模式**，Stream 模式会导致无法返回消息。 -- 公众号需区分已认证 / 未认证两条[工作流](../concepts/workflow.md)；未认证只能被动回复，5 秒超时。 -- 企业微信需配置企业可信 IP；若报“域名主体校验未通过”，需配置企业自有域名或通过 ECS / 托管实例 / 计算巢 Nginx 代理转发。 +- **身份凭证**： + - `App ID`：百炼应用唯一标识，在 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) 页面获取。 + - `API Key`：用于 AppFlow 或前端 SDK 调用百炼 API 的密钥，在 [API Key](https://bailian.console.aliyun.com/?tab=app#/api-key) 页面创建。 +- **平台凭证**（依目标平台而异）： + - 企业微信：`CorpID`、`AgentID`、`Secret`（文档 1）。 + - 微信公众号：`AppID`（文档 3），认证状态决定连接流模板选择。 + - 钉钉：`Client ID`、`Client Secret`（文档 4）。 + - 网站：无需平台凭证，依赖前端 SDK 注入（文档 2）。 +- **RAG 相关参数**： + - `召回片段数`、`相似度阈值`：控制知识检索精度，可在本地 RAG 应用中直接调整 [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md)。 + - `调用方式`（必定调用/按需调用）、`权重`、`文件处理方式`（全文引用/切片检索）：在百炼应用配置界面设置（文档 1、2、3、4）。 ## 使用方式 -网站接入：在 AppFlow AI 助手 Web 集成页复制悬浮挂件部署脚本，粘贴到网站 HTML 注释下方；可选启用图标拖拽。也可用函数计算 FC 一键部署示例网站。 - -企业微信 / 公众号 / 钉钉：发布连接流后，在目标平台配置 API 接收消息（URL 填 WebhookUrl，[Token](../concepts/token.md) / EncodingAESKey 填 AppFlow 凭证生成的值），配置可信 IP，即在聊天中 @机器人 或直接对话使用。 - -本地 RAG：解压 `local_rag.zip`，Python 3.8–3.12 环境安装依赖，配置百炼 [API Key](../concepts/api-key.md) 环境变量，运行 `uvicorn main:app --port 7866`，访问 `http://127.0.0.1:7866`。支持临时性文件上传（对话框直接传 pdf/docx/txt/xlsx/csv，刷新后失效）与长期知识库创建（上传到 File/Unstructured 或 File/Structured 后在界面创建知识库存于 VectorStore）。通过 Gradio 界面下方的“通过 API 使用”可获取 API 文档集成到业务。 - -## 日志与扩展 - -AppFlow 连接流可在百炼步骤后添加 SLS 日志云服务节点，将对话写入阿里云日志服务（需创建 Project/Logstore 并开启全文索引）。钉钉还支持通过卡片平台导入模板展示回答引用的文档来源，以及展示 DeepSeek 深度思考过程（在发送 AI 卡片阶段填入 `思考过程： {{Node2.reasoning}}` 与 `推理结果： {{Node2.text}}`）。 +1. **创建百炼应用**：进入百炼控制台 → 应用管理 → 创建智能体应用 → 选择模型、配置 Prompt → 发布。 +2. **配置目标平台连接**： + - **企业微信/钉钉/公众号**：使用 AppFlow 预置模板（文档 1、3、4 提供具体 URL），通过向导配置平台凭证与百炼凭证，生成 Webhook URL 或完成 OAuth 授权。 + - **网站**：在 AppFlow 创建 AI 助手 → 导入百炼应用 → 配置 Web 集成（悬浮挂件）→ 将生成的 JS 脚本嵌入 HTML（文档 2）。 +3. **启用知识增强（可选）**： + - 上传文件至百炼 [数据中心](https://bailian.console.aliyun.com/?tab=app#/data-center) 或使用 [数据连接](https://bailian.console.aliyun.com/cn-beijing?tab=app#/connector/list)（文档 2、4）。 + - 在 [知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base) 创建知识库并关联至应用。 +4. **平台侧配置**： + - 企业微信：配置 API 接收消息（填入 Webhook URL、[Token](../concepts/token.md)、EncodingAESKey）及可信 IP（文档 1）。 + - 公众号：开启服务器配置（需认证）或使用被动回复（文档 3）。 + - 钉钉：在应用后台启用 HTTP 模式机器人并填入 Webhook URL（文档 4）。 + - 网站：部署 JS 脚本后即可生效（文档 2）。 ## 限制和注意事项 -- 新用户免费额度可覆盖教程资源消耗，超额后按 token [计费](../concepts/billing.md)。 -- 百炼文件导入支持 pdf、doc、docx、txt、md、pptx、ppt、png、jpg、jpeg、bmp、gif、xls、xlsx，单文档最大 100MB 或 1000 页，单图片最大 20MB，最多 200 个文件；文件存储在新加坡区域，解析通常 1–6 分钟。 -- 钉钉应用需开通 `Card.Streaming.Write` 与 `Card.Instance.Write` 权限以发送卡片消息；应用供企业内其他用户使用需发布版本并设置可见范围。 -- 企业微信可信 IP 一个 IP 仅能用于一个企业，多企业共用会被识别为服务商导致通讯录 / 身份校验接口不可用。 -- 公众号开启服务器配置后自定义菜单会冲突关闭；未认证无法同时开启二者，已认证可通过接口重建菜单。 -- 本地 RAG Windows 系统若缺少 Microsoft Visual C++ Redistributable 需另行安装；报 `DLL load failed while importing _cext:` 时运行 `pip install msvc-runtime`。 -- 上线前建议组织业务人员参与应用[评测](../concepts/evaluation.md)，结合优化提示词、补充私有知识、调整切分策略改进效果。 +- **平台能力限制**： + - 未认证微信公众号仅支持 5 秒内被动回复，超时将失败；建议完成认证或选用 `qwen-turbo` 降低延迟（文档 3）。 + - 钉钉机器人必须选择 **HTTP 模式**，Stream 模式不被 AppFlow 支持（文档 4）。 + - 企业微信需配置可信 IP 与域名主体校验，否则 API 接收失败；若无备案域名，需通过 AppFlow 内网代理或 Nginx 转发解决（文档 1）。 +- **文件与知识库限制**： + - 百炼云端知识库：单文件 ≤ 100MB 或 1000 页，支持格式详见各文档（PDF/DOCX/TXT 等）；解析耗时 1–6 分钟（文档 1、2、3、4）。 + - 本地 RAG 应用：不建议上传 >100MB 文件，受限于 Embedding API 限流（文档 5）。 +- **调试与监控**： + - 所有 AppFlow 连接流均支持查看 **运行日志** 排查失败原因（文档 3）。 + - 可通过添加 SLS 日志节点记录完整对话（文档 1、3、4），用于效果分析与合规审计。 +- **安全要求**： + - 钉钉应用需显式开通 `Card.Streaming.Write` 和 `Card.Instance.Write` 权限（文档 4）。 + - 企业微信/钉钉的凭证（Secret/Client Secret）须严格保密，不可硬编码于前端。 ## 来源文档 -- [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [在企业微信中集成一个 AI 助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) +- [在网站上增加一个AI助手](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) - [10分钟让微信公众号成为智能客服](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [在钉钉上增加一个AI机器人](../../raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [基于本地知识库构建RAG应用](../../raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md index 5d694cf8..1bb4f160 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/bailian-application-calling.md @@ -1,195 +1,77 @@ # bailian [application call](../api/application-call.md)ing -阿里云百炼支持通过 [DashScope SDK](../concepts/dashscope-sdk.md) 或 HTTP API 将已创建的应用集成到业务系统中。可调用的应用类型包括**[智能体应用](../concepts/agent-application.md)**和**[工作流](../concepts/workflow.md)应用**（[智能体编排](../concepts/agent-orchestration.md)应用已被[工作流](../concepts/workflow.md)应用替代），二者调用方式一致，均通过 `Application.call` / `POST /apps/{app_id}/completion` 触发，区别仅在于应用内部编排逻辑和可附加的扩展能力（如自定义参数传递）。 +百炼应用调用（bailian [application call](../api/application-call.md)ing）是指通过 DashScope SDK 或标准 HTTP API，将阿里云百炼平台创建的智能体应用或工作流应用集成至第三方业务系统的能力。该机制统一使用 `/api/v1/apps/{app_id}/completion` 接口，支持单轮/多轮对话、自定义插件参数透传等核心能力，适用于从简单问答到复杂编排的各类 AI 应用场景。所有调用均需有效的 API Key 和已发布的应用 ID。 -## 前提条件 +## 支持的模型/功能 -无论调用哪种应用，都需要先完成以下准备： +- **应用类型**：同时支持**智能体应用**（单 Agent 应用）和**工作流应用**（原“智能体编排应用”，已在[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)中明确说明“智能体编排应用已被工作流应用替代”）。 +- **核心功能**： + - 单轮文本生成（`prompt` 输入） + - 多轮对话（通过 `session_id` 或显式 `messages` 数组管理上下文） + - 自定义插件参数透传（通过 `biz_params.user_defined_params` 字段，详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)） + - 调试信息返回（`debug` 字段可选启用） -1. **获取 [API Key](../concepts/api-key.md)**：在百炼控制台密钥管理页面创建 [API Key](../concepts/api-key.md)。 -2. **配置环境变量（推荐）**：将 [API Key](../concepts/api-key.md) 写入 `DASHSCOPE_API_KEY` 环境变量，避免在代码中硬编码。SDK 会自动读取该变量。 -3. **获取应用 ID**：在应用管理页面创建对应应用（[智能体应用](../concepts/agent-application.md) / [工作流](../concepts/workflow.md)应用），并从应用卡片复制 `APP_ID`。 -4. **安装 [DashScope SDK](../concepts/dashscope-sdk.md)**（HTTP 调用可跳过）：Python 通过 `python3 -m pip install -U dashscope`；Java 通过 Maven/Gradle 添加 `com.alibaba:dashscope-sdk-java` 依赖（建议版本 >= 2.12.0）；Node.js 安装 `axios`。 - -## 基本调用方式 - -[智能体应用](../concepts/agent-application.md)与工作流应用的调用接口完全相同，详见[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)与[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。核心请求结构如下： - -``` -POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion -Authorization: Bearer $DASHSCOPE_API_KEY -Content-Type: application/json - -{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} -} -``` - -### Python - -```python -import os -from http import HTTPStatus -from dashscope import Application - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='你是谁？' -) - -if response.status_code != HTTPStatus.OK: - print(f'request_id={response.request_id}') - print(f'code={response.status_code}') - print(f'message={response.message}') -else: - print(response.output.text) -``` - -### Java - -```java -import com.alibaba.dashscope.app.*; -import com.alibaba.dashscope.exception.*; - -ApplicationParam param = ApplicationParam.builder() - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .appId("YOUR_APP_ID") - .prompt("你是谁？") - .build(); - -Application application = new Application(); -ApplicationResult result = application.call(param); -System.out.printf("text: %s\n", result.getOutput().getText()); -``` - -### Node.js / curl - -Node.js 使用 `axios` 发起 POST 请求即可，结构与 curl 示例一致： - -```bash -curl -X POST https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header 'Content-Type: application/json' \ - --data '{ - "input": { "prompt": "你是谁？" }, - "parameters": {}, - "debug": {} - }' -``` - -响应统一为 `{"output": {"finish_reason", "session_id", "text"}, "usage": {...}, "request_id": "..."}` 结构，业务侧主要消费 `output.text`。 - -## 多轮对话 - -工作流应用支持多轮对话，详见[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)。两种实现方式： - -- **使用 `session_id`**：系统自动从云端加载历史对话，实现简单。`session_id` 有效期 1 小时，最多支持 50 轮对话。 -- **自行管理 `messages`（推荐）**：手动维护 `messages` 数组传递每轮历史，无需传 `prompt`，控制更灵活。 - -> **注意**：若请求中同时包含 `session_id` 和 `messages`，系统将优先使用 `messages`。 - -使用 `messages` 时，需先在工作流的大模型节点中配置提示词变量 `historyList` 并发布应用，再发起调用。 - -## 自定义参数传递 - -针对自定义插件与自定义节点，百炼支持通过 `biz_params` 的 `biz_params.user_defined_params` 透传业务参数，详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md)。该能力可用于[智能体应用](../concepts/agent-application.md)的自定义插件，以及工作流应用中的插件节点。 - -### 使用流程 - -1. **创建自定义插件**：在百炼控制台插件页面新增自定义插件，按需配置鉴权（如用户级鉴权 + Header + basic）。创建工具时，输入参数的**传参方式务必选择「业务透传」**，并发布插件。 -2. **关联应用**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联；工作流应用则在插件节点中引用。关联后发布应用。 -3. **API 调用**：通过 `biz_params.user_defined_params` 传递插件 ID 与入参键值对。 - -### 请求示例 - -```python -import os -from http import HTTPStatus -from dashscope import Application - -biz_params = { - "user_defined_params": { - "your_plugin_code": { # 替换为实际插件 ID（在插件卡片获取） - "article_index": 2 - } - } -} - -response = Application.call( - api_key=os.getenv("DASHSCOPE_API_KEY"), - app_id='YOUR_APP_ID', - prompt='寝室公约内容', - biz_params=biz_params -) -``` - -HTTP 请求体中 `biz_params` 位于 `input` 下： - -```json -{ - "input": { - "prompt": "寝室公约内容", - "biz_params": { - "user_defined_params": { - "your_plugin_code": { "article_index": 2 } - } - } - }, - "parameters": {}, - "debug": {} -} -``` - -Java SDK 通过 `JsonUtils.parse(...)` 将 JSON 字符串转为对象传入 `ApplicationParam.bizParams`。 +> **注意**：文档 2（调用工作流应用）声明“本文档仅适用于华北2（北京）地域”，而文档 1（调用智能体应用）及文档 3 均未限定地域。实际生产环境应以控制台所选应用部署地域为准，建议在调用前确认应用所在 Region 并匹配对应 endpoint —— 当前所有示例均使用 `https://dashscope.aliyuncs.com`，该域名默认路由至用户应用所在地域，无需手动切换。 ## 关键参数 -| 参数 | 位置 | 说明 | -| --- | --- | --- | -| `app_id` / `APP_ID` | URL path | 应用 ID，从应用卡片获取 | -| `prompt` | `input.prompt` | 单轮对话的输入指令（与 `messages` 二选一） | -| `messages` | `input.messages` | 自行维护的多轮对话历史（推荐，优先级高于 `session_id`） | -| `session_id` | `input.session_id` | 云端会话 ID，有效期 1 小时，最多 50 轮 | -| `biz_params` | `input.biz_params` | 业务透传参数，含 `user_defined_params` | -| `parameters` | 顶层 | 应用级参数 | -| `debug` | 顶层 | 调试信息 | +| 参数名 | 类型 | 必填 | 说明 | +|--------|------|------|------| +| `app_id` | string | 是 | 百炼控制台应用卡片上复制的唯一 ID，见[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)前提条件 | +| `prompt` | string | 否（若提供 `messages` 则不可用） | 单轮指令文本；若启用多轮且使用 `messages`，则此字段必须省略 | +| `messages` | array | 否（若提供则替代 `prompt`） | 显式对话历史数组，格式为 `[{"role": "user/system/assistant", "content": "..."}]`；优先级高于 `session_id` | +| `session_id` | string | 否（用于云端会话） | 由服务端生成并返回的会话标识符，有效期 1 小时，最多支持 50 轮；与 `messages` 同时存在时，以 `messages` 为准 | +| `biz_params` | object | 否 | 用于透传自定义插件参数，结构为 `{"user_defined_params": {"{plugin_code}": {...}}}`，详见[应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) | +| `parameters` | object | 否 | 模型级超参（如 `temperature`, `top_p`），当前对应用调用影响有限，建议保持空对象 `{}` | +| `debug` | object | 否 | 开启调试模式（如 `{"enable": true}`），返回详细执行链路信息 | + +## 使用方式 + +### 1. 准备工作 +- 获取 [API Key](https://bailian.console.aliyun.com/?tab=model#/api-key) 并配置为环境变量 `DASHSCOPE_API_KEY`（推荐，避免硬编码）； +- 在[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面获取目标应用的 `APP_ID`； +- 若使用 SDK，按语言安装对应版本（Python ≥1.14.0，Java ≥2.12.0，见[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)）。 + +### 2. 调用示例（SDK 与 HTTP 统一接口） +- **SDK 调用（Python）**： + ```python + from dashscope import Application + response = Application.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + app_id="YOUR_APP_ID", + prompt="你是谁？", + biz_params={"user_defined_params": {"plugin_abc": {"param1": "value1"}}} + ) + print(response.output.text) + ``` + +- **HTTP 调用（curl）**： + ```bash + curl -X POST "https://dashscope.aliyuncs.com/api/v1/apps/YOUR_APP_ID/completion" \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "input": { + "prompt": "你是谁？", + "biz_params": {"user_defined_params": {"plugin_abc": {"param1": "value1"}}} + } + }' + ``` + +> 所有语言（Java/PHP/Node.js/C#/Go）的完整示例请参考[调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md)和[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md)中的代码片段。 ## 限制和注意事项 -- **地域限制**：[调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) 文档明确仅适用于华北2（北京）地域。 -- **应用类型替代关系**：[智能体编排](../concepts/agent-orchestration.md)应用已被工作流应用替代，新场景应使用工作流应用。 -- **`session_id` 约束**：有效期 1 小时，最多 50 轮；与 `messages` 同时存在时优先使用 `messages`。 -- **插件业务透传**：自定义插件的输入参数传参方式必须选择「业务透传」，否则无法通过 `biz_params` 传递；插件 ID 在插件卡片获取，替换示例中的 `your_plugin_code`。 -- **[业务空间](../concepts/workspace.md)隔离**：插件工具只能与同一[业务空间](../concepts/workspace.md)内的[智能体应用](../concepts/agent-application.md)关联。 -- **密钥安全**：不要在生产环境硬编码 [API Key](../concepts/api-key.md)，统一通过 `DASHSCOPE_API_KEY` 环境变量注入。 -- **Responses API**：如需使用 OpenAI 兼容的 Responses API 调用工作流应用，需参阅 Responses API 文档，不在本文调用方式范围内。 +- **地域限制**：工作流应用调用受地域约束，当前仅支持华北2（北京），智能体应用无明确限制，但建议与应用部署地域一致； +- **会话管理**：`session_id` 有效期为 1 小时，最大轮次 50；生产环境推荐自行维护 `messages` 数组以获得确定性行为； +- **插件参数安全**：`biz_params.user_defined_params` 中的插件 ID 必须与应用内已关联的插件完全匹配，否则参数被忽略； +- **错误处理**：所有调用均返回 `request_id`，务必记录该字段用于问题排查；错误码含义请查阅[开发者参考错误码文档](https://help.aliyun.com/zh/model-studio/developer-reference/error-code)； +- **SDK 版本兼容性**：Python SDK 要求 ≥1.14.0（插件参数支持）、Java SDK 要求 ≥2.12.0（多轮对话稳定性），旧版本可能缺失关键功能。 ## 来源文档 -- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) -- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [调用智能体应用](../../raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - - - - - - - - - - - - - - - - - - - +- [调用工作流应用](../../raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) +- [应用的自定义参数传递](../../raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md index c0b98b7e..d8e4dff5 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/data-connection-overview.md @@ -1,80 +1,52 @@ # data connection overview -数据连接是阿里云百炼平台管理外部数据源的统一入口。通过创建数据连接器，百炼应用可以安全地访问企业数据库、文档系统和对象存储中的数据，并在对话中实时查询和引用这些数据。详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md)。 +数据连接是阿里云百炼平台统一管理外部数据源的核心能力，为应用提供安全、可控的实时数据访问入口。它支持结构化与非结构化数据源的接入，并通过平台托管或流处理两类模式实现数据读取与检索。开发者可基于业务场景选择适配的连接器类型，并在智能体或 API 调用中直接引用已配置的数据源。 -## 连接器类型 +## 支持的模型/功能 -数据连接器按数据的存储和访问方式分为两大类： +数据连接器按数据访问模式分为两类： -- **平台托管**：数据导入并存储在百炼平台（或自有 OSS）。 - - **文件**：管理非结构化文档（PDF、Word、Markdown 等）。 - - **表格**：导入并查询结构化表格数据（CSV、Excel 等）。 -- **流处理**：数据保留在原数据源，实时访问。 - - **MySQL / PostgreSQL / PolarDB-X 2.0**：连接对应数据库，支持执行 SQL 查询（仅 DMS 导入方式支持）。 - - **语雀**：访问语雀文档和知识库（仅公网版本）。 - - **OSS**：访问对象存储中的文件。 +- **平台托管型**：适用于文件（PDF/Word/Markdown）、表格（CSV/Excel）类非结构化与轻量结构化数据。数据导入后由百炼平台统一存储、解析并构建向量索引，支持语义检索与[多模态](../concepts/multi-modal.md)理解（如图表识别需启用[大模型文档解析](https://help.aliyun.com/zh/document-mind/product-overview/overview-of-document-understanding#9a4f5fb91fpps)）。详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“文件连接器”与“表格连接器”章节。 -各类型的适用场景与存储方式详见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的连接器类型表。 +- **流处理型**：适用于需实时查询的数据库与在线知识库，包括 MySQL、PostgreSQL、PolarDB-X 2.0、语雀和 OSS。此类连接器不导入数据副本，而是按需执行 SQL 查询或 API 检索，适用于动态数据场景。其中仅通过 DMS 导入方式创建的 MySQL/PostgreSQL/PolarDB-X 连接器支持 SQL 执行；自定义方式创建的同类连接器**不支持 SQL 查询**（参见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) “MySQL连接器”与“PostgreSQL连接器”说明）。 -## 前置条件 +> **注意**：OSS 连接器虽属流处理型，但其 `searchOSSFile` 和 `searchOSSFileByFileName` 工具依赖向量检索服务，该服务需手动开通；而文件/表格连接器的向量索引构建则由平台自动完成，无需额外开通——二者能力边界存在差异，不可混用。 -- **账号权限**：主账号或具有数据连接管理权限的 RAM 用户；RAM 用户需先获得主账号授权。 -- **数据源准备**（按连接器类型）： - - 文件/表格：准备好待上传文档/表格，或已创建 OSS Bucket。 - - MySQL：已有 MySQL 实例（RDS 或自建），网络可达（公网或私网）。 - - PostgreSQL：账号具备高权限（Superuser 或 REPLICATION），且 `wal_level` 设置为 `logical`；自建实例还需配置 `listen_addresses` 允许 `100.64.0.0/16` 网段访问。 - - PolarDB-X 2.0：已有阿里云 PolarDB-X 2.0 实例且所在地域支持私网访问；DMS 导入方式需先在 DMS 录入实例。 - - 语雀：已有公网版语雀知识库并获取访问 Token。 - - OSS：已创建 Bucket 并开通向量检索服务。 +## 关键参数 -## 数据库连接器关键差异 +| 连接器类型 | 必填参数 | 特殊要求 | 检测机制 | +|------------|----------|----------|----------| +| 文件/表格 | 连接器名称、描述、存储位置（平台存储/OSS） | 平台存储有额度限制（文件连接器限 200,000 文件/1 TB；表格连接器 1 TB 免费额度）；OSS Bucket 需添加 `bailian-connector-access` 标签（值 `ReadAndWrite`） | 无主动连通性检测，依赖上传/导入任务状态 | +| MySQL | 数据库地址、端口、用户名、密码；若为 RDS 实例则需实例 ID | 公网连接需白名单放行指定 IP 段；`wal_level` 无特殊要求 | EventBridge 服务检测 | +| PostgreSQL | 主机地址、端口、数据库名（`dbName`）、用户名、密码 | `wal_level=logical`；自建实例需配置 `listen_addresses` 允许 `100.64.0.0/16` 访问；用户需具备 `REPLICATION` 或 Superuser 权限 | DTS 服务检测 | +| PolarDB-X 2.0 | 数据库用户名、密码；仅支持私网连接 | 仅支持阿里云实例；首次使用需授权 `AliyunServiceRoleForSFMConnectorAccessDTS` 与 `AliyunServiceRoleForSFMAccessPolarDBX` 角色 | EventBridge 检测（同 MySQL） | +| 语雀 | Tenant access token | 仅支持公网语雀；[Token](../concepts/token.md) 需通过 [语雀开放 API](https://www.yuque.com/yuque/developer/api) 获取 | [Token](../concepts/token.md) 校验接口调用 | +| OSS | Bucket 名称 | Bucket 需添加 `bailian-datahub-access` 标签（值 `read`）；不支持归档/冷归档存储类型；Referer 防盗链需白名单 `*.console.aliyun.com` | Bucket 权限与标签校验 | -| 差异项 | MySQL | PostgreSQL | PolarDB-X 2.0 | -| --- | --- | --- | --- | -| 默认端口 | 3306 | 5432 | 自动获取 | -| 额外必填字段 | 无 | dbName（数据库名称） | 无 | -| 连通性检测服务 | EventBridge | DTS | DTS | -| 网络类型 | 公网 / 私网 | 公网 / 私网 | 仅私网 | -| 特殊配置 | 无 | `wal_level=logical` | 需 SLR 授权（DTS、PolarDB-X，DMS 方式加 DMS 角色） | +## 使用方式 -三类流处理数据库连接器均支持两种数据来源配置：**创建自定义数据源**（手动配置连接信息）与**从 DMS 导入数据源**（导入 DMS 中已有数据源）。数据库用户须具备读取权限，配置后可点击检测验证连通性。完整字段说明见 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 的各连接器配置章节。 - -> **注意**：仅通过**从 DMS 导入数据源**方式创建的 MySQL / PostgreSQL / PolarDB-X 2.0 连接器支持执行 SQL 查询；通过**创建自定义数据源**方式添加的连接器不支持直接执行 SQL。 - -## 创建连接器 - -1. 在[数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list)页面点击**创建连接器**。 -2. 选择连接器类型，填写**连接器名称**和**描述**（描述会用于指导应用调用的准确度，建议写明数据内容和用途）。 -3. 按类型填写存储位置或数据源信息，必要时执行连通性检测。 -4. 点击**确认**完成创建。 - -存储位置选择要点： - -- **文件连接器**：平台存储限时免费（最多 200,000 个文件、1 TB）；或使用自有 OSS，需为 Bucket 添加 `bailian-connector-access` 标签（值 `ReadAndWrite`）。 -- **表格连接器**：平台存储提供 1 TB 免费额度，用尽后转按量付费；自有 OSS 同样需上述标签。 -- **OSS 连接器**：从下拉列表选择 Bucket，需添加 `bailian-datahub-access` 标签（值 `read`）。 - -## 导入数据 - -- **导入文件**：进入文件连接器详情页，在**类目**下选择或新建类目后导入。平台暂不支持直接导入 JSON、CSV、YAML，需先转换为 XLSX/XLS。可选择**默认设置**或**自定义设置**解析（电子文档解析、文档智能解析、大模型文档解析、Qwen VL 解析、音视频解析等，具体能力取决于文件类型）。可为文件配置**标签**，API 调用时通过 `tags` 参数筛选以提升检索效率。 -- **导入表格**：进入表格连接器详情页，在**数据表管理**下选择或新建数据表。支持**直接上传 Excel**（自动识别表头）或**自定义表头**。数据表结构（列名、描述、类型）一旦确定不可修改，且上传文件的列数与列名必须与表结构一一对应，否则导入失败。 +1. **创建连接器**：进入 [数据连接](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/connector/list) 页面 → 单击“创建连接器” → 选择类型 → 填写基本信息与连接参数 → （可选）执行连通性检测 → 确认创建。 +2. **导入数据（仅平台托管型）**： + - 文件连接器：进入详情页 → 选择类目 → “导入数据” → 本地上传 → 选择解析方式（推荐默认设置；图表理解需选“大模型文档解析”）→ 配置标签（可选）→ 提交。 + - 表格连接器：进入详情页 → 新建或选择数据表 → 上传 Excel 或自定义表头（列名、类型必填；`image_url` 字段需确保 URL 公开可访问）→ 提交。 +3. **在应用中调用**：连接器创建成功后，可在智能体工作流中作为知识库数据源绑定，或通过 API 的 `knowledge_sources` 参数引用（格式：`{ "type": "connector", "id": "" }`）。具体集成方式请参考 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 中“导入数据”与“在应用中使用”部分。 ## 限制和注意事项 -- 数据库连接器执行 SQL 的限制见上文注意框（仅 DMS 导入方式支持）。 -- **OSS 连接器**：使用需开通[向量检索服务](https://help.aliyun.com/zh/oss/user-guide/vector-retrieval/)，否则无法使用 `searchOSSFile` / `searchOSSFileByFileName` 工具；不支持归档/冷归档/深度冷归档类型的 Bucket；支持内容加密与私有 Bucket；开启 Referer 防盗链时需将 `*.console.aliyun.com` 加入白名单。 -- **文件导入**：文件作为独立副本存储在平台免费空间（当前无容量限制），仅支持查看最近 **90** 天内导入的文件（超期不可查看但不删除），且仅供当前业务空间使用。请求高峰期解析可能耗时数小时甚至偶现超时，需耐心等待或重试。 -- **语雀连接器**：仅支持公网版本语雀，需提供有效的 Tenant access token。 - -以上流程、字段和限制的完整细节，请以原文 [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) 为准。 +- **权限约束**：RAM 用户需主账号授予 `AliyunBailianFullAccess` 或最小化自定义策略（含 `bailian:ListConnectors`, `bailian:CreateConnector` 等动作），否则无法创建或管理连接器。 +- **网络限制**： + - MySQL/PostgreSQL 支持公网与私网；PolarDB-X 2.0 **仅支持私网**，且必须与百炼服务同地域。 + - 自建数据库需确保出方向防火墙放行百炼服务 IP 段（如 `100.64.0.0/16`）。 +- **数据时效性**： + - 平台托管型：文件/表格导入后生成静态副本，更新需重新上传；90 天内可查看历史导入记录，超期仅保留索引不可预览。 + - 流处理型：数据始终实时，但语雀/OSS/数据库查询受目标服务稳定性影响。 +- **功能限制**： + - 不支持 JSON/YAML 直接导入（表格连接器需转为 XLSX/XLS）。 + - MySQL/PostgreSQL/PolarDB-X 的 SQL 执行能力**严格依赖 DMS 导入方式**，自定义方式创建的连接器仅支持元数据同步，不可执行查询。 + - OSS 连接器的向量检索能力（`searchOSSFile`）需单独开通 [向量检索服务](https://help.aliyun.com/zh/oss/user-guide/vector-retrieval/)，否则相关工具不可用。 ## 来源文档 - [数据连接](../../raw/application-user-guide/data-connection-overview/data-connection.md) - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md index 3781b4b1..8e2fe727 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/fine-tuning.md @@ -1,77 +1,57 @@ # fine tuning -模型微调（Fine-tuning）是阿里云百炼在 Prompt 工程、插件调用等手段仍无法满足效果时提供的深度定制手段。它覆盖文本生成、视觉理解（Qwen-VL）、图像/视频生成（万相）以及语音合成（CosyVoice）等多种模态，通过 SFT、CPT、DPO 等训练方式，把领域知识、任务能力、人类偏好或特定音色/风格直接写入模型参数。 +微调（Fine-tuning）是阿里云百炼平台提供的核心模型优化能力，通过在基础模型上注入领域知识、业务指令或人类偏好，显著提升模型在特定任务上的准确性、安全性与风格一致性。它适用于文本生成、图像生成、视频生成及语音合成等[多模态](../concepts/multi-modal.md)场景，支持高效微调（LoRA）与全参训练两种模式，兼顾效果与成本。 -> **注意**：以下所有微调、部署与调用能力均**仅在华北2（北京）地域可用**，且必须使用该地域的 API Key；子账号（RAM 用户）需预先被授予调用、训练和部署权限。 +## 支持的模型与功能 -## 支持的模型与训练方式 +百炼平台支持对多种模态模型进行微调，覆盖文本、视觉、语音和视频生成任务。所有微调任务均需在**华北2（北京）地域**执行，并使用该地域的 API Key [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -按模态划分，不同模型支持的训练方式差异明显： +- **文本生成**：支持 Qwen 系列大语言模型（如 `qwen3-8b`、`qwen2.5-7b-instruct`）的 SFT、CPT 和 DPO 训练；支持千问 VL 视觉理解模型的[多模态](../concepts/multi-modal.md) SFT [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 +- **图像生成**：仅支持万相系列模型（`wan2.7-image-pro`、`wan2.7-image`），采用 SFT-LoRA 方式，适用于文生图与图生图任务 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **视频生成**：支持 `wan2.7-i2v`、`wan2.2-kf2v-flash` 等万相视频模型，同样基于 SFT-LoRA，支持首帧/首尾帧驱动的特效定制。 +- **语音合成**：仅支持 `cosyvoice-v3-flash` 模型的 SFT 高效微调，用于高还原度专属音色定制，**控制台暂不支持，必须通过 API 发起** [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 -- **文本生成（千问系列）**：支持 CPT、SFT（全参 `sft` / 高效 `efficient_sft`）、DPO（全参 `dpo_full` / 高效 `dpo_lora`）。是否支持某种方式因模型而异，例如 Qwen3-32B、Qwen3-4B/1.7B/0.6B、Qwen2.5 系列支持全部 5 种；而 Qwen3.5-Plus/Flash、Qwen3.6/3.7 等新模型往往仅支持 `sft`。详见 [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md)。 -- **视觉理解（千问 VL）**：Qwen3-VL、Qwen2.5-VL 系列支持 SFT 全参与高效训练，不支持 CPT/DPO。 -- **图像生成（万相）**：`wan2.7-image-pro`、`wan2.7-image`，仅支持 SFT-LoRA 高效微调，见 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 -- **视频生成（万相）**：图生视频-基于首帧 `wan2.7-i2v`/`wan2.5-i2v-preview`/`wan2.2-i2v-flash`，基于首尾帧 `wan2.2-kf2v-flash`，同样仅支持 SFT-LoRA，见 [微调视频生成模型](../../raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md)。 -- **语音合成（CosyVoice）**：`cosyvoice-v3-flash`，仅支持 `efficient_sft`，且**当前只能通过 API 发起，控制台暂不支持**，见 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +> **注意**：文档 4 中称“阿里云百炼推荐您如果**模型支持全参训练，请优先选择全参训练**，因为全参训练效果比高效训练效果要好”，但文档 1、2、6 明确限定图像、视频、语音微调**仅支持 `efficient_sft`**（即 LoRA），且未提供全参训练选项。因此，对于非文本类模型，高效训练是唯一可用方式，不存在“优先选择”问题。 -## 三种调优方式（文本生成） +## 关键参数 -推荐按递进顺序组合使用：`CPT（可选）→ SFT → DPO（可选）`。 +不同模态模型的超参数命名与语义存在差异，开发者需按任务类型严格匹配： -| 方式 | 目标 | 数据量 | 数据形态 | -| --- | --- | --- | --- | -| CPT（持续预训练） | 补领域知识 | 1000 万+ Token | 无标签领域文本 `{"text":"..."}` | -| SFT（监督微调） | 学会遵循指令 | 1000+ 条 | ChatML「问-答」对 | -| DPO（直接偏好优化） | 对齐人类偏好 | 100+ 组 | 同指令下「更好/更差」回答对（`chosen`/`rejected`） | +| 参数名 | 文本模型（SFT） | 图像模型（wan） | 视频模型（wan） | 语音模型（CosyVoice） | 说明 | +|--------|----------------|-----------------|------------------|------------------------|------| +| `training_type` | `sft`, `efficient_sft`, `dpo_lora` | 固定为 `efficient_sft` | 固定为 `efficient_sft` | 固定为 `efficient_sft` | 必填，指定训练方法 | +| `learning_rate` | 默认 `3e-4`（高效训练）或 `1e-5`（全参） | `3e-5`（文生图）、`2e-5`（图生视频） | `2e-5` | 不直接暴露，由 `lm_max_epoch`/`fm_max_epoch` 间接影响 | 学习率过高易发散，过低收敛慢 | +| `max_steps` / `n_epochs` | `n_epochs`（循环次数），默认 `3` | `max_steps`（总步数），如 `800` | `n_epochs`（轮次），如 `50` | `lm_max_epoch` & `fm_max_epoch`（双网络轮次） | 控制训练强度的核心参数；图像用步数，其余多用轮次 | +| `batch_size` | 默认 `16` | 未显式暴露（由系统自动适配） | `1`（wan2.7-i2v）或 `4`（其他） | `lm_batch_size` / `fm_batch_size`（如 `1000`/`2000`） | 批次大小直接影响内存占用与训练稳定性 | +| `lora_rank` | 默认 `8` | `32`（必须为 2 的幂） | `32` | 不适用（CosyVoice 使用专用 LM/FM 架构） | LoRA 低秩矩阵维度，值越大拟合能力越强，但易过拟合 | +| `lora_alpha` | 默认 `16` | 未提及 | `32` | 不适用 | LoRA 缩放因子，控制修正项权重 | -训练模式分**全参训练**与**高效训练（LoRA）**：两者费用相同，官方建议在模型支持全参训练时优先选择全参（效果更好、性价比更高）；LoRA 适合对训练时间/成本敏感或数据集较小的场景。 - -## 关键超参数 - -文本生成调优的常用超参及默认值（以控制台实际显示为准）： - -- `learning_rate`：高效训练建议 `1e-4` 量级，全参/CPT 建议 `1e-5` 量级。 -- `n_epochs`：默认 `3`，范围 `[1, 200]`；数据量 <10000 建议循环 3~5 次，>10000 建议 1~2 次。 -- `batch_size`：一般 16/32。 -- `max_length`：建议设为模型支持的最大值；SFT 会**丢弃**超长数据，DPO 则**截断**后仍训练。 -- `lora_rank` / `lora_alpha` / `lora_dropout`：LoRA 专用，秩越大效果略好但更慢、更易过拟合。 -- 通过 API 创建任务时，`n_epochs`、`batch_size`、`max_length` 因影响计费而**必填**。 - -> **注意**：默认学习率各文档取值不一致。控制台参数面板列出的 `learning_rate` 默认值为 `3e-4`（对应高效训练默认场景），而 API 示例中 SFT 全参使用的是 `1.6e-5`。请以实际训练方式对应的量级为准，切勿照搬。 - -万相图像/视频与 CosyVoice 使用各自独立的超参集，例如万相有 `max_steps`/`generation_type`/`val_img_size`，CosyVoice 分 `lm_*`（影响韵律）与 `fm_*`（影响音色）两组网络的 `*_max_epoch`/`*_step`/`*_num`/`*_batch_size`（8 个子字段全部必填）。 +所有模型均需配置 `model`（基础模型 ID）和 `training_datasets`（数据源）。图像/视频模型还需指定 `generation_type`（`t2i`/`i2i`/`i2v`）或 `max_pixels`（分辨率上限）；语音模型则强制要求 `wav_fn` 字段路径以 `train/` 开头。 ## 使用方式 -**控制台（推荐入门）**：在[模型调优](https://bailian.console.aliyun.com/?tab=model#/efm/model_manager)页面创建训练任务 → 选训练方式与模型 → 配置训练集/验证集（可自动切分）→ 配置 Checkpoint 保存 → 开始训练 → 部署 → 评测。零代码场景可参考 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)，其中给出了以 Qwen3-8B 为例的完整安全对齐 SFT 流程与超参实验对照（全参 `n_epochs=3`/`lr=1e-5` 或 LoRA `n_epochs=3`/`lr=3e-4` 效果较好）。 - -**API / 命令行**：统一四步流程，详见 [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)： +微调流程统一为四步：**准备数据 → 上传文件 → 创建任务 → 部署调用**，但各模态入口与细节不同： -1. 上传数据集到 `POST /api/v1/files`（`purpose=fine-tune`），获取 `file_id`。 -2. `POST /api/v1/fine-tunes` 创建任务，关注返回的 `job_id`、`finetuned_output`、`status`。 -3. 轮询 `GET /api/v1/fine-tunes/` 直到 `status` 变为 `SUCCEEDED`。 -4. `POST /api/v1/deployments` 部署（`plan=lora`），轮询直到 `status` 为 `RUNNING`，再用 `deployed_model` 调用。 +- **控制台操作**：适用于文本与视觉理解模型。进入[模型调优](https://bailian.console.aliyun.com/?tab=model#/efm/model_manager)页面，选择模型、训练方式（SFT/CPT/DPO）、上传数据集（ZIP 或 OSS 路径），配置超参后启动 [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md)。 +- **API 操作**：通用方式，**语音模型强制要求**。先调用 `/api/v1/files` 上传 ZIP 数据包（`purpose="fine-tune"`），获取 `file_id`；再调用 `/api/v1/fine-tunes` 提交训练任务，`training_datasets` 中引用该 `file_id` [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md)。 +- **部署与调用**：训练成功（`status=SUCCEEDED`）后，调用 `/api/v1/deployments` 部署模型，获得 `deployed_model` 名称；调用时需将 `model` 字段设为该名称，并遵循对应模态的输入格式（如图像微调需含触发词 `s86b5p`，语音微调固定 `voice="default"`）。 -> **注意**：通过 API 创建的训练任务**仅支持按 Token 计费**，不支持模型训练单元（预付费/后付费）；如需使用训练单元，必须通过控制台创建。 +> **注意**：文档 2 中视频微调部署请求包含 `aigc_config` 字段（如 `lora_prompt_default`），而文档 1 图像微调部署请求无此字段；文档 5 的通用 API 文档也未要求该字段。这表明 `aigc_config` 是视频模型特有的部署配置，开发者需按模型类型查阅对应指南。 -数据集除 `file_id` 外还可用 OSS 挂载（`data_source_type=oss_mount`），OSS Bucket 地域支持 `cn-beijing` 与 `ap-southeast-1`，挂载时只需指定 `data.jsonl` 路径。 +## 限制和注意事项 -## 数据格式要点 - -- **SFT**：ChatML `{"messages":[...]}`，支持多轮；不支持 OpenAI 的 `name`/`weight`，所有 assistant 行都会被训练；思考模型（thinking）只训练**最后**一个 assistant 输出且须保留 `` 标签前后的换行。 -- **DPO**：在 `messages` 基础上追加 `chosen`/`rejected`。 -- **CPT**：纯文本 `{"text":"..."}`。 -- **视觉理解**：`content` 用数组，`image`/`video` 声明文件名（不含路径），打包为 ZIP（≤2GB），`data.jsonl` 必须在根目录，文件名全局唯一且仅含 ASCII。 -- **CosyVoice**：`data.jsonl` 每行 `{"wav_fn":"train/xxx.wav","text":"..."}`，`wav_fn` 必须以 `train/` 前缀；`text` 须为纯文本，禁止 SSML/LaTeX/情感标注。 - -## 限制与注意事项 - -- **成本与耗时高**：文本模型微调需构建大规模数据集，且调优后模型**必须部署才能使用**，部署费用较高；官方明确将模型调优定位为「最后的手段」，建议先充分尝试 Prompt 工程与插件调用。 -- **训练耗时差异大**：万相文生图约数十分钟，视频微调可达数小时；文本 LoRA 通常 15~30 分钟；部署一般需 3~10 分钟。 -- **计费方式各异**：文本/VL 按训练 Token 计费（`训练Token × 循环次数 × 单价`）；CosyVoice 按 `(lm_max_epoch+fm_max_epoch)×25×音频总秒数` 估算 Token，单价 0.2 元/千 Token，另加部署时长费用。 -- **能力边界不可突破**：CosyVoice 调优产物为单音色模型（`voice` 锁定 `default`），无法新增基础模型不支持的语种、也不支持 `instruction` 指令控制。 -- **过拟合/欠拟合判断**：观察 Training/Validation Loss 曲线，欠拟合可增大 `n_epochs`/`lora_rank`，过拟合则反向调整。 -- 万相 LoRA 调用需在提示词中包含**触发词**以激活风格；图像模型部署后当前仅支持异步调用。 +- **地域与权限**：所有微调服务仅限华北2（北京）地域，子账号需被授予 `AliyunBailianFullAccess` 或精细化的 `dashscope:CreateFineTuneJob` 等权限 [微调图像生成模型](../../raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md)。 +- **数据规范**： + - 文本/视觉数据必须为 ZIP 包，根目录含 `data.jsonl`，图片尺寸 ≤1024px； + - 语音数据 ZIP 内 `data.jsonl` 的 `wav_fn` 必须以 `train/` 开头； + - 视频微调数据集需严格按 `i2v` 或 `kf2v` 格式组织，不可混用。 +- **成本与耗时**： + - 计费按训练 [Token](../concepts/token.md) 总量 × 单价（如文本 `¥0.006/千Token`，语音 `¥0.2/千Token`）； + - 图像微调约 77 分钟（2K, 300 步），视频微调需“数小时”，语音微调最小化超参约 37 分钟 [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md)。 +- **效果边界**： + - 微调无法扩展基础模型能力（如 CosyVoice 不能通过微调支持新语种）； + - 过度增加 `n_epochs` 或 `lora_rank` 可能导致基础能力“遗忘”或过拟合； + - 安全合规微调需高质量拒答样本，单纯 [Prompt 工程](../concepts/prompt-engineering.md)无法替代参数层面的对齐 [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md)。 ## 来源文档 @@ -80,8 +60,7 @@ - [模型调优简介](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-overview.md) - [在控制台进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/model-training-on-console.md) - [使用 API 或命令行进行模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/fine-tuning-api-guide.md) -- [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - [CosyVoice模型调优](../../raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - +- [0 代码强化大模型安全合规能力](../../raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md index 56974632..a1f1f09a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/get-started-with-models.md @@ -1,73 +1,84 @@ # get started with models -阿里云百炼是一站式大模型开发与应用平台，集成千问（Qwen）全系列及 DeepSeek、Kimi、GLM 等主流第三方模型，并提供兼容 OpenAI 的 API。本页面帮助开发者快速完成从选择模型、获取 API Key、配置 Base URL 到发起首次调用的全流程，并梳理地域、接入域名与限流等关键约束。 +阿里云百炼提供开箱即用的大模型服务，支持通过 OpenAI 兼容 API、DashScope SDK 等方式快速调用千问（Qwen）及第三方模型。开发者无需部署运维，只需获取 API Key、配置 Base URL 并指定模型 ID 即可发起首次推理请求。本文聚焦核心接入路径，涵盖模型选择、参数配置、调用方式及关键约束。 -## 支持的模型与能力 +## 支持的模型与功能 -百炼提供开箱即用的模型服务，无需自行部署或运维即可调用。文本生成方面，千问旗舰模型按能力与成本分层选择（参见 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md)）： +百炼提供覆盖文本、[多模态](../concepts/multi-modal.md)及领域专用的全系列模型，其中千问（Qwen）为核心自研模型： -- **千问 Max**（如 `qwen3.7-max`）：Qwen 系列效果最好的模型，适合复杂、多步骤任务。 -- **千问 Plus**（如 `qwen3.7-plus`）：效果、速度和成本均衡，多数场景的**推荐选择**。 -- **千问 Flash**（如 `qwen3.6-flash`）：高性价比、低延迟，适合需要快速响应的简单任务。 +- **主力推荐模型**：`qwen3.7-plus`（效果、速度、成本均衡），`qwen3.7-max`（复杂任务首选），`qwen3.6-flash`（高并发低延迟场景）[什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) +- **模型可用性差异**：不同地域支持的模型集合不同。例如 `qwen3.7-max-preview` 仅限 [Token](../concepts/token.md) Plan 用户且仅在北京地域可用；DeepSeek 模型目前仅支持华北2（北京）地域 [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) +- **功能覆盖**：除基础文本生成外，还支持视觉理解、图像生成、语音合成、嵌入向量、长文本处理、法律/意图/角色扮演等细分领域模型 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) -此外还覆盖视觉理解、图像生成、视频生成、语音识别与合成、嵌入向量等多模态能力，以及长文本、翻译、法律等细分领域模型。平台同时支持模型调优（SFT / CPT / DPO）、模型部署与模型评测，详见 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md)。 +> **注意**：文档 3 中列出的 `qwen3.8-max-preview` 在文档 5 的限流表格中未出现，且文档 5 明确标注该模型“仅 [Token](../concepts/token.md) Plan 可用”，而文档 2 未提及此限制。实际使用前请以控制台实时模型列表为准。 -## 关键概念与参数 +## 关键参数 -调用前需要先确定四个维度（详见 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md)）： +### API Key +- 必须通过[阿里云百炼控制台 → API Key 页面](https://bailian.console.aliyun.com/?tab=model#/api-key)创建，**不可跨地域复用** +- 建议配置为环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露风险 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) -- **地域（Region）**：决定接入点和数据存储位置。目前提供华北2（北京，`cn-beijing`）、新加坡（`ap-southeast-1`）、日本（东京，`ap-northeast-1`）、德国（法兰克福，`eu-central-1`）、美国（弗吉尼亚，`us-east-1`）。就近选择可降低延迟。 -- **服务部署范围**：决定推理执行位置。有数据合规需求时选择特定地理边界（如中国内地、欧盟、美国），无合规需求可选全球部署（推理资源池更大）。德国、日本地域通过[业务空间（Workspace）](../concepts/workspace.md)区分部署范围；美国地域可用带 `-us` 后缀的模型名（如 `qwen-plus-us`）限定境内推理。 -- **接入域名**：影响并发上限、超时等服务保障，推荐使用**业务空间专属域名**（`{WorkspaceId}.{region}.maas.aliyuncs.com`，SLA 99.9%、请求超时 3600 秒、支持 HTTP/SSE/WebSocket/WebRTC），另有 Dashscope 域名（现有，超时 600 秒）和试用域名（限流小，不建议生产）。 -- **API Key**：各地域相互独立、不能跨地域混用；Base URL 也必须与同一计费方案的 API Key 配套使用，否则报 401。 +### Base URL +- **业务空间专属域名（生产推荐）**：`https://{WorkspaceId}.{region}.maas.aliyuncs.com/compatible-mode/v1`，需先在[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)获取 WorkspaceId +- **DashScope 共享域名（兼容存量）**：如 `https://dashscope.aliyuncs.com/compatible-mode/v1`（北京）、`https://dashscope-us.aliyuncs.com/compatible-mode/v1`（美国） +- **试用域名（非生产）**：`https://trial.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，RPM 限流严格（1000） +- 各域名鉴权范围不同：业务空间专属域名仅允许对应 Workspace 的 API Key 调用，而 DashScope 域名支持跨 Workspace [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) -> **注意**：各地域接入点（Base URL）、API Key 和模型列表均不能跨地域混用；使用北京、新加坡、日本、德国地域时，业务空间专属域名中的 `{WorkspaceId}` 需替换为真实业务空间 ID（可在业务空间管理页查看）。 +### 模型 ID +- 必须与所选地域和 Base URL 匹配。例如 `qwen3.7-plus-us` 仅适用于美国（弗吉尼亚）地域的 DashScope 域名，而 `qwen3.7-plus` 在北京地域需配合业务空间专属域名使用 +- 部分模型带时间后缀（如 `qwen-plus-2025-07-28`），其限流额度显著低于稳定版（见下文限制部分） ## 使用方式 -**1. 账号与凭证准备**（参见 [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md)）：注册并开通百炼 → 在 API Key 页面创建 Key → 获取业务空间 ID。建议将 Key 配置到环境变量 `DASHSCOPE_API_KEY`，避免硬编码泄露。 - -**2. 选择 Base URL**（详见 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md)）：百炼提供 OpenAI 兼容（`/compatible-mode/v1`）、Anthropic 兼容（`/apps/anthropic`）、DashScope（`/api/v1`）三类接口。以北京地域业务空间专属域名的 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)为例：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`。 - -**3. 发起调用**：兼容 OpenAI 接口规范，迁移现有代码只需调整 API Key、base_url 和模型名称。Python 示例： - -```python -import os -from openai import OpenAI - -client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen-plus", - messages=[ - {'role': 'system', 'content': 'You are a helpful assistant.'}, - {'role': 'user', 'content': '你是谁？'} - ] -) -print(completion.choices[0].message.content) -``` - -也可通过 DashScope SDK（`pip install -U dashscope`）、curl，或 Chatbox、Claude Code 等客户端/开发工具调用。OpenAI Python SDK 要求 Python ≥ 3.8。 - -## 限制与注意事项 - -- **限流按主账号维度合并计算**：账号下所有 RAM 子账号、业务空间和 API Key 的调用量合并统计，不同模型限流额度相互独立。超限请求被拒绝，通常一分钟内自动恢复。详见 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md)。 -- **RPM 与 TPM 双重约束**：`Requests rate limit exceeded` 表示触发每分钟请求数（RPM）限流；`Allocated quota exceeded` 表示触发每分钟 Token 数（TPM）限流；`Request rate increased too quickly` 表示请求瞬时激增触发稳定性保护。限流可能按秒级 RPS（RPM/60）、TPS（TPM/60）执行。 -- **规避限流**：优先选用高限流额度模型（稳定版比日期快照版更宽松）、平滑请求速率（匀速/指数退避/队列）、配置备选模型自动切换、拆分任务、无需实时响应时改用 Batch API（不受实时限流约束）。北京与新加坡地域支持在控制台申请临时 TPM 提额（生效 30 天）。 -- **计费独立**：模型推理按 Token 用量计费，知识库（RAG）按规格时长与调用独立计费，两者互不相通。限流只约束速率、不限制累计用量；如需控费可设置费用告警、开启"免费额度用完即停"或订阅 Coding Plan（固定月费）。 -- **各地域功能差异**：批量推理、模型调优、应用开发等能力目前主要在华北2（北京）支持，海外地域功能覆盖较少，选型前请核对目标地域的功能与模型列表。 - -> **注意**：Token Plan 与 Coding Plan 的 Base URL 及专属 API Key 仅限 Claude Code、Codex 等 AI 工具交互式使用，不能用于后端服务；使用非专属 Base URL 调用将按量付费。 +### 基础调用流程 +1. **开通服务**：使用阿里云主账号登录[百炼控制台](https://bailian.console.aliyun.com/?tab=model#/model-market)，同意协议开通服务 +2. **获取凭证**：创建 API Key，并在业务空间管理页获取 WorkspaceId（若使用专属域名） +3. **配置环境**：将 `DASHSCOPE_API_KEY` 设为环境变量（Linux/macOS/Windows 均有详细指南）[首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) +4. **发起请求**：使用 OpenAI SDK 或 DashScope SDK，示例（OpenAI 兼容）： + ```python + from openai import OpenAI + client = OpenAI( + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" + ) + response = client.chat.completions.create( + model="qwen3.7-plus", + messages=[{"role": "user", "content": "你是谁？"}] + ) + ``` + +### 多语言支持 +- 官方提供 Python、Node.js 示例，curl 命令亦可直接验证 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) +- 所有 OpenAI 兼容客户端（如 LangChain、LlamaIndex）均可通过调整 `base_url` 和 `api_key` 迁移使用 + +## 限制和注意事项 + +### 限流策略 +- **账号级聚合限流**：主账号下所有子账号、业务空间、API Key 的调用量合并计算 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- **双维度限制**：每分钟请求数（RPM）和每分钟 [Token](../concepts/token.md) 消耗（TPM）任一超限即触发 429 错误 +- **典型额度对比**（北京地域）： + - `qwen3.7-plus`：RPM 30,000 / TPM 5,000,000 + - `qwen-plus-2025-07-28`：RPM 60 / TPM 1,000,000 + - `qwen-long-2025-01-25`：RPM 3 / TPM 7,500 +- **临时提额**：可在控制台[限流提额页面](https://bailian.console.aliyun.com/?tab=model#/efm/temp_limit_raise)申请提升 TPM，有效期 30 天 [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) + +### 地域与域名约束 +- **地域隔离**：API Key、Base URL、模型列表均按地域独立，混用导致 401 错误 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) +- **域名适配**：业务空间专属域名要求 API Key 必须属于该 Workspace；DashScope 域名支持跨 Workspace，但不提供业务级流量隔离 [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) +- **功能差异**：批量推理（Batch API）仅在北京、新加坡地域支持；模型调优、应用开发等功能在北京地域独有 [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) + +### 其他关键约束 +- **免费额度**：新用户在北京地域享有专属免费额度，用尽后认证用户自动转按量付费，未认证用户需完成实名认证 [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) +- **数据合规**：若要求数据不出中国内地，必须选择华北2（北京）地域 + “中国内地”服务部署范围；国际业务推荐新加坡（国际）或美国（全球） [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) +- **Token Plan 限制**：Token Plan 专属 API Key 仅限 Claude Code 等交互式工具使用，**不可用于后端服务调用** [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) ## 来源文档 +- [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [什么是阿里云百炼](../../raw/model-user-guide/get-started-with-models/what-is-model-studio.md) - [选择模型](../../raw/model-user-guide/get-started-with-models/models.md) -- [首次调用千问API](../../raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) -- [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) -- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) - [Base URL总览](../../raw/model-user-guide/get-started-with-models/base-url.md) +- [限流](../../raw/model-user-guide/get-started-with-models/rate-limit.md) +- [选择地域、服务部署范围和接入域名](../../raw/model-user-guide/get-started-with-models/regions.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md index 8f2e139a..46ff8a40 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/knowledge-base.md @@ -1,91 +1,55 @@ # [knowledge](../api/knowledge.md) base -知识库（Knowledge Base）是阿里云百炼平台基于 RAG（检索增强生成）技术为大模型补充私有数据和最新信息的能力。大模型在生成回答前先从知识库中检索语义相关的内容，从而显著提升在特定领域问题上的准确性。围绕知识库，平台还提供了知识检索、知识问答、API 集成、日志监控与计费等一整套配套能力。 +知识库（Knowledge Base）是阿里云百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，用于为大模型注入私有、结构化或非结构化数据，提升其在垂直领域回答的准确性与时效性。它通过索引构建、语义检索与结果重排三阶段 pipeline 实现高效知识召回，并支持文档搜索、数据查询、图片问答、音视频搜索等多种类型。知识库需部署于华北2（北京）地域，且所有操作均基于业务空间隔离。 -> **注意**：知识库功能仅能在中国站 **华北2（北京）** 地域开通和使用，新加坡、德国（法兰克福）等其他地域均不支持。 +## 支持的模型/功能 -## 支持的模型与知识库类型 +知识库本身不直接运行模型，但其检索流程深度依赖以下模型能力，并与多种大模型协同工作： -预置模型（千问-QwQ/Long/Max/Plus/Turbo/Coder、千问VL 系列、Qwen 开源版，以及 DeepSeek-R1/V3.1、Llama3.1 等第三方文本生成模型）和部分调优后的自定义模型均可挂载知识库。具体可选模型以[应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center)页面创建应用时实际可选项为准。 +- **向量模型**：文档搜索类知识库默认使用 `text-embedding-v4` 或 `text-embedding-v3`（512维）；图片问答与音视频搜索类必须使用 `qwen3-vl-embedding`（1024维）[原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 +- **排序模型（Rerank）**：文档类支持 `qwen3-rerank`（含 hybrid 模式），[多模态](../concepts/multi-modal.md)类支持 `qwen3-vl-rerank`，可选关闭以降低成本 [原文标题](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)。 +- **路由模型**：当应用挂载多个知识库并启用路由时，调用 `qwen-plus` 判断目标知识库，产生独立 [Token](../concepts/token.md) 费用 [原文标题](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)。 +- **问答模型**：知识问答服务中由用户自主选择（如 `qwen3.7-plus`），费用按输入/输出 [Token](../concepts/token.md) 单独计费，不包含在知识库规格费中。 -创建知识库时按场景选择类型（创建后不可更改）： - -- **文档搜索**：企业内部文档、产品手册等非结构化数据检索。可细分为基础文档问答、图文并茂回复、视觉理解（富文本文档）、极速问答四种使用场景。选择视觉理解后向量模型自动切换为 qwen3 多模态向量（qwen3-vl-embedding），不可更改。 -- **数据查询（表格库）**：结构化 Excel/CSV，单库仅支持 1 篇。 -- **图片问答类**：仅支持 multimodal-embedding-v1 向量模型。 -- **音视频搜索类**：支持语音识别、视频帧提取与剧情解析。 - -不同解析方式（电子文档解析、文档智能解析、大模型文档解析、Qwen VL 解析、音视频解析）在速度与图表理解能力上有明显差异，详见[知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +> **注意**：文档 8 中列出的“预置模型”（如 QwQ/Long/Max 等）仅表示**可绑定知识库的大模型列表**，并非知识库自身使用的模型；而文档 4 明确规定向量模型与排序模型的选择范围及维度限制，二者存在功能层级差异，不可混淆。 ## 关键参数 -知识库的检索效果主要由以下参数决定，在命中测试、检索服务和问答服务中可反复调优： +知识库效果高度依赖以下可配置参数，需结合场景权衡精度与成本： -- **相似度阈值（0.01~1.0）**：仅语义相似度高于阈值的切片会被召回。阈值过高会导致相关切片被全部丢弃（例如调至 0.60 可能返回无召回结果）。 -- **初步向量检索 TopK / 初步关键词检索 TopK（1~100，默认各 50）**：控制初步召回的切片数量，直接影响送入 Rerank 模型的 Token 量与成本。 -- **最大召回数量 / 召回片段数（1~20）**：即多路召回的 K 值，最终提供给大模型的切片数。对总结、列举、比较类复杂问题应适当调大。 -- **权重**：多知识库联合召回时用于干预排序，但**仅在同类型知识库之间生效**。 -- **排序模型（Rerank）**：纯文本可选 qwen3-rerank、qwen3-rerank(hybrid)；多模态可选 qwen3-vl-rerank。支持问答模式与相似模式。 -- **Meta 信息抽取与标签过滤**：通过元数据（常量/变量/大模型/正则/关键词方式提取）和标签在向量检索前做结构化筛选，精准定位目标文件。注意元数据只能在创建时配置，创建后无法再开启。 +| 参数 | 取值范围 | 说明 | 关联环节 | +|------|----------|------|----------| +| `初步向量检索 TopK` | 1–100 | 向量召回阶段返回的切片数，默认 50；直接影响 Rerank 模型 [Token](../concepts/token.md) 消耗量 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) | 检索 | +| `初步关键词检索 TopK` | 1–100 | 关键词召回阶段返回的切片数，默认 50；混合检索时与向量结果合并去重 | 检索 | +| `相似度阈值` | 0.01–1.0 | 过滤 Rerank 后低分切片；值过高易漏召，过低引入噪声 | 检索 | +| `最大召回数量` | 1–20 | 最终返回给大模型的切片数上限；影响下游 Token 成本与回答完整性 | 检索 | +| `标签过滤` | — | 通过 `tags` 参数限定检索范围，支持单标签、多标签“或”/“与”逻辑 [原文标题](../../raw/application-user-guide/knowledge-base/rag-optimization.md) | 检索 | +| `Meta信息抽取` | — | 在创建知识库时配置，支持常量、变量、大模型、正则、关键词五种提取方式；**创建后不可修改** | 索引 | ## 使用方式 -**控制台快速构建**：进入[知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base)选择规格后，按"填写基础信息 → 配置数据来源 → 设置索引参数"三步完成创建，随后可关联到智能体应用、工作流应用或外部应用。工作流应用需将知识库节点接在开始节点之后、大模型节点之前，并在大模型提示词中引用 `result` 变量。 - -**知识检索服务**：面向多知识库联合检索（最多 15 个），提供 Query 改写、混合检索（向量+关键词）、Rerank 排序的流水线，支持知识库路由、混排模型模式等全局与单库独立参数配置，详见[知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md)。 - -**知识问答服务**：在检索基础上由大模型（如 qwen3.6-plus）生成自然语言回答，提供**极速模式**（单轮检索+生成）和**多轮智能模式**（Agentic 多轮规划搜索），并支持文件预解析、拒答、防泄漏、多模态回复、引用来源等生成控制。 - -**API/SDK 集成**：通过[阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)调用，典型创建流程为：申请上传租约（ApplyFileUploadLease）→ 上传文件 → AddFile → CreateIndex → SubmitIndexJob → 轮询 GetIndexJobStatus。子账号需先获取 AliyunBailianDataFullAccess 策略并加入业务空间，详见[知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 - -## 效果优化 - -RAG 效果由建立索引、检索召回、生成答案三个阶段决定。优化前建议用[自动评测](https://help.aliyun.com/zh/model-studio/application-auto-evaluation)建立至少 100 组用例的评估基线，再针对失败用例（打分 < 4）诊断改进： - -- **检索无效（没找到）**：补充知识、优化源文件排版（推荐转 Markdown、移除水印、避免复杂表格）、统一实体表述、启用多轮对话改写。 -- **召回不相关**：使用标签过滤或元数据做结构化搜索。 -- **切片不完整**：采用"智能切分"（基于语义自适应切分），并人工检查修正异常切片。 -- **重排不佳**：调整相似度阈值与召回片段数，在漏召回与噪声之间平衡。 - -## 日志与监控 - -所有检索调用都会以日志形式投递到日志服务（SLS），topic 为 `log_dispatch`，包含 `request_id`、`pipeline_id`（知识库 ID）、`workspace_id`、`latency`、`response_status_code`、`response_code`、`request_body`、`response_body` 等索引字段，可用于调用审计、用量统计、慢查询与错误率监控。首次使用需在知识库列表页的**监控配置**中授权 SLS 角色、开通日志服务并创建 LogStore。SLS 存储与流量单独计费，关闭检索日志开关只停止新投递，历史日志仍保留计费，需彻底停止请到 SLS 删除对应 LogStore。 +知识库可通过控制台、API 或集成至应用三种方式使用： -## 限制与配额 +- **控制台快速接入**：在知识库页面创建后，直接绑定至智能体应用（配置权重与相似度阈值）或工作流应用（拖拽知识库节点），无需编码即可启用 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md)。 +- **API 集成**：调用 `Retrieve` 接口进行单次检索；需子账号具备 `AliyunBailianDataFullAccess` 权限，且**仅支持华北2（北京）地域** [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md)。 +- **日志与监控**：开通 SLS 日志服务后，每条检索请求生成完整日志（含 `request_body`、`response_body.data.nodes[]` 等字段），可用于审计、召回分析与性能排查 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md)。 -| 类别 | 上限 | -| --- | --- | -| 知识库数量 | RDS 数据源 100，其它数据源无限制 | -| 存储容量 | 旗舰版 9,999 GB / 标准版 100 GB | -| 单个文档搜索类知识库文件数 | 无硬性上限（数据查询类仅 1 篇） | -| 单次控制台导入文件数 | 50（API 批量建议单次 ≤ 10,000） | -| 单文件标签数 | 32 | -| 文本切片长度 | 6,000 Token | -| 召回文本切片数量 | 20 | -| 检索并发 | 旗舰版 50-10,000 QPS（1-200 RCU）/ 标准版 1 QPS 固定 | +## 限制和注意事项 -文件格式限制：pdf/docx/ppt 等最大 150MB 且页数 ≤ 1,000；txt/markdown/html 最大 10MB；图片最大 20MB；音视频最大 512MB。向量模型仅支持 text-embedding-v3/v4（512 维）与 multimodal-embedding-v1（1024 维），维度不可更改。完整清单见[知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 - -## 计费注意事项 - -知识库服务自 **2026 年 1 月 4 日**起正式计费，总费用由**规格费用**（运行时长）和**模型调用费用**（向量化 + Rerank）两部分构成，扣费顺序为免费额度 > 资源包 > 按量付费。 - -- 规格费用：标准版 0.03 元/知识库/小时；旗舰版 0.2 元/RCU/小时（1 RCU ≈ 50 QPS）。 -- 模型调用费用独立计费，公式为 `(输入 Token 总数 / 1000) × 模型单价`。**Rerank 费用取决于初步召回的总切片数，而非最终返回数**，因此降低初步 TopK 或关闭排序可显著省钱。 -- 应用挂载多个知识库时，模型 Token 消耗按知识库数量倍增（N 个库则 × N）。 -- 平台提供一次性 720 小时免费额度（仅抵扣标准版规格费用，不含模型调用费）。删除知识库以停止计费，但删除会**永久清除数据且无法恢复**。 - -> **注意**：免费额度有效期存在新老用户差异——老用户统一截至 2026 年 2 月 3 日 23:59，新用户自开通起 30 天内有效，逾期作废。此外《知识库计费说明》以 2026 年计费规则描述，而《知识库》文档仍按旧的即时计费口径（0.03/0.2 元/小时）介绍创建流程，接入时以[知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)为准。 +- **地域限制**：知识库功能**仅在中国站华北2（北京）地域可用**，其他地域（如新加坡、法兰克福）完全不支持，此限制在文档 2 和文档 8 中一致强调。 +- **存储与配额**：标准版免费额度仅抵扣**标准版知识库规格费用**，不覆盖模型调用费；旗舰版存储上限 9,999 GB，标准版为 100 GB；单个知识库文件数量无硬性上限（非结构化类），但单次控制台上传上限为 50 个文件 [原文标题](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md)。 +- **元数据与切片**：`Meta信息抽取` 必须在创建知识库时完成配置，**创建后无法追加或修改**；文本切片长度上限为 6,000 Token，编辑切片内容长度限制为 10–6000 字符。 +- **计费关键点**：Rerank 费用取决于**初步召回总切片数**（即 `TopK` 之和），而非最终返回数；多知识库联合检索时，Query 向量化与 Rerank 调用量按知识库数量线性倍增 [原文标题](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md)。 ## 来源文档 -- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) -- [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](../../raw/application-user-guide/knowledge-base/rag-optimization.md) -- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - [知识库API指南](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) +- [知识库日志与监控](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) +- [知识库配额与限制](../../raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - [知识库计费说明](../../raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - [知识检索](../../raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](../../raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) +- [知识库](../../raw/application-user-guide/knowledge-base/rag-knowledge-base.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md index cf82ffab..3c854ccb 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/llm-application.md @@ -1,124 +1,58 @@ # llm application -阿里云百炼平台提供三种核心应用构建模式：智能体（Agent）、工作流（Workflow）和高代码应用，用于突破大模型在私有知识访问、实时信息获取和复杂任务规划方面的原生局限。开发者可根据开发门槛、控制粒度和业务场景选择合适的应用类型，并通过集成知识库、MCP 工具、插件等能力构建完整的 AI 应用。 +百炼平台的 LLM Application 是面向业务场景的 AI 应用构建范式，通过封装大模型能力与外部系统集成能力，支持开发者和业务人员以零代码、低代码或高代码方式快速构建可交付的 AI 服务。其核心价值在于突破大模型在私有知识访问、实时信息获取、流程控制与复杂任务规划等方面的原生局限，提供智能体（Agent）、工作流（Workflow）和高代码应用三种互补的技术路径。 -## 应用类型与选型 +## 支持的模型/功能 -| 对比维度 | 智能体（Agent） | 工作流（Workflow） | 高代码应用 | -|---------|---------------|-------------------|-----------| -| 开发方式 | 自然语言配置（零代码） | 可视化节点编排（低代码） | Python 编码 | -| 核心特点 | AI 自主决策、动态规划 | 预定义流程精确控制 | 完全由代码控制 | -| 适合人群 | 业务人员、产品经理 | IT 运维、业务分析师 | AI 工程师、开发者 | -| 开发门槛 | 低 | 中 | 高 | +百炼 LLM Application 支持三类主流构建模式，各自适配不同抽象层级与开发需求： -详细的类型介绍参见 [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md)。 +- **智能体（Agent）应用**：以提示词驱动，具备自主意图理解、多步规划与工具调用能力。新版智能体（Agent 2.0）将知识库、MCP 等统一为可调度工具，支持完整“规划-执行-反思”链路回溯，适用于开放式对话、动态任务助理等场景 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +- **工作流（Workflow）应用**：基于可视化节点编排（如开始、大模型、意图分类、变量处理、结束等），实现确定性、可复现的多步骤自动化流程，适合报告生成、订单审批、智能导购等固定路径任务 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 +- **高代码应用**：面向专业开发者，支持完整 Python 项目部署为 Serverless 或 K8s 后端服务，内置 MCP 工具接入、可观测性与自定义前端能力，适用于需深度定制、私有算法集成或企业级运维的场景 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 -## [智能体应用](../concepts/agent-application.md) +所有类型均支持文件问答能力，提供**全文引用**、**切片检索（RAG）** 和**自定义处理**三种模式，覆盖文档、图片、音视频等[多模态](../concepts/multi-modal.md)输入 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 -### 新版智能体（Agent 2.0） +> **注意**：文档 3（旧版智能体）与文档 2（新版智能体）存在明确架构不兼容声明：“不支持将旧版智能体升级到新版本”，且新版已将知识库作为工具纳入统一调度体系，而旧版仍将其视为独立能力模块。实际开发应优先采用 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)，避免依赖已弃用的 Agent 1.0 路径。 -新版智能体将知识库、MCP 等能力统一为工具，由智能体自主规划调用顺序，支持完整的"规划-执行-反思"链路展示。推荐在无旧版依赖时使用新版。 +## 关键参数 -核心能力配置： +| 参数类别 | 参数名 | 说明 | 适用场景 | +|----------|--------|------|----------| +| **模型配置** | `temperature` | 控制生成随机性，值越高越发散 | 所有应用类型通用 | +| | `enable_thinking` | 是否开启思考模式（仅支持模型可用） | 新版智能体中用于增强反思能力 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) | +| | `ReAct 最大轮次`（1–50） | 单次会话中工具调用最大次数 | 新版智能体防止无限循环 | +| **文件处理** | `单文件最大解析长度（token）` | 全文引用模式下截断位置（从末尾） | 防止超上下文，见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) | +| | `召回片段数` / `最大拼装长度` | 切片检索模式下控制 RAG 输入规模 | 平衡精度与 [Token](../concepts/token.md) 成本 | +| **会话控制** | `短期记忆轮数`（0–30） | 多轮对话中保留的历史轮数 | 新版智能体，0 表示禁用上下文 | +| | `historyList` 变量 | 工作流中预置的全局对话历史变量 | 工作流节点间共享上下文 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) | -- **模型选择**：推荐具备强工具调用能力的模型（如千问-Max 系列）。支持配置最长回复长度、temperature、enable_thinking（思考模式）等参数。 -- **提示词**：定义角色、行为指令与能力边界，支持自定义变量嵌入。 -- **内置工具**：沙箱环境中的 bash、write、read、edit、glob、grep、download_file 等工具，默认关闭需按需开启。 -- **知识库**：作为工具由智能体自主调用，支持标签过滤限定查询范围。 -- **MCP**：外部工具以 MCP 协议接入，支持动态非固定顺序调用。 -- **记忆**：短期记忆支持 0-30 轮上下文；长期记忆暂未支持。 -- **ReAct 最大轮次**：取值 1-50，限制单次会话中工具调用最大次数。 +## 使用方式 -详细配置方法参见 [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +- **创建入口**：统一通过百炼控制台 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → **创建应用**，按类型选择入口。 +- **配置核心**： + - 智能体：在模型选择器中指定 `千问-Max` 等强工具调用模型；通过系统提示词定义角色与工具使用规则；在“规划”模块启用知识库、MCP、技能等能力 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 + - 工作流：拖拽节点（大模型、意图分类、变量处理等）至画布，配置各节点的模型、提示词、输入/输出变量（如 `${sys.query}`、`大模型1/result`），并连线形成执行流 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 + - 高代码：选择模板或上传 `.whl` 包，配置部署方式（Serverless/K8s）、资源规格及环境变量；通过“工具”Tab 关联知识库/MCP，通过“网关”Tab 发布生产 API [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 +- **发布与调用**：**所有应用必须发布后方可调用**。发布后可在“发布渠道”页签获取 API Endpoint 与鉴权方式（API Key 或网关 [Token](../concepts/token.md)），支持 HTTP POST 调用标准 JSON 接口。文件需通过 `file_list`（URL）或 `session_file_id`（上传 API 返回）传递，不可在请求体中直接嵌入二进制内容 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 -### 旧版智能体（Agent 1.0） +## 限制和注意事项 -旧版智能体通过知识库（RAG）和插件扩展能力，适合意图单一、流程固定的简单任务。知识库检索后再决策是否调用其他工具。 - -> **注意**：新版智能体与旧版智能体基于不同技术架构，不支持直接升级或版本切换。需要迁移时必须重新创建新版应用。 - -旧版智能体的自定义插件有 5 秒超时限制。详情参见 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 - -## 工作流应用 - -工作流通过可视化节点编排将多步骤任务串联为稳定可控的执行链路，适合固定流程自动化场景。 - -主要节点类型： - -- **开始/结束节点**：定义输入输出参数结构，预置 query、historyList、imageList 变量 -- **大模型节点**：配置模型、提示词和用户提示词 -- **意图分类节点**：根据用户输入分发到不同处理分支 -- **智能体群组节点**：将子智能体作为工具组合调用 -- **变量处理节点**：文本输出或变量转换 - -工作流支持会话变量作为全局参数在节点间传递，支持记忆功能（本节点缓存或自定义缓存）。 - -创建和配置详情参见 [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md)。 - -## 高代码应用 - -面向专业开发者，支持基于 Python 项目部署 AI 后端服务。 - -关键特性： - -- **部署方式**：Serverless Function（无状态快速拉起）和 K8s（高性能有状态长程任务） -- **MCP 工具接入**：控制台直接关联知识库、工作流、插件等 MCP 服务 -- **前端体验**：支持直接体验、自定义交互卡片、基于 Spark Design 的自定义 WebUI -- **企业级能力**：自动化运维、可观测、日志服务、API 网关 -- **代码提交**：支持控制台模板创建或命令行上传 .whl 代码包 - -生产环境建议开启网关功能，通过自定义域名访问。时延敏感业务建议最小实例数大于等于 1。 - -详细开发和部署流程参见 [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md)。 - -## 文件问答 - -[智能体应用](../concepts/agent-application.md)支持上传文件进行智能问答，提供三种处理模式： - -| 模式 | 适用场景 | 特点 | -|------|---------|------| -| 全文引用 | 文档总结、全文翻译 | 简单直接，受上下文长度限制 | -| 切片检索（RAG） | 长文档问答、知识库检索 | 能处理超长文件，效果依赖检索策略 | -| 自定义处理 | 图片转换、视频分析等需工具介入的任务 | 功能灵活，依赖配置的工具 | - -文件限制：单会话最多 10 个文件，单文件不超过 10MB。超过 10MB 需使用文件上传 API。 - -支持格式：文档（doc/docx/pdf/md/txt 等）、图片（png/jpg/bmp/gif）、视频（mp4/mkv/avi 等）、音频（mp3/wav/flac 等）。 - -详细使用方式参见 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 - -## 发布与调用 - -所有应用类型均需先发布才能通过 API 集成： - -1. 在应用配置页点击"发布"，确认变更后完成发布 -2. 在"发布渠道"页签查看 API 调用方式 -3. [智能体应用](../concepts/agent-application.md)还支持发布到钉钉、微信公众号等第三方平台 - -RAM 账号发布前需确认拥有 `ram:CreateServiceLinkedRole` 权限。 - -## 计费说明 - -- **模型调用**：按模型类型和 [Token](../concepts/token.md) 用量计费 -- **知识库**：按量付费，召回的文本切片会增加输入 [Token](../concepts/token.md) -- **MCP/插件**：部分官方 MCP 按调用计费，第三方 MCP 费用由第三方收取 -- **高代码应用**：部署后函数计算、API 网关、存储均按量计费 -- **文件上传**：上传本身不收费，问答消耗按所选模型标准计费 +- **模型兼容性**：`enable_thinking` 参数仅对支持思考模式的模型生效（如千问-Max 系列），非支持模型在配置界面中不可见；千问-VL 系列模型在关闭预解析时仍可直接解析图片/视频，但其他模型必须开启预解析才能处理非文本文件 [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md)。 +- **文件限制**：单次会话最多上传 10 个文件，单文件 ≤ 10MB；聊天窗口上传的文件**仅在当前会话有效**，刷新即失效；生产环境推荐使用文件上传 API 获取 `session_file_id` 以支持更大文件与稳定传输 [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md)。 +- **计费关键点**： + - 模型调用费用 = 输入 [Token](../concepts/token.md)（含知识库召回内容、文件解析文本、记忆体内容） + 输出 Token； + - 知识库检索本身不额外计费，但召回文本计入输入 Token； + - [长期记忆](../concepts/long-term-memory.md)存储免费，但其内容参与 Prompt 构建，**占用的 Token 暂不计费**（见文档 3）； + - MCP 工具调用费用由第三方或按模型调用计费，百炼不加收 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 +- **权限与部署**：高代码应用部署需提前授权函数计算（FC）与 API 网关服务角色；工作流中“自定义缓存”记忆功能需在节点级显式启用；RAM 子账号发布应用前须确保拥有 `ram:CreateServiceLinkedRole` 权限 [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md)。 ## 来源文档 - [应用类型介绍](../../raw/application-user-guide/llm-application/application-introduction.md) -- [新版智能体应用（Agent 2.0）](../../raw/application-user-guide/llm-application/new-single-agent-application.md) +- [新版智能体应用](../../raw/application-user-guide/llm-application/new-single-agent-application.md) - [智能体应用](../../raw/application-user-guide/llm-application/single-agent-application.md) -- [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) - [工作流应用](../../raw/application-user-guide/llm-application/workflow-application.md) +- [高代码应用](../../raw/application-user-guide/llm-application/rich-code-application.md) - [文件问答](../../raw/application-user-guide/llm-application/file-q-a.md) - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md index 6ad7d7d1..9685d7ad 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/managed-agents.md @@ -1,75 +1,51 @@ # managed agents -Managed Agents 是百炼提供的智能体托管运行时，面向多步工具调用、代码执行、文件处理等长时运行任务。与无状态的[智能体应用](../concepts/agent-application.md)不同，它由平台在服务端托管会话状态、沙箱环境与工具执行，智能体在独立云端容器中自主执行命令、读写文件、安装依赖，事件历史在服务端持久化并支持中断与续接。 +Managed Agents 是百炼平台提供的智能体托管运行时，专为多步工具调用、代码执行、文件处理等长时运行任务设计。平台统一托管会话状态、沙箱环境与工具执行生命周期，智能体在隔离的云端容器中自主执行命令、读写文件、安装依赖，并通过服务端持久化的事件流反馈全过程。相比无状态的智能体应用，Managed Agents 本质是“有状态会话 + 托管沙箱 + 事件驱动”的组合能力。 -## 与[智能体应用](../concepts/agent-application.md)的区别 +## 支持的模型与功能 -| 维度 | [智能体应用](../concepts/agent-application.md) | Managed Agents | -| --- | --- | --- | -| 运行模式 | 无状态调用，应用侧维护上下文 | 服务端维护会话状态，支持中断与续接 | -| 执行环境 | 共享运行时 | 独立沙箱，云端容器 | -| 事件模型 | 响应级[流式输出](../concepts/streaming.md) | 会话级 SSE 事件流，事件历史持久化 | -| 典型场景 | 问答、对话、轻量任务 | 多步工具调用、代码执行、文件处理等长时任务 | +- **模型支持**：当前支持 `qwen3-max`、`qwen3.7-plus` 等 Qwen 系列大模型（见 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)）；模型需显式指定 ID，不支持通配符或别名。 +- **内置工具**：默认提供 7 类工具：`bash`（命令执行）、`read`/`write`/`edit`（文件操作）、`glob`/`grep`（文件搜索）、`download_file`（URL 下载）。所有工具均在沙箱内受限执行，不可绕过权限控制。 +- **扩展能力**： + - **MCP 服务**：可接入符合 MCP 协议的外部工具服务； + - **Skill**：复用预置的端到端任务流程（如数据清洗、PDF 解析）； + - **自定义沙箱环境**：支持 `apt`/`pip` 包预装、网络策略（`unrestricted` 或 `restricted`）配置（见 [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md)）。 -## 核心概念 +> **注意**：文档 2 示例中 Python SDK 创建 Agent 时使用 `system_prompt` 字段，而文档 5 明确说明字段名为 `system`；实际 API 以 `system` 为准（[构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md)），SDK 封装可能存在命名差异，请以 OpenAPI Schema 为准。 -四个核心对象构成完整的运行链路，详见 [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md)： +## 关键参数 -- **智能体（Agent）**：模型、系统提示词、工具、MCP 服务和 Skill 的组合配置。创建后通过 ID 引用，可在多个会话中复用。 -- **运行环境（Environment）**：会话运行的沙箱配置，由百炼托管的云端容器，独立于智能体管理，可被多个会话复用。 -- **会话（Session）**：智能体在指定环境中的一次运行实例，承载任务执行与输出。 -- **事件（Event）**：应用与智能体之间交换的消息，包括用户消息、工具调用结果和状态变更。 - -## 支持的工具 - -智能体通过以下工具与运行环境交互： - -- **命令执行**：在沙箱中运行 shell 命令（`bash`）。 -- **文件操作**：`read`、`write`、`edit`、`glob`、`grep`，以及从 URL 下载的 `download_file`；也可上传本地文件挂载到沙箱。 -- **MCP 服务**：接入外部工具服务，详见 [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md)。 -- **Skill**：挂载预置的工具组合，封装端到端任务流程。 - -快速开始阶段默认全选 7 个内置工具（`bash`、`read`、`write`、`edit`、`glob`、`grep`、`download_file`），可按需取消勾选。 +| 参数 | 类型 | 必填 | 说明 | +|------|------|------|------| +| `agent` | string | 是 | 智能体 ID（创建后返回），非名称 | +| `environment_id` | string | 是 | 运行环境 ID，非名称；同一环境可被多个会话复用 | +| `title` | string | 否 | 会话标题，仅用于管理标识，不影响执行 | +| `resources` | array | 否 | 资源挂载列表，格式为 `[{"resource_id": "...", "mount_path": "/mnt/session/uploads/data"}]`；挂载路径必须以 `/mnt/session/uploads/` 开头（见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)） | +| `timeout`（SSE 流） | number | 否 | 推荐设为 `120.0` 秒以上，避免长任务被意外中断 | ## 使用方式 -典型流程分为四步（控制台向导或 API 均可完成），参见 [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md)： - -1. **配置智能体**：指定名称、模型（如 `qwen3-max`）、系统提示词与工具。API 为 `POST /api/v1/agentstudio/agents`。 -2. **配置运行环境**：默认云端托管沙箱，可通过 `config.packages` 预装 apt / pip 依赖并设置网络策略。API 为 `POST /api/v1/agentstudio/environments`。 -3. **发起会话**：绑定智能体 ID 与环境 ID 创建会话实例。API 为 `POST /api/v1/agentstudio/sessions`。 -4. **发送事件并接收响应**：向会话写入用户消息触发处理（`POST /sessions/{id}/events`），通过 SSE 事件流实时接收工具调用过程与输出（`GET /sessions/{id}/events/stream`）。 - -控制台的**预览调试**标签页可直接对话并按事件类型（User、Agent、Tool、Tool_output、Error、Model、System）筛选查看执行过程。 - -### 上下文与资源挂载 - -上下文中的挂载资源独立于会话管理，可被多个会话复用，详见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)： - -- **挂载时机**：创建会话时在 `resources` 字段指定，或运行时通过 `POST /sessions/{session_id}/resources` 追加，实时生效且无需重启会话。 -- **路径约定**：挂载资源统一放在 `/mnt/session/uploads` 前缀下，可在系统提示词中直接引用完整路径。 -- **会话隔离**：平台为挂载资源做内部拷贝放入沙箱，会话内的修改不影响原始资源，也不影响挂载同一资源的其他会话；卸载后副本被清理。 - -## 限制与注意事项 +1. **创建智能体**：调用 `POST /api/v1/agentstudio/agents`，指定 `model.id`、`system` 提示词和 `tools` 列表（含 `builtin_toolkit` 结构）； +2. **创建环境**：调用 `POST /api/v1/agentstudio/environments`，配置 `config.type="cloud"` 及 `packages`； +3. **发起会话**：调用 `POST /api/v1/agentstudio/sessions`，绑定 `agent` 和 `environment_id`，可选传入 `resources`； +4. **发送用户消息**：调用 `POST /api/v1/agentstudio/sessions/{session_id}/events`，`input` 中包含 `role: "user"` 的 message 块； +5. **订阅事件流**：`GET /api/v1/agentstudio/sessions/{session_id}/events/stream`，使用 `text/event-stream` 头解析 SSE，关注 `message`、`tool_output`、`session_status` 等事件类型。 -- 单个上传文件不超过 **10 MB**。 -- 沙箱内文件路径遵循 `/mnt/session/uploads` 约定，代码中引用文件应使用该完整路径。 -- 会话状态、中断续接与工具审批由会话状态机管理，详见 [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md)。 +## 限制和注意事项 -> **注意**：文档中出现的模型名称不一致——控制台向导示例填写 `qwen3.7-plus`，而 API 代码示例使用 `qwen3-max`。请以控制台模型下拉列表中实际可选的模型 ID 为准。 +- **文件限制**：单个上传文件 ≤ 10 MB；挂载后路径固定为 `/mnt/session/uploads/xxx`，不可自定义根目录（见 [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md)）； +- **沙箱隔离**：每个会话独占容器实例，资源挂载为副本，修改互不影响； +- **超时控制**：会话默认最长运行 2 小时，超时自动终止；SSE 流需客户端主动维持连接，建议设置 `timeout` ≥ 120 秒； +- **工具调用安全**：`bash` 工具默认禁用危险命令（如 `rm -rf /`、`sudo`），且无法访问宿主机文件系统； +- **状态管理**：会话支持中断（`PATCH /sessions/{id}/status` → `"interrupted"`）与续接，但重启后仅恢复事件历史，不恢复内存状态。 ## 来源文档 - [概述](../../raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](../../raw/application-user-guide/managed-agents/managed-agents-quick-start.md) -- [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) +- [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md) - [配置 Agent 环境](../../raw/application-user-guide/managed-agents/managed-agents-environment.md) +- [构建 Agent](../../raw/application-user-guide/managed-agents/managed-agents-agent.md) - [Agent 上下文管理](../../raw/application-user-guide/managed-agents/managed-agents-context.md) -- [委派任务给 Agent](../../raw/application-user-guide/managed-agents/managed-agents-session.md) - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md index 3ee13beb..4c95a3cd 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/memory-library-overview.md @@ -1,166 +1,68 @@ # memory library overview -百炼记忆库（Memory Library）通过长期记忆 API 解决大模型跨会话上下文丢失的问题：自动从对话中提取关键信息并持久化存储，再在后续对话中基于语义检索召回相关记忆注入 Prompt，使智能体能够持续理解用户偏好与历史信息。该能力既可在百炼控制台可视化管理，也提供开放的 HTTP API 接入任意应用，并支持通过 OpenClaw 插件以"自动捕获 / 自动召回"的方式零侵入接入 Agent。详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md)、[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 与 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md)。 +记忆库是百炼平台提供的[长期记忆](../concepts/long-term-memory.md)能力核心组件，用于解决大模型跨会话上下文丢失问题。它通过自动从对话中提取关键信息并结构化存储为“记忆片段”和“用户画像”，支持语义检索与跨会话上下文注入，使智能体具备持续理解用户偏好与历史的能力。该能力以开放 API 形式提供，适用于自研应用、OpenClaw Agent 等各类集成场景。 -## 核心能力 +## 支持的模型/功能 -记忆库提供两类持久化记忆内容，二者可独立或组合使用： +- **记忆片段（Memory Nodes）**：自动从 `messages` 中提炼事件性、意图性内容（如“每天上午9点提醒我喝水”），支持手动写入 `custom_content`；适用于通用[长期记忆](../concepts/long-term-memory.md)场景。 +- **用户画像（User Profile）**：基于预定义的 `profile_schema` 从对话中抽取结构化属性（如年龄、职业、兴趣），需先调用 `CreateProfileSchema` 创建模板，再在 `AddMemory` 中传入 `profile_schema` ID 触发抽取。 +- **双模式接入**：既可通过直接调用 [长期记忆 API (新)](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 实现细粒度控制，也支持通过 [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) 实现零代码自动捕获与召回。 +- **多应用共享**：同一 `memory_library_id` 可被多个应用共用，`user_id` 作为隔离维度保障数据边界。 -- **记忆片段**：从对话中自动提取的关键事件和信息（如"用户每天上午9点需要喝水提醒"），适用于大多数长期记忆场景。支持自动去重、动态更新，也可通过 `custom_content` 直接写入指定内容。 -- **用户画像**：基于自定义画像模板从对话中提取的结构化属性（如年龄、职业、偏好等），适用于需要固定属性持久化存储的场景。属性字段及描述应清晰具体，避免"姓名/名称/名字"等同义字段并存，且不应期望一次对话就提取全部信息。 - -> **注意**：记忆有效期在不同入口存在差异。[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 文档指出"生成的记忆片段与用户画像暂无失效日期"，而 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) 控制台的默认记忆片段规则预置了"默认有效期 180 天"，并支持按规则配置 7/30/180 天或永不过期。以控制台记忆规则配置为准；通过 API 直写且不指定 `project_id` 时使用默认规则。 - -## 接入方式 - -### 方式一：API 直连 - -通过 HTTPS 调用 `https://dashscope.aliyuncs.com/api/v2/apps/memory/*` 系列接口，需在环境变量中配置 `DASHSCOPE_API_KEY`。典型流程为：对话结束调用 `AddMemory` 写入记忆 → 调用 `SearchMemory` 语义检索 → 将结果注入 Prompt。 - -```bash -# 写入记忆（从对话自动提取） -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "messages": [ - {"role": "user", "content": "每天上午9点提醒我喝水"}, - {"role": "assistant", "content": "好的，已记录"} - ], - "user_id": "user_001" - }' - -# 语义检索记忆 -curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ - --header "Authorization: Bearer $DASHSCOPE_API_KEY" \ - --header "Content-Type: application/json" \ - --data '{ - "user_id": "user_001", - "messages": [{"role": "user", "content": "我需要做什么？"}], - "top_k": 5 - }' -``` - -Python 用户可安装 `agentscope-runtime`，使用 `AddMemory`、`SearchMemory`、`ListMemory`、`CreateProfileSchema`、`GetUserProfile` 等封装类（均需在 `finally` 中调用 `close()`）。 - -### 方式二：OpenClaw 记忆插件 - -OpenClaw Agent 可通过插件实现零侵入的[跨会话记忆](../concepts/cross-session-memory.md)。插件在 Gateway 内通过 `before_agent_start`（自动召回）和 `agent_end`（自动捕获）两个生命周期钩子与长期记忆 API 交互，所有读写均由百炼服务端完成提炼、向量化和语义检索。 - -```bash -# 安装 -openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw - -# 验证 -openclaw plugins info modelstudio-memory-for-openclaw -openclaw modelstudio-memory stats -openclaw gateway restart -``` - -插件配置写入 `~/.openclaw/openclaw.json`，关键项：`slots.memory` 注册为记忆槽位（会自动禁用内置 `memory-core` 和 `memory-lancedb`）；`apiKey` 填 DashScope [API Key](../concepts/api-key.md)；`userId` 用于隔离不同用户记忆空间。 - -> **注意**：OpenClaw 记忆插件为统一配置，所有 Agent 共享同一记忆，暂不支持按 Agent 独立配置；不支持阿里云百炼 Coding Plan 的 [API Key](../concepts/api-key.md)。 +> **注意**：文档 3 提到默认记忆库中“默认项目”规则“默认有效期 180 天”，而文档 1 明确说明“生成的记忆片段与用户画像暂无失效日期”。该矛盾源于规则配置项（`memory_expiration_time`）与实际存储行为的差异——**记忆本身永不过期，但规则可配置过期策略，且仅对新写入生效；已存在的记忆不受影响**。详见 [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) 中“配置记忆片段规则”部分。 ## 关键参数 -### AddMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `messages` | 与 `custom_content` 二选一 | 对话内容，系统自动从中提取记忆片段 | -| `custom_content` | 与 `messages` 二选一 | 直接指定要存入的记忆内容，不经过对话提炼 | -| `user_id` | 是 | 记忆空间用户标识，同 `user_id` 共享命名空间，不同 `user_id` 完全隔离 | -| `memory_library_id` | 否 | 记忆库 ID，不填使用默认记忆库 | -| `project_id` | 否 | 记忆片段规则 ID，不填使用默认规则 | -| `profile_schema` | 否 | 用户画像规则 ID，传入后同时提取画像 | -| `meta_data` | 否 | 自定义元数据，用于分类管理 | - -### SearchMemory 请求参数 - -| 参数 | 必填 | 说明 | -| --- | --- | --- | -| `user_id` | 是 | 记忆空间用户标识 | -| `messages` | 是 | 查询对话内容 | -| `memory_library_id` | 否 | 限定检索的记忆库 | -| `top_k` | 否 | 返回记忆条数，建议 3–10 | - -### OpenClaw 插件配置项 - -| 配置项 | 类型 | 默认值 | 说明 | -| --- | --- | --- | --- | -| `apiKey` | string | - | 必填，以 `sk-` 开头 | -| `userId` | string | - | 必填，记忆空间用户标识 | -| `autoCapture` | boolean | `true` | 对话后自动提取并存储记忆 | -| `autoRecall` | boolean | `true` | 对话前自动检索并注入记忆 | -| `topK` | number | `5` | 每次召回返回的记忆条数 | -| `minScore` | number | `0` | 最小相似度阈值（0–100） | -| `profileSchema` | string | - | 用户画像 ID | -| `memoryLibraryId` | string | - | 记忆库 ID | -| `projectId` | string | - | 记忆片段规则 ID | - -## 记忆库与记忆规则管理 - -每个账号自带一个无法删除的默认记忆库，预置一条"默认项目"记忆片段规则（默认有效期 180 天，可编辑但不可删除）。可按业务场景创建新记忆库并配置记忆规则，每个记忆库最多 50 条记忆片段规则和 50 条用户画像规则。 - -- **记忆片段规则**：定义从对话中提取关键事件和信息的策略，可选择默认或自定义规则指令，支持自动更新和过期时间（7/30/180 天或永不过期）。 -- **用户画像规则**：定义画像字段名称、描述和初始值。当用户尚未通过对话提供信息时，系统使用初始值作为属性值。 - -控制台记忆检索页支持配置最大召回数量（1–100）、意图判别召回（建议开启）、查询改写（口语化提问时开启）和排序（使用 `gte-rerank-v2` 模型，相似度阈值建议 0.5–0.7）。 +| 参数名 | 类型 | 是否必填 | 说明 | +|--------|------|----------|------| +| `user_id` | string | ✅ | 用户唯一标识，用于隔离记忆空间；不同 `user_id` 数据完全隔离 | +| `memory_library_id` | string | ❌ | 记忆库 ID；不填则使用默认记忆库（每个账号自带一个） | +| `project_id` | string | ❌ | 记忆片段规则 ID；不填则使用默认规则或记忆库内首个可用规则 | +| `profile_schema` | string | ❌ | 用户画像规则 ID；仅当需触发画像抽取时传入 | +| `top_k` | number | ❌（默认 5） | 检索返回的最大记忆条数；OpenClaw 插件默认为 `5`，API 推荐设为 `3–10` 平衡效果与性能 | +| `min_score` | number | ❌（默认 0） | 相似度阈值（0–100），低于此值的结果将被过滤；OpenClaw 插件单位为百分制，API 返回分数范围为 `[0,1]`，需注意单位转换 | -## OpenClaw 插件工具 +## 使用方式 -除自动捕获/召回外，插件向 Agent 注册四个可主动调用的工具： +### 1. 基础 API 调用（推荐用于自研应用） +- **写入记忆**：调用 `POST /api/v2/apps/memory/add`，传入 `messages` 或 `custom_content` + `user_id`。 +- **检索记忆**：调用 `POST /api/v2/apps/memory/memory_nodes/search`（推荐）或 `/api/v2/apps/memory/search`（旧路径，功能一致），传入 `user_id` 和 `messages` 或 `query`。 +- **管理记忆**：支持 `GET /memory_nodes`（分页列表）、`PATCH /memory_nodes/{id}`（更新）、`DELETE /memory_nodes/{id}`（删除）。 +- **用户画像流程**：`CreateProfileSchema` → `AddMemory`（含 `profile_schema`）→ `GetUserProfile`（需等待约 3 秒后查询）。 -- **memory_search**：语义检索记忆库，返回相似度最高的记忆列表，适用于"之前讨论过什么"等回顾性问题。 -- **memory_store**：直接写入记忆，不经过对话提炼，适用于"记住我的服务器 IP 是 192.168.1.x"等显式记忆请求。 -- **memory_list**：分页列出当前 `userId` 下所有记忆条目。 -- **memory_forget**：按记忆 ID 删除指定记忆，通常先 `memory_search` 定位再删除。 +### 2. OpenClaw 插件集成（推荐用于 OpenClaw Agent） +- 安装插件：`openclaw plugins install @modelstudio/modelstudio-memory-for-openclaw` +- 配置 `~/.openclaw/openclaw.json`，设置 `apiKey` 和 `userId`，启用 `autoCapture`/`autoRecall`（默认开启） +- 插件自动注册工具：`memory_search`、`memory_store`、`memory_list`、`memory_forget`，Agent 可在运行时动态调用 -CLI 等效：`openclaw modelstudio-memory search|list|stats`。 +### 3. 控制台操作（调试与验证） +- 在 [记忆库](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 页面查看/编辑记忆库、配置规则、调试检索效果 +- “记忆详情”页按 `user_id` 查看记忆实体，“记忆检索”页模拟语义查询并调整 `改写`、`排序`、`相似度阈值` 等参数优化召回质量 -## 配额与限制 +## 限制和注意事项 -长期记忆 API 速率限制（阿里云账号级别）： +- **配额限制**（阿里云账号级别）： + - 所有 API 合计 ≤ 3000 QPM + - `AddMemory` ≤ 120 QPM + - `SearchMemory` ≤ 300 QPM + （详见 [长期记忆 API (新)](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md)） -| API 操作 | 速率上限 | -| --- | --- | -| AddMemory（写入） | 120 次/分钟 | -| SearchMemory（查询） | 300 次/分钟 | -| 所有操作合计 | 3000 次/分钟 | +- **延迟表现**： + - `SearchMemory` 端到端延迟：200–500ms + - `AddMemory` 延迟：500–1000ms；OpenClaw 插件中 `autoCapture` 为异步执行，不影响主响应流 -性能指标：SearchMemory 端到端延迟 200–500ms；AddMemory 延迟 500–1000ms；自动捕获异步执行，不影响响应速度。该功能与 API 调用限时免费。 +- **元数据与分类**：建议在 `AddMemory` 的 `meta_data` 字段中添加业务标签（如 `"category": "reminder"`），便于后续 `ListMemory` 过滤与管理 -## 排错要点 +- **用户画像最佳实践**： + - 字段名需语义唯一（避免同时定义“年龄”“年纪”“岁数”） + - 不应期望单轮对话提取全部画像字段，需通过多轮交互逐步完善 -- **OpenClaw 插件重启后状态为 not loaded**：检查 `openclaw.json` 中 `plugins.entries.modelstudio-memory-for-openclaw.enabled` 是否为 `true`，以及 `plugins.slots.memory` 是否指向该插件，修正后重新 `openclaw gateway restart`。 -- **日志出现 InvalidApiKey**：DashScope [API Key](../concepts/api-key.md) 无效或过期，到百炼控制台确认状态或重新创建；若用环境变量引用，确认 `DASHSCOPE_API_KEY` 已设置且 Gateway 进程可读取。 -- **查看插件日志**：日志按日期存储在系统临时目录，文件名 `openclaw-YYYY-MM-DD.log`，可用 `grep modelstudio-memory` 过滤。 - -> **注意**：[长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) 为新版本，相比旧版长期记忆 API 在延迟、自动提取、语义检索准确性和用户画像能力上均有改进，建议新接入直接使用新版接口。 +- **环境依赖**：所有方式均需配置 `DASHSCOPE_API_KEY` 环境变量或显式传入 `apiKey`，且 Key 必须为百炼平台生成的有效密钥（**不支持 Coding Plan 的 API Key**） ## 来源文档 +- [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - [为 OpenClaw 配置长期记忆插件](../../raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [记忆库](../../raw/application-user-guide/memory-library-overview/memory-library.md) -- [长期记忆 API](../../raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md index 7b9ed189..67cc2c05 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-compression.md @@ -1,77 +1,40 @@ # model compression -模型压缩是百炼平台提供的量化功能，用于将全精度微调模型转换为低精度版本，在保持模型能力的前提下降低部署所需的 MU 规格，从而减少推理成本。该功能属于模型生产链路中的可选环节，位于模型调优与模型部署之间。需要注意的是，百炼平台的模型压缩特指量化，不涉及结构剪枝或知识蒸馏。 +模型压缩是百炼平台提供的量化能力，用于将全精度微调模型转换为低精度版本，在保持业务可用性的前提下显著降低推理部署所需的 MU 规格与成本。该功能仅作用于通过百炼平台完成的微调模型，不支持基础模型或第三方模型，且压缩操作不可逆。详细背景与设计边界请参见 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 -## 核心概念 +## 支持的模型与功能 -模型压缩通过量化技术降低模型参数精度来实现部署成本优化。完整的模型生产链路为：模型调优 → 模型压缩（可选）→ 模型部署。以 qwen3.5-flash-2026-02-23 微调模型为例，压缩前部署规格为 MU1*2（108 元/小时），压缩后降至 MU8*1（47 元/小时），部署成本节省约 56%。详细说明参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 +- **支持模型**：当前仅支持百炼平台微调产出的自定义模型，例如 `qwen3.5-flash-2026-02-23`；基础模型、OSS 导入模型或非百炼训练的模型均不可压缩。 +- **功能范围**：百炼模型压缩特指**后训练量化（Post-Training Quantization, PTQ）**，不包含结构剪枝、知识蒸馏等其他压缩技术。该限定在 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) 中明确说明。 +- **地域限制**：仅华北2（北京）地域可用。 -> **注意**：压缩操作不可逆。压缩后的模型不支持继续微调，也不支持二次压缩。如需调整，必须从上游全精度微调模型重新压缩。 +> **注意**：文档中示例 `qwen3.5-flash-2026-02-23` 的压缩前后规格（MU1\*2 → MU8\*1）为示意性数据，实际支持的模型列表及对应规格请以控制台实时展示为准；部分新发布模型可能尚未同步至压缩模板库，具体兼容性请参考最新版 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 -## 支持的模型 +## 关键参数 -当前支持压缩的模型以控制台展示为准。已知支持的模型系列及规格如下： +| 参数 | 是否必填 | 说明 | +|------|----------|------| +| **任务名称** | 是 | ≤50 字符，建议含模型简称、量化方式和版本号（如 `qwen35-ft-awq-v2`） | +| **量化产出模型名后缀** | 是 | 仅小写字母+数字，≤8 位，将拼接至源模型名后（如源模型 `my-qwen-ft` + 后缀 `awq` → `my-qwen-ft-awq`） | +| **量化模板** | 是 | 卡片式选择，模板名中 MU 编号越大，部署规格越小、成本越低，但潜在精度损失风险越高；**切换源模型会自动清空已选模板** | +| **校准数据** | 条件选填 | 仅当所选模板要求校准时显示；最多选 5 个已发布数据集（不支持 OSS 挂载），推荐与目标场景语义一致的数据（如客服问答场景选用对话数据集） | -| 模型系列 | 基础模型 | 压缩前部署规格 | 压缩后部署规格 | -|---------|---------|--------------|--------------| -| Qwen | qwen3.5-flash-2026-02-23 | MU1*2（108 元/小时） | MU8*1（47 元/小时） | +## 使用方式 -仅支持通过百炼平台微调产出的自定义模型，不支持基础模型或第三方模型。 +1. **前提**：确保工作空间中存在状态为「成功」的百炼微调模型；若无可选模型，请先完成 [模型调优](https://help.aliyun.com/zh/model-studio/fine-tuning/#73749f1ee5634)。 +2. **入口**：控制台 → **模型** > **模型训练** > **模型压缩** → **创建压缩任务**。 +3. **配置**：按上述关键参数填写，特别注意量化模板需在选定源模型后才可加载；所有必填项完成后「开始压缩」按钮方可点击。 +4. **监控**：任务创建后不可修改配置，可通过详情页查看状态与日志；失败时优先检查「详情」页错误信息，并在「日志」页搜索 `ERROR` 级别条目。完整操作流程详见 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 -## 使用限制 +## 限制和注意事项 -- 仅华北2（北京）地域可用。 -- 仅支持百炼平台微调产出的自定义模型。 -- 压缩后的模型不支持继续微调或二次压缩。 -- 压缩任务创建后不可修改配置。 - -## 创建压缩任务 - -在使用模型压缩前，需确保工作空间中已有微调训练完成的自定义模型。操作路径：控制台 → 模型 → 模型训练 → 模型压缩 → 创建压缩任务。 - -创建任务时需配置以下参数（详见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)中的操作步骤）： - -| 参数 | 必填 | 说明 | -|-----|------|-----| -| 任务名称 | 是 | 最长 50 字符，建议包含模型简称、量化方式和版本号 | -| 任务描述 | 否 | 最长 200 字符 | -| 选择源模型 | 是 | 仅展示可压缩的微调模型，切换源模型会清空已选量化模板 | -| 量化产出模型名后缀 | 是 | 仅支持小写字母和数字，最长 8 位 | -| 量化模板 | 是 | 须先选择源模型，MU 编号越大表示部署规格越小、成本越低 | -| 校准数据 | 条件选填 | 仅当量化模板包含校准输入参数时显示，最多选择 5 个数据集 | - -## 量化模板与校准数据选择 - -**量化模板**决定了压缩后模型的部署规格。模板名称中 MU 编号越大，部署规格越小、成本越低，但精度损失可能越大。建议根据业务对成本和精度的权衡来选择。 - -**校准数据**用于提升量化精度，建议选择与目标推理场景语义相近的数据集。例如用于客服问答场景时，应选择包含客服对话样本的数据集。 - -## 任务状态与管理 - -压缩任务有 7 种状态：PENDING（待开始）→ QUEUING（排队中）→ RUNNING（运行中）→ SUCCEEDED（成功）/ FAILED（失败）/ CANCELING（停止中）→ CANCELED（已取消）。 - -任务管理操作： -- **停止任务**：仅 PENDING 和 RUNNING 状态可停止，停止后不可恢复。 -- **删除任务**：仅终态（SUCCEEDED、FAILED、CANCELED）可删除，删除任务记录不影响已产出的压缩模型。 - -任务失败时，可在详情页查看错误信息，或切换到日志页签搜索 ERROR 级别日志进行排查。如仍无法解决，可提交工单并附上任务 ID 和日志文件。更多管理细节参见[模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 - -## 计费说明 - -- 压缩任务本身限时免费，截止时间以控制台公告为准。 -- 压缩后的模型在部署阶段按 MU 规格计费。 -- 建议在免费期内对同一微调模型尝试多个量化模板，分别部署后用业务测试集验证推理效果，选择最优方案后再正式上线。 +- **不可逆性**：压缩后的模型**不支持继续微调**，也**不支持二次压缩**；如需调整，必须基于原始全精度微调模型重新发起任务。 +- **任务管理**：仅 `PENDING` / `RUNNING` 状态可停止；仅终态（`SUCCEEDED` / `FAILED` / `CANCELED`）可删除；删除任务记录不影响已产出模型。 +- **计费**：压缩任务当前限时免费（截止时间以控制台公告为准），但压缩后模型的部署费用按 MU 规格正常计费，与免费期无关。 +- **精度权衡**：量化必然引入精度损失，强烈建议在免费期内对同一源模型尝试多个模板，并使用真实业务测试集验证效果后再正式部署——该实践建议亦出自 [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md)。 ## 来源文档 - [模型压缩](../../raw/model-user-guide/model-compression/model-compression-introduction.md) - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md index d021cb14..11f2b674 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-context-protocol.md @@ -1,82 +1,56 @@ # model context protocol -模型上下文协议（Model Context Protocol, MCP）是 Anthropic 提出的开源标准协议，用于在大模型与外部工具之间搭建统一的信息传递通道。阿里云百炼基于 MCP 提供全周期服务：开发者无需为每个外部工具编写专用接口，即可让智能体、工作流应用接入海量第三方工具，也能通过外部调用集成到第三方应用或个人项目中。 +模型上下文协议（Model Context Protocol, MCP）是阿里云百炼平台提供的标准化接口协议，用于在大模型应用（如智能体、工作流）与外部工具服务之间建立安全、可扩展的上下文交互通道。它屏蔽了底层通信细节，使开发者无需为每个工具单独开发适配层，即可复用官方或自定义的 MCP 服务。该协议基于 Anthropic 提出的开源标准实现，当前在百炼中以 Streamable HTTP 和 SSE 两种传输方式落地 [MCP 协议](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -## 服务类型 +## 支持的模型/功能 -百炼将 MCP 服务分为两大类，均需先在 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market) 开通或部署后使用： +MCP 协议本身不绑定特定模型，但其能力需通过百炼平台的**智能体应用**和**工作流应用**触发和编排： -- **官方 MCP 服务**：百炼官方云端部署，开通即用（如 Amap Maps、Sequential Thinking、QuickChart、联网搜索 WebSearch 等）。详见 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 -- **自定义 MCP 服务**：由开发者自行部署，支持三种方式（详见 [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)）： - - **使用脚本部署**：面向遵循 MCP 协议的代码包，托管到函数计算 FC，支持 `npx`（Node.js）、`uvx`（Python）、`http`（远程服务）三种安装方式。 - - **从 AI 网关导入**：把已有的 RESTful API 通过 AI 网关升级为 MCP 服务后导入。 - - **从阿里云 OpenAPI 导入**：通过 OpenAPI 开发者门户将官方 OpenAPI 发布为 MCP 服务，用于操作 OSS、ECS 等阿里云产品。 +- **智能体应用**：支持自动决策调用（基于提示词理解），最多同时接入 5 个 MCP 服务；适用于多步推理、动态工具选择场景（如路径规划、气温趋势分析）[官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 +- **工作流应用**：需显式配置 MCP 节点并指定具体工具（如 `maps_weather`），适合确定性、单步调用流程；常配合大模型节点完成参数提取与结果总结 [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md)。 +- **不支持直接接入千问 API**：MCP 服务无法在调用 `qwen-*` 等基础模型 API 时启用，仅限平台内智能体/工作流应用使用 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 -## 使用方式 - -MCP 服务既可在平台内部集成，也可通过外部调用集成到第三方应用。 - -### 平台内部：智能体与工作流 - -- **[智能体应用](../concepts/agent-application.md)**：大模型根据对话内容自动判断是否调用 MCP 服务，单个智能体最多可同时添加 **5 个** MCP 服务。适合路径规划、逻辑推理、多工具组合（如天气查询 + 图表绘制）等场景。 -- **工作流应用**：每个 MCP 节点只能使用一个工具，需手动指定输入参数并将输出传递到下一节点。通常需先用大模型节点把自然语言解析为 MCP 工具所需的输入参数（在 System Prompt 中描述工具的名称、功能、输入输出格式），再接入 MCP 节点。 - -> **注意**：在工作流中仅使用单一工具（如 Amap Maps 的 `maps_weather`）时，工作流只能回答与该工具相关的问题。 - -### 外部调用 - -百炼 MCP 服务支持集成到第三方应用或个人项目，详见 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)： - -- **集成至第三方应用**：支持一键自动配置或手动配置到 Cherry Studio、Cursor 等客户端。手动配置需获取 `DASHSCOPE_API_KEY` 并替换配置文件中的对应变量。 -- **集成至个人项目**：通过 MCP SDK 灵活编码。可结合 OpenAI SDK 调用百炼 MCP 服务，典型端点如 `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp`，鉴权使用 `Authorization: Bearer `。 +> **注意**：文档 3 提到“百炼 MCP 服务已从旧版 SSE 协议升级为新版 Streamable HTTP 协议”，但文档 4 的错误码表（如 `11200058`）仍明确要求区分 `"sse"` 与 `"streamableHttp"` 类型配置，且文档 5 的脚本部署模板也保留 `type: "sse/streamableHttp"` 字段。这表明两种协议并存，**并非完全替代关系**，实际部署需严格匹配服务端端点（`/sse` 或 `/mcp`）。 -## 关键参数与配置 +## 关键参数 -- **传输协议**：`type` 字段须与端点路径一致，`"sse"` 对应 GET `/sse`，`"streamableHttp"` 对应 POST `/mcp`。配置不匹配会触发 405/404 等错误。 -- **自定义服务配置示例**（脚本部署）： +| 参数 | 说明 | 示例值 | 来源依据 | +|------|------|--------|----------| +| `type` | 传输协议类型，决定请求方法与端点路径 | `"sse"`（GET `/sse`）或 `"streamableHttp"`（POST `/mcp`） | [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) 错误码 11200058/11200059 | +| `url` | MCP 服务接入地址 | `https://dashscope.aliyuncs.com/api/v1/mcps/WebSearch/mcp` | [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) SDK 示例 | +| `Authorization` | 认证头，值为 `Bearer ` | `Bearer sk-xxx` | [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) SDK 示例 | +| `command` / `args` | 自定义部署时启动命令（`npx`/`uvx`）及参数 | `"npx", ["-y", "@modelcontextprotocol/server-memory"]` | [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) | -```json -{ - "mcpServers": { - "memory": { - "command": "npx", - "args": ["-y", "@modelcontextprotocol/server-memory"] - } - } -} -``` - -- **敏感信息加密**：涉及敏感数据的服务在创建时使用 KMS 凭据加密管理。 -- **部署后可修改项**：部署完成后仅支持编辑服务名称和描述；修改部署方式、地域、安装方式或服务配置须先停止部署再重新部署。 - -> **注意**：百炼 MCP 服务已从旧版 SSE 协议升级为新版 **Streamable HTTP** 协议。已开通用户需在 MCP 广场执行"取消开通 → 立即开通"完成协议升级；SDK 调用请使用 `streamablehttp_client` 连接。 +## 使用方式 -## 计费 +### 1. 接入官方 MCP 服务 +- **开通**：前往 [MCP 广场](https://bailian.console.aliyun.com/?tab=mcp#/mcp-market)，点击服务卡片（如 Amap Maps）→ “立即开通” → “确认开通”。 +- **配置到智能体**：创建智能体 → “添加 MCP 服务” → 从已开通列表选择，最多 5 个。 +- **配置到工作流**：拖入 MCP 节点 → 选择具体工具（如 `maps_weather`）→ 手动绑定输入参数（如引用上游节点输出）。 -- **云部署 MCP 服务**：限时免部署费用；部分服务涉及第三方 API 调用，费用由第三方收取。联网搜索 MCP 服务免费额度 2000 次，用尽后按 29 元/千次计费，限流 15 QPS（主账号与 RAM 子账号共享）。 -- **自定义部署 MCP 服务**： - - **基础模式**：无部署费用，按调用时长计费（0.000156 元/秒），首次调用有冷启动延迟，适合偶尔调用。 - - **极速模式**：有部署费用（0.000036 元/秒）+ 调用费用（0.000156 元/秒），适合长时间在线、调用频繁的场景。 +### 2. 外部调用集成 +- **第三方工具**：支持 Cherry Studio、Cursor 一键配置，自动注入 `DASHSCOPE_API_KEY` 和服务元信息。 +- **自有项目**：使用 MCP SDK（如 `mcp.client.streamable_http`）连接，配合 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)实现工具调用循环 [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md)。 -## 限制与注意事项 +### 3. 部署自定义 MCP 服务 +支持三种方式： +- **脚本部署**：上传 `npx`/`uvx` 启动配置（如 `@modelcontextprotocol/server-memory`），托管至函数计算 FC； +- **AI 网关导入**：将现有 RESTful API 封装为 MCP 服务后，从 AI 网关导入； +- **OpenAPI 导入**：将阿里云产品 OpenAPI（如 OSS、ECS）快速发布为 MCP 工具 [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md)。 -结合 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)，接入时需注意以下限制： +## 限制和注意事项 -- **不能直连千问 API**：MCP 服务必须集成在智能体或工作流应用中使用，无法在直接调用千问 API 时接入。 -- **无法访问本地资源**：自定义 MCP 服务托管在函数计算 FC，暂不支持访问用户本地数据库、文件、硬件等资源；需要访问本地资源的 MCP Server 建议在本地部署。 -- **访问远程资源需配置网络**：FC 无固定出口公网 IP，访问云数据库等远程资源需配置 FC 的 IP 白名单或打通 VPC 网络。 -- **仓库与版本限制**：私有 npm 仓库暂不支持，需发布到公共仓库或改用 SSE；通过 npx/uvx 部署的服务在源版本更新后不会自动更新，需手动重新部署。 -- **调用增加 Token 消耗**：MCP 返回的内容会作为上下文传入模型，增加输入 Token，并可能间接增加输出 Token。 -- **调用失败排查**：优先确认已开通/升级服务、API Key 有效、额度未用尽；若模型无报错但不调用工具，应在提示词中明确工具名称与能力，必要时更换更强的推理模型（如千问 3 系列）。自定义服务的连接、超时、鉴权、协议等错误可对照 `11200044`~`11200060` 系列错误码逐项排查。 +- **网络与权限**：自定义 MCP 服务运行在函数计算 FC，**无固定出口 IP**，访问云数据库等资源需配置 IP 白名单或 VPC 打通；**不支持访问本地资源**（文件、硬件、本地数据库）[MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **[Token](../concepts/token.md) 消耗**：MCP 返回内容作为上下文输入模型，**直接增加输入 [Token](../concepts/token.md) 数量**；模型响应可能因信息更丰富而间接增加输出 [Token](../concepts/token.md) [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **协议兼容性**：若服务返回非标准 JSON-RPC 格式或版本不匹配，将触发 `MCP_PROTOCOL_ERROR`（错误码 11200054），需检查服务端实现是否符合 [MCP 官网](https://modelcontextprotocol.io/) 规范 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 +- **部署约束**：私有 npm 仓库包暂不支持 `npx` 部署；自定义服务版本更新后需手动重新部署，不会自动同步 [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md)。 ## 来源文档 - [模型上下文协议（MCP）](../../raw/application-user-guide/model-context-protocol/mcp-introduction.md) - [官方 MCP 服务](../../raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) -- [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) - [外部调用](../../raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](../../raw/application-user-guide/model-context-protocol/mcp-faq.md) - - +- [自定义 MCP 服务](../../raw/application-user-guide/model-context-protocol/custom-mcp.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md index a2d4f8ac..47d9fdee 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-data-overview.md @@ -1,173 +1,60 @@ # model data overview -百炼平台的数据管理功能用于在[模型调优](../concepts/fine-tuning.md)和[评测](../concepts/evaluation.md)前创建、清洗、增强训练集与[评测](../concepts/evaluation.md)集。它统一管理[业务空间](../concepts/workspace.md)下的大模型相关数据集，分为训练集（用于[模型调优](../concepts/fine-tuning.md)）和[评测](../concepts/evaluation.md)集（用于模型评测）两类，并支持基于数据流的可视化数据处理能力。本文汇总训练集与评测集的格式规范、关键参数以及数据清洗与增强的使用方式。 +百炼平台的模型数据管理功能为开发者提供统一的数据集创建、清洗、增强与回流能力，支撑大模型训练（SFT/DPO/CPT）、评测及[多模态](../concepts/multi-modal.md)任务。所有数据操作均通过控制台 [数据管理](https://bailian.console.aliyun.com/#/efm/model_data) 统一入口进行，当前功能**仅适用于华北2（北京）和新加坡地域**。本文档整合训练集、评测集、数据处理与日志回流的核心规范，面向开发者提供可直接落地的结构化参考。 -> **注意**：本文涉及的数据管理与数据处理能力**仅适用于华北2（北京）地域**。此外，阿里云百炼目前暂未提供可用的数据处理 API，所有数据处理操作需在控制台完成。 +## 支持的模型/功能 -## 支持的数据集类型 +百炼支持以下模型类型与对应的数据功能： -数据集分为训练集和评测集两类，详见 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 +- **文本生成类模型**：全面支持 SFT（监督微调）、DPO（直接偏好优化）、CPT（持续预训练）三类训练方式，以及文本生成评测集构建； +- **[多模态](../concepts/multi-modal.md)理解模型（如 Qwen-VL 系列）**：支持 SFT 训练，需遵循 ChatML 格式并严格满足图像/视频字段要求（如 `content` 必须为数组格式）； +- **图生视频模型（如 wan-i2v / wan-kf2v）**：支持基于首帧或首尾帧的训练集与验证集构建，需按指定 ZIP 结构组织 `data.jsonl`、图像及视频文件； +- **思考模型（Thinking）**：仅对最后一条 `assistant` 输出进行训练，且必须用 `` 标签包裹思考内容，详见 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md)。 -| 类型 | 用途 | 支持的子类型 | -| --- | --- | --- | -| 训练集 | 用于[模型调优](../concepts/fine-tuning.md)，通过在特定任务上进行有监督训练提升模型表现 | 文本生成、[多模态](../concepts/multimodal.md)理解、图生视频（首帧）、图生视频（首尾帧） | -| 评测集 | 用于评估模型在未见过数据上的泛化能力 | 文本生成 | +> **注意**：文档 2 明确指出数据清洗与增强**暂不支持 SFT-图片理解训练集和 DPO 训练集**，而文档 1 中未限定该限制。实际使用时请以 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md) 的说明为准——即仅支持 SFT-文本生成（ChatML 格式）训练集。 -## 训练集格式 +## 关键参数 -### SFT 训练集（文本生成） +| 参数 | 适用场景 | 类型 | 取值范围 | 说明 | +|------|----------|------|-----------|------| +| `loss_weight` | SFT（所有 assistant 行）、SFT-Thinking（仅最后 assistant 行）、DPO（`chosen` 字段） | float | `0.0 ~ 1.0` | 控制单条样本/输出在训练中的相对重要性；属邀测功能，需联系商务经理开通 | +| `resized_width` / `resized_height` | [多模态](../concepts/multi-modal.md) SFT（图像/视频） | int | ≥ 1 | 图像/视频帧缩放目标尺寸（像素），影响坐标标注基准（Qwen2.5-VL 用绝对像素，Qwen3-VL 用 `[0,999]` 相对坐标） | +| `fps` / `sample_fps` | 多模态 SFT（视频） | float | > 0 | 视频输入帧率（`fps`）或图片帧序列帧率（`sample_fps`），仅 Qwen3.5+ VL 模型支持 | +| `foreignKey` | 数据增强节点输出 | string | 自动生成 | 增强后自动添加的标识字段，**无需删除，不影响训练** | -采用 ChatML 格式，支持多轮对话和多种角色设置。一行训练数据为一个 JSON 对象，结构如下： +## 使用方式 -```json -{"messages": [ - {"role": "system", "content": "系统输入1"}, - {"role": "user", "content": "用户输入1"}, - {"role": "assistant", "content": "期望的模型输出1"}, - {"role": "user", "content": "用户输入2"}, - {"role": "assistant", "content": "期望的模型输出2"} -]} -``` +1. **数据集创建**： + - 训练集/评测集上传：ZIP 包（≤2 GB）或 Excel 文件（仅评测集），`data.jsonl` 必须位于 ZIP 根目录； + - 日志回流：通过 [模型监控](https://bailian.console.aliyun.com/#/model-telemetry) 或 [数据管理](https://bailian.console.aliyun.com/#/efm/model_data) 入口配置时间范围、API Key、模型等参数，单次上限 10 万条，支持追加版本；详情见 [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md)。 -不支持 OpenAI 的 `name`、`weight` 参数，所有的 assistant 输出都会被训练。单条训练数据的所有 assistant 行支持 `loss_weight` 参数（设置范围 `0.0~1.0`，数值越大重要性越高），该参数属于邀测参数，如需使用请联系商务经理。 +2. **数据清洗与增强**： + - 仅支持 SFT-文本生成训练集（ChatML 格式）； + - 在控制台「数据流」中编排节点：先清洗（如敏感信息打码），再增强（如 Few-Shot 生成）； + - 处理结果自动生成新版本（如 V1 → V2），原数据集不受影响。 -### SFT 思考模型（thinking） +3. **格式校验要点**： + - SFT ChatML：`system` 消息中 `content` 必须为 `[{ "text": "..." }]` 数组格式； + - DPO：`chosen`/`rejected` 必须为单个 `{"role": "assistant", "content": "..."} ` 对象； + - 图生视频：ZIP 内路径名仅支持 ASCII 字符（a-z, A-Z, 0-9, `_`, `-`），图片/视频文件名全局唯一。 -只能针对**最后**的 assistant 输出进行训练，思考内容必须用 `导读` / `` 标签包裹，且思考标签前后的若干个 `\n` 必须保留。中间的 assistant 输出不应添加思考标签。也可以在训练样本中设置模型不输出 `导读` 标签，但训练完成后不建议再开启思考模式调用。 +## 限制和注意事项 -### SFT 视觉理解（千问 VL） - -支持图片、视频文件路径和图片帧列表三种输入。如需传入 `system` 消息，对应 `content` 必须使用数组格式 `[{"text":"..."}]`，不能使用字符串格式。关键参数： - -| 字段 | 类型 | 必填 | 说明 | -| --- | --- | --- | --- | -| `image` | str | 是 | 图片文件路径 | -| `resized_width` / `resized_height` | int | 否 | 图片目标缩放尺寸（像素） | -| `video`（路径模式） | str | 是 | 视频文件路径，仅 qwen3.5 及以后 VL 模型支持 | -| `video`（帧列表模式） | `List[str]` | 是 | 图片帧列表，需配合 `sample_fps` | -| `fps` / `sample_fps` | float | 否 | 训练输入频率 / 帧率 | -| `video_start` / `video_end` | float | 否 | 视频截取起止时间（秒） | - -关于物体定位坐标：Qwen2.5-VL 使用相对缩放后图像左上角的绝对像素坐标；Qwen3-VL 使用相对坐标，坐标值会缩放到 `[0, 999]` 范围。 - -### DPO 数据集 - -DPO ChatML 格式将 `messages` 内所有内容作为输入，通过 `chosen` 与 `rejected` 字段训练模型对最后一条用户输入的正负反馈。针对深度思考内容需使用 `导读` 标签包裹。`chosen` 模块支持 `loss_weight` 参数（邀测参数）。 - -### CPT 训练集 - -纯文本格式，一行训练数据结构为 `{"text":"文本内容"}`。 - -### 图生视频训练集 - -分为基于首帧和基于首尾帧两种。训练集必须提供，验证集可选（无需提供视频，训练任务会在评估节点自动调用模型生成预览视频）。 - -- 标注文件固定命名为 `data.jsonl`，最大 20MB，每行为一个 JSON 对象。 -- 字段包括 `prompt`、`first_frame_path`、`last_frame_path`（仅首尾帧）、`video_path`（仅训练集）。 -- 图像最大分辨率 4096*4096，支持 BMP、JPEG、PNG、WEBP；视频支持 MP4、MOV。 - -## 文本生成评测集 - -单轮对话评测数据，使用 Excel 格式，每行包含 Prompt（用户输入）和 Completion（模型期望输出）。参评模型基于评测集中每条 Prompt 进行推理，评分员或自动化评分系统参考 Completion 对推理结果评分。 - -## 数据集构建规模要求 - -不同调优方式对训练集规模有不同最低要求： - -| 调优方式 | 最低规模 | -| --- | --- | -| CPT | 一千万 [Token](../concepts/token.md) 优质预训练数据 | -| SFT | 上千条优质微调数据 | -| DPO | 上百条人类偏好数据 | - -> 如果调优后评测结果不佳，最简单的改进方法是收集更多数据进行训练。 - -## 数据清洗与增强 - -在模型调优前，可使用数据处理功能对训练集进行数据清洗和数据增强，从而获得更高质量的训练集，详见 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 - -| 处理方式 | 适用场景 | -| --- | --- | -| 数据清洗 | 修正训练数据中的规范性、合规性、一致性及重复等问题（如特殊内容移除、敏感信息打码等十种清洗操作） | -| 数据增强 | 增加训练数据的多样性和均衡性，或扩展数据规模 | - -> **注意**：如果训练集数据不适合清洗与增强（如法律文件、医学记录、文学作品、方言汇总、用户评论、技术手册等），建议直接跳过数据处理。此外，数据处理目前仅支持 SFT-文本生成训练集，不支持图片理解训练集和 DPO-文本生成训练集。 - -### 创建数据流任务 - -数据处理通过在控制台搭建自定义数据流完成。典型流程为「先清洗后增强」——确保增强操作在干净、高质量的数据集上进行，保证模型调优数据源的准确性。两步操作： - -1. **创建数据流**：在数据管理页面创建空白数据流，将数据清洗节点（开启所需算子）和数据增强节点拖入画布并依次连接，发布后即可使用。也可直接使用预置的数据流模板。 -2. **创建数据流任务**：选择已发布的数据流，输入任务名称，选择模型数据作为数据来源并指定训练集，任务自动执行。执行期间不支持手动终止。 - -任务处理状态包括：**处理中**（执行中，高峰时段需排队）、**已完成**（成功完成，可查看处理结果）、**处理失败**（建议提交工单咨询）。 - -> 训练集在清洗或增强后会自动生成一个新版本，新版本独立保存，不会覆盖原训练集。建议检查清洗后的训练集，确保数据完整性和真实性未被破坏。 - -### 节点说明 - -数据流由若干节点组成，每个数据流必须包含一个开始节点和一个结束节点： - -| 节点 | 作用 | -| --- | --- | -| 开始/结束 | 开始节点接收待处理训练集（`对话文本`参数无法更改）；结束节点输出处理结果，自动生成新版本 | -| 条件判断 | 设置条件分支，支持且/或配置，多条件自上而下顺序执行 | -| 数据清洗 | 选择清洗算子（如文章相似度去重、敏感词过滤、毒性消除等），按编排顺序执行，输出清洗后训练集及 `dataSetCount` 变量 | -| 数据增强 | 增加数据多样性和规模，分为通用、文本分类、文本抽取、文本创作四种场景 | - -### 数据增强节点关键参数 - -数据增强节点本质上是基于 `千问-Max` 大模型的 Few-Shot 生成器，暂不支持选择其他模型。每次最多生成 2000 条样本。 - -| 参数 | 说明 | -| --- | --- | -| 生成样本数 | 需要生成的数据量。原训练集 N 条 + 生成 M 条 = 增强后 N+M 条 | -| 指令生成依赖样本数 | 从原训练集中选出的种子数量，拼入 Prompt 提供给大模型。若种子+Prompt 总长度超过千问-Max 最大输入 [Token](../concepts/token.md)，系统会自动调整 | -| 过滤相似度阈值 | 控制生成数据的相似度过滤 | -| Prompt 配置 | 定义增强任务输入输出要求，支持 `few_shot_examples` 参数。提供默认模板 | - -Few-Shot 策略示例：训练集 1000 条、指令生成依赖样本数 5、生成样本数 200，则每次从 1000 条中抽样 5 条拼入 `few_shot_examples`，请求千问-Max 生成 1 条，重复 200 次。 - -> 输出数据中 `foreignKey` 为系统后添加的标识字段，增强后的训练集可直接用于模型调优，无需删除该字段。 - -### 数据增强建议 - -- **任务相关性**：确保生成数据与目标任务高度相关，避免引入不相关变体。 -- **多样化策略**：使用同义词替换、随机抽样、翻译变换等多种策略提升数据多样性。 -- **平衡增强**：生成数据应在类别、难度和结构上相对平衡，避免过度接触特定类型数据导致过拟合。 - -## 限制与注意事项 - -- 数据管理与数据处理能力仅适用于华北2（北京）地域。 -- 数据处理暂未提供可用 API，需在控制台完成；仅支持 SFT-文本生成训练集（ChatML 格式），不支持图片理解和 DPO 训练集。 -- 压缩包格式为 ZIP，最大 2GB；VL 训练集图片单张尺寸宽高均不超过 1024px、单张不超过 10MB；图生视频训练集图像/视频最大分辨率 4096*4096。 -- `loss_weight`、思考模式相关参数等部分能力属于邀测参数，如需使用请联系商务经理。 -- 数据流任务执行期间暂不支持手动终止，处理失败建议提交工单咨询。 - -> **注意**：构建有效的 SFT 训练集通常需要 1000+ 样本；数据增强节点暂不支持选择模型，固定使用千问-Max。更多信息参考 [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) 与 [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md)。 +- **地域限制**：所有功能（训练集管理、数据处理、日志回流）均**仅限华北2（北京）和新加坡 Region**，其他地域不可用； +- **格式与容量**： + - ZIP 包最大 2 GB（训练集）或 20 MB（图生视频 `data.jsonl`）； + - 图片单张 ≤ 1024×1024 px 且 ≤ 10 MB；视频 ≤ 4096×4096 px； + - 文本生成评测集仅支持单轮对话 Excel 格式（`.xlsx`），多轮不生效； +- **版本与覆盖**：数据清洗/增强、日志回流均生成**独立新版本**，不会覆盖原始数据集，但需手动切换版本用于训练； +- **模型兼容性**： + - 视频参数（`fps`, `video_start` 等）仅 Qwen3.5+ VL 模型支持； + - 思考模型训练后，若样本中存在无 `` 标签的 `assistant` 输出，则**不建议开启思考模式调用**； +- **权限与费用**：日志回流需提前开通 SLS 并授权服务角色；推理日志开启后将产生 SLS 存储与读写费用，长期不用应及时关闭。 ## 来源文档 - [训练集与评测集](../../raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - [数据清洗或增强](../../raw/model-user-guide/model-data-overview/data-processing.md) - - - - - - - - - - - - - - - - - - - +- [日志回流](../../raw/model-user-guide/model-data-overview/model-log-backflow.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md index 0491c527..1015ade9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-deployment-1.md @@ -1,100 +1,48 @@ # model deployment 1 -模型部署让你为平台预置模型或调优后的自定义模型获得独立、资源专享的推理服务，以满足高并发、低延迟等生产需求。本页汇总三种计费方式的选型、PTU 长输入与前缀缓存机制、LoRA 模型导入约束，以及通过控制台或 API 完成部署的完整流程，面向需要落地专属推理服务的开发者。 +`model deployment 1` 是百炼平台面向生产环境的核心推理服务部署能力，提供三种正交计费与资源模型：预置吞吐（PTU）、模型单元（MU）和按 [Token](../concepts/token.md) 用量（LoRA）。开发者可根据业务对吞吐稳定性、延迟确定性、成本敏感度及模型定制深度的需求，选择最适配的部署方式。所有部署均通过统一 API 接口管理，支持自动化扩缩容与细粒度监控。 -## 三种计费方式与选型 +## 支持的模型/功能 -百炼提供三种互斥的部署计费方式，计费方式在服务创建后无法更改，如需切换必须先下线已部署的模型再重新部署（详见 [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md)）： +- **预置吞吐（PTU）**：适用于高并发、可预测流量场景，支持长输入（最高 256K token）与前缀缓存，当前覆盖 `qwen3.7-plus-2026-05-26`、`deepseek-v4-pro`、`glm-5.1` 等主流模型，详见[预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)。 +- **模型单元（MU）**：适用于需独占算力、自定义性能指标（如首 [Token](../concepts/token.md) 延迟、TPM 上限）的场景，支持 PD 分离计算模式、思考/非思考推理模式切换及最长上下文配置，覆盖全部千问系列、GLM、DeepSeek 及千问 VL 模型，详见[模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md)。 +- **按 [Token](../concepts/token.md) 用量（LoRA）**：仅支持经百炼平台完成 LoRA 微调后的自定义模型，按实际输入/输出 token 计费，不支持全参微调模型；该模式下模型必须满足严格格式约束（如 `adapter_model.safetensors` + `adapter_config.json`、rank ∈ {8,16,32,64}、未修改 vocab/chat_template 等），详见[模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。 -- **预置吞吐（PTU，Provisioned Throughput Unit）**：平台预留资源保障特定 TPM 吞吐能力，额度内不限速。相比按 Token 计费，TPS 通常提升约 1.5～2.0 倍，适合流量可预估的高负载生产环境（智能客服、实时内容审核）。支持预付费（按天）与后付费（按小时），可自助增减吞吐量并设置自动续费。 -- **模型单元（MU）**：按使用时长 × 模型单元数量计费，资源独占，延迟/吞吐等性能指标可自定义。支持部分预置模型与所有调优后模型，可自助增减模型单元数量，支持 PD 分离计算模式（拆分 Prefill 与 Decode 阶段以降低首 Token 延迟、提高吞吐）。 -- **按 Token 使用量**：以每次调用的输入/输出 Token 计量，不使用不计费。仅支持对基础模型完成 SFT 高效训练后的自定义模型，主要用于调优后模型的效果验证；扩缩容需在控制台提交申请等待人工审核。 +> **注意**：文档 3 中表格显示 `glm-5.1` 输入上限为 64K，但文档 1 明确其支持 200K token 长输入且阶梯系数生效。以[预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)为准，该模型实际支持 200K。 -关键计费公式： +## 关键参数 -- 预置吞吐（按时长）：`费用 = 使用时长 × (输入 TPM 单价 × 输入 TPM + 输出 TPM 单价 × 输出 TPM)` -- 模型单元（按时长）：`费用 = 使用时长（小时）× 模型单元数量 × 模型单元单价`；预付费按月时改为 `包月数 × 模型单元数量 × 月单价` -- 按 Token：`费用 = 输入 Token 数 × 输入单价 + 输出 Token 数 × 输出单价` +| 参数名 | 适用部署类型 | 说明 | 示例值 | +|--------|--------------|------|--------| +| `ptu_capacity.input_tpm` / `output_tpm` | PTU | 预置输入/输出吞吐量（单位：token/min），决定额度购买量 | `"input_tpm": 10000` | +| `deploy_spec` / `capacity` | MU | 模型单元规格（如 `MU1`, `MU3`）与副本数，直接影响算力与并发 | `"deploy_spec": "MU1", "capacity": 4` | +| `enable_thinking` | MU | 是否启用思考模式（影响推理逻辑与计费单价） | `true` | +| `max_context_length` | MU | 最长上下文长度（部分模型支持，单位：token） | `10000` | +| `rpm_limit` / `tpm_limit` | MU | 服务级限流阈值（RPM/TPM），用于保障服务质量 | `"rpm_limit": 500` | +| `plan: "lora"` | LoRA | 必须显式指定，`capacity` 字段在该模式下无效但需填写 | `"plan": "lora", "capacity": 1` | -## PTU 长输入与前缀缓存 +## 使用方式 -PTU 部署支持长输入请求（部分模型最高 200K token）和前缀缓存，通过阶梯容量系数和缓存折扣管理额度消耗，详见 [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md)： +- **控制台操作**：前往[模型部署控制台](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/efm/model_deploy/create)，选择模型、计费方式及对应参数后提交。 +- **API 调用**：使用 DashScope REST API 创建部署任务： + - PTU：`POST /api/v1/deployments`，`"plan": "ptu"` + `ptu_capacity` 对象； + - MU：`"plan": "mu"` + `deploy_spec`, `capacity`, `enable_thinking` 等字段； + - LoRA：`"plan": "lora"` + `model_name`（必须为已导入的 LoRA 模型 ID）。 + 全流程示例见[使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。 +- **状态管理**：通过 `GET /api/v1/deployments/{deployed_model}` 查询状态（`PENDING` → `RUNNING` 表示就绪），`DELETE /api/v1/deployments/{deployed_model}` 下线服务。 -- **长输入阶梯系数**：超过 32K token 的输入按更高阶梯系数折算 TPM。例如 glm-5.1 在 `[32K, 200K]` 区间输入系数为 1.33、输出为 1.17；deepseek-v4-pro 与 qwen3.7-plus-2026-05-26 无阶梯（1.0）。 -- **前缀缓存折扣**：命中缓存的输入 token 按折扣系数消耗额度（glm-5.1 为 0.2，deepseek-v4-pro 为 0.08，qwen3.7-plus 为 0.2），可显著降低多轮对话和重复前缀场景的额度消耗。 -- **自动转按量计费**：超出 PTU 额度或输入超过模型上限时，请求自动转为按量计费，无需修改调用代码，业务不中断。 +## 限制和注意事项 -API 响应关键字段：`service_tier`（值为 `ptu-standard` 表示使用 PTU 额度，`default` 或不返回表示按量计费）、`provisioned_tokens`（折算后实际消耗的额度 token 数）、`cached_tokens`（前缀缓存命中数）。不同 API 格式（OpenAI Chat / Responses、Anthropic、DashScope）下这些字段的 JSON 路径不同，需按对应格式取值。 - -> **注意**：模型输入上限存在两处口径。[模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) 的价格表中千问系列多为 128K、部分新模型达 256K，而 PTU 文档明确将「千问 128K / DeepSeek 64K」作为触发自动转按量计费的上限。请以控制台实际展示与所选具体模型为准。 - -> **注意**：长输入场景下 PTU 利用率可能超过 100%，这是阶梯系数导致折算消耗高于原始 token 数的正常现象，超出部分自动转按量计费，不影响服务可用性。 - -## 模型导入（LoRA） - -通过**我的模型**页面可将本地训练的 LoRA 模型从 OSS 导入百炼平台，详见 [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md)。当前版本**仅支持 LoRA 模型，不支持全参微调模型**。 - -导入前提与约束： - -- **OSS Bucket**：需为目标 Bucket 添加 `bailian-datahub-access` 标签（标签值 `read`）；不支持归档/冷归档类存储；不支持访问 Bucket 根目录文件，需放入子目录。首次导入需先完成 OSS 服务关联角色授权（子账号还需主账号授予 `ram:CreateServiceLinkedRole` 权限）。 -- **必需文件**：`adapter_model.safetensors`（权重）与 `adapter_config.json`（含 rank、alpha 等配置）。 -- **rank 限制**：必须为 8、16、32、64 之一，且同一模型所有 LoRA 层使用相同 rank。 -- **词汇表与对话模板**：不得修改原始 vocab 或 chat_template，必须与开源基础模型默认配置一致，否则无法导入。 -- **VL 模型**：必须冻结 VIT，若 adapter 中包含 `visual` 开头的权重参数则无法导入。 - -支持导入的基础模型涵盖千问3、千问3-VL、千问2.5、千问2.5-VL 系列的指定版本。导入后模型状态包括创建中、创建成功（可部署）、创建失败、已失效。 - -> **注意**：导入模型若与本地 vLLM/SGLang 推理效果不一致，通常是推理引擎参数默认值差异所致。可将 `temperature`、`top_p`、`repetition_penalty` 设为 1.0、`presence_penalty` 设为 0 以对齐 vLLM 默认行为。 - -## 使用 API/命令行部署 - -除控制台外，可通过 DashScope HTTP API 完成部署，**仅适用于华北2（北京）地域**，需先获取并配置 API Key，详见 [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md)。核心接口为 `POST/GET/DELETE https://dashscope.aliyuncs.com/api/v1/deployments`，通过 `plan` 字段区分计费方式： - -- **PTU**：`plan: "ptu"`，配合 `ptu_capacity.input_tpm` / `output_tpm`。 -- **模型单元**：`plan: "mu"`，配合 `deploy_spec`（如 `MU1`）、`capacity`（副本数）、`enable_thinking`、`max_context_length`、`rpm_limit`、`tpm_limit`。 -- **按 Token（LoRA 自定义模型）**：`plan: "lora"`，`capacity` 必填但设置无效，扩缩容需在控制台申请。 - -典型部署命令（模型单元）： - -```bash -curl "https://dashscope.aliyuncs.com/api/v1/deployments" \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data '{ - "name": "my_qwen_plus", - "model_name": "qwen-plus-2025-12-01", - "plan": "mu", - "deploy_spec": "MU1", - "enable_thinking": true, - "capacity": 4, - "max_context_length": 10000, - "rpm_limit": 500, - "tpm_limit": 1000 -}' -``` - -部署流程：创建部署 → 返回 `deployed_model`（专属服务唯一 ID）→ 轮询 `GET /deployments/{id}` 直到 `status` 为 `RUNNING` → 通过 DashScope SDK 或兼容 API 发起推理 → 不再使用时 `DELETE /deployments/{id}` 下线并停止计费。 - -## 部署配置与列表管理 - -在控制台部署时可配置：服务名称、选择模型、模型单元类型（部署规格）、部署副本数、部署模板（如「单机部署」，仅模型单元模式可用）、推理模式（Instruct 非思考 / Thinking 思考）、最长上下文、服务限流（RPM/TPM）。 - -部署列表页展示服务名称、模型名称、**模型 Code**（API 调用时指定模型的唯一标识）、部署状态（待部署、部署中、运行中、部署失败、下线中、已停止、变配中等）、计费方式、部署详情与限流详情。 - -## 限制与注意事项 - -- **计费不可逆变更**：计费方式创建后不可改；预付费按天/按月无法提前退费，首月内提前退订按日单价 1.2 倍计费。 -- **部署即计费**：模型部署成功后即产生费用，即便尚未发起任何调用；后付费欠费后资源保留并继续计费 24 小时，超时后停止计费并删除底层资源（部署任务保留）。 -- **按 Token 模式约束**：仅支持 LoRA 调优后模型，一个月内不使用将自动释放。 -- **权限**：API 部署报错 `Workspace ... does not have deployment privilege` 或 `Workspace access denied` 时，需检查 API Key 归属业务空间的模型部署授权与账号操作权限。 -- **删除不可恢复**：执行 DELETE 后服务立即下线且不可恢复。 +- **PTU 溢出策略**：创建时必须选择「自动溢出」（默认）或「仅使用 PTU 容量」。前者超限转按量计费（响应头含 `x-dashscope-ptu-overflow:true`），后者直接返回 HTTP 429；两种策略下超出模型原生 token 上限（如 Qwen 128K）均自动转按量计费。 +- **LoRA 导入硬约束**：仅支持 LoRA 微调模型，不支持全参微调；要求 `adapter_config.json` 中 `rank` 值为 8/16/32/64，且所有层一致；禁止修改基础模型 vocab 或 `chat_template`；VL 模型必须冻结 VIT（即 `adapter_model.safetensors` 中不得含 `visual.*` 权重）。 +- **地域与权限**：API 部署仅支持华北2（北京）地域；调用方 API Key 所属业务空间必须已获目标模型的部署权限，否则报错 `Workspace xxx does not have deployment privilege for model xxxx`。 +- **计费生效时机**：部署成功（状态变为 `RUNNING`）即开始计费，与是否发起推理请求无关。PTU 和 MU 为预付费/后付费按使用时长计费，LoRA 为随用随付按 token 计费。 ## 来源文档 -- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [预置吞吐长输入与缓存](../../raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) - [模型导入](../../raw/model-user-guide/model-deployment-1/model-import.md) +- [模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [使用 API或命令行进行模型部署](../../raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md index 674be24b..f28f4a59 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-evaluation-introduction.md @@ -1,97 +1,55 @@ # model evaluation introduction -模型评测是百炼平台提供的模型能力评估功能，支持自定义评测和基线评测两种方式，通过评测维度对模型推理结果进行打分和对比，帮助开发者选择最优模型或验证调优效果。当前仅支持文本生成类模型评测。 +模型评测是百炼平台提供的模型能力量化评估功能，支持通过自定义或基线方式对文本生成类模型的推理结果进行多维度打分与对比。它服务于模型选型、调优验证、能力归因和持续质量监控等核心场景，提供 AI 自动评测、规则评估和人工评估三种评分范式。所有评测均基于明确的评分标准（评测维度）执行，并生成可下载的结构化报告。 -## 核心概念 +## 支持的模型/功能 -模型评测涉及两个容易混淆的核心概念： +- **支持模型类型**：当前仅支持文本生成类模型（包括预置模型与调优后模型），不支持[多模态](../concepts/multi-modal.md)、语音或结构化输出模型。具体可用模型列表请参见[模型评测产品概览](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 +- **核心评测范式**： + - **大模型评估**（AI 自动评测）：使用裁判模型（如千问-Max）对输出进行语义级评分或分类，适用于问答质量、内容安全等无确定性答案的场景； + - **规则评估**（自动化指标）：基于算法（ROUGE/BLEU/Cosine/字符串匹配等）直接计算分数，适用于翻译、摘要、Function Calling 等有明确标准的场景； + - **人工评估**（人工标注）：由人工逐条标注 Pass/Fail，适用于创意写作、合规审核等主观性强的场景。 +- **评测模式**： + - **自定义评测**：用户自主上传评测数据集（EvaluationSet 类型）、创建评测维度、配置任务，支持全地域、结果下载与排行榜； + - **基线评测**：使用平台预置公开数据集（如 C-Eval、GSM8K、BBH），仅限北京地域可用，不支持自定义维度与结果下载，详见[创建基线评测任务](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 -- **评分器 Prompt**：配置于评测维度，指导裁判模型如何给被评测模型的回答打分。 -- **System Prompt**：配置于评测任务，为被评测模型设定角色定位或行为规范，通常可留空。 - -两者作用对象和费用归属不同，详见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)。 - -## 评测方式 - -### 自定义评测 - -使用自有数据集和自定义评测维度，支持三种评分方式： - -| 评分方式 | 说明 | 适用场景 | -|---------|------|---------| -| 大模型评估（AI 自动评测） | 由裁判模型（推荐千问-Max）对回答评分 | 问答质量、内容安全等语义理解场景 | -| 规则评估（自动化指标） | ROUGE、BLEU、Cosine 等算法直接计算 | 翻译、摘要、Function Calling 等确定性场景 | -| 人工评估（人工标注） | 人工逐条标注 Pass/Fail | 创意性写作、专业领域判断 | - -### 基线评测 - -使用公开标准数据集（C-Eval、MMLU、GSM8K、BBH 等）快速评测模型基础能力，无需自行准备数据集或配置维度。 - -> **注意**：基线评测仅北京地域可用，不支持下载评测结果，不支持推理结果集数据来源。 - -## 评测维度类型 - -评测维度定义模型的评分规则，创建为模板后可被多个评测任务复用。百炼提供五种评分器类型，详细说明参见[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 - -| 维度类型 | 评分方式 | 需要参考答案 | 裁判模型费用 | -|---------|---------|------------|------------| -| 大模型评估-数值型 | 裁判模型打分（整数，如 0-5） | 否 | 有 | -| 大模型评估-分类型 | 裁判模型标签（Pass/Fail） | 否 | 有 | -| 规则评估-文本相似度 | ROUGE/BLEU/Cosine 等算法 | 是 | 无 | -| 规则评估-字符串匹配 | 相等/不相等/包含 | 是 | 无 | -| 人工评估-分类型 | 人工标注 Pass/Fail | 否 | 无 | - -**选型决策路径**：有标准答案且格式固定用字符串匹配；有标准答案但表述多样用文本相似度；无标准答案需语义理解用大模型评估；需主观判断用人工评估。 +> **注意**：文档 1 称“基线评测仅北京地域可用”，文档 2 未提及地域限制，但未否定该约束；以文档 1 的明确声明为准，开发者在非北京地域控制台将不显示基线评测选项，属正常行为。 ## 关键参数 -### 评测维度参数 - -- **评分范围**（数值型）：裁判模型打分区间，整数，默认 0-5。建议不超过 10，范围过大会降低 LLM 评分一致性。 -- **通过阈值**（数值型/相似度型）：判定 Pass/Fail 的分界线，步长 0.1（数值型）或 0.01（相似度型）。 -- **评分器 Prompt**（大模型评估类型）：至少包含一个变量（`${prompt}`、`${output}`、`${completion}`），长度不超过 50000 字符。 -- **Pass/Fail 标签**（分类型/人工评估）：两组标签不可重复。 - -> **注意**：维度类型创建后不可修改，选错需删除重建。 - -### 评测任务参数 - -- **数据来源**：评测数据集（含 Prompt + Completion，会产生推理费用）或推理结果集（已含 Output，不产生推理费用）。 -- **推理参数**：Temperature、TopP、System Prompt 等，按所选模型动态加载。 -- **排行参与**：开启后须绑定排行榜，结果加入排名对比。 - -## 使用流程 - -端到端流程分四步： - -1. **准备数据集**：在数据管理模块上传评测集类型数据（含 Prompt 和 Completion 列）。 -2. **创建评测维度**：定义评分标准和方式，选择评分器类型并配置参数。 -3. **创建评测任务**：选择被评测模型、关联数据集和维度，提交评测。 -4. **查看结果**：在指标统计 Tab 查看综合得分和通过率，在数据明细 Tab 查看逐条评分。 - -数据量建议：小规模验证 50-100 条，正式评测 200-500 条，全面评估 500 条以上。 - -## 计费说明 - -费用由两部分构成：被评测模型推理费用 + 裁判模型评分费用。 - -- 使用推理结果集可免去推理费用。 -- 规则评估和人工评估无裁判模型费用。 -- 已部署的调优模型评测不额外计费（推理费用包含在部署算力费用中）。 - -**成本优化**：先用 50-100 条小规模验证 → 保存推理结果集复用 → 有确定性标准的场景优先用规则评估。 - -## 限制与注意事项 - -- 当前仅支持文本生成类模型评测。 -- 基线评测仅北京地域可用。 -- 任务提交后不可更换目标模型，需删除重建。 -- 评测维度类型创建后不可修改。 -- 模型评测当前仅支持控制台操作，不提供公开 API/SDK，如需编程化评测可参考 PAI Judge Model API。 -- 1-3% 的分数差异通常为评测噪声，不建议仅凭微小分差做决策。 -- LLM 评分器存在位置偏差和自我偏好偏差，建议定期人工抽查校准。 - -更多操作细节及常见问题排查请参见[模型评测](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)和[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 +| 参数类别 | 参数名 | 说明 | 必填性 | 取值示例 | +|----------|--------|------|--------|-----------| +| **通用** | 维度名称 | 最长 20 字符，建议采用“评估方面+评估方式”命名（如`回答准确性-LLM评分`） | 是 | `安全性判定-分类型` | +| **大模型评估** | 裁判模型 | 执行评分的 LLM，影响准确率与费用 | 大模型类型必填 | `qwen-max` | +| | 评分器 Prompt | 指导裁判模型打分的提示词，**必须包含至少一个变量**：`${prompt}`、`${output}` 或 `${completion}` | 大模型类型必填 | `请判断${output}是否符合${prompt}要求且无事实错误，仅输出Pass或Fail` | +| | 评分范围（数值型） | 整数区间，最小值 ≥ 0，最大值 ≥ 1 | 数值型必填 | `0~5` | +| | 通过阈值 | 判定 Pass 的最低分（数值型）或相似度（规则型），步长 0.1（数值）/0.01（相似度） | 数值型/相似度型必填 | `3.0`（数值）、`0.75`（相似度） | +| **规则评估** | 匹配规则 / 相似度算法 | 字符串匹配支持相等/不相等/包含；文本相似度支持 ROUGE-1/ROUGE-L/BLEU/Cosine 等 7 种 | 是 | `ROUGE-L`, `包含` | +| **人工评估** | Pass/Fail 标签 | 标签互斥且穷尽，各标签 ≤ 20 字符 | 是 | `合规`, `不合规` | + +完整参数说明与配置逻辑请参考[评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)。 + +## 使用方式 + +1. **准备数据**：在数据管理模块上传 `EvaluationSet` 类型数据集（含 `Prompt` 和 `Completion` 两列），或复用已有的 `InferenceResultSet`（含 `Prompt`、`Output`、`Completion`）。 +2. **创建维度**：在「模型评测 > 评测维度」中创建至少一个维度模板。类型一经创建不可修改，选错需删除重建（[创建评测维度](../../raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md)）。 +3. **创建任务**： + - 自定义评测：选择模型、指定数据来源（评测数据集 or 推理结果集）、关联维度、设置 System Prompt（可选）； + - 基线评测：仅北京地域可见，选择模型与预置数据集（如 `MMLU`），无需配置维度。 +4. **查看结果**：任务状态为「评测完成」后，在详情页的「指标统计」Tab 查看综合得分、通过率及分布图；「数据明细」Tab 查看每条样本的逐项评分。支持结果下载（待执行/基线任务不支持）。 + +## 限制和注意事项 + +- **模型限制**：仅支持文本生成类模型；[多模态](../concepts/multi-modal.md)、语音、代码生成等非纯文本生成模型暂不支持。 +- **地域限制**：基线评测功能仅在北京地域可用，其他地域控制台不展示该选项。 +- **维度不可变性**：评测维度的类型、裁判模型、评分范围等核心配置创建后不可修改，需删除重建（[评测维度设计建议](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md)）。 +- **费用说明**： + - 使用「评测数据集」会触发被评测模型推理，按 [Token](../concepts/token.md) 计费； + - 大模型评估维度（数值型/分类型）额外产生裁判模型评分费用； + - 规则评估与人工评估无裁判模型费用； + - 推理结果集方式可避免重复推理费用。 +- **数据量建议**：小规模验证用 50–100 条；正式评测建议 200–500 条；全面评估建议 ≥500 条。 +- **结果解读**：综合得分是各维度平均分，易掩盖维度间差异；应结合「分数分布图」与「数据明细」逐维度分析短板；1–3% 分差通常属评测噪声，不宜作为决策依据。 ## 来源文档 @@ -99,9 +57,3 @@ - [评测维度](../../raw/model-user-guide/model-evaluation-introduction/evaluation-metrics.md) - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md index 5cedd681..db4424ea 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-experience.md @@ -1,96 +1,70 @@ # model experience -百炼平台按模态和场景组织了一整套可直接调用的模型能力：文本生成、视觉理解、图片/视频/3D 生成、语音合成与识别、语音转语音、全模态、音乐生成，以及向量与重排序。本页汇总各场景的推荐模型、关键参数和选型要点，帮助开发者快速定位到合适的模型；模型的实时上下文窗口、计费等详细参数请以模型广场为准。 +`model experience` 是百炼平台统一的模型能力体验层，面向开发者提供覆盖文本、视觉、音频、3D、视频等[多模态](../concepts/multi-modal.md)任务的标准化调用接口与能力矩阵。所有模型均通过统一的 DashScope API 接入，支持同步/异步、流式/非流式、HTTP/WebSocket 等多种交互模式，并在功能支持（如 Function Calling、思考模式、结构化输出）、上下文长度、输入模态和计费维度上形成清晰的分层体系。选型应优先依据任务类型与核心能力需求，再结合成本、延迟与地域约束综合决策。 -## 文本生成 +## 支持的模型与功能 -通用文本场景（聊天机器人、内容生成、摘要、文档处理、办公任务）推荐从 `qwen3.7-plus` 起步——能力与成本均衡，具备 1M 上下文、Function Calling 和内置工具；效果确认后可切到 `qwen3.6-flash` 降本，功能与上下文一致；需要最强推理时用 `qwen3.7-max`。超长文档（多合同审阅、大规模文献）可用 `qwen-long`（10M 上下文）。AI 编程 / Agent 开发推荐 `qwen3.7-plus`（工具调用完整、1M 上下文适合大代码库）。 +百炼平台按模态与任务类型组织模型能力，主要分为以下五类： -关键能力： +- **文本生成**：以 `qwen3.7-plus` 为旗舰，支持 1M 上下文、Function Calling、内置工具（联网搜索/代码解释器）、结构化 JSON 输出及 `enable_thinking` 控制的深度推理模式；轻量场景可选用 `qwen3.6-flash` 或 `deepseek-v4-flash` [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **视觉理解**：`qwen3.7-plus` 同时支持图像、视频（最长2小时）、OCR 及多图理解；专用 OCR 模型 `qwen3.5-ocr` 在文档/手写识别上更优；Qwen3-VL 系列（如 `qwen3-vl-plus`）专为图文联合建模优化 [原文标题](../../raw/model-user-guide/model-experience/vision-model.md)。 +- **图片生成与编辑**：`wan2.7-image-pro` 支持文生图（4096×4096）、多图参考编辑与角色一致性；`qwen-image-3.0-pro`（邀测中）支持负向提示词与多语言字体渲染；`z-image-turbo` 适用于低成本快速生成 [原文标题](../../raw/model-user-guide/model-experience/image-model.md)。 +- **3D/视频/音频生成**：Tripo 系列（`Tripo/Tripo-P1.0`）支持文/图/多图生3D；视频生成推荐 `happyhorse-1.1-t2v`（文生视频）或 `wan2.7-i2v-2026-04-25`（首尾帧续写）；Fun-Music（`fun-music-v1`）支持歌词/提示词驱动的歌曲生成，但需申请邀测 [原文标题](../../raw/model-user-guide/model-experience/fun-music.md)。 +- **语音与全模态**：语音识别推荐 `fun-asr`（支持说话人分离）或 `qwen3.5-omni-plus`（支持 Prompt 注入与情感识别）；语音合成首选 `qwen-audio-3.0-tts-plus`（支持声音复刻与指令控制）；全模态模型 `qwen3.5-omni-plus` 支持音视频+文本联合理解与 Function Calling [原文标题](../../raw/model-user-guide/model-experience/omni.md)。 -- **思考模式**：通过 `enable_thinking` 参数开启（Responses API 用 `reasoning.effort` 控制开关与深度），所有 Qwen3 及以上模型均支持，多为混合模式可按请求切换。 -- **Function Calling**：所有通用模型均支持自定义工具调用；**内置工具**（联网搜索、代码解释器、网页抓取）免复杂配置。 -- **结构化输出**：可强制返回有效 JSON，适合信息抽取。 -- **批量推理**：适合大量、低时延要求的请求以降本。 +> **注意**：文档 1 与文档 2 均将 `qwen3.7-plus` 列为视觉理解首选，但文档 2 明确其支持“最长2小时视频”，而文档 1 未提及视频能力——该差异源于文档 1 聚焦通用文本生成，视觉能力属延伸支持，实际调用需以文档 2 的视觉输入规格为准。 -详细的模型对位（从 GPT / Claude / Gemini 迁移）与完整模型清单见 [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +## 关键参数 -## 视觉理解与 OCR +不同模态模型共用核心参数，但语义与约束各异： -图像 / 视频理解推荐 `qwen3.7-plus`（1M 上下文、最长 2 小时视频、Function Calling + 内置工具），稳定后可降本到 `qwen3.6-flash`。要点见 [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md)： +- `model`：必填，指定模型 ID（如 `qwen3.7-plus`, `wan2.7-image-pro`, `Tripo/Tripo-P1.0`）。 +- `input`：结构因模态而异： + - 文本/语音：`{"text": "..."}` 或 `{"audio_url": "..."}`； + - 视觉：`{"image": "url"}`, `{"images": ["url1", "url2"]}`, `{"video": "url"}`； + - 3D：`{"prompt": "..."}`, `{"image": "url"}`, 或 `{"images": [...]}`（三者互斥）[原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)； + - 音乐：`{"prompt": "...", "gender": "female"}` 或 `{"lyrics": "...", "is_instrumental": true}`。 +- `parameters`：控制生成行为： + - `texture_quality`（Tripo）、`format`（Fun-Music）、`geometry_quality`（Tripo-H3.1）； + - `reasoning.effort`（文本思考深度）、`enable_thinking`（布尔开关）； + - `max_output_tokens`（部分模型限制输出长度）。 +- `X-DashScope-Async: enable`：异步任务必需头（如 Tripo 3D、批量视频生成）。 -- **图像分辨率**：多数模型支持每张最高 1600 万像素，Token 数按 `h x w / (32 x 32) + 2` 计算。 -- **视频支持**：`qwen3.7-plus` / `qwen3.6-plus` / `qwen3.6-flash` / `qwen3.5-plus` / `qwen3.5-flash` 最长 2 小时 / 2GB；`qwen3-vl-plus` / `qwen3-vl-flash` 最长 1 小时。 -- **OCR / 文档提取**：`qwen3.5-ocr` 针对文档、表格、试卷、手写内容优化；通用图片文字提取也可用旗舰模型。 +## 使用方式 -## 图片生成与编辑 +统一采用 RESTful API，基础流程如下： -见 [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md)。推荐 `wan2.7-image-pro`，集成文字渲染、品牌色控制、角色一致性多图生成和图片编辑（文生图最高 4096x4096，编辑最高 2048x2048，支持最多 9 张输入图参考）。仅需生图且追求速度/成本时用 `z-image-turbo`（约快 10 倍、价格约 1/5，写实人像与产品照）；需要负向提示词或单次最多 6 张变体时用 `qwen-image-2.0-pro`（生成和编辑同一模型 ID）。 +1. **认证**：通过 `Authorization: Bearer $DASHSCOPE_API_KEY` 传入 API Key（需在华北2北京地域开通）； +2. **端点**：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/{service}/{action}`，其中 `service` 如 `aigc/text-generation`、`audio/music/generation`、`aigc/video-generation/3d-generation`； +3. **同步调用**：HTTP POST 直接返回结果（适用于文本、TTS、ASR 小文件）； +4. **异步调用**： + - 先 POST 创建任务获 `task_id`； + - 再 GET `/api/v1/tasks/{task_id}` 轮询状态（建议间隔 ≥15 秒）[原文标题](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md)； +5. **流式响应**：WebSocket 连接（实时 TTS/ASR/S2S）或 HTTP chunked encoding（部分文本/语音模型）。 -## 视频生成与编辑 +## 限制和注意事项 -见 [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md)，按子场景选型： - -- **文生视频**：`happyhorse-1.1-t2v`（1080P、单片段最长 15 秒、有声）；需传入自定义音频文件用 `wan2.7-t2v-2026-06-12`。 -- **图生视频**：首帧生视频用 `happyhorse-1.1-i2v`；首尾帧串联长视频用 `wan2.7-i2v-2026-04-25`。 -- **参考生视频**：`happyhorse-1.1-r2v`（保持角色一致性）；需自定义音色或视频参考主体用 `wan2.7-r2v-2026-06-12`。 -- **视频编辑 / 角色动画**：编辑用 `happyhorse-1.0-video-edit`，特效/运镜复刻用 `wan2.7-videoedit`；动作迁移用 `wan2.2-animate-move`，人物替换用 `wan2.2-animate-mix`（均支持 wan-std / wan-pro 两种模式）。 - -## Tripo 3D 模型生成 - -支持文生 3D、单图生 3D、多图生 3D 三种模式，通过 `input` 中互斥的 `prompt` / `image` / `images` 字段区分。`Tripo/Tripo-H3.1` 面向高精度（最高 200 万面，较慢），`Tripo/Tripo-P1.0` 面向快速预览（最高 2 万面，更快）。贴图质量用 `parameters.texture_quality`（`standard` / `detailed`）控制，几何精度用 `parameters.geometry_quality`（仅 H3.1 支持，`standard` 最高 150 万面 / `ultra` 最高 200 万面）。调用为异步任务，需轮询任务状态（`PENDING` → `RUNNING` → `SUCCEEDED` / `FAILED`），建议间隔 15 秒。 - -> **注意**：Tripo 3D 生成仅适用于**华北2（北京）**地域，且必须使用该地域的 API Key。 - -## 语音合成（TTS） - -见 [语音合成](../../raw/model-user-guide/model-experience/tts-model.md)。先确定内置音色还是自定义音色：标准合成推荐 `qwen-audio-3.0-tts-plus` / `MiniMax/speech-2.8-hd`；自定义音色分**声音复刻**（提供音频样本，用 `qwen-audio-3.0-tts-flash` / `MiniMax/speech-2.8-hd`）和**声音设计**（用文字描述音色，用 `cosyvoice-v3.5-plus` / `cosyvoice-v3.5-flash`），音色统一由 `voice-enrollment` 服务注册管理。接入方式上，WebSocket 双向流式延迟最低（实时交互），HTTP 适合有声阅读等；Qwen 系列以 `-realtime` 后缀区分 WebSocket / HTTP。指令控制可用自然语言动态调节语速、情绪和风格。 - -## 语音识别（ASR） - -见 [语音识别](../../raw/model-user-guide/model-experience/asr-model.md)，按维度选型： - -- **实时 vs 非实时**：实时（WebSocket）用 `fun-asr-realtime` 或 `qwen3.5-omni-plus-realtime`；非实时文件转写（HTTP）用 `fun-asr` 或 `qwen3.5-omni-plus`。 -- **专业术语**：Prompt 上下文注入（Qwen3.5-Omni，无需预配置）或热词表（Fun-ASR，适合稳定术语列表）。 -- **说话人分离**：仅 Fun-ASR 非实时模型（`fun-asr`、`fun-asr-mtl`）支持。 -- **情感识别**：Qwen-ASR 与 Qwen3.5-Omni 系列支持，推荐 `qwen3-asr-flash-realtime` / `qwen3-asr-flash-filetrans`。 - -Paraformer 为较早一代模型，新业务建议迁移到 Fun-ASR 或 Qwen-ASR。 - -## 语音转语音（S2S）与全模态 - -构建语音应用可选 **S2S 单模型**（延迟低、端到端感知语调情绪）或 **Pipeline（ASR + LLM + TTS）**（可自定义音色、各阶段独立选优）。S2S 路线推荐：语音助手/客服用 `qwen3.5-omni-plus-realtime`，成本敏感用 `qwen3.5-omni-flash-realtime`，同传/直播翻译用 `qwen3.5-livetranslate-flash-realtime`，视频配音/播客翻译用 `qwen3-livetranslate-flash`。全模态（同时理解文本/音频/图片/视频）三大系列为 Qwen3.5-Omni（旗舰）、Qwen3-Omni-Flash（轻量、支持思考模式）、Qwen3.5-Livetranslate（专业翻译，开箱即用，60 种语言）。 - -> **注意**：Function Calling 与联网搜索能力在不同接入模式下差异较大——例如 `qwen3-omni-flash` 在 HTTP 模式支持 Function Calling 和思考模式，但其 WebSocket（`-realtime`）版本均不支持；联网搜索仅 Qwen3.5-Omni（HTTP / WebSocket）支持，且联网搜索与 Function Calling 不可同时开启；思考模式下不输出语音。选型前务必核对目标模型的具体接入模式。 - -## 音乐生成 - -Fun-Music 是端到端音乐生成模型，通过 `prompt` 描述风格/场景/情绪自动作词谱曲，或通过 `lyrics` 提供自定义歌词，用 `gender` 选男女声（仅 `fun-music-v1`），`is_instrumental=true` 生成纯音乐（此时 `lyrics` / `gender` 被忽略），`format` 指定 mp3 / wav 输出。 - -> **注意**：Fun-Music 处于邀测阶段，需在模型广场申请开通，且服务仅在**华北2（北京）**地域可用。 - -## 向量与重排序 - -见 [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md)。纯文本搜索 / RAG / 聚类推荐 `text-embedding-v4`（维度 64~2048，默认 1024，最大 8192 Token；迁移旧索引可用 `text-embedding-v3`）；跨模态检索用 `qwen3-vl-embedding`（融合向量）或 `tongyi-embedding-vision-plus`（独立向量）。重排序用于 Embedding 检索后对 Top-N 结果精排：纯文本用 `qwen3-rerank`（100+ 语言、最多 500 文档），多模态用 `qwen3-vl-rerank`（文本/图片/视频混排）。 - -## 选型与使用注意事项 - -- 旧版模型（如旧版 Qwen、Paraformer、`qwen-omni-turbo`、`qwen-tts` 等）不再作为首选，新项目建议使用各系列最新版本。 -- 上下文窗口、计费、地域可用性等以模型广场实时信息为准；本页面数据可能滞后。 -- 图片/视频/3D/音乐等生成类模型多为异步任务，需按文档轮询或配置回调。 +- **地域与服务开通**：Tripo 3D、Fun-Music 仅限华北2（北京）；部分模型（如 `qwen3.8-max-preview`）需 [Token](../concepts/token.md) Plan 权限 [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **输入约束**： + - 图像：单图 ≤1600万像素，[Token](../concepts/token.md) 数 = `h × w / (32 × 32) + 2`； + - 视频：`qwen3.7-plus` 最长2小时/2GB，`qwen3-vl-plus` 限1小时； + - 音频：`fun-asr` 非实时最大 12小时/2GB，`qwen3.5-omni-plus` 非实时限3小时/2GB。 +- **能力冲突**：Qwen3.5-Omni 的联网搜索与 Function Calling 不可同时启用；思考模式下不支持语音输出 [原文标题](../../raw/model-user-guide/model-experience/s2s-model.md)。 +- **版本管理**：快照版本（如 `qwen3.7-plus-2026-05-26`）保障稳定性，但旧版模型（Qwen3、Qwen2.5 系列）已停止更新，新项目应使用 Qwen3.6+ [原文标题](../../raw/model-user-guide/model-experience/text-generation-model.md)。 +- **跨模态兼容性**：[多模态](../concepts/multi-modal.md) Embedding（`qwen3-vl-embedding`）与重排序（`qwen3-vl-rerank`）需配套使用，不可混用文本 Embedding 模型。 ## 来源文档 - [文本生成](../../raw/model-user-guide/model-experience/text-generation-model.md) - [视觉理解](../../raw/model-user-guide/model-experience/vision-model.md) -- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) - [图片生成与编辑](../../raw/model-user-guide/model-experience/image-model.md) - [Tripo 3D模型生成](../../raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) +- [视频生成与编辑](../../raw/model-user-guide/model-experience/video-generate-edit-model.md) +- [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) - [语音合成](../../raw/model-user-guide/model-experience/tts-model.md) -- [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) - [语音识别](../../raw/model-user-guide/model-experience/asr-model.md) -- [音乐生成](../../raw/model-user-guide/model-experience/fun-music.md) -- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) +- [语音转语音](../../raw/model-user-guide/model-experience/s2s-model.md) - [全模态](../../raw/model-user-guide/model-experience/omni.md) +- [向量与重排序](../../raw/model-user-guide/model-experience/embedding-rerank-model.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md index 712b5a20..43707941 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-high-speed-inference.md @@ -1,93 +1,55 @@ # model high speed inference -百炼平台针对高吞吐、高速率的推理场景提供两类能力：**TPM 预留**用于为指定模型锁定专属推理容量，避免业务高峰期受公共限流影响；**快速模式（Fast mode）**则针对输出速度敏感的场景提升 TPS。两者都通过替换或指定 `model` 参数接入，无需大改代码。 +百炼平台提供两种面向高吞吐、低延迟场景的推理加速能力：**TPM 预留（[Token](../concepts/token.md) Per Minute Reservation）** 和 **快速模式（Fast Mode）**。前者通过预购专属容量保障确定性吞吐与稳定性，适用于流量可预估、不可接受限流的关键业务；后者通过底层调度与硬件优化提升单请求输出速度（TPS），适用于对响应延迟敏感的实时交互场景。二者可独立使用，也可组合部署（例如在 TPM 预留实例上启用 fast 模型）。 -## TPM 预留：锁定专属容量 +## 支持的模型/功能 -TPM（Tokens Per Minute）预留为指定模型锁定专属的推理吞吐量，预留容量内的调用不受公共资源限流影响，容量为业务专属、不与其他用户共享。详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md)。 +- **TPM 预留**：为指定模型锁定专属推理吞吐量（单位：kTPM），支持输入/输出维度独立配置，确保高峰期调用不受公共资源池限流影响。支持模型详见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档中的“支持的模型”表格，覆盖千问、GLM、DeepSeek、Kimi 等主流模型的多个版本（如 `qwen3.7-max-2026-05-20`、`glm-5.2`、`deepseek-v4-flash` 等），按地域（华北2/新加坡）分列定价与阶梯系数。 -核心特性： +- **快速模式**：当前仅支持 `glm-5.2-fast-preview` 模型（[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md)），处于 preview 阶段，提供 1.5~2 倍于标准 API 的 TPS（达 80~100 TPS），并引入排队机制缓解瞬时超载，不立即返回 429。 -- **专属模型 code**：创建预留后系统自动生成专属模型 code，需将 API 请求中的 `model` 参数替换为该 code 才能使用预留容量。 -- **超额不中断**：超出预留容量的请求自动降级为按量计费处理，服务不中断，无需修改代码。可在详情页的**超额降级统计**查看降级次数。 -- **计费单位**：按 kTPM 预付费（1 kTPM = 1,000 Tokens/分钟），一次性支付，从购买成功起连续生效。 +> **注意**：两篇文档对 `glm-5.2` 的缓存折扣描述存在差异——TPM 预留文档称其支持 `0.25` 缓存命中折扣（即缓存部分按 25% 折算容量），而快速模式文档未提及缓存折扣，且计费单价中“缓存命中”列为 `4元`（疑似指缓存命中时的输入单价）。实际缓存行为以 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中明确列出的参数为准，快速模式因处于 preview 阶段，其缓存策略可能尚未同步或未开放配置。 -### 方案选型 +## 关键参数 -[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 文档给出了多种容量方案的对比，便于按业务诉求选型： +| 参数 | TPM 预留 | 快速模式 | +|------|----------|----------| +| **核心指标** | 输入/输出 kTPM（1 kTPM = 1,000 tokens/min） | TPS（tokens per second），实测 80~100 | +| **计费单位** | 预付费（按天，kTPM × 天数） | 按 token 计费（与标准 API 一致） | +| **溢出策略** | 可选：自动溢出至按量计费（默认）或仅预留容量（返回 429） | 请求排队，不立即限流 | +| **专属标识** | 生成唯一 `dedicated model code`，需替换 API 中 `model` 字段 | 使用固定模型 ID（如 `glm-5.2-fast-preview`） | +| **接入域名** | 通用 DashScope 域名（`https://dashscope.aliyuncs.com/...`） | 地域专属域名（`https://{workspace_id}.{region}.maas.aliyuncs.com/...`） | -| 方案 | 计费单位 | 容量保障 | 适用场景 | 超额处理 | 代码改动 | -| --- | --- | --- | --- | --- | --- | -| 按量付费 | 按 token | 无（共享公共池） | 流量波动大/短期 | 自动服务，受公共限流 | 无需改动 | -| 资源包/节省计划 | 预付费额度 | 承诺用量折扣（非专属） | 费用优化 | 超出转按量 | 无需改动 | -| TPM 预留 | 按 kTPM 预付费 | 专属容量刚性兑付 | 流量可预估、不能接受限流 | 超出自动降级公共池按量，不中断 | 替换 model 参数 | -| PTU 专属部署 | 按 kTPM 预付费 | 专属部署实例 | 高吞吐高性能 | 超出转按量 | 替换 model 参数 | +## 使用方式 -### 创建与接入 +- **TPM 预留**： + 1. 在百炼控制台创建预留实例，选择目标模型、输入/输出 kTPM、购买时长及溢出策略； + 2. 获取详情页中的 **专属模型 code**； + 3. 将 API 请求中的 `model` 参数替换为该 code（[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中提供了 Python/curl 示例）； + > 注意：实例需处于“运行中”状态方可生效；首次大流量请求前存在短暂预热期，建议实现客户端重试。 -1. 登录百炼控制台创建 TPM 预留，填写预留名称、选择模型、付费周期（按天）、输入/输出 TPM（单位 kTPM）、购买时长（支持 1~30、60、90、120、365 天）等参数。建议先用 **TPM 容量计算器**（根据 RPM、平均输入/输出长度、缓存命中率估算）确认所需额度。 -2. 确认费用后完成支付。 -3. 在详情页**概览** Tab 复制**专属模型 code**。 -4. 将 API 请求的 `model` 参数替换为该 code 即可： +- **快速模式**： + 1. 确保业务空间已开通对应地域（如华北2）的 MaaS 服务； + 2. 从 [业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management) 获取 `workspace_id`； + 3. 构造请求 URL：`https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions`； + 4. 设置 `model: "glm-5.2-fast-preview"`（[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 提供了 OpenAI 兼容 SDK 调用示例）； + > 注意：`stream: true` 时需分别处理 `delta.reasoning_content` 和 `delta.content` 字段。 -```python -import dashscope +## 限制和注意事项 -response = dashscope.Generation.call( - api_key="your-api-key", - model="your-dedicated-model-code", # 替换为专属模型 code - messages=[{"role": "user", "content": "你好"}], -) -print(response.output.text) -``` +- **TPM 预留**： + - 预留容量按日计费，缩容/退订按 1.5 倍系数结算违约金； + - 服务到期后 2 小时内仍可调用，14 小时后实例删除且不可恢复； + - 专属模型 code 在退订后立即失效，回退至公共资源（按量计费）。 -> **注意**：短时间内请求量快速拉升时，系统需短暂预热以匹配算力，预热期间部分请求可能出现延迟波动，请做好请求排队或重试机制。 +- **快速模式**： + - 当前仅 `glm-5.2-fast-preview` 可用，其他模型暂不支持； + - 处于 preview 阶段，接口行为、性能指标及计费规则可能调整，不建议用于生产环境长期依赖； + - 不支持与 TPM 预留直接绑定（即无法为 `glm-5.2-fast-preview` 创建 TPM 预留），但可在同一业务空间下并行使用两者。 -### 容量换算参数 - -部分模型支持长输入阶梯系数和缓存折扣，容量计算器会自动应用。例如 glm-5.2 输入长度上限 1M、缓存折扣 0.25、无阶梯；glm-5.1 缓存折扣 0.2，且在 [32K, 200K] 区间输入系数 1.33 / 输出 1.17；deepseek-v4-pro 缓存折扣低至 0.08。Qwen3.6-flash-2026-04-16 不支持缓存。 - -### 管理与生命周期 - -- **扩缩容**：在详情页调整输入/输出 TPM，变配期间服务不中断。利用率持续接近 100% 或频繁降级时建议扩容。 -- **续费**：可手动续订，或开启**到期自动续费**（到期前一天 08:00 自动扣款）。 -- **退订**：跳转费用中心完成，退订后专属模型 code 失效、请求回退公共资源，不可恢复。缩容/退订退费按已使用部分 1.5 倍系数结算：`退款 = 降量部分预付费 - (降量部分预付费 × 已用时长/购买时长 × 1.5)`。 - -预留实例状态：服务到期后 2 小时内仍为**运行中**（可调用、可续费）；2~14 小时转为**已停止**（不可调用、仍可续费）；到期 14 小时后**已过期/删除**，不可恢复。其余状态包括待生效、变配中、已取消。 - -## 快速模式（Fast mode）：提升输出速度 - -[快速模式](../../raw/model-user-guide/model-high-speed-inference/fast-mode.md) 面向对输出速度敏感的场景（AI 编程助手、Agent 多步推理、实时对话等），当前处于 **preview 阶段**，能力与规格可能随版本调整。 - -关键特性： - -- **高速输出**：TPS 提升至标准 API 的 1.5~2 倍，达 80~100 TPS。 -- **按 token 计费**：计费逻辑与标准 API 一致，按输入/输出 token 计费。 -- **特殊限流**：超出 TPM 额度不会立即限流，请求进入排队队列。 - -### 接入方式 - -将 `model` 参数指定为支持的模型 ID（如 `glm-5.2-fast-preview`）即可开启，无需额外参数。接入域名格式为 `https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`，其中 `{workspace_id}` 可在业务空间管理页面切换到对应地域后查看。 - -```bash -curl -X POST https://{workspace_id}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ - -H "Authorization: Bearer $API_KEY" \ - -H "Content-Type: application/json" \ - -d '{ - "model": "glm-5.2-fast-preview", - "messages": [{"role": "user", "content": "你是谁"}], - "stream": false -}' -``` - -glm-5.2 默认返回 `reasoning_content` 思考字段；[流式输出](../concepts/streaming-output.md)时思考内容与回答内容分别通过 `delta.reasoning_content` 与 `delta.content` 推送。 - -## 限制与注意事项 - -- **两者定位不同**：TPM 预留解决"容量保障/不受限流"，快速模式解决"输出速率提升"。快速模式仍为 preview，生产环境需评估稳定性。 -- **接入差异**：TPM 预留通过替换为专属模型 code 接入标准 dashscope 域名；快速模式使用带 `{workspace_id}` 的 maas 域名并指定 fast-preview 模型 ID。 -- **限流行为差异**：TPM 预留超额自动降级按量、不中断；快速模式超额不立即限流而是排队。 -- **计费口径**：具体价格、容量换算与费用以百炼控制台为准，[TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 与快速模式的价格表可能随时间调整，请以控制台实时展示为准。 +- **共性限制**： + - 两种模式均要求请求符合百炼 API 规范（如 `messages` 格式、`stream` 参数等）； + - 缓存能力依赖模型本身支持（如 `glm-5.2` 支持缓存，`qwen3.6-flash` 不支持），具体参见 [TPM 预留](../../raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) 中的“长输入阶梯系数和缓存折扣”表格。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md index 6aa8dc00..5875c22a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/model-monitoring.md @@ -1,79 +1,89 @@ # model monitoring -阿里云百炼提供两套互补的用量与监控能力：**模型用量**（控制台聚合视图，用于查看调用量、Token 消耗和费用）与**模型监控**（面向指标、告警、日志的可观测体系）。前者侧重成本与免费额度管理，后者侧重性能、错误、安全指标的采集、告警与对话审计，二者共同覆盖从成本控制到线上运维的完整链路。 +模型监控是百炼平台提供的核心可观测性能力，用于实时跟踪模型调用行为、性能表现、成本消耗及安全合规性。它支持从基础调用统计到高级指标告警的全链路监控，并提供推理日志回溯与 Prometheus 数据对接能力，适用于生产环境下的稳定性保障与精细化成本治理。 -## 支持的模型与功能范围 +## 支持的模型与功能 -- **用量查看**：模型列表中的所有模型均支持查看用量，包括基于它们调优后的自定义模型。详见 [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 -- **模型监控**： - - **普通监控**支持所有模型（含调优后的自定义模型），延迟通常为小时级。 - - **高级监控**支持北京、新加坡、弗吉尼亚地域下的所有模型，可提供分钟级数据洞察。 - - **告警功能**支持北京、新加坡地域下的所有模型。 -- 监控可查看调用记录、指标监控与告警（Token、延时、调用时长、RPM、TPM、失败率）、以及 Token 消耗统计。详见 [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)。 +- **监控覆盖范围**： + - 普通监控（免费）支持[选择模型](https://help.aliyun.com/zh/model-studio/models)中的所有模型，包括基于其调优的[自定义模型](https://help.aliyun.com/zh/model-studio/model-deployment-introduction#f17bf700c06k5)； + - 高级监控（收费）仅支持北京、上海、新加坡、弗吉尼亚地域下的模型； + - 告警功能仅支持北京、新加坡、弗吉尼亚地域（见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)）。 -## 用量与费用查看 +- **关键功能模块**： + - 调用记录追踪（含 Request ID、状态码、错误码）； + - 四类核心指标监控：**安全**（如内容安全错误次数）、**成本**（如平均单次请求调用量）、**性能**（如调用时长、首[Token](../concepts/token.md)延时、TPS）、**错误**（如失败率、限流错误次数）； + - [Token](../concepts/token.md) 消耗细粒度统计（按业务空间、API Key、单次调用）； + - 推理日志（输入/输出内容）查看与回流为训练数据集； + - 主动告警（支持短信/邮件/钉钉/企业微信/Webhook）； + - Prometheus HTTP API 对接，支持 Grafana 可视化或自建系统集成。 -数据按[业务空间](https://help.aliyun.com/zh/model-studio/use-workspace)维度统计，不支持按阿里云账号维度统计。 +> **注意**：文档 1 中称“高级监控支持北京、上海、新加坡、弗吉尼亚地域”，而文档 2 未提及上海地域支持情况；但文档 1 的告警说明明确限定为“北京、新加坡、弗吉尼亚”，且[模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md)中所有配置入口链接均未包含上海控制台路径。因此，**上海地域暂不支持告警与高级监控功能**，以文档 1 实际配置为准。 -- **数据延迟约 1 小时**；不支持查看 30 天以前的统计数据，更早数据需前往「费用与成本」页面查询。 -- **时间精度**：支持分钟 / 小时 / 天三种精度。时间跨度超过 1 天时分钟精度不可选，超过 7 天时仅支持按天查看。 -- **筛选维度**：仅「大语言模型」页签支持按推理类型（实时推理 / 批量推理）筛选；支持按 API-KEY、模型名称（如 `qwen-plus`）筛选。 -- **费用概览**：可查看当前账期总消费、订阅费用、账单趋势（按月/按天，可按产品分类、API Key ID、模型筛选），并可设置**费用告警**。 +- **推理日志支持模型**（仅限开通后生效）： + 包括 qwen3-max 系列、qwen-plus 系列、qwen-flash/turbo/coder 系列、部分开源模型（如 qwen3-235b-a22b）及三方模型（deepseek-v3.1/v3.2），详见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) 中“支持请求和响应的模型”列表。不支持模型界面将提示“当前模型暂不支持日志”。 -不同模型的用量统计口径不同：大语言模型 / 全模态 / 向量模型按 **Token**，图像生成按**张**，视频生成按**秒**，语音模型按**秒、字符或 Token**（视模型而定）。完整口径见 [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md)。 +## 关键参数与指标 -## 免费额度管理 +| 类别 | 指标名 | 说明 | 来源 | +|--------|---------|------|------| +| **调用统计** | `model_call_count` | 调用总次数 | [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) | +| **性能** | `model_first_token_duration_p99` | 首[Token](../concepts/token.md)延时P99值 | [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) | +| **性能** | `model_tps_per_request` | 单次请求输出TPS（每秒Token数），**仅高级监控支持** | [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) | +| **用量** | `model_usage` | Token 总用量（输入+输出） | [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) | +| **错误** | `model_call_failure_count` | 失败次数（含 429 限流、内容安全拦截等） | [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) | -「免费额度」页面提供使用概览（按模型总数、额度充沛、使用超 50%/80%、无免费额度等维度汇总）及「即将用尽 Top 3」列表。 +> 所有指标均支持按 `workspace_id`、`model`、`apikey_id`、`protocol`（HTTP/SSE/WS）、`sub_protocol`（DEFAULT/ASYNC）等 Label 过滤，详见 [模型监控 (raw/model-user-guide/model-monitoring/model-telemetry.md)](../../raw/model-user-guide/model-monitoring/model-telemetry.md) 中“支持的过滤条件”。 -- **免费额度用完即停**：开启后免费额度用尽时服务自动停止（返回 `403 AllocationQuota.FreeTierOnly`），避免产生额度外费用。 -- 支持批量开启/关闭、一键开启/关闭所有模型；账号未绑定有效支付方式时批量操作会失败。 +## 使用方式 -> **注意**：「免费额度用完即停」只能在账户仍有未消耗免费额度时开启；一旦开启，需在免费额度完全消耗后才能关闭。控制台免费额度数据为分钟级更新，账单记录按分钟汇总，请以控制台显示数值为准。 +1. **启用监控**： + - 普通监控默认开启，数据延迟约 **1–2 小时**； + - 高级监控需手动开通：进入目标业务空间的[模型监控配置](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)，开启“性能和用量指标监控”； + - 推理日志需额外开通（审计日志 + 推理日志），开通后才记录请求/响应内容。 -## 监控指标与告警 +2. **查看数据**： + - 在[模型监控列表](https://bailian.console.aliyun.com/?tab=model#/model-telemetry)中点击目标模型右侧「监控」进入详情页，分「调用统计」与「性能指标」页签； + - 点击「日志」可查看带 Token 用量、输入/输出内容的实时推理记录（仅支持模型且已开通日志）； + - Token 消耗可在「调用统计」页签的「调用量」区域直接查看，或通过[模型用量](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/usage-statistics)页面按业务空间聚合查询。 -在模型监控列表中点击目标模型操作列的**监控**，可查询 4 类指标： +3. **配置告警**： + - 需先开启高级监控； + - 进入[模型告警页面](https://bailian.console.aliyun.com/?tab=model#/model-alert)，点击「创建告警规则」，选择模型、指标模板、阈值与通知方式； + - 告警等级（INFO/WARNING/ERROR/CRITICAL）决定可用通知渠道，不可自定义。 -- **安全**：如 `内容安全错误次数`（输入/输出被内容安全服务拦截）。 -- **成本**：如 `平均单次请求调用量`。 -- **性能**：`调用时长`、`首 Token 延时`、RPM、TPM、非首 Token 延时等。 -- **错误**：`失败次数`、`失败率`，其中**限流错误次数**指因 [429 状态码](https://help.aliyun.com/zh/model-studio/error-code)导致的失败。 +4. **对接外部系统**： + - 获取 Prometheus HTTP API 地址后，可通过标准 PromQL 查询指标，例如： + ```http + GET {API}/api/v1/query_range?query=model_usage{workspace_id="llm-xxx",model="qwen-plus"}&start=2025-11-20T00:00:00Z&end=2025-11-20T23:59:59Z&step=60s + ``` + - Authorization 使用 Base64 编码的 `AccessKey:AccessKeySecret`，且必须与 Prometheus 实例归属同一阿里云账号。 -**创建告警**（仅限新加坡、华北2（北京）地域）：需先开启高级监控（性能和用量指标监控），再在模型告警页面创建规则。 +## 限制和注意事项 -- **通知方式**：短信、电子邮件、电话、钉钉群机器人、企业微信机器人、Webhook。 -- **告警等级**：紧急（电话/短信/邮件）、错误（短信/邮件）、警告（短信/邮件）、普通（邮件），不支持自定义。 +- **地域限制**： + - 高级监控、告警、推理日志功能**仅在北京、新加坡、弗吉尼亚地域可用**；上海地域目前不支持告警与高级监控（见上文注意项）； + - 推理日志查看功能在文档 1 中明确标注支持“华北2（北京）、新加坡、弗吉尼亚”，文档 2 未提及其地域约束，以文档 1 为准。 -## Token 消耗与历史对话 +- **数据延迟**： + - 普通监控（调用次数、Token 总量）延迟 **1–2 小时**； + - 高级监控与推理日志延迟为**分钟级**； + - 免费额度数据更新为**分钟级**，账单费用数据按分钟汇总生成。 -- **历史 Token 消耗**：最近 30 天可在监控页调用统计的「调用量」区域查看；更早数据前往「费用与成本」页面。 -- **单次调用 Token 消耗 / 历史对话（模型日志）**：需在「模型监控配置」中依次开通审计日志和推理日志，之后在**日志**页签查看每次调用的输入、输出与用量。开通后从调用到记录存在分钟级延迟。 +- **权限与可见性**： + - 默认业务空间成员可查看所有业务空间数据；子业务空间成员**仅能查看当前空间数据**，无法切换； + - 开通推理日志需主账号或具备 `AliyunBailianFullAccess` 权限的子账号。 -> **注意**：查看某次调用的 Token 消耗及历史对话（模型日志）功能**目前仅适用于华北2（北京）地域的部分模型**，且仅覆盖特定模型/快照版本（如 qwen3-max、qwen-plus、qwen3-coder 系列、部分开源与三方模型）。详见 [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md)。 +- **历史数据限制**： + - 模型监控列表仅显示**最近 30 天**的 Token 消耗；更早数据需通过[费用与成本](https://billing-cost.console.aliyun.com/finance/expense-report/expense-detail-by-instance)页面查询； + - **未开通推理日志前的调用无请求/响应内容记录，且不可补录**。 -## 接入 Grafana 与自建应用 - -高级监控的指标数据存储在私有 Prometheus 实例中，支持标准 Prometheus HTTP API，可接入 Grafana 或自建应用做可视化分析。 - -1. 确保已开启高级监控，在模型监控配置中查看 Prometheus 实例详情，按网络环境（公网/VPC）复制 HTTP API 地址。 -2. 通过 `GET {HTTP API}/api/v1/query_range?query=<指标名>&start=...&end=...&step=60s` 查询，`Authorization` 需用 `Basic base64Encode(AccessKey:AccessKeySecret)`。 - -常用指标名包括 `model_call_count`（调用次数）、`model_call_duration`（调用时长均值）、`model_usage`（用量总和）等；可用 `{workspace_id="...",model="qwen-plus"}` 形式追加过滤条件（支持 `user_id`、`apikey_id`、`workspace_id`、`model`、`protocol`、`status_code`、`usage_type` 等 LabelKey）。 - -> **注意**：`status_code`、`error_code` 仅 `model_call_count` 支持；`usage_type` 仅 `model_usage` 支持。AccessKey 必须与 Prometheus 实例归属同一阿里云账号。 - -## 生产环境实践建议 - -- **控制输出长度**：合理设置 `max_tokens` 与限制思考长度以控制费用。 -- **按任务选模型**：分类、摘要等简单任务优先用轻量级模型。 -- **监控与告警**：通过模型监控跟踪用量趋势并配置告警。 -- **优化 Prompt**：简洁清晰的 Prompt 可减少输入 Token 消耗。 -- **使用批量推理**：非实时大批量任务用批量推理更具成本优势。 +- **模型兼容性**： + - 并非所有模型支持推理日志（如部分[多模态](../concepts/multi-modal.md)模型不支持），是否支持由模型本身决定，与是否为[多模态](../concepts/multi-modal.md)无关； + - TPS 指标（`model_tps_per_request`）**仅高级监控提供**，且其物理意义为“单次请求输出速度”，与 TPM（账号级限流）不同，排查慢响应需结合 TTFT、非首Token延时与输入长度综合分析。 ## 来源文档 -- [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md) - [模型监控](../../raw/model-user-guide/model-monitoring/model-telemetry.md) +- [模型用量](../../raw/model-user-guide/model-monitoring/model-usage-statistics.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md index 5a84d01d..13b24492 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/plug-in.md @@ -1,158 +1,46 @@ # plug in -百炼插件是一个工具集合，用于扩展大模型的能力边界。一个插件下可包含多个工具（API），每个工具实现特定功能。通过将插件集成到大模型应用中，可弥补大模型在获取最新信息、精确计算、图像处理等方面的不足。百炼支持官方插件、三方插件和自定义插件三类。详见[插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 +插件是百炼平台用于扩展大模型能力的核心机制，通过将外部工具（API）集成到大模型应用中，弥补其在实时信息获取、精确计算、代码执行、图像生成等场景下的固有局限。插件以“工具集合”形式组织，支持官方预置、三方市场及完全自定义三种类型，可被智能体应用、工作流应用或 Assistant API 主动调用或自动规划调用。 -## 插件分类 +## 支持的模型/功能 -- **官方插件**：组件广场预置，无需配置输入输出参数即可直接调用。 -- **三方插件**：涵盖商业服务、图像视频、学习教育等领域，经过效果测试，开通后直接调用，无需额外配置。 -- **自定义插件**：当官方和三方插件无法满足业务需求时，用户可创建或从云市场导入自定义插件，集成到应用中。 +当前插件能力已覆盖以下通义千问系列模型：`qwen-turbo`、`qwen-plus`、`qwen-max`、`qwen-vl-max`、`qwen-vl-plus`。各模型对插件的兼容性存在差异，**实际可用性请以控制台运行结果为准**，而非静态列表 [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md)。 +插件功能分为三类： +- **官方插件**：开箱即用，无需配置参数，包括 `code_interpreter`（Python 执行）、`calculator`（数学计算）、`text_to_image`（文生图）、`quark_search`（实时搜索）、`generate_qrcode`（二维码生成）、`github_search`（GitHub 项目检索）； +- **三方插件**：来自云市场，覆盖商业服务、图像视频、教育等领域，需开通后使用； +- **自定义插件**：用户自主定义插件 URL、工具路径、鉴权方式及输入/输出参数，支持从零创建或从云市场导入 [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 -## 官方插件列表 +> **注意**：文档 1 和文档 2 均列出 `quark_search` 插件说明，但文档 2 明确指出其“目前支持检索出网页标题、关键词和摘要，但不支持直接访问网页详情”，而文档 1 仅简述为“查找公开的网络知识和信息”。应以文档 2 的限定描述为准，避免误判能力边界。 -| 插件名称 | 工具 ID | 说明 | [计费](../concepts/billing.md)方案 | -| --- | --- | --- | --- | -| Python 代码解释器 | `code_interpreter` | 执行 Python 代码片段，如数学计算、数据分析与可视化、数据处理 | 免费 | -| 计算器 | `calculator` | 进行复杂数学计算，例如计算 `12313x13232` | 免费 | -| 图片生成 | `text_to_image` | 基于文本生成图片，例如"请画一只在笑的小狗" | 限时免费，需申请开通 | -| 夸克搜索 | `quark_search` | 搜索实时信息，查找公开网络知识和信息 | 限时免费，需申请开通 | -| 生成二维码 | `generate_qrcode` | 根据网站链接地址生成二维码 | 免费 | -| GitHub 搜索 | `github_search` | 在 GitHub 中搜索相关项目列表 | 免费 | +## 关键参数 -> **注意**：夸克搜索插件目前支持检索网页标题、关键词和摘要，但不支持直接访问网页详情。GitHub 搜索插件支持检索项目标题、链接和摘要，不支持访问项目详情。Python 代码解释器不支持对外访问网络以及上传本地文件。 - -Python 代码解释器可用依赖包括：matplotlib、pandas、scipy、seaborn、sympy、pillow、pydantic~=1.10.8、requests~=2.31.0、oss2~=2.18.1、pdfminer-six、pypdf、python-pptx、wordcloud 等。详见[官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 - -## 支持的模型 - -插件调用对模型有一定要求，目前支持以下模型： - -| 模型 | 模型标识符 | -| --- | --- | -| 通义千问-Turbo | qwen-turbo | -| 通义千问-Plus | qwen-plus | -| 通义千问-Max | qwen-max | -| 通义千问VL-Max | qwen-vl-max | -| 通义千问VL-Plus | qwen-vl-plus | - -> **注意**：各模型对插件的兼容性可能有差异，最新兼容性状态以控制台实际执行结果为准。 - -## 插件调用机制 - -调用插件的本质是调用插件下的工具。百炼支持通过[智能体应用](../concepts/agent-application.md)、[工作流](../concepts/workflow.md)应用以及 Assistant API 调用插件。 - -- **[智能体应用](../concepts/agent-application.md) / Assistant API**：大模型根据用户输入内容、工具名称和工具描述判断是否调用工具。需要调用时，模型选择合适工具，应用内部完成调用后将工具返回结果与用户内容合并再次输入模型，由模型生成最终结果；无需调用时直接生成结果输出。 -- **[工作流](../concepts/workflow.md)应用**：插件作为[工作流](../concepts/workflow.md)的一个节点，按用户编排的方式执行特定任务，而非由模型主动规划和调用。 +插件调用依赖以下核心参数： +- **工具 ID**：唯一标识工具（如 `calculator`），用于 API 请求中指定目标工具，获取方式见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)； +- **输入参数**：由大模型从用户输入中识别提取（`传参方式 = 大模型识别`）或由业务系统透传（`传参方式 = 业务透传`），需明确定义名称、类型（String/Number/Object）、描述及必填性； +- **输出参数**：定义 API 返回数据中哪些字段将被大模型用于生成最终回复，所有出参均为必填项，嵌套层级应尽量扁平； +- **鉴权配置**：针对自定义插件，支持 Header 或 Query 方式传递 [Token](../concepts/token.md)，鉴权类型包括 `basic`、`bearer`、`appcode`；云市场插件通常自动注入 AppKey/AppSecret，无需手动配置。 ## 使用方式 -### 首次访问授权 - -主账号或 RAM 用户首次访问插件页面时，需授权服务关联角色 `AliyunServiceRoleForSFMAccessCloudAPI`（权限策略 `AliyunServiceRolePolicyForSFMAccessCloudAPI`），用于授权百炼访问云市场商品清单并根据插件配置进行 API 调用。 - -- **主账号**：在插件页面勾选条款，单击"授权并进入"即可。 -- **RAM 用户（子账号）**：会因缺少创建服务关联角色权限报错（错误码 140052）。需先由主账号在 RAM 控制台创建自定义权限策略（Action 为 `ram:CreateServiceLinkedRole`，Condition 中 `ram:ServiceName` 为 `cloundapi-access.sfm.aliyuncs.com`），并授予子账号该策略后，再完成授权。 - -### 调用插件 - -- **方式一（插件页面）**：在插件页面将工具添加至[智能体应用](../concepts/agent-application.md)。官方插件只能与位于相同[业务空间](../concepts/workspace.md)里的[智能体应用](../concepts/agent-application.md)关联。每个[智能体应用](../concepts/agent-application.md)最多支持添加 10 个工具，应用会根据输入选择调用一个或多个工具。 -- **方式二（应用管理页面）**：在指定智能体或工作流应用内添加插件，测试效果并发布应用。 -- **方式三（Assistant API）**：通过 Assistant API 调用工具，需正确传递工具 ID。 - -子[业务空间](../concepts/workspace.md)调用官方插件前，需先在插件详情页为子[业务空间](../concepts/workspace.md)授权；默认[业务空间](../concepts/workspace.md)无需此步骤。 - -### 获取工具 ID - -通过 API 调用工具时需正确传递工具 ID。在插件页面找到目标插件，单击"查看详情"，在"插件工具"下获取工具 ID（例如 `calculator`）。 - -## 自定义插件 - -当官方和三方插件无法满足业务需求时，可创建自定义插件。详见[自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md)。 - -### 工作流程 - -1. **创建/导入插件**：定义插件基础信息，或直接从云市场导入。 -2. **添加工具**（导入插件无需此步）：配置 API 路径、请求参数和返回数据。 -3. **调试与发布**：在线测试 API 连通性，功能正常后发布。只有已发布的工具才能在应用中被调用。 -4. **在应用中使用**：将插件关联到智能体，通过对话测试或 API 集成调用。 - -### 创建插件关键参数 - -- **插件名称**：支持中英文，需具语义。 -- **插件描述**：对插件功能和使用场景的简要说明，帮助大模型判断是否调用，使用自然语言描述。 -- **插件 URL**：插件访问地址。同一域名下不同路径拆分为不同 API（工具路径）。 -- **是否鉴权**：支持服务级鉴权和用户级鉴权。鉴权信息可放在 Header（默认参数名 `Authorization`）或 Query 中；Type 支持 `basic`（[Token](../concepts/token.md) 前不增加内容）、`bearer`（[Token](../concepts/token.md) 前增加 "Bearer"）、`appcode`（[Token](../concepts/token.md) 前增加 "APPCODE"）。 - -### 创建工具关键参数 - -- **工具名称**：支持中英文，有字符数限制（20 字符）。 -- **工具描述**：帮助大模型判断是否调用该工具，尽量给出使用示例。 -- **工具路径**：指向插件 URL 的相对路径，必须以正斜杠（/）开头。 -- **请求方法**：GET 或 POST。 -- **提交方式**：`application/json` 或 `application/x-www-form-urlencoded`。 -- **输入参数传参方式**： - - **大模型识别**：参数值由大模型从用户输入中提取。 - - **业务透传**：参数值从外部主动透传，通过 `biz_params` 和 `user_defined_params` 传递。 -- **高级配置**：提供调用示例（Query 与期望入参 Value），减少漏召回和误召回。 - -> **注意**：Object 类型下的子属性不能为空，需点击对象行末图标新增子属性。GET 请求方法下的输入参数不支持 Object 类型。 - -### 从云市场导入 - -云市场提供丰富 API，可在云市场开通后导入至百炼插件列表。导入的插件为草稿状态，需测试、发布后使用。导入时系统自动填充出入参，但可能存在信息缺失，发布时需根据错误提示修正。 - -### 使用自定义插件 +插件可通过三种方式接入： +1. **控制台可视化集成**：在 [插件市场](https://bailian.console.aliyun.com/#/plugin-market) 页面，为智能体应用添加工具（最多 10 个），或通过“发布为 MCP 服务”后在智能体编排页的 MCP 区块中引入； +2. **工作流应用节点**：将插件作为独立节点编排进工作流，按预设逻辑顺序执行，不依赖大模型自主决策； +3. **API 调用**：通过 Assistant API 的 `tools` 字段声明可用工具，并在 `tool_choice` 中控制调用策略；若含业务透传参数或用户级鉴权，需通过 `biz_params` 传递 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 -控制台内可将插件发布为 MCP 服务，再在[智能体应用](../concepts/agent-application.md)编排页面的 MCP 区块添加该服务；也可直接在应用管理页面的[智能体应用](../concepts/agent-application.md)编排中添加 MCP 服务。无鉴权插件可直接对话测试；用户级/服务级鉴权需在对话前配置鉴权 [Token](../concepts/token.md)；业务透传参数需配置变量值。从云市场导入的插件无需在对话页输入鉴权 [Token](../concepts/token.md)。 - -通过 API 调用时，若应用关联的插件存在业务透传参数或开启了用户级鉴权，需通过 `biz_params` 传递鉴权信息或透传参数。 +> **注意**：官方插件仅支持与**同业务空间内的智能体应用**关联；子账号（RAM 用户）首次使用插件前，必须由主账号授予 `ram:CreateServiceLinkedRole` 权限，否则授权失败（错误码 140052），详见 [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md)。 ## 限制和注意事项 -- 官方插件只能与位于相同[业务空间](../concepts/workspace.md)的[智能体应用](../concepts/agent-application.md)关联。 -- 每个[智能体应用](../concepts/agent-application.md)最多添加 10 个工具。 -- 只有已发布且启用状态、调试状态为成功的工具才能用于调用。 -- 删除插件会删除其下所有工具，调用该插件的应用会失效，操作不可撤回。 -- 编辑插件信息后立即生效；修改 URL、Header、鉴权信息可能影响工具调用，需重新测试并发布工具。 -- 工具信息修改后需重新测试并发布才能生效。 - -### 常见错误码 - -| 错误码 | 错误信息 | 说明 | -| --- | --- | --- | -| 130040 | xx 缺少参数描述信息 | 参数描述缺失，补充后重新发布 | -| 130022 | 保存工具信息异常/请检查示例参数是否正确 | 原因一：Object 类型参数子属性为空，需新增子属性；原因二：GET 请求下存在 Object 类型输入参数，需选择其他类型 | - -## 常见问题 - -**夸克搜索和联网搜索（enable_search）有什么区别？** - -- **夸克搜索插件**：模型直接调用插件执行搜索，将搜索结果以文本形式返回，可直接用于生成最终输出。 -- **联网搜索（enable_search）**：同样基于夸克搜索，但模型仅利用互联网信息丰富生成内容，不会完全依赖或返回互联网搜索结果。 +- **权限约束**：主账号或 RAM 子账号需具备 `AliyunServiceRoleForSFMAccessCloudAPI` 服务关联角色权限，否则无法访问插件市场或导入云市场 API； +- **功能限制**：`code_interpreter` 不支持网络访问与本地文件上传，可用依赖库已固化（如 `pandas`、`matplotlib`、`requests` 等）；`quark_search` 和 `github_search` 均仅返回摘要信息，不支持跳转详情页； +- **配置要求**：自定义插件的 Object 类型参数**子属性不能为空**，否则发布失败（错误码 130022）；GET 请求方法下**禁止配置 Object 类型入参**； +- **生命周期管理**：删除插件将导致其下所有工具及关联应用失效；工具修改后必须重新测试并发布才生效； +- **计费提示**：`text_to_image` 与 `quark_search` 为限时免费且需单独申请开通，其余官方插件默认免费；三方插件按所选套餐计费。 ## 来源文档 -- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) - [插件概述](../../raw/application-user-guide/plug-in/plug-in-overview.md) +- [官方和第三方插件](../../raw/application-user-guide/plug-in/plugins.md) - [自定义插件](../../raw/application-user-guide/plug-in/custom-plug-ins.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md index 1a567cf3..5b868301 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/prompt.md @@ -1,112 +1,54 @@ # prompt -阿里云百炼提供了一套完整的 Prompt 工程工具链，帮助开发者高效管理和优化提示词。核心能力包括 Prompt 模板（预置与自定义）、Prompt 自动优化、Prompt 样例库以及基于输入输出样例的 Prompt 反馈优化。这些功能覆盖了从模板创建、结构优化到少样本学习引导的完整流程，适用于文本生成、图片生成、智能客服等多种场景。 +Prompt 是百炼平台中用于引导大模型生成预期输出的核心指令载体。它既可作为静态文本直接调用，也可通过模板化、工程化、反馈优化等方式进行结构化管理与持续迭代。平台提供从零构建、自动优化、样例增强到模板复用的全链路支持，覆盖文本生成、图片生成等[多模态](../concepts/multi-modal.md)场景，适用于通用任务快速启动和复杂业务深度定制。 -## Prompt 模板 +## 支持的模型/功能 -Prompt 模板将提示词的固定结构与动态变量分离，实现可复用的统一管理。模板分为**预置模板**和**自定义模板**两类，详见 [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md)。 +- **基础模型支持**：所有百炼接入的大语言模型（如通义千问系列）均支持原始 Prompt 输入；图片生成类模型（如万相）支持正向/负向 Prompt 分离输入。 +- **核心功能**： + - **自定义Prompt模板**：支持文本生成与图片生成两类模板，提供“自定义创建”和“基于Prompt工程创建”两种模式，内置 ICIO、CRISPE、RASCEF 等结构化框架 [原文标题](../../raw/application-user-guide/prompt/prompt-custom-template.md)； + - **预置Prompt模板**：开箱即用的场景化模板（如营销文案生成、摘要抽取），效果稳定，适用于无Prompt设计经验的用户 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)； + - **Prompt自动优化**：基于大模型对原始 Prompt 进行结构重组、角色注入、指令增强与边界约束，提升输出稳定性 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)； + - **Prompt反馈优化**：利用用户提供的输入-输出样例（few-shot）驱动多轮评估与迭代，显著提升特定任务准确率，推荐使用千问-max 作为推理模型 [原文标题](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)； + - **Prompt样例库**：已**停止维护**，官方明确建议迁移至 RAG 表格库 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。> **注意**：该功能虽仍可访问，但不再更新且不推荐新项目使用，详见文档末尾迁移指引。 -### 预置模板 +## 关键参数 -由百炼平台提供，涵盖营销文案、摘要抽取、文案润色、商品评论等通用场景，已经过优化，效果稳定，无需额外开发即可通过控制台或 API 调用。预置模板不支持修改，但可通过"复制模板"创建自定义副本后编辑。 +| 参数 | 说明 | 约束 | +|------|------|------| +| `promptTemplateId` | 模板唯一标识符，用于 API 调用获取模板内容 | 必填，需与 `workspaceId` 配对使用 | +| `workspaceId` | 业务空间 ID，是模板归属和权限控制的基础 | 必填，通过[获取APP ID 和 Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id)获取 | +| `variables` | 模板中声明的占位符列表（如 `${topic}`、`${num1}`），用于运行时动态填充 | 模板创建时自动解析，不可在 API 调用中新增 | +| `max_tokens`（上下文） | 单次请求总 [Token](../concepts/token.md) 上限（含 Prompt + 输入 + 输出） | 文本生成默认 ≤ 6144 字符（约 8K tokens），具体依模型而定；图片生成受分辨率与提示词长度双重限制 | -### 自定义模板 +## 使用方式 -支持两种创建方式： +### 控制台操作 +- **创建模板**：进入[提示词](https://bailian.console.aliyun.com/?tab=app#/component-manage/prompt)页面 → 单击 **创建提示词** → 选择类型（文本/图片生成）→ 选择输入模式（自定义 or Prompt工程）→ 编辑并保存。 +- **调用模板**：在智能体应用配置中点击 **使用prompt** → **创建应用**，模板变量（如 `${name}`）将自动填充至提示词编辑框；调试时可直接输入测试问题验证效果。 +- **反馈优化**：在[提示词 > 反馈优化](https://bailian.console.aliyun.com/?tab=app#/component-manage/prompt/feedback-optimization)页面 → 新增优化任务 → 上传样例数据（5–10条）与评测数据（≥20条）→ 启动优化 → 保存为模板或直接创建应用。 -- **控制台创建**：在"提示词"页面直接创建，或从预置模板复制后修改。支持"自定义创建"和"基于 Prompt 工程创建"两种输入模式。 -- **API 创建**：通过 `CreatePromptTemplate` 接口创建，需要提供 `workspaceId`（[业务空间](../concepts/workspace.md) ID）。 +### API/SDK 调用 +- **获取模板**：调用 `GetPromptTemplate` 接口，传入 `workspaceId` 和 `promptTemplateId`，响应中返回 `content`（含变量）及 `variables` 列表。 +- **生成最终 Prompt**：将业务数据替换模板中的 `${variable}` 占位符，再作为 `system` 或 `user` 消息发送至目标模型 API。 +- **优势**：实现逻辑与内容分离，支持控制台热更新 Prompt 而无需重发代码，保障多服务间一致性 [原文标题](../../raw/application-user-guide/prompt/prompt-template.md)。 -自定义模板支持文本生成和图片生成两种类型。文本生成模板可选择 ICIO、CRISPE、RASCEF 等 Prompt 工程框架进行结构化设计；图片生成模板支持分别定义正向和负向提示词。具体创建流程参见 [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md)。 +## 限制和注意事项 -### 模板使用方式 - -**控制台**：在模板卡片上点击"创建应用"，模板内容自动填充到[智能体应用](../concepts/agent-application.md)的提示词编辑框中。提示词最大支持 6144 个字符。 - -**API/SDK**：通过 `GetPromptTemplate` 接口拉取模板内容（需 `workspaceId` 和 `promptTemplateId`），将业务数据填入模板变量后生成最终 Prompt，再发送给目标模型。返回内容包含 `variables`（变量列表）、`content`（模板内容）等字段。 - -> **注意**：Prompt 模板功能目前仅适用于**华北2（北京）**地域。 - -## Prompt 自动优化 - -当手动编写高质量 Prompt 成本较高时，可使用自动优化功能。该功能利用大模型对原始 Prompt 进行分析和重写，优化策略包括： - -- **结构重组**：调整整体结构使其更符合逻辑 -- **角色扮演引导**：为模型设定明确的专家角色 -- **指令增强**：将模糊指令具体化、步骤化 -- **安全与边界注入**：增加输出格式、内容限制等边界条件 - -操作路径：**应用开发 > 组件管理 > 提示词 > 自动优化**。优化结果可直接复制使用或保存为模板。该功能**不计费**，且提交的数据不会被存储或用于模型训练。详见 [Prompt自动优化](../../raw/application-user-guide/prompt/optimize-prompt.md)。 - -## Prompt 反馈优化 - -相比普通自动优化，Prompt 反馈优化基于用户提供的**输入输出样例**进行多轮自动化评估和迭代，生成更贴合实际业务场景的 Prompt。其工作流程为： - -1. 选择推理模型（推荐千问-max） -2. 输入初始 Prompt（描述任务目标） -3. 上传样例数据（建议 5-10 条，每种场景至少 1 条） -4. 上传评测数据（建议至少 20 条，数据越多效果越好） -5. 系统自动进行多轮评测与优化 - -优化后的 Prompt 包含三部分：原始 Prompt、添加的样例（few-shot）、以及自动生成的内容提示（对分类边界等的补充说明）。优化结果可保存为模板或直接创建[智能体应用](../concepts/agent-application.md)。详见 [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md)。 - -## Prompt 样例库 - -> **注意**：Prompt 样例库功能已不再维护，推荐将样例库数据迁移到 RAG 表格库中。 - -Prompt 样例库采用少样本学习（Few-shot learning）思路，从预定义的高质量问答对中检索相关样例注入模型上下文，引导模型生成更准确、风格更一致的回复。适用于智能客服、特定领域知识问答、格式化内容生成等场景。 - -### 使用限制 - -| 限制项 | 上限 | -|--------|------| -| 单个样例库容量 | 300 条样例 | -| 单应用关联样例库数 | 5 个 | -| 单次召回片段数 | 最多 10 个 | -| 批量导入文件大小 | 20MB（Excel） | -| 单次导入条数 | 100 条 | - -### 计费说明 - -样例库功能本身不收费，但启用后会增加大模型调用的 [Token](../concepts/token.md) 消耗。总输入 [Token](../concepts/token.md) 约等于：用户查询 [Token](../concepts/token.md) + 所有召回样例的总 [Token](../concepts/token.md) + 系统指令 [Token](../concepts/token.md)。 - -详见 [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 - -## Prompt 工程框架 - -百炼平台内置三种 Prompt 工程框架，可在创建自定义文本生成模板时选用： - -| 框架 | 组成要素 | 适用场景 | -|------|----------|----------| -| **ICIO** | 指令、背景信息、补充数据、输出格式 | 简单明确的任务，如数据分析、内容生成、文本摘要 | -| **CRISPE** | 角色与能力、背景信息、任务、输出风格、输出范围 | 需要 AI 扮演特定角色的交互，如客服、创意写作 | -| **RASCEF** | 角色、行动、步骤、上下文、示例、格式 | 多步骤复杂业务流程，如项目规划、战略分析 | - -## 常见问题 - -**使用 `GetPromptTemplate` 接口和直接在代码中拼接字符串有什么区别？** - -通过接口管理 Prompt 的优势在于：逻辑与内容分离（可在控制台更新 Prompt 无需重新部署代码）、集中管理与协作（团队共享和版本管理）、一致性保障（避免手动维护导致的不一致）。 - -**Prompt 自动优化失败的可能原因？** - -输入内容超出 [Token](../concepts/token.md) 限制、触发安全审核策略、或网络/服务临时不可用。 +- **地域限制**：所有 Prompt 模板功能（包括创建、管理、优化）**仅支持华北2（北京）地域**，跨地域调用将失败。 +- **模板容量**：单个 Prompt 模板内容最大支持 6144 字符（控制台界面显示字符计数）；API 层面受模型最大上下文窗口限制。 +- **样例数据要求**： + - 反馈优化中，样例数据集建议 **5–10 条**，且覆盖全部类别；评测数据集建议 **≥20 条**，越多效果越优； + - 样例库功能虽保留，但已明确废弃，新项目请勿依赖 [原文标题](../../raw/application-user-guide/prompt/prompt-sample-optimization.md)。 +- **安全与隐私**：Prompt 自动优化过程中的输入数据**不会被存储或用于模型训练**，符合阿里云数据隐私政策 [原文标题](../../raw/application-user-guide/prompt/optimize-prompt.md)。 +- **图片生成特殊性**：正向 Prompt 定义期望内容，负向 Prompt 排除干扰元素；二者共同作用，需避免语义冲突（如正向写“高清”，负向写“模糊”）。 ## 来源文档 -- [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md) - [自定义Prompt模板](../../raw/application-user-guide/prompt/prompt-custom-template.md) +- [Prompt模板概述](../../raw/application-user-guide/prompt/prompt-template.md) +- [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) - [Prompt自动优化](../../raw/application-user-guide/prompt/optimize-prompt.md) - [使用Prompt样例库优化模型输出](../../raw/application-user-guide/prompt/prompt-sample-optimization.md) -- [基于大模型输入输出样例的Prompt自动优化](../../raw/application-user-guide/prompt/prompt-feedback-optimization.md) - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md index f686e492..2f1daf08 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/release-notes.md @@ -1,45 +1,37 @@ # release notes -阿里云百炼平台的 release notes 由两条互补的时间线组成：一是**平台功能更新**（计费、部署、调优、知识库 RAG、SDK/接入工具、API 等能力的演进），二是**模型上下架与更新**（各模型的上架时间、服务部署范围、模型规格与能力说明）。开发者可据此追踪能力可用性、模型可调用状态与计费/下线变更，避免因模型下线或网关变更导致线上调用中断。 +本页汇总百炼平台近期模型与功能更新，涵盖新模型上线、已有模型能力演进、平台功能增强及关键使用变更。所有信息均基于官方发布记录整理，面向开发者提供可直接落地的参考依据。建议结合具体模型文档与 API 参考手册进行集成。 -## 两类更新的定位 +## 支持的模型/功能 -release notes 覆盖两个正交的维度，查询时应按需求选择对应文档： +- **新增模型**：2026年7月起，Qwen-Image-3.0-Pro（[原文标题](../../raw/model-user-guide/release-notes/newly-released-models.md)）、Kimi K3（2.8万亿参数，100万token上下文）、PixVerse系列视频模型（lipsync/motioncontrol/upscale）、Qwen-Audio-3.0-realtime-plus 与 flash 版本、Vidu 多版本 reference2image/reference2video/img2video、Wan2.7-r2v-2026-06-12 等集中上线，覆盖图像生成、视频对口型、动作模仿、超清增强、实时语音交互等场景。 +- **[多模态](../concepts/multi-modal.md)能力扩展**：Qwen3.7-plus 及 Qwen3.7-max-2026-06-08 明确支持视觉模态理解；qwen3.5-ocr 提供128K上下文与多卡证识别；Tripo-H3.1/P1.0 支持文生/图生/多图生3D；Fun-music-v1 支持歌词驱动的中英文歌曲生成。 +- **平台级功能新增**：2026年6月起，知识检索服务（[原文标题](../../raw/model-user-guide/release-notes/model-release-notes.md)）与知识问答服务上线，支持多知识库联合检索与混合排序；智能体托管运行时 API（6月29日）、Responses API 异步调用（6月1日）、模型导入 API（6月3日）均已正式可用；5月起模型调优支持强化学习训练（RL，邀约制）、图像/视频/视觉理解模型类型定制训练。 -- **平台功能动态**：记录平台侧能力的上线与调整，按 `年 → 月 → 日` 组织，每条含"功能模块 / 功能点 / 功能说明"。详见 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md)。 -- **模型上架清单**：按地域（如华北2（北京））列出每个模型的上架时间、服务部署范围、模型规格（如 `qwen3.7-max`、`kimi/kimi-k2.7-code`）及能力说明。详见 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md)。 +> **注意**：文档1中多次出现 `kimi/kimi-k2.6` 与 `kimi-k2.6` 两种命名（如2026-04-26与2026-04-21条目），且后者链接指向 `https://help.aliyun.com/zh/model-studio/kimi-api`（非月之暗面专属页），而前者统一链接至 `https://help.aliyun.com/zh/model-studio/kimi-api-by-moonshot-ai`。建议以 `kimi/kimi-k2.6` 为准，该命名与文档2中“Kimi-月之暗面”官方标识一致，且符合平台模型命名规范。 -> **注意**：模型**上架**信息在上表中维护，而模型**下线**规则与清单不在此列，需单独参考 [模型下线机制说明](https://help.aliyun.com/zh/model-studio/model-depreciation)。功能更新文档中也频繁出现"部分老旧模型下线通知""网关变更通告"等公告，建议在集成前定期核对，防止依赖的模型或域名失效。 +## 关键参数 -## 平台功能更新的关键脉络 +- **上下文长度**：Kimi K3 支持 100 万 token；qwen3.5-ocr、glm-5.1 分别支持 128K、200K；qwen3.7-max 系列仅支持纯文本输入，不接受图像/视频。 +- **部署规格**：Qwen-Audio-3.0-realtime-plus 与 flash 版本分别面向高品质专业场景与低延迟实时交互；PixVerse 系列明确区分 lipsync/motioncontrol/upscale 功能边界；Vidu 模型按 `reference2image`/`reference2video`/`img2video` 后缀标识任务类型。 +- **计费与资源**：deepseek-v4-pro 的 `cached_token` 单价调整为 1 元/百万 token（文档1，2026-04-29）；qwen-turbo 资源包已启动退市（文档2，2026-06-28）；[模型部署](../concepts/model-deployment.md)支持按模型单元（MU）时长计费（文档2，2026-01-23）。 -从 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) 可提炼出几条对开发者影响较大的主线： +## 使用方式 -- **[模型调优与部署](../concepts/fine-tuning-and-deployment.md)**：陆续新增视觉理解（VL）、视频生成、图像生成模型的定制训练支持；DPO 偏好训练、强化学习（RL，邀约制）训练；以及按模型单元（MU）时长计费的部署方式与预置模型部署（qwen-flash / qwen-plus 等）。 -- **API 能力**：文本生成 API 入口聚合 OpenAI Responses、Anthropic Messages 等分类；Responses API 新增 `background=true` 异步调用；异步任务支持事件总线 EventBridge 回调，无需轮询。 -- **知识库 RAG**：上线知识检索服务、知识问答服务，检索调用全量投递 SLS 日志，Retrieve 接口新增排序模型与指令干预模式。 -- **接入与计费**：新增 Codex、Kilo CLI 等客户端接入；Token Plan 团队版团队管理与共享用量包；Coding Plan 联网搜索 MCP 升级；API Key 加密存储与业务空间专属推理域名升级。 -- **地域接入**：新增美国、德国、日本等地域与服务部署范围，跨地域部署时需确认目标模型的服务部署范围。 +- **API 调用**：文本生成 API 已聚合 OpenAI Responses 与 Anthropic Messages 接口分类（文档2，2026-05-15）；Responses API 新增 `background=true` 异步模式（文档2，2026-06-01）；异步任务支持事件总线 HTTP 回调与 RocketMQ 主动推送（文档2，2026-04-23），避免轮询。 +- **SDK 与客户端**：[多模态](../concepts/multi-modal.md)交互开发套件提供 Linux C++ SDK（2026-02-28）、Android/iOS Lite SDK（2026-02-06）、Java SDK（2026-04-28）及 RTOS C SDK（2026-04-09）；Kilo CLI 支持 [Token](../concepts/token.md) Plan/Coding Plan/按量计费三种接入方式（文档2，2026-02-22）；Spring AI Alibaba 文档已上线（文档2，2026-06-01）。 +- **安全与鉴权**：新增生成临时 API Key 文档（文档2，2026-06-03），适用于不可信环境；API Key 加密存储与业务空间专属推理 API 域名已完成升级（文档2，2026-06-29）。 -## 模型上架的读法与关键字段 +## 限制和注意事项 -[模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 的每一行都对应一个可调用的模型规格，集成时重点关注： - -- **模型规格**：即调用时使用的 `model` 名称，例如推理模型 `qwen3.7-max`、`qwen3.7-plus`、`deepseek-v4-pro`、`kimi/kimi-k2.7-code`、`ZHIPU/GLM-5.1`；文字提取 `qwen3.5-ocr`；语音合成 `qwen-audio-3.0-tts-plus/flash` 等。三方厂商模型通常带 `厂商/` 前缀（如 `vidu/`、`pixverse/`、`stepfun/`、`xiaomi/`）。 -- **快照版本**：形如 `qwen3.7-max-2026-05-20`、`wan2.7-t2v-2026-04-25` 的带日期后缀模型为快照版，能力与主版本一致但版本锁定，适合对稳定性敏感的生产环境。 -- **服务部署范围**：当前上架清单多为"中国内地"，跨地域调用前务必核对。 -- **模态与模式限制**：部分模型有明确约束，例如 `qwen3.6-max-preview` 仅支持纯文本输入、不支持图像与视频输入且默认开启思考模式；`kimi/kimi-k2.7-code` 仅支持思考模式。 - -## 限制与注意事项 - -- **计费与优惠会随时间调整**：如 `deepseek-v4-pro` 的 `cached_token` 单价曾调整为 1 元/百万 token（标准 `input_token` 不变），GLM-5.2 Fast mode 降价、上下文缓存降价等均以对应公告为准，release notes 中的价格描述可能滞后。 -- **模型可能延期或提前下线**：功能更新中同时存在"部分老旧模型下线通知"与"部分老旧模型延期下线通知"，同一批模型的下线时间可能被修订，务必以最新公告为准。 -- **两份文档存在时间粒度差异**：功能更新文档到具体功能点，模型清单文档到具体模型规格；排查某能力是否可用时，建议交叉比对 [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) 与 [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) 两处。 -- **以控制台/API 实际返回为准**：release notes 是变更记录而非实时状态，最终的可调用模型列表、配额与地域支持应以控制台模型广场和 API 返回为准。 +- **模型下线机制**：老旧模型分批次下线，包括“部分老旧模型下线通知”（2026-07-10）、“部分老旧长尾模型下线通知”（2026-07-09）及“延期下线通知”（2026-07-06）。具体清单需参考 [模型下线机制说明](https://help.aliyun.com/zh/model-studio/model-depreciation)，该机制在两篇原始文档中均被引用（[原文标题](../../raw/model-user-guide/release-notes/newly-released-models.md) 和 [原文标题](../../raw/model-user-guide/release-notes/model-release-notes.md)）。 +- **地域与部署范围**：新增美国、德国、日本地域（文档2，2026-06-12），但文档1中所有模型当前仅标注“中国内地”服务范围，跨地域调用需确认模型是否已同步部署。 +- **功能兼容性**：qwen3.6-max-preview 明确注明“> 不支持图像与视频输入”（文档1，2026-04-20），而同系列 qwen3.7-plus 则明确支持视觉-语言能力，版本间能力差异显著，不可混用。 +- **免费额度策略**：新人免费额度启用“用完即停”功能后，耗尽将返回 `AllocationQuota.FreeTierOnly` 错误（文档2，2025-07-29），避免意外计费；2026年7月起企业知识库（旧）已下线（文档2，2026-07-16），迁移需使用新版知识库RAG服务。 ## 来源文档 -- [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) - [模型上下架与更新](../../raw/model-user-guide/release-notes/newly-released-models.md) +- [模型平台功能更新](../../raw/model-user-guide/release-notes/model-release-notes.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md index f9850221..90abaf77 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/security-and-compliance.md @@ -1,189 +1,76 @@ # security and compliance -阿里云百炼围绕"身份权限、内容安全、合规备案、传输加密、私网与安全存储"五个维度构建安全合规体系，覆盖从控制台到 API 调用、从公网到 VPC 私网的完整链路。本文面向开发者，按"权限与身份—内容安全—合规备案—传输加密—[私网访问](../concepts/vpc-private-access.md)—安全存储"的顺序梳理关键能力、参数与注意事项。 - -## 身份与权限管理 - -百炼的权限管理基于[业务空间](../concepts/workspace.md)（workspace）这一最小管理单元，支持控制台页面级与模型级的多维度权限控制，满足多地域、多用户的复杂组织架构需求，详见 [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md)。 - -### 三种角色 - -- **超级管理员**：阿里云主账号，或拥有 `AliyunBailianFullAccess` 系统策略的 RAM 用户，可跨空间统一管理用户权限、空间可用模型、模型限流和 [API Key](../concepts/api-key.md)。建议 AI 安全护栏、模型监控、应用观测等功能使用主账号一次性开通。 -- **[业务空间](../concepts/workspace.md)管理员**：拥有访问某个[业务空间](../concepts/workspace.md)"权限管理"页面的 RAM 用户，可管理该空间内的用户与资源，"管理员"权限包含该空间所有页面的访问权限。 -- **普通用户**：根据分配的权限使用资源，不能管理用户或模型授权。 - -[业务空间](../concepts/workspace.md)按地理区域划分，**单个[业务空间](../concepts/workspace.md)不能跨地域存在**，各地域的默认[业务空间](../concepts/workspace.md)也是不同的空间。 - -### 模型与 [API Key](../concepts/api-key.md) 权限 - -[业务空间](../concepts/workspace.md)内可对模型进行三类精细化授权，**默认[业务空间](../concepts/workspace.md)无法设置这些限制**（所有模型均可调用、调优、部署）： - -| 权限项 | 控制范围 | 配置入口 | -| --- | --- | --- | -| 限制模型调用 | 是否可调用（控制台 & API）+ 请求数限流 + [Token](../concepts/token.md) 限流 | 模型列表 → 模型调用列开关 + 当前空间限流列 | -| 限制模型训练 | 是否可调优（控制台 & API）及调优后部署 | 模型列表 → 模型授权 → 模型训练列 | -| 限制[模型部署](../concepts/model-deployment.md) | 是否可直接部署 | 模型列表 → 模型授权 → [模型部署](../concepts/model-deployment.md)列 | - -单个 [API Key](../concepts/api-key.md) 只能归属一个地域内的一个[业务空间](../concepts/workspace.md)和一个用户，且不能转移。[API Key](../concepts/api-key.md) 的可调用功能与模型限流与**归属[业务空间](../concepts/workspace.md)**的权限保持一致，不受用户控制台权限影响，也无需为不同模型（文生文、文生图、语音合成）创建不同 [API Key](../concepts/api-key.md)。自 2026 年 3 月 25 日起，华北2（北京）地域所有新创建的 [API Key](../concepts/api-key.md) 均归属主账号，并支持设置 IP 访问白名单。 - -> **注意**：[API Key](../concepts/api-key.md) 的有效性受账号操作影响——将 RAM 账号移出[业务空间](../concepts/workspace.md)会使其 [API Key](../concepts/api-key.md) 失效（重新加入后恢复），而在 RAM 控制台删除账号/角色则会使 [API Key](../concepts/api-key.md) 永久失效、不可恢复。 - -### OpenAPI 接口权限 - -RAM 用户默认无权调用百炼应用的数据、[知识库](../concepts/knowledge-base.md)、Prompt 工程、长期记忆等 Open API。需阿里云主账号在 RAM 控制台添加以下系统策略之一： - -- `AliyunBailianDataFullAccess`：可调用应用 API 目录下的所有 API。 -- `AliyunBailianDataReadOnlyAccess`：仅可调用只读类 API（如 `DescribeFile`、`GetIndexJobStatus`）。 - -### 生产环境实践 - -- **空间规划**：推荐按环境（dev/test/prod）划分[业务空间](../concepts/workspace.md)实现隔离，或按业务线划分便于权限与成本管理。 -- **限流策略**：将主账号总配额按比例分配给各[业务空间](../concepts/workspace.md)并预留缓冲。例如总配额 1000 QPM，可分配 prod 600 / test 200 / dev 100，预留 100。 - -## 内容安全：AI 安全护栏 - -大模型输入输出可能包含涉黄、涉政、广告等敏感内容。除模型自有合规检查外，百炼支持接入 AI 安全护栏服务，进一步识别违规信息，保障安全合规，详见 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 - -### 开通步骤 - -1. **开通内容审核服务**：访问 AI 安全护栏购买页面，创建服务关联角色并购买。 -2. **授权内容安全设置**：在百炼"安全管理"页面单击"去授权"开启内容安全设置。若已显示"已开通（不可取消）"及《自建安全机制承诺函》全文，可跳过此步。 -3. **设置请求头**：调用百炼时在请求头设置 `X-DashScope-DataInspection`，开启输入输出检测： - -```json -{"X-DashScope-DataInspection": {"input": "cip", "output": "cip"}} -``` - -目前支持文本和图片类型的模型，模型与护栏服务的对应关系及[计费](../concepts/billing.md)请参见官方说明。命中护栏时返回 `data_inspection_failed`（DashScope 为 `DataInspectionFailed`，HTTP 400），提示输入可能包含不当内容。 - -## 合规资质与备案 - -### 资质与隐私 - -百炼以无保留意见通过 SOC 2 审计，安全、可用性、保密性控制符合国际标准。阿里云承诺**不会将您的数据用于模型训练**，应用构建与模型训练过程中的传输数据均经过 AES-256 加密。根据法律法规，百炼会存储模型与应用调用产生的数据，详见《阿里云百炼服务协议》中的数据处理、隐私和安全条款，参考 [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md)。 - -### 模型备案信息公示 - -百炼公示所接入大模型的算法备案号与大模型备案号，详见 [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)。部分示例如下： - -| 模型 | 算法名称 | 算法备案号 | -| --- | --- | --- | -| 千问 | 达摩院交互式多能型合成算法 | 网信算备330110507206401230035号 | -| 万相 | 达摩院图像合成算法 | 网信算备330110507206401230027号 | -| 万相 | 通义万相视频生成算法 | 网信算备330106003156001240091号 | -| DeepSeek | DeepSeek 大语言模型算法 | 网信算备110108970550101240011号 | -| 智谱 AI | 智谱交互式内容生成算法 | 网信算备110108105858001230027号 | - -大模型备案号（如通义千问 `ZheJiang-TongYiQianWen-20230901`、DeepSeek `Beijing-DeepseekChat-202404280016` 等）同样公示。第三方模型备案信息由提供方负责，阿里云百炼不作额外承诺。 - -### 应用上架及合规备案 - -接入千问、万相等模型的应用/小程序上架前，需依《生成式人工智能服务管理暂行办法》完成合规备案，详见 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 - -典型场景与所需材料： - -| 场景 | 是否面向 C 端 | 舆论/动员能力 | 主要材料 | -| --- | --- | --- | --- | -| 场景 1 | 是 | 不具有 | 大模型算法备案信息 + 应用主体与阿里云的合作协议 | -| 场景 2 | 是 | 具有 | 场景 1 全部材料 + 企业自主安全评估报告 + 企业自主算法备案 | -| 场景 3 | 企业内部 | — | 关注数据安全、保密等内部合规要求 | - -算法备案信息查询步骤：打开互联网信息服务算法备案系统（beian.cac.gov.cn），以"备案编号"为搜索类别输入对应备案号（如千问 `网信算备330110507206401230035号`），截图保存查询页面完整内容。合作协议（需含算法名称、应用产品或备案编号）请联系商务经理获取。 - -> **注意**：即使使用阿里云提供的模型及备案信息，应用/小程序开发者仍是法规定义的"服务提供者"，需独立履行内容审核、用户保护、数据安全、标识规范等全部法定义务。备案号应以算法备案系统实时查询结果为准，建议定期核验。 - -## 传输加密：以加密方式接入模型推理 - -当请求涉及敏感信息或经公网传输时，可对请求体 `input` 字段值加密，防止传输过程中被窃听或篡改，详见 [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 - -### 加解密机制 - -采用混合加密：数据由 AES 对称算法加密，AES 密钥通过 RSA 非对称加密实现安全传输。流程为：生成 AES 密钥 → 用 AES 密钥加密 `input` → 用 RSA 公钥加密 AES 密钥 → 将加密后的 `input` 与密钥信息（封装在 `X-DashScope-EncryptionKey` 请求头）传入百炼 → 平台推理链路全程加密、解密数据并用相同 AES 密钥加密答案返回 → 用户侧用 AES 密钥解密响应。 - -### 两种接入方式 - -**1. [DashScope SDK](../concepts/dashscope-sdk.md)（自动加密·开箱即用）** - -仅支持 Java 和 Python，不支持自定义密钥： - -- Java SDK：`GenerationParam.builder().enableEncrypt(true)` -- Python SDK：`dashscope.Generation.call(..., enable_encryption=True)` - -SDK 自动完成加解密，响应为明文，无需手动处理。 - -**2. HTTP 调用（手动密钥管理）** - -需额外完成三步：添加 `X-DashScope-EncryptionKey` 请求头、对 `input` 内容加密、对响应数据解密。该请求头为 JSON 字符串，含 `public_key_id`（公钥 ID）、`encrypt_key`（RSA 公钥加密后的 AES 密钥）、`iv`（初始向量）三个字段。AES 密钥长度支持 128/192/256 位（32 字节），长度越长安全性越高但开销越大。 - -> **注意**：手动加密调用仅适用于 DashScope 的 Endpoint，OpenAI 兼容（Chat Completions API 和 Responses API）的 Endpoint **不支持**此加密机制。 - -### 获取 RSA 公钥 - -加密前需调用接口获取当前最新公钥 ID 及其值，详见 [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md)。 - -- 接口：`GET /api/v1/public-keys/latest` -- 鉴权：`Authorization: Bearer ` -- 返回：`request_id`、`data.public_key`（RSA 公钥值）、`data.public_key_id`（公钥 ID） -- 前提：已开通百炼服务并获得 API-KEY，建议配置到环境变量以降低泄漏风险。 - -## [私网访问](../concepts/vpc-private-access.md)：通过 PrivateLink 终端节点访问 API - -为在 VPC 内直接调用百炼模型/应用 API 且流量不经公网，可创建私网终端节点（PrivateLink），将通信限制在阿里云内网，详见 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)。 - -### 工作原理与地域 - -终端节点连接为**单向设计**，仅允许 VPC 内资源主动访问百炼，百炼无法反向访问 VPC 内资源。百炼公共云服务所在地域：华北2（北京）、新加坡。**美国（弗吉尼亚）地域暂不支持[私网访问](../concepts/vpc-private-access.md)**。 - -### 配置步骤 - -1. **创建接口终端节点**：在终端节点控制台创建，终端节点服务选择"阿里云服务"并筛选 `com.aliyuncs.dashscope`，开启"自定义服务域名"。建议至少选择两个不同可用区的交换机以实现高可用；安全组入方向需允许 80（http）、443（https）。 -2. **获取终端节点服务域名**：默认服务域名格式 `ep-{实例ID}.privatelink.aliyuncs.com`（仅 HTTP）；自定义服务域名格式 `vpc-{实例ID}.{地域ID}.dashscope.aliyuncs.com`（支持 HTTPS）。 -3. **调用验证**：将 API base_url 中的域名替换为终端节点服务域名后在对应 VPC 发起调用。支持 HTTP/curl、OpenAI Python SDK、DashScope Python SDK（建议 ≥1.14.0）、DashScope Java SDK（建议 ≥2.12.0）。 - -### 跨地域访问 - -- **同境内或同境外跨地域**（如华东1杭州 VPC → 华北2北京百炼）：推荐启用跨地域端点。 -- **跨境跨地域**（中国内地与境外之间，如华北2北京 VPC → 新加坡百炼）：通过 CEN 跨地域 VPC 互通。 - -## 安全存储[业务空间](../concepts/workspace.md) - -安全存储[业务空间](../concepts/workspace.md)通过反向终端节点与 VPC 私网连接，让部署其中的应用访问同 VPC 下的 ElasticSearch、AnalyticDB（ADB）、OSS 等云组件，避免公网访问风险。该能力需联系商务人员申请开通。 - -### 配置流程总览 - -| 步骤 | 说明 | 参考 | -| --- | --- | --- | -| 1. 创建安全存储[业务空间](../concepts/workspace.md) | [业务空间](../concepts/workspace.md)管理 → 新增[业务空间](../concepts/workspace.md)，空间类型选"安全存储空间" | [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) | -| 2. 创建反向终端节点 | 终端节点控制台创建反向终端节点，终端节点服务选描述为"百炼公共云生产环境-北京站点-安全存储空间专网通道接入点"的服务（VPC NAT 网关、反向、IPv4），配置 VPC/安全组/可用区与交换机 | 同上 | -| 3. 在百炼确认连接 | [业务空间](../concepts/workspace.md)管理 → 管理安全存储空间 → 选择终端节点 → 连接，等待状态变为"已连接" | 同上 | -| 4. 配置可用区 IP | 创建 MSE 云原生网关（2核4G、2 节点、启用 TLS 硬件加速、私网、至少两可用区），获取 NLB 各可用区 VIP 与交换机网段，在百炼配置对应可用区 IP，并将 VIP 加入反向终端节点安全组入方向（全部端口） | [配置可用区IP](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) | -| 5. 配置私有网络资源 | 配置 OSS（Bucket 名 `bailian-safe-workspace-oss-access`、特定标签、CORS 来源 `*bailian.console.aliyun.com`）、ADB（6.0 标准版、高可用版、开启向量引擎优化）、ElasticSearch（7.10、内核增强版、两可用区，交换机网段加入 VPC 私网访问白名单） | [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) | -| 6. 配置 MSE 云原生网关 | 在网关创建 DNS 域名服务（指向 ES 私网地址/端口，TLS 关闭）、创建路由（域名为 ES 域名、路径 `/`、单服务），然后回百炼"资源配置"页激活安全存储[业务空间](../concepts/workspace.md) | [配置MSE云原生网关](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) | - -### 关键约束与注意事项 - -- 专有网络地域须为华北2（北京），可用区须在 G/H/L（或 ADB 的 G/H/I）中按要求选择，每个可用区至少一个交换机；反向终端节点的安全组不要放入其他云组件、无需配置出入网规则。 -- 阿里云百炼只能访问客户授权过且带特定标签（标签名 `bailian-safe-workspace-oss-access`，标签值 `ReadAndWrite`）的 OSS Bucket。 -- ADB 配置需授权服务关联角色 `AliyunServiceRoleForSFMAccessADB`（策略 `AliyunServiceRolePolicyForSFMAccessADB`）。 -- **资源不可中断**：OSS Bucket 停止服务会导致安全存储空间、[知识库](../concepts/knowledge-base.md)、审计日志、历史记录等模块不可用，恢复 Bucket 后可恢复；但 Bucket 被释放会造成安全存储空间不可用且**无法恢复**，需重建。ES 停止[计费](../concepts/billing.md)/被释放的后果与 OSS 相同。 -- 激活前安全存储空间不可用，激活成功后才可用。 - -## 限制与注意事项汇总 - -- 默认[业务空间](../concepts/workspace.md)无法设置模型调用/训练/部署限制，所有模型均可调用、调优、部署，且无法限流。 -- [API Key](../concepts/api-key.md) 不可跨地域、跨业务空间、跨用户转移；账号移出空间会使其 [API Key](../concepts/api-key.md) 失效（重新加入恢复），删除账号/角色则永久失效。 -- AI 安全护栏目前仅支持文本和图片类型模型。 -- [DashScope SDK](../concepts/dashscope-sdk.md) 自动加密仅支持 Java/Python 且不支持自定义密钥；HTTP 手动加密仅适用于 DashScope Endpoint，OpenAI 兼容 Endpoint 不支持。 -- PrivateLink 私网访问美国（弗吉尼亚）地域暂不支持；跨地域访问需区分同境内/同境外与跨境两种方式。 -- 安全存储业务空间的 OSS/ES 等底层资源一旦释放，安全存储空间不可恢复，需重建。 -- 模型与应用的合规备案信息应以算法备案系统实时查询结果为准，建议定期核验；开发者作为"服务提供者"需独立承担全部法律责任。 +阿里云百炼平台提供多层次的安全与合规能力，覆盖模型输入输出内容安全、权限管控、数据传输加密、私网隔离、算法与模型备案、隐私保护及安全存储等关键维度。所有功能均面向生产环境设计，开发者需根据业务场景选择组合使用，以满足《生成式人工智能服务管理暂行办法》等监管要求及企业内部安全策略。 + +## 支持的模型/功能 + +- **AI 安全护栏服务**：支持文本和图片类模型的输入/输出内容实时检测，识别涉黄、涉政、广告等高风险内容。该服务需显式启用，不默认生效，且仅对已开通服务的模型生效 [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md)。 +- **加密传输（AES+RSA）**：支持对请求体 `input` 字段进行端到端加密，防止公网传输中敏感数据泄露。该机制仅适用于 DashScope 原生 Endpoint（如 `/api/v1/services/aigc/text-generation/generation`），**OpenAI 兼容模式（`/compatible-mode/v1`）不支持** [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)。 +- **私网访问能力**：提供两种私网方案： + - **终端节点（PrivateLink）**：适用于普通业务空间，通过接口终端节点将 VPC 流量直连百炼 API，支持华北2（北京）和新加坡地域 [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md)； + - **安全存储业务空间**：专为高敏感场景设计，需配合反向终端节点、MSE 网关及私有云资源（OSS/ADB/ES）部署，实现数据不出私网、存储完全隔离 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md)。 +- **模型与算法备案信息**：公示千问、万相等自研模型及智谱、DeepSeek、Moonshot 等第三方模型的算法备案号与大模型备案号，供开发者用于上架合规材料 [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)。 + +> **注意**：文档 8 与文档 9 描述的私网方案存在适用范围差异——前者面向通用 API 调用，后者仅限“安全存储业务空间”这一特定空间类型，二者不可混用。安全存储空间必须使用反向终端节点（而非文档 8 中的接口终端节点），且强制要求华北2（北京）地域及指定可用区。 + +## 关键参数 + +| 参数名 | 作用 | 必填 | 示例值 | 来源 | +|--------|------|------|--------|------| +| `X-DashScope-DataInspection` | 启用 AI 安全护栏 | 是（启用时） | `{"input":"cip","output":"cip"}` | [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) | +| `X-DashScope-EncryptionKey` | 传递 RSA 加密后的 AES 密钥 | 是（加密调用时） | `{"public_key_id":"1","encrypt_key":"...","iv":"..."}` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `enable_encryption=True` (Python) / `enableEncrypt(true)` (Java) | SDK 层启用自动加解密 | 是（SDK 方式） | `True` / `true` | [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) | +| `base_url` (OpenAI SDK) / `dashscope.base_http_api_url` (DashScope SDK) | 替换为私网终端节点域名 | 是（私网访问时） | `https://vpc-cn-beijing.dashscope.aliyuncs.com/compatible-mode/v1` | [通过终端节点私网访问阿里云百炼模型或应用 API](../../raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) | + +## 使用方式 + +### 启用内容安全检测 +1. 开通 [AI 安全护栏服务](https://common-buy.aliyun.com/?commodityCode=lvwang_guardrail_public_cn)； +2. 在控制台 [安全管理](https://bailian.console.aliyun.com/?globalset=1#/efm/global_set) 页面完成授权； +3. 在请求 Header 中添加 `X-DashScope-DataInspection: {"input":"cip","output":"cip"}`。 + > 响应返回 `400` + `data_inspection_failed` 错误码表示拦截成功，无需额外解析响应体。 + +### 启用传输加密（SDK 方式） +- **Python**：调用 `dashscope.Generation.call(..., enable_encryption=True)`； +- **Java**：构建 `GenerationParam` 时调用 `.enableEncrypt(true)`； +- SDK 自动处理 AES 密钥生成、RSA 加密、请求体加密及响应体解密，返回明文结果。 + +### 配置私网访问（终端节点方式） +1. 在 [VPC 终端节点控制台](https://vpc.console.aliyun.com/endpoint/cn-beijing/endpoints) 创建接口终端节点，服务选择 `com.aliyuncs.dashscope`； +2. 获取终端节点服务域名（如 `vpc-cn-beijing.dashscope.aliyuncs.com`）； +3. 将 SDK 或 HTTP 请求的 `base_url` 替换为该域名。 + +### 部署安全存储业务空间（高隔离场景） +需严格按顺序执行： +1. 创建安全存储类型业务空间 → +2. 创建反向终端节点并绑定 → +3. 配置 MSE 网关及可用区 VIP → +4. 授权并配置 OSS/ADB/ES 私有资源 → +5. 激活空间。 +完整流程见 [配置终端节点并发起连接](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) 及后续文档。 + +## 限制和注意事项 + +- **AI 安全护栏**：仅支持部分模型，具体兼容性请查阅 [面向阿里云百炼用户的AI安全护栏服务](https://help.aliyun.com/zh/document_detail/2923687.html)，不支持所有第三方模型。 +- **加密传输**： + - 不支持 OpenAI 兼容模式（`/compatible-mode/v1`）Endpoint； + - Java/Python SDK 提供开箱即用支持，其他语言需手动实现 [HTTP调用（手动密钥管理）](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md)； + - AES 密钥长度建议使用 256 位。 +- **私网访问**： + - 美国（弗吉尼亚）地域暂不支持终端节点私网访问； + - 安全存储业务空间仅支持华北2（北京）地域，且专有网络必须包含可用区 G/H/L 中至少两个； + - 反向终端节点的安全组需放行 MSE NLB 的 VIP（非整个交换机网段）。 +- **备案信息**： + - 第三方模型（如 DeepSeek、Moonshot）的备案信息由其提供方负责，阿里云百炼仅作公示，不承担验证责任 [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md)； + - 应用上架前，开发者须自行完成算法备案及安全评估，阿里云不替代履行《生成式人工智能服务管理暂行办法》规定的“服务提供者”主体责任 [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md)。 +- **数据隐私**：百炼承诺不将用户数据用于模型训练，传输过程默认启用 TLS 1.2+，静态数据采用 AES-256 加密 [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md)。 ## 来源文档 -- [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md) - [输⼊输出AI安全护栏](../../raw/model-user-guide/security-and-compliance/content-security.md) -- [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) +- [权限管理](../../raw/model-user-guide/security-and-compliance/permission-management-overview.md) - [千问大模型应用上架及合规备案](../../raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) +- [模型备案信息公示](../../raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) - [合规资质与隐私说明](../../raw/model-user-guide/security-and-compliance/privacy-notice.md) - [以加密的方式接入模型推理功能](../../raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - [获取RSA的公钥](../../raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) @@ -193,23 +80,4 @@ SDK 自动完成加解密，响应为明文，无需手动处理。 - [配置私有网络中的资源](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - [配置MSE云原生网关](../../raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md index c3b17ca0..d8ba356d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/skill.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/skill.md @@ -1,135 +1,44 @@ # skill -Skill 是百炼平台[智能体应用](../concepts/agent-application.md)的可扩展能力包，让智能体在对话中自动识别并处理特定类型的任务，如文件处理、数据分析等，无需额外编码或接入外部工具。详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)。 +Skill 是百炼平台提供的可插拔能力包，用于扩展智能体在对话中自动处理特定任务的能力（如文件解析、数据清洗等），无需开发者编写集成代码。官方 Skill 由平台预置并维护，自定义 Skill 则通过符合规范的 ZIP 包上传实现。Skill 的调用完全由智能体根据 `SKILL.md` 中的 `description` 描述自主决策，因此描述质量直接影响匹配准确率 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md)。 -## Skill 类型 +## 支持的模型/功能 -百炼提供两类 Skill： +- **官方 Skill**：覆盖常见文件处理场景（如 `xlsx`、`pdf`、`csv` 等），开箱即用，无需配置，版本由平台统一更新 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md)。 +- **自定义 Skill**：支持用户上传 ZIP 包实现业务定制能力，例如行业专属格式解析、私有 API 封装等。ZIP 包必须包含根目录下的 `SKILL.md` 文件，且整体大小 ≤ 10 MB [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md)。 +- 当前所有 Skill 均不依赖特定大模型底座，而是作为独立能力模块被智能体调度；调用过程对底层模型透明。 -- **官方 Skill**：平台预置的通用 Skill，覆盖常见文件处理场景，由平台统一维护，添加后即可使用，且会自动更新到最新版本。官方 Skill 列表持续更新，可在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面查看最新清单。 -- **自定义 Skill**：通过上传 ZIP 技能包创建，适用于官方 Skill 未覆盖的业务场景（如特定行业数据处理、自定义文件格式解析等）。更新方式为重新上传同名 ZIP 包生成新版本。 - -## 创建自定义 Skill - -当官方 Skill 无法满足业务需求时，可上传 ZIP 技能包创建自定义 Skill。ZIP 包需满足以下要求，详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)： - -| 要求 | 说明 | -| --- | --- | -| 必须包含 SKILL.md | ZIP 包根目录下必须有 `SKILL.md` 文件，定义 Skill 元信息 | -| 大小限制 | 整个 ZIP 包不超过 10 MB | -| 名称唯一 | SKILL.md 中的 `name` 字段不可与当前账号下已有 Skill 重名 | - -### SKILL.md 编写规范 - -`SKILL.md` 使用 YAML 格式定义 Skill 的名称和描述： - -```yaml -name: my-custom-skill -description: "Skill 的功能描述，包含触发条件、适用场景和不适用场景。" -``` +## 关键参数 | 字段 | 必填 | 说明 | -| --- | --- | --- | -| name | 是 | Skill 的唯一标识名称，建议使用小写英文和连字符（如 `data-cleaner`、`invoice-parser`） | -| description | 是 | 描述 Skill 的触发条件和处理能力。智能体据此判断是否调用该 Skill，描述质量直接影响调用准确率 | - -### description 编写建议 - -description 的质量决定了智能体调用 Skill 的准确性，建议包含： - -1. **适用的输入类型**：明确 Skill 处理的文件格式或数据类型。 -2. **支持的操作**：列出可执行的具体操作。 -3. **触发关键词**：用户对话中可能出现的、应触发该 Skill 的关键词或表达方式。 -4. **不适用的场景**：标注不应触发该 Skill 的场景，避免误调用。 - -以下为官方 xlsx Skill 的 SKILL.md 示例： - -```yaml -name: xlsx -description: "Use this skill any time a spreadsheet file is the primary input or output. This means any task where the user wants to: open, read, edit, or fix an existing .xlsx, .xlsm, .csv, or .tsv file (e.g., adding columns, computing formulas, formatting, charting, cleaning messy data); create a new spreadsheet from scratch or from other data sources; or convert between tabular file formats. Trigger especially when the user references a spreadsheet file by name or path — even casually (like \"the xlsx in my downloads\") — and wants something done to it or produced from it. Also trigger for cleaning or restructuring messy tabular data files (malformed rows, misplaced headers, junk data) into proper spreadsheets. The deliverable must be a spreadsheet file. Do NOT trigger when the primary deliverable is a Word document, HTML report, standalone Python script, database pipeline, or Google Sheets API integration, even if tabular data is involved." -``` - -该示例明确了支持的文件格式（.xlsx、.xlsm、.csv、.tsv）、适用操作（读取、编辑、创建、格式转换、数据清洗等）、触发场景（用户提到文件名或路径时也应触发），并标注了不适用场景（产出物为 Word、HTML、Python 脚本等）。 - -### 上传并创建 - -1. 在控制台左侧导航栏，选择 **组件** > **Skill 管理**。 -2. 点击右上角 **自定义 Skill** 按钮。 -3. 在弹窗中点击上传区域选择 ZIP 文件，或直接将文件拖拽到上传区域。 -4. 点击 **确认** 提交。 - -提交后系统自动审查 Skill 内容，预计耗时约 2 分钟。审查通过后 Skill 出现在 **自定义 Skill** 标签页中，可添加到[智能体应用](../concepts/agent-application.md)；未通过则根据提示修改 SKILL.md 后重新上传。 - -### 更新自定义 Skill - -重新上传同名 Skill 的 ZIP 包时，系统会创建新版本，流程与首次创建一致： - -1. 修改本地 ZIP 包内容（如更新 SKILL.md 中的 description）。 -2. 在 **自定义 Skill** 标签页重新上传 ZIP 包。 -3. 审查通过后，已添加该 Skill 的智能体会自动使用最新版本。 - -## 添加 Skill 到智能体 - -添加后，智能体在对话中遇到匹配 Skill 描述的任务时会自动调用该 Skill。支持两种添加方式： - -**方式一：从 Skill 详情页添加** - -1. 在控制台左侧导航栏选择 **组件** > **Skill 管理**，点击目标 Skill 卡片进入详情页。 -2. 点击右上角 **添加到智能体**。 -3. 在应用列表中选择目标应用，确认添加。 +|------|------|------| +| `name` | 是 | Skill 唯一标识符，需全账号唯一，建议使用小写英文+连字符（如 `invoice-parser`） | +| `description` | 是 | 决定 Skill 是否被调用的核心字段。必须明确说明：适用输入类型、支持操作、典型触发关键词、**不适用场景**（避免误触发） | -**方式二：在应用配置中添加** +> **注意**：`description` 不是 UI 展示文案，而是供智能体推理的语义指令。模糊或缺失“不适用场景”将显著增加误调用概率——该要求在 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中被多次强调，但部分早期示例未严格遵循。 -1. 进入目标[智能体应用](../concepts/agent-application.md)的 **应用配置** 页面。 -2. 在左侧配置面板找到 **技能** 区域，点击 Skill 右侧的加号。 -3. 从 Skill 列表中选择需要的 Skill，确认添加。 +## 使用方式 -## 查看 Skill 详情 +1. **创建** + - 官方 Skill：直接在 [Skill 管理](https://bailian.console.aliyun.com/?tab=app#/skill) 页面添加。 + - 自定义 Skill：按规范编写 `SKILL.md` → 打包 ZIP → 控制台「组件 > Skill 管理 > 自定义 Skill」上传。审查约 2 分钟，通过后即可使用。 -在 **组件** > **Skill 管理** 的 **官方 Skill** 或 **自定义 Skill** 标签页中点击目标 Skill 卡片进入详情页，详见 [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md)。详情页包含两个标签： +2. **绑定到智能体** + - 方式一：从 Skill 详情页点击「添加到智能体」，选择目标应用。 + - 方式二：进入智能体「应用配置」→「技能」区域 → 点击对应 Skill 右侧 `+` 添加。 -- **概览**：展示 Skill 名称、当前版本号、功能描述和属性信息，可通过版本下拉框切换查看历史版本。 -- **更新记录**：展示该 Skill 全部版本的发布时间和变更内容。 - -官方 Skill 由平台统一维护和更新，已添加到智能体的官方 Skill 会自动使用最新版本；自定义 Skill 通过重新上传同名 ZIP 包更新版本。 - -## 测试 Skill 效果 - -添加 Skill 后，可在应用配置页面右侧的对话窗格中测试效果。例如发送： - -``` -帮我创建一个包含本月销售数据的表格，按地区分列统计 -``` - -智能体将调用 xlsx Skill，生成包含分列统计的 .xlsx 文件并提供下载。 +3. **测试** + 在应用配置页右侧对话窗格输入典型用户指令（如 `帮我清洗这个 CSV 中的重复行`），观察是否触发 Skill 并返回预期结果（如下载清洗后的文件）。 ## 限制和注意事项 -- ZIP 包整体大小上限为 10 MB，超出将无法上传。 -- 自定义 Skill 的 `name` 字段在当前账号下必须唯一，重名会导致创建失败。 -- description 编写质量直接决定智能体调用 Skill 的准确率，务必明确触发条件、适用操作和不适用场景。 -- 上传后需等待约 2 分钟的系统审查，审查未通过需修改后重新上传。 -- 官方 Skill 自动更新到最新版本；自定义 Skill 需手动重新上传 ZIP 包才能升级。 +- 自定义 Skill ZIP 包内禁止包含可执行二进制文件（`.exe`, `.so`, `.dll` 等），仅允许文本、脚本（Python）、配置及静态资源。 +- 同名 Skill 重新上传会创建新版本，已绑定该 Skill 的智能体**自动升级至最新版**（官方 Skill 同理）。 +- `description` 中若未明确排除冲突场景（例如“产出 Word 报告时不触发 xlsx Skill”），可能导致 Skill 被错误调用——这是当前最常见的配置失误，详见 [原文标题](../../raw/application-user-guide/skill/introduction-to-skill.md) 中的完整示例对比。 +- Skill 无状态设计：每次调用均为独立上下文，不共享内存或临时文件。 ## 来源文档 - [Skill](../../raw/application-user-guide/skill/introduction-to-skill.md) - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md index 93ce31e7..f14c0038 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/start-using.md @@ -1,81 +1,54 @@ # start using -阿里云百炼提供零代码方式快速构建基于私有知识的问答应用，同时持续迭代应用、[知识库](../concepts/knowledge-base.md)、[工作流](../concepts/workflow.md)等核心能力。本页汇总从创建第一个[智能体应用](../concepts/agent-application.md)到跟踪功能动态所需的关键信息，帮助开发者快速上手并了解平台最新能力。 +阿里云百炼平台提供低门槛、高灵活性的 AI 应用构建能力，开发者可通过零代码配置或 API 集成快速启动智能体、工作流及高代码应用。核心路径包括：选择模型与 Prompt 定义角色、接入知识库增强领域理解、配置技能与参数后发布应用。所有操作均在控制台可视化完成，亦支持全链路 API 调用。 -## 快速构建私有知识问答应用 +## 支持的模型/功能 -借助百炼的[智能体应用](../concepts/agent-application.md)构建能力，可在约 5 分钟内零代码完成一个能回答私有领域问题的大模型问答应用，完整流程见 [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)。整体分为三步： +- **模型支持**： + - 智能体应用支持 `qwen-max`、`qwq-plus`、`qwq-32b`、`qwen-vl-plus-latest`、`qwen-vl-plus-2025-01-25` 及 DeepSeek 系列模型（如 DeepSeek-V2、DeepSeek-Coder）； + - 工作流应用支持 `qwq-plus`、`qwq-32b`、DeepSeek 系列及[多模态](../concepts/multi-modal.md)生成节点（图像/视频/音频生成）； + - 知识库向量化默认使用 `text-embedding-v4`，兼容 `v3`，图片解析可选 `qwen-vl-max` 或 `qwen-vl-plus` [原文标题](../../raw/application-user-guide/start-using/application-release-notes.md)。 +- **核心功能**： + - 零代码构建私有知识问答应用，含 Prompt 设计、欢迎语/预设问题配置、知识库绑定与发布全流程 [原文标题](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)； + - 知识库类型覆盖**文档**、**数据**（RDS/DMS/自建 MySQL）、**图片**、**音视频**四类，支持 HTML、Excel、PDF、DOCX、MP4、MP3 等格式； + - 新版智能体应用（Agent 2.0）统一知识库与 MCP 为工具，支持自主规划调用顺序与过程可视化 [原文标题](../../raw/application-user-guide/start-using/application-release-notes.md)。 -1. **构建第一个[智能体应用](../concepts/agent-application.md)（约 1 分钟）**：在应用管理页面创建空白[智能体应用](../concepts/agent-application.md)，选择大语言模型（建议千问-Max），编写 System Prompt 定义角色与任务，并配置欢迎语和预设问题。此阶段由于缺少私有知识，回答较为笼统甚至可能无中生有。 -2. **构建[知识库](../concepts/knowledge-base.md)（约 3 分钟）**：在数据连接页面创建文件类型连接器并上传知识文档（如 docx），等待 1~6 分钟解析完成；随后在[知识库](../concepts/knowledge-base.md)页面创建标准版知识库，选择默认类目与智能切分策略，等待 1~2 分钟完成解析。智能切分为系统预置策略，经[评测](../concepts/evaluation.md)对多数文档可获得最佳检索效果。 -3. **添加知识库并发布应用（约 1 分钟）**：进入应用配置界面，通过「技能 > 知识库」旁的「+」按钮为应用挂载知识库，验证检索增强效果后点击「发布」。 +> **注意**：文档 1 中提及“建议选择千问-Max 模型”，但文档 2 明确指出智能体应用已支持 `qwq-plus`、`qwen-vl-plus-latest` 等更多模型，且 `qwq` 系列具备更强推理能力（数学/代码/IFEval 指标达 DeepSeek-R1 满血版水平）。实际选型应以控制台实时可用模型为准，旧文档中“仅推荐千问-Max”的表述已过时。 -> **注意**：使用大模型会产生[计费](../concepts/billing.md)，百炼提供限时免费额度，可在模型广场查看；知识库服务自 2026 年 1 月 4 日起正式[计费](../concepts/billing.md)，费用由规格费用和模型调用费用两部分组成。 +## 关键参数 -## 支持的模型与功能 +- **知识库检索参数**： + - `初步向量检索 TopK` 与 `初步关键词检索 TopK` 可手动调低，以减少送入排序模型的 [Token](../concepts/token.md) 量，显著降低模型调用费用 [原文标题](../../raw/application-user-guide/start-using/application-release-notes.md)； + - 多知识库场景下支持按信息源重要性设置**权重**，系统优先召回高权重知识库内容； + - 检索配置中可开启“[多模态](../concepts/multi-modal.md)回复增强”，启用后智能体可解析知识库内图表/图像并结合视觉信息生成回答。 +- **应用级参数**： + - System Prompt（角色定义）直接影响模型行为边界，需明确任务范围与输出约束； + - “知识检索增强”开关启用后，可配置回答范围（如“仅基于知识库回答”）、是否展示引用来源等； + - 工作流应用支持异步运行模式：请求中设置 `background=true`，立即返回 Task ID，后续通过 [任务中心](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/app-task-center) 查询结果。 -百炼应用支持多种模型系列，详见 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md)： +## 使用方式 -- **千问系列**：千问-Max 为构建问答应用的推荐模型；[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用均支持 QwQ 系列（具备强推理能力，先输出思考过程再输出回答，数学/代码能力达 DeepSeek-R1 满血版水平，但不包括插件、流程、音视频交互能力）；视觉模型支持 qwen-vl-plus-latest、qwen-vl-plus-0125（Qwen2.5-VL 系列，128k 上下文）以及 qwen-vl-max/plus 用于图片解析。 -- **DeepSeek 系列**：[智能体应用](../concepts/agent-application.md)与[工作流](../concepts/workflow.md)应用均可集成 DeepSeek 系列模型，结合知识库、长期记忆和 Prompt 模板构建私有知识问答应用。 -- **嵌入模型**：知识库支持 text-embedding-v3、v4 模型，v4 在语种支持、代码片段向量化效果和向量维度选择上较 v3 全面升级。 +1. **零代码快速启动（推荐入门）**： + - 访问 [应用管理](https://bailian.console.aliyun.com/?tab=app#/app-center) → 创建智能体应用 → 选择模型 → 设置 System Prompt → 配置欢迎语与预设问题 → 发布前绑定知识库（支持直接上传文件创建，无需预导入数据连接器）[原文标题](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md)； + - 知识库创建流程已简化：进入 [知识库](https://bailian.console.aliyun.com/?tab=app#/knowledge-base) 页面 → 选择类型（文档/数据/图片）→ 直接上传文件或配置数据源 → 启用“智能切分” → 完成。 -## 应用类型与关键能力 +2. **API 集成调用**： + - 同步调用：使用 Responses API（兼容 OpenAI 格式），适用于实时交互场景； + - 异步调用：设置 `background=true`，通过 Task ID 轮询结果； + - 知识库管理：支持 `CreateIndex`（含音视频）、`UpdateIndex`、`GetIndexMonitor` 等 API； + - [长期记忆](../concepts/long-term-memory.md)：新版[长期记忆](../concepts/long-term-memory.md) API 支持多应用共享、自动信息提取与语义检索 [原文标题](../../raw/application-user-guide/start-using/application-release-notes.md)。 -百炼提供多种应用类型以适配不同场景： +## 限制和注意事项 -- **[智能体应用](../concepts/agent-application.md)**：2025 年 12 月 26 日上线新版[智能体应用](../concepts/agent-application.md)（Agent 2.0），将知识库、MCP 统一为工具，由智能体自主规划调用时机与顺序，并完整展示模型思考与工具调用全过程。文件问答支持全文引用、切片检索和自定义处理三种模式。 -- **工作流应用**：支持批量节点、[多模态](../concepts/multimodal.md)生成节点（生成图像/视频/音频）、异步运行模式（文本生成模式下后台执行并返回 Task ID）、Dify 工作流一键导入、[多模态](../concepts/multimodal.md)数据节点（文档/图片/视频/音频解析）等。 -- **高代码应用**：2025 年 9 月 24 日上线，支持基于 Python 项目结构部署 AI 后端服务，内置自动化运维、可观测性及日志服务等企业级能力。 -- **MCP 服务**：2025 年 4 月 9 日新增 MCP 市场与 MCP 管理功能，可开通预置 MCP 服务或部署自定义 MCP 服务；8 月 13 日新增外部调用功能，支持一键配置到第三方应用或通过 MCP SDK 调用。 - -## 知识库核心能力 - -知识库是构建私有知识问答应用的关键，能力持续扩展： - -- **类型与数据源**：分为文档、数据、图片三类；结构化知识库数据源支持云数据库 RDS、自建 MySQL、DMS；非结构化知识库支持导入 docx、pdf、Excel、离线 HTML，并支持自定义 metadata 与标签分类。 -- **音视频知识库**：2025 年 12 月 25 日上线，支持上传音视频文件实现智能检索问答（直播回放问答、课程助教、客服质检）与二次创作（脚本、字幕、剪辑建议）；2026 年 1 月 30 日支持通过 API 创建音视频知识库。 -- **检索与调优**：知识库节点支持必定调用、智能调用和旧版调用三种方式；支持权重设置（多知识库按重要性召回）；支持调整初步向量检索 TopK 和关键词检索 TopK 以降低成本；提供在线调试面板实时验证召回效果；支持图文检索与[多模态](../concepts/multimodal.md)回复增强。 -- **监控与管理 API**：2026 年 1 月新增 GetIndexMonitor（监控数据）、UpdateIndex（更新配置）等 API，并支持子账号开通知识库与基于标签的分账管理。 - -## 应用调用与发布 - -- **API 调用**：2025 年 11 月 3 日起支持通过 Responses API 调用百炼应用，提供同步调用 API（实时交互，可复用 OpenAI 代码库）与[异步调用](../concepts/async-invocation.md) API（设置 `background=true` 立即返回任务 ID）。调用工作流和[智能体编排](../concepts/agent-orchestration.md)应用时需传入自定义参数。 -- **发布渠道**：支持微信、钉钉分享渠道（创建钉钉 AI 机器人或微信公众号 AI 机器人）；支持音视频实时互动（将图文对话应用转为音视频实时互动应用，提供 H5/APP 调试窗口，通过音视频 SDK 发布到 WEB/iOS/Android）。 -- **应用观测**：2024 年 10 月 24 日新增应用观测能力，支持端到端查看应用处理流程；2026 年 2 月 6 日上线新版应用[评测](../concepts/evaluation.md)，支持智能体、工作流和自定义三种类型[评测](../concepts/evaluation.md)集。 -- **长期记忆**：2026 年 1 月 31 日上线新版长期记忆与用户画像管理 API，支持多应用共享同一记忆库、自动提取关键信息、语义检索优化及完整用户画像管理。 - -## 限制与注意事项 - -- 知识库自 2026 年 1 月 4 日起正式[计费](../concepts/billing.md)，提供后付费（按量付费）和资源包两种方式，总费用由规格费用与模型调用费用组成，详情参见 [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md) 中的[计费](../concepts/billing.md)公告。 -- 大模型调用产生[计费](../concepts/billing.md)，平台提供限时免费额度，可在模型广场查看各模型系列详情。 -- QwQ 系列模型在[智能体应用](../concepts/agent-application.md)中不支持插件、流程、音视频交互能力。 -- 文档解析耗时与文档大小相关，知识文档导入通常 1~6 分钟，知识库解析通常 1~2 分钟，需耐心等待。 -- [智能体编排](../concepts/agent-orchestration.md)应用已于 2025 年 8 月 12 日随工作流应用界面升级而下线，相关需求请使用新版智能体应用或工作流应用。 -- Assistant API 处于下线中状态，如需全代码开发高度定制化 RAG 应用请关注官方公告。 +- **计费变更**：知识库服务自 2026 年 1 月 4 日起正式计费，费用 = 规格费 + 模型调用费；支持后付费与 RAG 资源包（标准版/旗舰版）两种模式 [原文标题](../../raw/application-user-guide/start-using/application-release-notes.md)； +- **权限与隔离**：知识库支持子账号开通与标签分账，便于部门级成本归属； +- **调试与验证**：编辑智能体应用时，可使用内置**调试面板**在线调整知识库参数并实时验证召回效果； +- **模型能力边界**：QwQ 系列模型虽推理能力强，但当前不支持插件、流程编排及音视频交互能力（见文档 2 2026 年 4 月条目）； +- **文件处理限制**：非结构化知识库导入 Excel 时，若含复杂公式或宏，可能无法完整解析；音视频知识库依赖 ASR/OCR 能力，原始音画质量直接影响检索精度。 ## 来源文档 - [0代码构建私有知识问答应用](../../raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - [应用功能动态](../../raw/application-user-guide/start-using/application-release-notes.md) - - - - - - - - - - - - - - - - - - - diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/support.md b/skills/bailian-docs-llm-wiki/wiki/guides/support.md index e235c257..e6d38682 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/support.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/support.md @@ -1,106 +1,56 @@ # support -本页汇总阿里云百炼平台的常见问题解答、服务协议与技术支持渠道，帮助开发者快速定位使用中遇到的问题并找到对应解决方案。内容涵盖[计费](../concepts/billing.md)、API/SDK、模型训练、模型幻觉处理以及平台相关协议。 - -## [计费](../concepts/billing.md)常见问题 - -百炼平台采用按量后付费模式（分钟级出账、按月结算），部分模型支持预付费（节省计划与资源包）。关键要点： - -- 模型调用价格与模型部署/训练[计费](../concepts/billing.md)分开计算，具体单价参见百炼控制台模型市场 -- 开通服务要求阿里云账户余额不小于 0 元 -- 万相会员与百炼 API [计费](../concepts/billing.md)体系相互独立，会员权益不适用于 API 调用 -- 费用明细与发票申请通过阿里云费用与成本控制台操作 - -详细[计费](../concepts/billing.md)说明参见[常见问题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)中的[计费](../concepts/billing.md)相关章节。 - -## API/SDK 使用 - -百炼支持 Java 和 Python SDK，API 调用返回标准状态码标识结果。开发者常遇问题： - -| 问题 | 解决方式 | -|------|----------| -| Completion API 报错 100004（参数缺失） | 检查必填参数是否完整、格式是否正确（注意 JSON body 字段名大小写） | -| doc_reference_type 不生效 | 该参数仅旧版应用有效；新版应用需在控制台开启"展示回答来源"开关 | -| Assistant API 不支持多函数串行调用 | 当前限制，可创建多个 Assistant 分别处理 | -| Assistant API 无 memory 能力 | 当前暂不支持 | - -错误码完整列表与 SDK 安装指引请参见[常见问题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)中的 API/SDK 相关章节。 - -## 模型训练与选型 - -### 模型选择 - -- **qwen-turbo**：侧重速度与资源效率，费用更低，适合对响应速度要求高的场景 -- **qwen-max**：侧重顶级性能与全面知识，适合对精度和复杂任务处理能力要求严格的场景 -- 千问系列支持 14 种语言（中文、英文、阿拉伯语、西班牙语、法语等） -- qwen-vl-plus 已支持图片训练微调 - -### 训练注意事项 - -- 仅使用垂直领域数据做 SFT 可能导致模型遗忘通用知识 -- 训练数据需保证：任务定义清晰、数据质量高（准确简洁）、数据多样性（同一语义多种 [prompt](prompt.md) 表达） -- 循环次数与数据量无固定规律，需通过实验确定；不应仅通过 loss 判断是否过拟合，最终效果以人工评估为准 -- 训练后的模型不支持导出；本地训练的模型不支持上传 - -### 模型幻觉处理 - -降低幻觉的主要手段（按实施难度排序）： - -1. **选择更强模型**：Max > Plus > Turbo -2. **[提示词工程](../concepts/prompt-engineering.md)**：限定回答范围、要求引用来源、分步骤引导 -3. **RAG（[检索增强生成](../concepts/rag.md)）**：让模型基于检索到的知识回答，严格限制范围 -4. **插件/MCP**：数值计算等任务通过工具完成，避免模型直接处理 -5. **参数调优**：降低 temperature/top_k/top_p，降低 max_tokens 防止过度生成 -6. **后处理验证**：通过 AI 二次校验回复正确性（增加成本和延迟） - -## 产品使用要点 - -- 百炼服务需**分地域开通**，使用主账号在控制台切换目标地域后自动开通 -- 服务开通后暂不支持关闭；删除 API-Key 即可停止调用 -- 数据隔离通过[业务空间](../concepts/workspace.md)权限管理实现，不同子账号分配不同空间权限 -- 阿里云不会将用户数据用于模型训练，传输数据经 AES-256 加密；根据法规要求会存储调用数据 -- 控制台最多展示 100 条历史对话记录，不设时间限制 -- 模型生成速度非固定值，受服务负载和请求并发影响；限流触发后等待时间取决于具体限流值（如 120 RPM 则约等待 0.8 秒） - -## 服务协议 - -百炼平台涉及的主要协议包括： - -- 阿里云百炼服务协议（平台总协议） -- 阿里云百炼模型推理服务等级协议（SLA） -- 阿里云百炼服务特别说明 -- 开源模型协议条款说明 -- 三方模型服务协议和使用条款清单 - -完整协议链接请参见[相关协议](../../raw/model-user-guide/support/related-agreements.md)。 - -## 技术支持渠道 - -| 需求类型 | 联系方式 | -|----------|----------| -| 业务合作/售前咨询 | 服务热线 4008013260 或官网售前咨询 | -| 产品使用问题/售后 | 官网售后服务 | -| 合作协议申请 | 提交阿里云工单 | +阿里云百炼平台的 `support` 体系涵盖计费、API/SDK、产品使用、模型能力及法律合规等多个维度，为开发者提供从开通、调用到问题排查的全链路支持。核心支持渠道包括控制台、官方文档、错误码中心、工单系统及7×24小时客服（95187/4008013260）。所有服务均以《[阿里云百炼服务协议](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=a2ty02.30260209.aillm.1.d8bb74a10sknig)》为法律基础，数据隐私与安全严格遵循协议约定 [原文标题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 + +## 支持的模型/功能 + +- **模型类型**：支持千问系列（Qwen-Turbo、Qwen-Max、Qwen-Plus、Qwen-VL-Plus等）、万相（图像生成）及其他第三方模型；Qwen-VL-Plus 已支持图片微调训练 [原文标题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **语言支持**：千问系列支持中文、英文、阿拉伯语、西班牙语等共14种语言。 +- **关键能力**： + - RAG（[检索增强生成](../concepts/rag.md)）：可显著降低幻觉，需配合高质量检索系统与来源标注； + - Function Calling：当前**不支持单次调用中依次执行多个本地函数**，需拆分为多个 Assistant API 实例 [原文标题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)； + - Memory：Assistant API 当前**暂不支持 memory 配置功能**； + - 结构化数据对接：当前**不支持直接对接 MySQL/Hive 等外部数据库**，RDS 接入正在开发中。 + +> **注意**：文档1中提及“qwen-plus-latest 属于 Qwen3 系列”，但未明确其与 Qwen3.5/Qwen3.7 的关系；而实际模型命名体系中 Qwen3.5、Qwen3.7 为独立并列系列，并非 Qwen3 子版本——该表述易引发歧义，应以控制台模型市场实时展示为准。 + +## 关键参数 + +- `doc_reference_type`：仅在旧版应用中生效；新版应用需在控制台开启「展示回答来源」开关，否则该参数无效 [原文标题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- `temperature` / `top_k` / `top_p`：用于抑制幻觉，降低值可使输出更保守（但可能牺牲多样性）。 +- `max_tokens`：合理设置可防止模型在关键信息后继续捏造内容。 +- 必填参数（Completion API）：`AppId`、`Prompt`、`RequestId`、`User`、`Bot`；缺失或格式错误将返回错误码 `100004`。 + +## 使用方式 + +- **开通服务**：需使用阿里云主账号，在目标地域的[百炼控制台](https://bailian.console.aliyun.com/?tab=model#/model-market)开通；未实名认证用户需先完成[实名认证](https://help.aliyun.com/zh/account/verify-your-identity-individual-account)。 +- **API 调用**： + - 支持 Python/Java SDK，安装指引见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)； + - 请求示例需严格遵循 JSON 格式与 Header（含 `Authorization: Bearer `）； + - 错误码详情请查阅[错误码文档](https://help.aliyun.com/zh/model-studio/error-code)。 +- **售后支持渠道**： + - 基础服务：7×24 小时电话（95187、4008013260）、智能在线、标准工单； + - 技术问题诊断范围覆盖 API/SDK、控制台、账号、计费等 [原文标题](../../raw/model-user-guide/support/after-sales-service-scope.md)； + - 业务合作类需求请提交[阿里云工单](https://smartservice.console.aliyun.com/service/create-ticket?spm=a2c4g.2667824.0.0.6a2f6f83Ivpy5F)。 + +## 限制和注意事项 + +- **数据隔离**：通过业务空间（Business Space）实现租户级数据隔离，子账号权限需按空间粒度授权 [原文标题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md)。 +- **数据留存**： + - 控制台体验对话最多保留 **100 条历史记录**，无时间限制；未登录或推理报错对话不保存； + - 所有调用数据经 AES-256 加密传输，**绝不用于模型训练**，但依法依规存储用于服务审计与合规要求。 +- **第三方工具支持边界**： + - 阿里云仅对百炼服务端状态、API 可达性、调用明细提供支持； + - **不承担第三方工具（如 Cursor、Windsurf 等）的安装、配置、故障排查责任**，也不对其显示的 [Token](../concepts/token.md)/费用统计差异负责 [原文标题](../../raw/model-user-guide/support/after-sales-service-scope.md)。 +- **其他限制**： + - 万相会员权益**不适用于百炼 API 调用**，二者计费体系完全独立； + - 自定义模型仅支持平台内训练产出模型的二次微调，**不支持上传本地训练模型**； + - 训练完成的开源模型**暂不支持导出**。 ## 来源文档 - [常见问题](../../raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - [相关协议](../../raw/model-user-guide/support/related-agreements.md) - - - - - - - - - - - - - - - - +- [阿里云百炼平台售后服务范围说明](../../raw/model-user-guide/support/after-sales-service-scope.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md index 6f95466a..bacbbf6b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/test-1.md @@ -1,78 +1,43 @@ # test 1 -本页汇总阿里云百炼平台在计费与成本方面的核心规则，涵盖新人免费额度、模型调用价格、训练与部署计费、节省计划与资源包，以及账单查询与成本管理。面向开发者，帮助你在调用、微调、部署模型时准确预估费用并规避意外扣费。 +`test 1` 是阿里云百炼平台面向开发者提供的核心计费与资源管理主题，涵盖模型调用、训练、部署的费用结构、成本优化工具（如免费额度、节省计划、资源包）以及账单治理机制。其核心逻辑是：**免费额度优先抵扣实时推理费用 → 其次按需使用资源包或节省计划 → 最终回退至按量付费**。所有计费行为均严格绑定地域、服务部署范围及模型快照版本，开发者需在调用前明确配置并监控额度状态。 -## 计费优先级与整体逻辑 +## 支持的模型/功能 -无论使用哪种付费方式，系统在实时调用时按固定优先级自动抵扣，无需手动指定： +- **支持免费额度的模型**：仅限华北2（北京）地域且服务部署范围为“中国内地”的模型（如 `qwen3.7-plus-2026-05-26`、`qwen-max`），以及新加坡地域且服务部署范围为“国际”的模型；带日期后缀的快照版本（如 `qwen3.7-max-2026-06-08`）与不带日期的最新版（如 `qwen3.7-max`）视为独立模型，各自拥有独立的 100 万 [Token](../concepts/token.md) 免费额度 [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **支持阶梯计费的模型**：千问 Max、Plus 等系列模型按单次请求输入 [Token](../concepts/token.md) 总量分档计价（如 `0 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费** +## 关键参数 -- 免费额度仅抵扣模型**实时推理**费用，不抵扣 Batch 调用、模型调优、模型部署等场景。 -- 超出各类额度后的用量自动转为按量付费，从阿里云账户余额扣除。 +- **免费额度有效期**：90 天，自开通百炼、模型发布或模型申请通过之日三者中较晚者起算；但 **2025年9月8日11点前已开通的用户，有效期可能不足90天** [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **抵扣优先级**：系统严格遵循 `免费额度 > 资源包 > 其他模型节省计划 > AI 通用型节省计划 > 按量付费` 的顺序进行费用抵扣，该逻辑直接影响成本控制效果 [原文标题](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 +- **账单出账延迟**：大模型推理账单为分钟级出账（通常 2~10 分钟），而批量推理、模型训练、知识库等为小时级出账，高峰期可能存在进一步延迟，查询账单时需预留缓冲时间 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 -> **注意**：账户欠费（可用额度 < 0）时，即使某模型仍有免费额度也无法调用；请提前配置余额预警或消费限额。 +> **注意**：文档 5 中 `qwen3.7-max` 在华北2（北京）的输入单价标注为“原价12元 限时5折”，而文档 2 中同模型在“[模型部署](../concepts/model-deployment.md)计费”表格里未体现折扣，仅列出原价 ¥28.8/10K TPM。二者适用场景不同（文档 5 针对实时推理调用，文档 2 针对预置吞吐部署），但需警惕混淆——**部署计费不享受文档 5 所述的限时折扣**。 -## 新人免费额度 +## 使用方式 -首次开通阿里云百炼时，平台自动为各模型发放新人专属免费额度，详见 [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md)。关键规则： +- **启用免费额度**：无需额外操作，开通百炼后系统自动发放；调用时使用通用 API Key（非 Token Plan/Coding Plan 专属 Key），系统将自动优先抵扣 [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **配置成本防护**：建议开启「免费额度用完即停」功能，避免额度耗尽后意外扣费；该功能可在[免费额度页面](https://bailian.console.aliyun.com/?tab=costing-balance#/costing-balance/free-quota)或模型详情页单独/批量开启。 +- **选购成本优化方案**： + - 通用型场景：优先购买 [AI 通用型节省计划](https://common-buy.aliyun.com/?commodityCode=sfm_GenAI_spn_cn)，承诺月消费金额换取最高 5.3 折，覆盖全部阿里直供模型； + - 专项高频调用：针对特定模型（如语音、向量、万相）可选「其他模型节省计划」或「资源包」，但折扣力度和灵活性低于通用型； + - 所有方案购买后立即生效，无需手动绑定模型或 API Key。 -- **有效期**：30～90 天，自开通或模型申请通过之日起计算。自 2025 年 9 月 8 日 11 点起首次开通的用户统一调整为 90 天。到期或耗尽后自动失效，不支持补发、延期或重置。 -- **地域限制**：仅华北2（北京）地域且服务部署范围为中国内地的模型享有免费额度（新加坡国际部署同理），其他地域和部署范围无免费额度。 -- **额度独立**：每个模型（含不同快照版本，如 `qwen-max` 与 `qwen-max-2026-05-17`）拥有独立额度（通常 100 万 Token），不可跨模型合并或转移；主账号与 RAM 子账号共享同一模型的额度。 -- **免费额度用完即停**：开启后额度耗尽即停止响应并返回 `AllocationQuota.FreeTierOnly`，不再扣费。全新未认证用户默认额度耗尽后无法继续调用，需完成认证并充值。 +## 限制和注意事项 -> **注意**：Token Plan / Coding Plan 专属 API Key **不消耗免费额度**，会直接按量付费；如需使用免费额度请改用通用 API Key。 - -## 模型调用价格（按量付费） - -模型调用默认按量计费，价格表见 [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md)。要点： - -- **阶梯计费**：部分模型的单价取决于单次请求的输入 Token 总量，该请求的所有 Token 均按落入的阶梯单价结算（K=1,000，M=1,000,000）。例如 `qwen3-max` 分 0 **注意**：免费额度和节省计划均**不抵扣**模型训练与部署费用；后付费部署在账户欠费后仍会继续保留并计费 24 小时。 - -## 节省计划与资源包 - -成本优化方案见 [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md)。 - -- **AI 通用型节省计划（推荐）**：承诺每月消费金额换取阶梯折扣，最高 5.3 折，可抵扣阿里直供的全部模型。以**动态月**为周期发放额度（非自然月），当月未用完自动清零、不累积。支持抵扣模型调用、原生工具调用、上下文缓存、批量推理；不抵扣调优、部署、联网搜索插件、MCP 广场等。 -- **其他模型节省计划**：一次性购买固定金额，抵扣特定模型系列（如大语言模型、语音模型、向量排序、万相），适合用量集中场景，折扣一般不如通用型。 -- **资源包**：预购具体 Token / 张数等资源量，抵扣单个特定模型超出免费额度后的实时推理用量，到期作废。 - -> **注意**：若开启了**免费额度用完即停**（安心模式），免费额度耗尽后服务停止，节省计划将无法继续抵扣；需手动关闭该功能后才能切换到节省计划。 - -## 账单查询与成本管理 - -账单查询、分账与欠费处理见 [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 - -- **出账时效**：大模型推理分钟级出账（通常 2~10 分钟），批量推理、模型训练、知识库等小时级出账。 -- **账单详情**：核心是"实例 ID（出账粒度）"字段，格式为 `ApiKeyID;业务空间ID;模型名称;输入/输出类型;调用渠道;免费额度用完即停标识`，可据此定位产生费用的模型与渠道（`app` 代码调用、`bmp` 控制台体验、`assistant-api`）。 -- **分账管理**：给业务空间绑定标签，按部门/项目归集费用，配置后 T+1 生效。 -- **欠费与停止计费**：欠费按商品维度判定，Coding Plan / Token Plan 套餐额度独立于余额、欠费期间仍可用。停止计费需停止调用、下线部署模型、删除 API Key 或退订套餐。 - -> **注意**：按量付费采用"预占+月结"模式，并非实时扣款；系统先冻结额度，月账期结束后（次月初）生成最终账单实际扣款。此外，`enable_search` 等联网搜索附加功能按次单独计费，可能在未主动操作时仍产生费用。 +- **免费额度不覆盖场景**：明确不抵扣 Batch 调用、模型调优、[模型部署](../concepts/model-deployment.md)、自定义模型（调优后/已部署）产生的费用，且仅限实时推理 [原文标题](../../raw/model-user-guide/test-1/new-free-quota.md)。 +- **账户欠费影响全局服务**：即使某模型仍有剩余额度，只要账户整体欠费（可用额度 < 0），所有按量付费模型调用均会暂停，包括免费额度、节省计划、资源包的抵扣能力 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 +- **地域与部署范围强约束**：免费额度、部分模型价格、节省计划适用地域均存在严格限制（如华北2仅支持中国内地部署范围），跨地域调用或错误配置部署范围将导致额度不可用、价格不匹配等问题。 +- **模型单元部署的计费连续性**：[模型部署](../concepts/model-deployment.md)状态为「运行中」即开始计费，与是否发生 API 调用无关；若不再使用，必须主动下线部署，否则持续产生费用 [原文标题](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md)。 ## 来源文档 - [新人免费额度](../../raw/model-user-guide/test-1/new-free-quota.md) - [模型训练与部署计费](../../raw/model-user-guide/test-1/model-training-and-deployment-billing.md) -- [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [节省计划与资源包](../../raw/model-user-guide/test-1/savings-plan-and-resource-package.md) +- [账单查询与成本管理](../../raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [模型调用价格](../../raw/model-user-guide/test-1/model-pricing.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md index 3bb3d847..1e43a14c 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/token-plan-guide.md @@ -1,102 +1,70 @@ # token plan guide -Token Plan 团队版与 Coding Plan 是百炼面向 AI 编程/智能体工具的两类订阅服务：Token Plan 团队版以 Credits 统一计量、按 Token 消耗抵扣，支持文本与图像生成模型并提供团队管理后台；Coding Plan 面向个人开发场景，按模型调用次数计费并设有请求限额。两者的 API Key 与 Base URL 完全隔离、互不相通，接入前需先明确使用的是哪种套餐。 +[Token](../concepts/token.md) Plan 是阿里云百炼面向开发者推出的 AI 大模型统一订阅服务，以 Credits 为计量单位，支持文本、[多模态](../concepts/multi-modal.md)生成及 Harness 工具调用。它分为个人版（面向单人交互式开发）和团队版（面向多人协作与企业级管理），均仅限华北2（北京）地域使用。服务通过专属 `sk-sp-` 开头的 API Key 与隔离 Base URL 实现计费隔离，严禁用于生产环境自动化调用。 -## 两种套餐对比 +## 支持的模型/功能 -| 维度 | Token Plan 团队版 | Coding Plan | -| --- | --- | --- | -| 适用场景 | 一人公司/团队/企业日常办公 | 个人开发场景 | -| 支持模型 | 文本生成 + 图像生成 | 文本生成模型 | -| 计费方式 | 按 Token 消耗抵扣 Credits | 按模型调用次数 | -| 使用频次 | 无每 5 小时/每周限额 | 有每 5 小时/每周/每月限额 | -| 高峰期性能 | 多租户隔离，不排队 | 高峰期可能排队 | -| 数据安全 | 承诺不使用数据训练模型 | 用户数据授权用于服务改进 | +[Token](../concepts/token.md) Plan 支持覆盖推理、视觉理解、图像生成、视频生成、语音处理等能力的[多模态](../concepts/multi-modal.md)模型，以及联网搜索、代码解释器等 Harness 工具： -> **注意**：两个计划互相独立，不支持互转（即使补差价也不行），可同时订阅、各自计费。详见 [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) 与 [Coding Plan概述](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md)。 +- **主流模型**：`qwen3.8-max-preview`（预览版，享限时 1 折+夜间 0.2 折优惠）、`qwen3.7-plus`、`qwen3.6-flash`、`glm-5.2`、`deepseek-v4-pro`、`wan2.7-image`、`happyhorse-1.1-t2v` 等（[完整列表见文档](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md)）。 +- **Harness 工具**：仅 `qwen3.7` 及以上系列模型原生支持，包括 `web_search`、`code_interpreter`、`t2i_search`、`i2i_search`、`web_extractor`（[接入说明详见](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md)）。 +- **[多模态](../concepts/multi-modal.md)生成**：图像生成（`qwen-image-2.0`、`wan2.7-image`）和视频生成（`happyhorse-1.1-t2v` 等）需通过工具扩展机制（如 Slash Command、Skill、Agent）调用独立 API（[接入指南见](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)）。 +- **视觉理解**：`qwen3.7-plus`、`qwen3.6-plus`、`kimi-k2.5` 等模型原生支持图片输入；纯文本模型（如 `glm-5`）需通过 Skill 或 Agent 借助视觉模型实现（[配置方法见](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md)）。 -## 支持的模型 +> **注意**：文档 8（Coding Plan 概述）中列出的 `qwen3-coder-next`、`qwen3-coder-plus` 等模型虽在 Coding Plan 中支持，但**未出现在 [Token](../concepts/token.md) Plan 个人版或团队版的任一支持模型列表中**，实际不可用。请以 [token-plan-personal-overview.md](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) 和 [token-plan-team-overview.md](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) 的白名单为准。 -Token Plan 团队版的模型清单为**精确字符串白名单**，必须逐字符完全匹配，版本号/子型号任何差异均视为不支持，禁止版本兼容推理。 +## 关键参数 -- **千问**：qwen3.7-max（限时活动）、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash、qwen-image-2.0、qwen-image-2.0-pro -- **万相**：wan2.7-image、wan2.7-image-pro -- **DeepSeek**：deepseek-v4-pro、deepseek-v4-flash、deepseek-v3.2 -- **月之暗面**：kimi-k2.7-code、kimi-k2.6、kimi-k2.5 -- **智谱 AI**：glm-5.2、glm-5.1、glm-5 -- **MiniMax**：MiniMax-M2.5 +| 参数 | 说明 | 示例值 | +|------|------|--------| +| **API Key** | 必须为 `sk-sp-` 开头的 Token Plan 专属密钥，与百炼通用 `sk-` Key 及 Coding Plan Key 完全隔离 | `sk-sp-xxxxxxxx` | +| **Base URL** | OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`；Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | 同上 | +| **Credits 消耗** | 动态计算，取决于模型类型、输入/输出 token 数、思考模式启用状态及 Harness 工具调用次数 | `qwen3.6-plus` 单次请求约 3.18 Credits（[计算示例见](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md)） | +| **并发 Agent 数** | 个人版按档位限制：Lite（1–2）、Standard（3–4）、Pro（6–8）；团队版无硬性并发上限，依赖席位额度与系统负载 | Standard 套餐支持 3–4 个并发 Agent | -Coding Plan Pro 套餐的推荐模型为 qwen3.7-plus、qwen3.6-plus、kimi-k2.5（均支持图片理解）、glm-5、MiniMax-M2.5，更多模型见 [Coding Plan概述](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md)。 +## 使用方式 -> **注意**：两套清单的白名单不完全一致，且 Coding Plan Lite 套餐已于 2026 年 3 月 20 日停止新购、4 月 13 日停止续费与升级。调用时务必以对应套餐控制台的实时清单为准。 +1. **订阅与授权** + - 访问 [Token Plan 控制台（华北2）](https://bailian.console.aliyun.com/cn-beijing?tab=plan#/efm/subscription/overview) 完成购买。 + - RAM 用户需由主账号授予 `AliyunTokenPlanFullAccess`（或 `ReadOnlyAccess`）及 `AliyunBSSReadOnlyAccess`（个人版）或 `AliyunBSSFullAccess`（团队版）策略，并在百炼控制台分配订阅权限（[授权步骤详见](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md)）。 -## 快速接入（三步） +2. **获取凭证** + - 个人版：在「我的订阅」页面生成 API Key（仅显示一次）。 + - 团队版：在成员管理页为成员分配席位后，为其生成专属 API Key（[操作流程见](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md)）。 -以 Token Plan 团队版为例，详见 [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md)： +3. **配置工具** + - 将 API Key 与对应协议的 Base URL 配置至 Cursor、Claude Code、Qwen Code、Qoder 等兼容工具。 + - 多模态生成与 Harness 工具需额外配置（如 Claude Code 的 Slash Command、OpenCode 的 Agent、MCP 服务等），具体路径与脚本参见各实践文档。 -1. **订阅**：在购买页选择坐席类型、数量和订阅周期。RAM 子账号订阅前需主账号授予 `AliyunBailianFullAccess` 权限。 -2. **获取 API Key 和 Base URL**：分配席位后为成员生成专属 API Key（Token Plan 以 `sk-sp-` 开头，与通用 `sk-` 不可混用；仅首次显示一次，需立即保存）。 -3. **接入 AI 工具**：支持 Claude Code、Qwen Code、OpenCode、OpenClaw、Cursor、Codex、Qoder、Cline、Kilo CLI 等。 +## 限制和注意事项 -### Base URL 对照 - -| 套餐 / 协议 | Base URL | -| --- | --- | -| Token Plan · OpenAI 兼容 | `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | -| Token Plan · Anthropic 兼容 | `https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | -| Coding Plan · OpenAI 兼容 | `https://coding.dashscope.aliyuncs.com/v1` | -| Coding Plan · Anthropic 兼容 | `https://coding.dashscope.aliyuncs.com/apps/anthropic` | -| 按量付费 · OpenAI 兼容 | `https://dashscope.aliyuncs.com/compatible-mode/v1` | - -> **注意**：Token Plan、Coding Plan、按量付费三者的 API Key 与 Base URL 必须配套使用。混用会导致走按量计费通道产生意外扣费，或返回 401/403 鉴权失败。 - -## 工具调用与扩展能力 - -- **模型内置工具**：qwen3.7-max、qwen3.7-plus、qwen3.6-plus、qwen3.6-flash 通过 Responses API 内置联网搜索、代码解释器、网页抓取、以图搜图、文搜图 5 个工具，不额外收费，token 消耗统一从套餐 Credits 抵扣。 -- **MCP 服务**：其他模型（如 deepseek-v3.2、glm-5）通过百炼 MCP 广场接入工具。联网搜索 MCP 前 2000 次调用免费，之后按 29 元/千次计费。接入 MCP 用的是**百炼通用 API Key（`sk-xxx`）**，而非套餐专属 Key。 -- **图像生成模型**：不在文本模型清单展示，需通过工具的 Skill / Slash Command / Agent 机制调用 `multimodal-generation` API，详见 [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md)。 -- **视觉理解**：qwen3.6-plus、qwen3.5-plus、kimi-k2.5 原生支持视觉；glm-5、MiniMax-M2.5 等纯文本模型可通过 Skill/Agent 辅助获得视觉能力。OpenCode/OpenClaw 需在配置中显式声明 `modalities`/`input` 为 `["text","image"]`。 - -## Credits 计费与额度 - -Token Plan 团队版单次请求消耗的 Credits **并非固定值**，由模型类型、Token 用量、思考模式及工具调用动态决定。多轮对话中上下文持续累积，消耗会随之上升；部分模型按上下文长度阶梯计费，长上下文可能进入更高价位档。 - -抵扣顺序：坐席套餐月度额度 → 共享用量包（多个时优先扣最近到期的）→ 全部用尽后服务暂停至下一计费周期。 - -> **注意**：续费/续订只延长有效期或预定下期额度，**不会叠加补充到当前计费周期**。当期额度用尽需立即恢复时，应购买共享用量包、升级坐席或加购坐席（加购后需分配给成员才能使用）。 - -控制消耗建议：任务切换时及时开启新会话、清理无关历史；对长文档/大代码库按需拆分输入；在控制台订阅页用量明细关注实时消耗趋势。 - -## 常见报错速查 - -- **401 Invalid API-key / invalid access token**：误用了通用 Key 或其他套餐的 Key/Base URL、订阅过期、或 Key 复制不完整含空格。核对套餐专属 Key 与配套 Base URL，必要时重置。 -- **404 model not found / model not [support](support.md)ed**：模型名拼写或大小写错误，或不在套餐白名单内。 -- **400 url error / Range of input length**：Base URL 路径与协议不匹配（Anthropic 端点以 `/apps/anthropic` 结尾，OpenAI 端点以 `/compatible-mode/v1` 或 `/v1` 结尾），或输入超出上下文长度（新建会话或切换更长上下文模型）。 -- **429 quota exceeded**：套餐额度用尽（加购/等待重置）或触发 TPS/TPM 限流（限流按主账号维度合并计算，等待约一分钟后平滑重试）。 -- **Coding Plan 限额类**：`hour/week/month allocated quota exceeded` 分别对应每 5 小时（滚动恢复）、每周一 00:00 重置、每月订阅日重置。 - -完整报错表见 [Token Plan 常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md) 与 [Coding Plan 常见问题](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md)。 - -## 团队管理与使用限制 - -- **角色**：所有者、管理员（权限同所有者，可被移除/降级）、成员（仅使用分配的 Key 调用）。 -- **成员接入**：支持手动添加（仅供 API 调用）、SAML 2.0（SSO）、钉钉登录三种方式。 -- **席位操作**：分配后自动生成 API Key；回收后席位释放、原 Key 失效；加购/升级按剩余时长折算费用；退订按席位维度，已消耗用量的席位不可退订。 -- **使用范围**：仅限在兼容的 AI 编程和智能体工具中**交互式**使用，禁止用于自动化脚本或应用后端，违规可能导致订阅暂停或 API Key 封禁。 - -> **注意**：Token Plan 团队版目前仅支持**华北2（北京）**地域；每个阿里云账号限购一个订阅，共享用量包需先订阅坐席套餐后才能购买、有效期 1 个月且到期清零。 +- **地域限制**：个人版与团队版均**仅支持华北2（北京）地域**，控制台需手动切换（[原文标题](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)）。 +- **额度机制差异**： + - 个人版采用 **5 小时 + 7 天双窗口限额**（非日历周期，自首次调用起计时），任一窗口触顶即暂停服务； + - 团队版采用 **月度总额度制**（无窗口限制），支持加购共享用量包（625,000 Credits/个）补充额度（[对比详见](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md)）。 +- **使用场景限制**： + - 严禁用于 API 自动化调用（如后台服务、定时任务、批量脚本），仅限交互式开发工具内使用；违规将导致 API Key 封禁（[原文标题](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md)）。 + - 个人版数据授权条款允许用于服务改进；团队版明确承诺**不使用对话数据训练模型**。 +- **模型与工具兼容性**： + - qwen3.8-max-preview 为预览模型，能力持续迭代，预览结束后可能下线或替换（[原文标题](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md)）。 + - 联网搜索 MCP 需使用**百炼通用 API Key（`sk-`）**，而非 Token Plan 专属 Key（[原文标题](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md)）。 +- **其他**：API Key 重置或退订重购后会变更，需在工具中重新配置；同一账号可同时持有个人版与团队版，额度与计费完全独立。 ## 来源文档 -- [Token Plan（团队版）概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) -- [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-quickstart.md) -- [团队管理](../../raw/model-user-guide/token-plan-guide/token-plan-team.md) -- [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-faq.md) -- [工具调用](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-tool.md) -- [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) +- [Token Plan 概述](../../raw/model-user-guide/token-plan-guide/token-plan-overview.md) +- [概述](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) +- [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md) +- [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md) +- [快速开始](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md) +- [团队管理](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md) +- [常见问题](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md) - [Coding Plan概述](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) -- [联网搜索](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/web-search-for-coding-plan.md) -- [添加视觉理解能力](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/add-vision-skill.md) - [常见问题](../../raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) +- [接入 Harness 工具](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md) +- [接入多模态生成模型](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) +- [联网搜索](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md) +- [添加视觉理解能力](../../raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md) +- [概述](../../raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md index c59a6d33..37e1809a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-cases.md @@ -1,84 +1,61 @@ # use cases -本页汇总阿里云百炼平台的典型使用场景与实践指南，覆盖三大方向：Prompt 设计技巧（文生文、文生图、文生视频）、第三方/多供应商模型接入（DeepSeek、Kimi、GLM、MiniMax、MiMo、Stepfun 等），以及工程化最佳实践（RAG、限流应对、显式缓存、模型调优、端到端解决方案）。面向开发者，以下内容按主题组织，便于快速定位到对应的参数、调用方式和注意事项。 +百炼平台的 use cases 覆盖从[多模态](../concepts/multi-modal.md)内容生成、智能体构建、深度研究到教育辅助等全场景落地实践。这些方案均基于百炼统一模型服务与编排能力，支持开箱即用的部署流程和面向生产环境的工程化配置（如限流、缓存、RAG集成），开发者可快速验证业务逻辑并规模化上线。 -## Prompt 设计与生成类场景 +## 支持的模型/功能 -针对不同模态，百炼提供了结构化的提示词方法论： +百炼提供两类核心能力：**原生模型服务**与**第三方模型直供接入**。 +- **原生模型**包括 `qwen3-vl-plus`（用于AI解题与批改）、`wan2.7`（文生视频/图生视频）、`qwen-deep-research`（深度研究）、`qwen3.7-*` 系列（通用文本生成）等，均深度集成于百炼控制台与API体系。 +- **第三方模型**通过标准化接口接入，覆盖 DeepSeek（[DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md)、[DeepSeek-硅基流动](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md)、[DeepSeek (快手万擎)](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md)）、Kimi（[Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md)、[Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md)）、GLM（[GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md)、[GLM-智谱](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md)）、MiniMax（[MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md)、[MiniMax (稀宇科技)](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md)）、MiMo（[MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md)）、Stepfun（[Stepfun-阶跃星辰](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md)）及 Vidu（[Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)）。 +> **注意**：多个第三方模型文档（如 DeepSeek、Kimi、GLM、MiniMax）均声明部分旧版本将于 2026 年中下架，并统一推荐迁移至 `qwen3.7-plus`/`qwen3.7-max`/`qwen3.6-flash`。该迁移路径具有一致性，但各供应商的地域支持范围存在差异（例如硅基流动仅限华北2，而阿里云百炼版支持多地域），需按实际部署需求选择。 -- **文生文**：推荐使用「背景 / 目的 / 风格 / 语气 / 受众 / 输出」六要素的 Prompt 框架，任务描述越清晰具体，模型表现越贴近预期。控制台还提供 Prompt「自动优化」工具，可自动扩写和补充细节（该功能调用大模型，按推理费用计费）。详见 [文生文Prompt指南](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md)。 -- **文生图**：核心参数为正向提示词 `prompt`、反向提示词 `negative_prompt`；文生图 V2 额外支持 `prompt_extend`（默认 `true`，开启大模型智能改写）。提示词公式分基础版（主体 + 场景 + 风格）与进阶版（增加镜头语言、氛围词、细节修饰），并配有景别、视角、风格、光线等提示词词典。详见 [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md)。 -- **文生视频 / 图生视频**：正向提示词描述画面内容与运动过程。基础公式为「主体 + 场景 + 运动」，进阶公式增加「美学控制 + 风格化」，图生视频则以「运动 + 运镜」为主。较新的 wan2.7 / wan2.6 还支持声音公式（人声/音效/BGM）、多镜头公式（镜头序号 + 时间戳 + 分镜内容）和参考生视频公式。详见 [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md)；第三方视频模型可参考 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)（含大动态、运镜、风格等触发关键词词典）。 +## 关键参数 -> **注意**：wan2.7 模型不再支持通过 `shot_type` 指定单镜头/多镜头，改由模型结合提示词自行发挥；如需一镜到底，中文写「生成单镜头」、英文写「Generate single shot.」。 +不同模态任务依赖特定参数组合： +- **文生图（万相）**：必填 `prompt`（正向提示词），可选 `negative_prompt`（反向提示词）与 `prompt_extend`（是否启用大模型智能扩写，默认 `true`）[文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md)。 +- **文生视频/图生视频（万相）**：支持结构化公式，关键参数包括 `prompt`（主体+场景+运动）、`prompt_extend`（同上）、`enable_thinking`（控制思考模式，非OpenAI标准参数，需通过 `extra_body` 传入）[文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md)。 +- **第三方模型调用**：普遍支持 `enable_thinking` 或 `reasoning_effort` 控制推理过程输出；Vidu 支持 `大动态`/`固定镜头` 等运镜关键词 [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md)。 +- **缓存与限流**：显式缓存需在请求中注入 `cache_control` 标记；限流应对需配置 `X-DashScope-Wait-Timeout` 请求头或客户端令牌桶策略 [限流应对最佳实践](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md)、[显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)。 -## 第三方与多供应商模型接入 +## 使用方式 -多篇教程介绍了在百炼平台通过 **[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)** 或 **DashScope SDK** 调用第三方模型，通用要点如下： +典型工作流分为三类： +1. **低代码编排**：通过无限画布（[HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)）或节点式工作流（[高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md)）可视化连接模型、工具与数据源，无需编写代码即可构建端到端应用。 +2. **SDK/API 集成**：使用 OpenAI 兼容 SDK（如 `openai` Python 包）或 DashScope SDK，按模型文档指定 `base_url`、`model` 名称及参数（如 `extra_body={"enable_thinking": True}`）发起调用。所有第三方模型均提供 Python/Node.js 示例代码。 +3. **RAG 与知识库增强**：基于 LlamaIndex 构建检索增强应用，通过 `DashScopeCloudIndex` 创建知识库，再以 `as_query_engine()` 封装为可调用接口 [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md)。 -- **前置条件**：先[获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key) 并配置到环境变量；部分模型需在控制台模型广场「立即开通」后才能调用。 -- **思考模式**：多数模型通过 `enable_thinking` 参数控制是否输出推理过程（`reasoning_content`）。注意 `enable_thinking` 非 OpenAI 标准参数——OpenAI Python SDK 需通过 `extra_body` 传入，Node.js SDK 作为顶层参数传入。 -- **地域差异**：不同地域的 Base URL 不同，部分供应商（硅基流动、快手万擎、月之暗面、智谱、MiniMax、小米、阶跃星辰）仅限特定地域（多为华北2（北京））。详见 [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) 与 [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md)。 +## 限制和注意事项 -各供应商的默认思考模式行为并不一致，接入时需按模型区分： - -- **默认开启思考**：`mimo-v2.5-pro`（[MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md)）、`kimi/kimi-k2.6`/`kimi-k2.5` 默认开启，可关闭。 -- **仅思考模型**：`kimi/kimi-k2.7-code` 系列 `enable_thinking` 始终为 `true`，无法关闭；`kimi-k2.7-code-highspeed` 功能与 `kimi-k2.7-code` 一致但速度提升 5~6 倍（见 [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md)）。 -- **默认关闭思考**：`stepfun/step-3.7-flash` 默认关闭，需显式开启，并可用 `reasoning_effort`（`low`/`medium`/`high`）控制深度。 -- **供应商差异**：同为 DeepSeek，硅基流动供应商支持更长上下文；阿里云百炼供应商限流更宽松，并支持联网搜索与上下文缓存。GLM 智谱直供的 `glm-5.2` 支持 1M 上下文，并可用 `reasoning_effort`（`max`/`high`/`none`）。 - -> **注意**：多篇教程标注 deepseek-v3/v3.1/v3.2/r1 系列、`MoonshotKimi-K2` 与 `kimi-k2-thinking`、`glm-4.6`/`glm-4.7`、`MiniMax-M2.1` 等模型将于 **2026年7月9日** 下架，推荐转用 `qwen3.7-plus` / `qwen3.7-max` / `qwen3.6-flash`。同时不同教程示例中出现的模型版本号存在差异（如 deepseek-v3.2 与 deepseek-v4-pro、MiniMax-M2.5 与 MiniMax-M2.7、kimi-k2.5 与 kimi-k2.7），以模型广场实际可用列表为准。 - -## RAG 与知识库 - -[基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) 演示了在 LlamaIndex 中使用百炼检索增强服务的完整链路： - -- 安装 `llama-index-core`、`llama-index-llms-dashscope`、`llama-index-indices-managed-dashscope`（Python 版本要求 >=3.8 且 <=3.12）。 -- 使用 `DashScopeParse` 在线解析 .doc/.docx/.pdf 文件（单文件 <100M、页数 <1000），再通过 `DashScopeCloudIndex.from_documents` 创建知识库，`index.as_retriever()` / `index.as_query_engine()` 获取检索器与查询引擎。 - -## 工程化最佳实践 - -- **限流应对**：百炼 API 按 RPM/TPM（分钟级）、RPS/TPS（瞬时）、Traffic Burst（增速）三种规则限流，按主账号维度、模型独立计算，触发后通常 1 分钟恢复。方案按改动成本由低到高分为平台配置（服务端排队等待、提升额度、PTU、Batch API）、客户端流控（重试、令牌桶、平滑限速、自适应拥塞控制）、架构兜底（模型降级、MQ 削峰）。针对突发限流推荐首选在请求头添加 `X-DashScope-Wait-Timeout`（建议 3~120 秒），并相应调大客户端超时时间。详见 [限流应对最佳实践](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md)。 -- **显式缓存**：通过在请求中添加缓存标记实现 100% 确定性命中，适合高频复用相同 Prompt、长上下文 Agent 等场景。首次写入约产生标准价格 25% 的额外开销，后续命中可节省约 90% 成本。Claude Code、OpenCode、OpenClaw 等工具通过 Anthropic 兼容端点（`/apps/anthropic`）接入时原生支持。详见 [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)。 -- **自定义模型**：创建自定义模型分为模型调优、模型部署、模型评测三个主步骤，配套训练数据准备、评测模板设计、调整训练策略。数据需编排为「Prompt-Completion」格式，建议至少准备 500 条并做脱敏处理。注意**调优后的模型必须先部署才能调用和评测**。详见 [自定义模型调优、部署与评测](../../raw/model-user-guide/use-cases/model-training-best-practices.md)。 - -## 端到端解决方案 - -多篇实践方案展示了如何组合百炼模型能力构建完整应用，多数基于函数计算 FC、开箱即用并提供免费试用额度： - -- **文档转视频**：结合大模型与多模态技术，将文档自动切片、生成演示文稿、语音字幕并合成视频，依赖 FFmpeg 与 Marp 工具，提供完整代码包。详见 [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md)。 -- **AI 智能体与工作流**：以 AI 电商客服为例，覆盖智能问答、RAG、自主决策 Agent、对话流四种应用形态（[高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md)）。 -- **视觉创作平台**：集成 Wan2.7 图像生成与 HappyHorse 视频生成，提供节点式编排、AI 导演与在线剪辑（[HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md)）。 -- **深度研究报告**：Qwen-Deep-Research 自动规划检索路径、多源交叉验证并生成结构化洞察报告（[深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md)）。 -- **AI 解题批改**：基于 Qwen3-VL 视觉模型实现拍照解题与作业自动批改，支持 33 种语言（[AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md)）。 +- **地域与权限约束**：多数第三方模型（如硅基流动、月之暗面、快手万擎、小米、阶跃星辰）仅支持华北2（北京）地域，且需对应地域的 API Key；部分模型（如 Kimi、GLM）在新加坡/美国等地域需替换 `WorkspaceId` 到 Base URL 中。 +- **限流维度**：百炼 API 同时受 RPM（每分钟请求数）、TPM（每分钟 [Token](../concepts/token.md) 数）、RPS/TPS（瞬时速率）及 Traffic Burst（增速突增）四重限制，单一重试策略无效，必须结合服务端排队（`X-DashScope-Wait-Timeout`）或客户端自适应拥塞控制 [限流应对最佳实践](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md)。 +- **缓存生效条件**：显式缓存要求输入内容完全一致（含 system [prompt](prompt.md) 动态字段），Claude Code 等工具默认注入当前目录/日期等变量，会降低跨会话命中率，需启用 `--exclude-dynamic-system-prompt-sections` 参数优化 [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md)。 +- **模型生命周期**：所有第三方模型文档均明确标注下架时间（集中于 2026 年 7–10 月），且推荐路径统一指向 Qwen 系列，开发者应规划迁移节奏，避免依赖已标记为 deprecated 的模型。 ## 来源文档 +- [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md) +- [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) +- [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md) +- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](../../raw/model-user-guide/use-cases/prompt-engineering-guide.md) - [文生图Prompt指南](../../raw/model-user-guide/use-cases/text-to-image-prompt.md) -- [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [文生视频/图生视频Prompt指南](../../raw/model-user-guide/use-cases/text-to-video-prompt.md) +- [基于LlamaIndex构建RAG应用](../../raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) - [自定义模型调优、部署与评测](../../raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](../../raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - [限流应对最佳实践 ](../../raw/model-user-guide/use-cases/rate-limiting-best-practices.md) +- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) - [DeepSeek-阿里云](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) - [DeepSeek-硅基流动](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/siliconflow-deepseek-api.md) - [DeepSeek](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api-by-vanchin.md) - [Kimi](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api.md) - [Kimi-月之暗面](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/kimi-api-by-moonshot-ai.md) - [GLM](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm.md) -- [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [GLM-智谱](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) +- [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [MiniMax](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) - [MiMo-小米](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) -- [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [Stepfun-阶跃星辰](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) -- [显式缓存最佳实践](../../raw/model-user-guide/use-cases/explicit-cache-guide.md) -- [HappyHorse 打造一站式影视创作平台](../../raw/model-user-guide/use-cases/infinite-canvas.md) -- [高效搭建 AI 智能体与工作流应用](../../raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) -- [深度研究：生成你的独家洞察报告](../../raw/model-user-guide/use-cases/deep-research.md) -- [AI 解题 + 批改：推动课程教学智变](../../raw/model-user-guide/use-cases/ai-homework-helper.md) - - +- [Vidu视频生成Prompt指南](../../raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md index 6f73a29b..58d2b5e1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md +++ b/skills/bailian-docs-llm-wiki/wiki/guides/use-chat-client-or-development-tool.md @@ -1,77 +1,53 @@ # use chat client or development tool -阿里云百炼支持将平台上的模型接入各类第三方 AI 聊天客户端、编程工具与应用开发平台。这些工具本身不由百炼提供，接入方式统一为「填入 Base URL + API Key + 模型 ID」，通过 **OpenAI 兼容协议**或 **Anthropic 兼容协议**访问百炼网关。本文汇总不同工具的接入要点、共用的凭证规则以及常见限制。 +阿里云百炼支持通过多种主流 AI 开发工具和客户端接入，包括终端编程助手（如 Hermes Agent、Claude Code）、IDE 插件（如 Cline、Qoder）、桌面应用（如 Cursor、Cherry Studio）以及开源 Agent 框架（如 OpenClaw、QwenPaw）。所有工具均通过 OpenAI 或 Anthropic 兼容协议对接，开发者可基于自身技术栈选择合适工具，快速集成百炼模型能力。 -## 支持的工具类型 +## 支持的模型/功能 -按形态大致分为三类： +百炼支持的模型因计费方案而异，**[Token](../concepts/token.md) Plan 个人版**与**团队版**均支持 `qwen3.8-max-preview`（强制开启思考模式）、`qwen3.7-max`、`qwen3.7-plus`、`qwen3.6-flash`、`glm-5.2`、`deepseek-v4-pro` 等文本生成模型；**Coding Plan** 主要面向开发场景，推荐 `qwen3.7-plus`；**按量计费**覆盖最全模型集，包括文生图（`wan2.6-t2i`）、文生视频（`wan2.1-t2v-turbo`）等 AIGC 模型。视觉与[多模态](../concepts/multi-modal.md)能力需显式启用（如 Qwen-VL、QVQ），详见 [使用Postman或cURL调用图像/视频生成API](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md)。 -- **终端 / CLI 编程工具**：[Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md)、[Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md)、[OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md)、[Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md)、[Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md)、[Kilo CLI](../../raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md)、Qoder CLI。 -- **IDE / 编辑器插件**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)、[Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md)（VSCode）、Qoder（IDE / JetBrains 插件）、Qoder CN（原 Lingma）。 -- **桌面 / 跨平台聊天客户端与助手**：[Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)、[Chatbox](../../raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md)、[OpenClaw](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md)、QwenPaw。 -- **应用开发 / 工作流平台**：[Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md)。 +> **注意**：文档中多次提及 `qwen3.8-max-preview` 的思考模式为“始终开启，不支持关闭”，但不同工具对参数暴露程度不一：OpenClaw 配置中未显式声明 `thinking` 字段 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md)，而 Claude Code 和 Qwen Code 明确要求设置 `enable_thinking: true` 或 `thinking.type: "enabled"`。实际调用时应以具体工具配置为准。 -此外，任何兼容 OpenAI / Anthropic 协议且支持自定义服务端点的工具（如 Trae）都可参照[更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)接入。若只想快速验证图像/视频生成 API，可用 [Postman 或 cURL](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) 直接调用。 +部分工具（如 Cursor、Chatbox、Cherry Studio）要求模型 ID 使用别名格式（如 `kimi-k2.6` → `kimi-k2-6`，`glm-5.2` → `glm-5-2`），否则报错 `Named models unavailable` 或 `Unknown Custom model Exception` [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)。Dify 等低代码平台则需通过插件（如“通义千问”）或 HTTP 节点间接接入万相模型，不支持直接配置 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md)。 -## 三种计费方案与凭证 +## 关键参数 -绝大多数工具的接入差异只在「Base URL 属于哪个方案」。百炼提供三种计费方案，各自有独立的 API Key，**互不通用**： +| 参数 | 说明 | 典型值 | +|------|------|--------| +| `base_url` | API 端点地址，协议决定兼容性 | OpenAI 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`；Anthropic 兼容：`https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | +| `api_key` | 方案专属密钥，**不可跨方案复用** | [Token](../concepts/token.md) Plan 个人版 Key 仅适用于 [Token](../concepts/token.md) Plan Base URL，与 Coding Plan Key 不互通 | +| `model_id` | 模型标识符，需与套餐支持列表严格匹配 | `qwen3.8-max-preview`、`wan2.6-t2i`、`text-embedding-v4` | +| `reasoning_effort` | 仅 `qwen3.8-max-preview` 支持，控制推理深度 | `xhigh`（默认）、`high`、`low` | +| `enable_thinking` | 思考模式开关，部分工具强制为 `true` | Qwen Code 中需在 `extra_body` 中显式设为 `true` | -| 方案 | 说明 | OpenAI 兼容 Base URL | Anthropic 兼容 Base URL | -| --- | --- | --- | --- | -| Token Plan 团队版 | 按坐席订阅，按 token 消耗抵扣 Credits | `https://token-plan.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` | `https://token-plan.cn-beijing.maas.aliyuncs.com/apps/anthropic` | -| Coding Plan | 固定月费订阅，按模型调用次数计量 | `https://coding.dashscope.aliyuncs.com/v1` | `https://coding.dashscope.aliyuncs.com/apps/anthropic` | -| 按量计费（华北2·北京） | 按实际调用量后付费 | `https://dashscope.aliyuncs.com/compatible-mode/v1` | `https://dashscope.aliyuncs.com/apps/anthropic` | +地域相关参数需严格对齐：按量计费的 `base_url` 与 `api_key` 必须同属华北2（北京）、新加坡或美国（弗吉尼亚）任一地域，否则返回 401 错误 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md)。 -按量计费还支持多地域，需保证 API Key 与 Base URL 地域一致： +## 使用方式 -- 新加坡：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1`（`WorkspaceId` 替换为真实值） -- 美国（弗吉尼亚）：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` +1. **安装工具**：根据操作系统选择安装方式（如 `npm install -g`、一键脚本、GUI 安装包），确保依赖版本满足要求（Node.js ≥ 18，Python 3.10–3.13）。 +2. **配置凭证**： + - CLI 工具（Hermes Agent、Claude Code）通过命令行或配置文件（`~/.hermes/config.yaml`、`~/.claude/settings.json`）设置； + - GUI 工具（Cursor、Cherry Studio）在设置界面填写 API Key、Base URL 和 Model ID； + - Agent 框架（OpenClaw、QwenPaw）通过交互式向导或 Web Console 配置提供商。 +3. **验证连接**：发送简单请求（如 `"你好"`）确认响应正常；AIGC 类任务需遵循异步流程（创建任务 → 轮询 `task_id`）[原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md)。 +4. **高级能力**：启用思考模式、[多模态](../concepts/multi-modal.md)输入（图片）、自定义 Skill（如百炼 CLI）需按各工具文档单独配置。 -> **注意**：OpenAI 协议的 Base URL 以 `/compatible-mode/v1`（或 `/v1`）结尾，Anthropic 协议以 `/apps/anthropic` 结尾。部分工具（如 OpenCode、Kilo CLI）要求在 Anthropic 端点后再追加 `/v1`。以各工具原文为准。 +## 限制和注意事项 -## 协议选择与配置形态 - -不同工具选用的协议和配置载体各异： - -- **Anthropic 协议**：[Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) 通过 `~/.claude/settings.json` 的 `ANTHROPIC_BASE_URL` / `ANTHROPIC_AUTH_TOKEN` 环境变量配置；[Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) 默认使用 Anthropic 协议（`api_mode: anthropic_messages`），也可切到 OpenAI 协议。 -- **OpenAI 协议**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md)、[Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md)、Cherry Studio、Chatbox 等在 GUI 中选择「OpenAI Compatible / 兼容」并填入 Base URL、API Key、模型 ID。 -- **配置文件**：Hermes（`~/.hermes/config.yaml`）、OpenCode（`~/.config/opencode/opencode.json`）、Kilo CLI（`~/.config/kilo/config.json`）、Qwen Code（`~/.qwen/settings.json`）、Codex（`~/.codex/config.toml` + `OPENAI_API_KEY` 环境变量）。 -- **原生下拉选择**：Qoder / Qoder CN 在设置中选择「阿里云百炼 - 国内」提供商 + 计费方案「类型」，仅需填 API Key。 - -## 关键参数与注意事项 - -- **思考模式**：许多模型（如 Qwen3 思考模式、QwQ）需显式开启思考。OpenCode / Kilo CLI 用 `thinking.budgetTokens`，Qwen Code 用 `extra_body.enable_thinking: true`，Cline 需勾选 **Enable R1 messages format**。若报错 `enable_thinking parameter is restricted to True`，说明该模型仅支持思考模式运行，需在客户端开启。 -- **模型名称别名**：[Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) 因内置模型名冲突，需改写模型名，如 `kimi-k2.6` → `kimi-k2-6`、`glm-5` → `glm-5-0`。其他工具一般直接使用原始模型 ID。 -- **上下文窗口**：Claude Code 默认 200K，可通过 `CLAUDE_CODE_MAX_CONTEXT_TOKENS=1000000` 或模型名后缀 `[1m]` 扩展到 1M（需模型支持）。 -- **Codex 版本差异**：仅 qwen3.7-max/plus、qwen3.6-plus/flash 支持 Responses API（可用最新版 Codex）；其他模型需用 Chat/Completions API，须安装旧版本（如 `@openai/codex@0.80.0`）。 -- **401 认证失败**：几乎都是「API Key 与 Base URL 不属于同一方案」或「按量计费 Key 与地域不匹配」，逐项核对即可。 - -## 套餐使用范围限制 - -> **注意**：Token Plan 团队版与 Coding Plan **仅限**在 AI 编程工具和 OpenClaw 类 Agent 中使用。以下类型不支持接入，误用可能导致订阅暂停或 API Key 被封禁（详见[更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)）： -> -> - 工作流/自动化平台：如 Dify、n8n、Coze 等； -> - API 测试工具：如 Postman、Insomnia 等； -> - 自定义应用程序：脚本或后端代码中直接调用 API。 - -因此 [Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md) 这类应用开发平台只能通过**按量计费**（`get-api-key` 获取的 API Key）接入，且在 Dify 中通过安装「通义千问」或「OpenAI-API-compatible」插件配置。免费额度仅适用于华北2（北京）地域，且各模型额度独立、不可跨模型共享。 - -## 快速验证 - -配置完成后统一用一句问候验证连通性，例如：`claude "你好"`、`hermes chat -q "你好"`，或在 GUI 客户端对话框发送「你好」。模型正常返回响应即表示接入成功。若为 RAM 子账号，需确保在业务空间中已获得目标模型的调用权限。 +- **套餐适用范围严格受限**：Token Plan 个人版/团队版及 Coding Plan **仅允许用于 AI 编程工具和 OpenClaw 类 Agent**，禁止用于 Dify、n8n、Postman 等工作流平台或自动化脚本，违规可能导致订阅暂停 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md)。 +- **模型兼容性差异**：`qwen3.8-max-preview` 在 Anthropic 协议下需 `api_mode: anthropic_messages`，而在 OpenAI 协议下需 `extra_body: {enable_thinking: true}`；部分工具（如 Codex）对 `qwen3.8-max-preview` 要求 Responses API，其他模型需降级至旧版 Codex 使用 Chat API。 +- **免费额度约束**：按量计费新用户免费额度**仅限华北2（北京）地域**，跨地域调用（如新加坡 Workspace）将立即计费 [原文标题](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md)。 +- **错误排查优先级**：报错 `401 Incorrect API key provided` 应首先核对 Key 与 Base URL 是否来自同一方案及地域；`400 InternalError.Algo.InvalidParameter` 通常需启用 R1 messages 格式（Cline）或思考模式开关（Cherry Studio）。 ## 来源文档 -- [Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [OpenClaw](../../raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) +- [Hermes Agent](../../raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [Claude Code](../../raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) -- [OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [Cursor](../../raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) -- [Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) -- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) +- [OpenCode](../../raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - [QwenPaw](../../raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) -- [Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) +- [Codex](../../raw/model-user-guide/use-chat-client-or-development-tool/codex.md) - [Chatbox](../../raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Cline](../../raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - [Qoder](../../raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) @@ -80,6 +56,7 @@ - [使用Postman或cURL调用图像/视频生成API](../../raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) - [Dify](../../raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](../../raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) - +- [Qwen Code](../../raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) +- [Cherry Studio](../../raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/index.md b/skills/bailian-docs-llm-wiki/wiki/index.md index f339c226..cfebd71a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/index.md +++ b/skills/bailian-docs-llm-wiki/wiki/index.md @@ -8,7 +8,7 @@ - [application monitoring](guides/application-monitoring.md) — 1 篇源文档 - [application permission management](guides/application-permission-management.md) — 1 篇源文档 - [application publishing and sharing](guides/application-publishing-and-sharing.md) — 3 篇源文档 -- [application support](guides/application-support.md) — 2 篇源文档 +- [application support](guides/application-support.md) — 3 篇源文档 - [application use cases](guides/application-use-cases.md) — 5 篇源文档 - [bailian application calling](guides/bailian-application-calling.md) — 3 篇源文档 - [data connection overview](guides/data-connection-overview.md) — 1 篇源文档 @@ -20,7 +20,7 @@ - [memory library overview](guides/memory-library-overview.md) — 3 篇源文档 - [model compression](guides/model-compression.md) — 1 篇源文档 - [model context protocol](guides/model-context-protocol.md) — 5 篇源文档 -- [model data overview](guides/model-data-overview.md) — 2 篇源文档 +- [model data overview](guides/model-data-overview.md) — 3 篇源文档 - [model deployment 1](guides/model-deployment-1.md) — 4 篇源文档 - [model evaluation introduction](guides/model-evaluation-introduction.md) — 2 篇源文档 - [model experience](guides/model-experience.md) — 11 篇源文档 @@ -32,9 +32,9 @@ - [security and compliance](guides/security-and-compliance.md) — 12 篇源文档 - [skill](guides/skill.md) — 1 篇源文档 - [start using](guides/start-using.md) — 2 篇源文档 -- [support](guides/support.md) — 2 篇源文档 +- [support](guides/support.md) — 3 篇源文档 - [test 1](guides/test-1.md) — 5 篇源文档 -- [token plan guide](guides/token-plan-guide.md) — 10 篇源文档 +- [token plan guide](guides/token-plan-guide.md) — 14 篇源文档 - [use cases](guides/use-cases.md) — 24 篇源文档 - [use chat client or development tool](guides/use-chat-client-or-development-tool.md) — 17 篇源文档 @@ -42,10 +42,10 @@ - [3d generation](api/3d-generation.md) — 1 篇源文档 - [application call](api/application-call.md) — 5 篇源文档 -- [application component api reference](api/application-component-api-reference.md) — 57 篇源文档 +- [application component api reference](api/application-component-api-reference.md) — 58 篇源文档 - [file management api](api/file-management-api.md) — 1 篇源文档 - [frameworks](api/frameworks.md) — 3 篇源文档 -- [image generation](api/image-generation.md) — 26 篇源文档 +- [image generation](api/image-generation.md) — 27 篇源文档 - [knowledge](api/knowledge.md) — 1 篇源文档 - [long term memory new](api/long-term-memory-new.md) — 1 篇源文档 - [managed agents api](api/managed-agents-api.md) — 7 篇源文档 @@ -56,29 +56,29 @@ - [omni realtime api](api/omni-realtime-api.md) — 6 篇源文档 - [preparations](api/preparations.md) — 4 篇源文档 - [qwen api reference](api/qwen-api-reference.md) — 1 篇源文档 +- [realtime api user guide](api/realtime-api-user-guide.md) — 15 篇源文档 - [toolkits and frameworks](api/toolkits-and-frameworks.md) — 10 篇源文档 - [vector and sort](api/vector-and-sort.md) — 4 篇源文档 - [video generation api](api/video-generation-api.md) — 34 篇源文档 ## 横切概念 -- [API Key 鉴权](concepts/api-key.md) — 关联 6 个主题 -- [MCP 与工具扩展](concepts/mcp-and-tools.md) — 关联 5 个主题 -- [OpenAI 兼容接口](concepts/openai-compatible-interface.md) — 关联 6 个主题 -- [Token 与计费](concepts/token-and-billing.md) — 关联 6 个主题 -- [业务空间（Workspace）](concepts/workspace.md) — 关联 5 个主题 -- [函数调用（Function Calling）](concepts/function-calling.md) — 关联 4 个主题 -- [异步调用与任务轮询](concepts/async-invocation.md) — 关联 5 个主题 -- [检索增强生成（RAG）](concepts/rag.md) — 关联 6 个主题 -- [模型调优与部署](concepts/fine-tuning-and-deployment.md) — 关联 5 个主题 -- [流式输出](concepts/streaming-output.md) — 关联 3 个主题 +- [OpenAI 兼容接口](concepts/openai-compatible-interface.md) — 关联 5 个主题 +- [Prompt 工程](concepts/prompt-engineering.md) — 关联 5 个主题 +- [Token](concepts/token.md) — 关联 5 个主题 +- [函数调用](concepts/function-calling.md) — 关联 5 个主题 +- [多模态](concepts/multi-modal.md) — 关联 5 个主题 +- [安全与合规](concepts/security-and-compliance.md) — 关联 5 个主题 +- [检索增强生成](concepts/rag.md) — 关联 5 个主题 +- [模型部署](concepts/model-deployment.md) — 关联 5 个主题 +- [流式输出](concepts/streaming-output.md) — 关联 5 个主题 +- [长期记忆](concepts/long-term-memory.md) — 关联 5 个主题 ## 对比分析 -- [图像、视频与 3D 生成对比](comparisons/image-vs-video-vs-3d-generation.md) — 对比 3 个主题 -- [应用评估与应用监控对比](comparisons/app-evaluation-vs-monitoring.md) — 对比 2 个主题 -- [托管智能体：指南与 API 对比](comparisons/managed-agents-guide-vs-api.md) — 对比 2 个主题 -- [模型微调、压缩与部署对比](comparisons/fine-tuning-vs-compression-vs-deployment.md) — 对比 3 个主题 -- [模型评估与模型监控对比](comparisons/model-evaluation-vs-monitoring.md) — 对比 2 个主题 -- [知识库与记忆库对比](comparisons/knowledge-base-vs-memory-library.md) — 对比 2 个主题 +- [多模态生成能力对比：图像、视频与3D生成](comparisons/generation-apis-comparison.md) — 对比 3 个主题 +- [实时 API 方案对比：Omni Realtime API vs Realtime API User Guide](comparisons/realtime-api-comparison.md) — 对比 2 个主题 +- [应用构建框架对比：Managed Agents vs Application Component API vs Frameworks](comparisons/application-frameworks-comparison.md) — 对比 3 个主题 +- [模型部署方案对比：Model Deployment 1 vs Model Production](comparisons/model-deployment-options.md) — 对比 2 个主题 +- [知识库与记忆库功能对比](comparisons/knowledge-base-vs-memory-library.md) — 对比 2 个主题 From 9f376274f2c945f15d45c5d9a250cdcd74f38636 Mon Sep 17 00:00:00 2001 From: gujieye Date: Thu, 23 Jul 2026 20:20:54 +0800 Subject: [PATCH 04/13] Add GitHub Actions workflow for skill publishing --- .github/workflows/publish-skills.yml | 32 ++++++++++++++++++++++++++++ 1 file changed, 32 insertions(+) create mode 100644 .github/workflows/publish-skills.yml diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml new file mode 100644 index 00000000..65cade1d --- /dev/null +++ b/.github/workflows/publish-skills.yml @@ -0,0 +1,32 @@ +name: publish-skills + +on: + push: + branches: [main] + paths: ['skills/**'] # 只有技能目录变化才触发 + +concurrency: + group: publish-skills # 串行化：避免连续 push 并发触发对账 + cancel-in-progress: false + +jobs: + poke: + runs-on: ubuntu-latest + steps: + - name: Poke FC publisher (publish-skills) + run: | + # FC 侧为对账式发布：请求体不可信也不需要，action 通过 URL 路径声明。 + # curl -f 只能捕获 HTTP 层错误；FC 业务失败时仍返回 200 + success:false， + # 所以额外解析响应体，把业务失败也变成 workflow 失败（红叉可见）。 + RESP=$(curl -fsS --retry 3 --retry-delay 5 -X POST \ + "${{ vars.FC_TRIGGER_URL }}/publish-skills") + echo "$RESP" | jq . + if [ "$(echo "$RESP" | jq -r '.success')" != "true" ]; then + echo "::error::publish-skills failed: $(echo "$RESP" | jq -r '.error // "unknown"')" + exit 1 + fi + + echo "### publish-skills result" >> "$GITHUB_STEP_SUMMARY" + echo '```json' >> "$GITHUB_STEP_SUMMARY" + echo "$RESP" | jq '{upserted, deleted, unchanged: (.unchanged | length), durationMs}' >> "$GITHUB_STEP_SUMMARY" + echo '```' >> "$GITHUB_STEP_SUMMARY" From 289e4092e5ffa25ea150d73461020b3009467246 Mon Sep 17 00:00:00 2001 From: gujieye Date: Thu, 23 Jul 2026 20:23:51 +0800 Subject: [PATCH 05/13] Update publish-skills workflow branch to feat/bailian-docs-update Change the branch for triggering the publish-skills workflow. --- .github/workflows/publish-skills.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.github/workflows/publish-skills.yml b/.github/workflows/publish-skills.yml index 65cade1d..1446cff1 100644 --- a/.github/workflows/publish-skills.yml +++ b/.github/workflows/publish-skills.yml @@ -2,7 +2,7 @@ name: publish-skills on: push: - branches: [main] + branches: [feat/bailian-docs-update] paths: ['skills/**'] # 只有技能目录变化才触发 concurrency: From c5e1c6251886e957cbbc5fddcb9841deab01328f Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E6=95=85=E7=92=83?= Date: Thu, 23 Jul 2026 20:26:47 +0800 Subject: [PATCH 06/13] feat: trigger git action --- .../application-api-reference/more/how-to-use-search-filters.md | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md index 3f7490cf..24c1ae25 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md @@ -2565,4 +2565,4 @@ RAM用户（子账号）请先获取阿里云百炼的数据权限再调用[Retr ## 错误码 -如果调用失败并收到报错信息，请参见[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)进行解决。 +如果调用失败并收到报错信息，请参见[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)进行解决。 \ No newline at end of file From 90c404bb073ea8c796d414f351201aa4de27b161 Mon Sep 17 00:00:00 2001 From: bailian-bot Date: Thu, 23 Jul 2026 20:16:52 +0000 Subject: [PATCH 07/13] chore: update bailian-docs-llm-wiki (2026-07-24) --- skills/bailian-docs-llm-wiki/SKILL.md | 1 + skills/bailian-docs-llm-wiki/llms.txt | 867 +++--- .../models/families.jsonl | 6 +- .../models/groups/Kimi-K2.json | 36 + .../models/groups/MiniMax-M2.1.json | 12 + .../groups/MiniMax-speech-market-place.json | 20 + .../models/groups/aitryon-parsing-v1.json | 4 + .../models/groups/aitryon-plus.json | 4 + .../models/groups/aitryon-refiner.json | 10 + .../models/groups/aitryon.json | 4 + .../groups/animate-anyone-detect-gen2.json | 4 + .../models/groups/animate-anyone-gen2.json | 4 + .../groups/animate-anyone-template-gen2.json | 4 + .../models/groups/cosyvoice.json | 28 + .../models/groups/deepseek.json | 251 +- .../models/groups/embedding.json | 15 + .../models/groups/emo-detect-v1.json | 4 + .../models/groups/emo-v1.json | 5 + .../models/groups/emoji-detect-v1.json | 4 + .../models/groups/emoji-v1.json | 4 + .../models/groups/facechain-generation.json | 4 + .../models/groups/farui-plus.json | 5 + .../models/groups/fun-asr-flash.json | 4 + .../models/groups/fun-asr-realtime.json | 8 + .../models/groups/fun-asr.json | 8 + .../models/groups/fun-music.json | 8 + .../models/groups/glm-4.5.json | 60 + .../models/groups/glm-fast.json | 48 +- .../models/groups/gui-plus.json | 5 + .../models/groups/gummy-chat-v1.json | 4 + .../models/groups/gummy-realtime-v1.json | 4 + .../models/groups/happyhorse-i2v.json | 6 + .../models/groups/happyhorse-r2v.json | 6 + .../models/groups/happyhorse-t2v.json | 6 + .../models/groups/happyhorse-video-edit.json | 5 + .../groups/kimi-models-market-place.json | 25 + .../groups/kling-models-market-place.json | 27 + .../models/groups/liveportrait-detect.json | 4 + .../models/groups/liveportrait.json | 4 + .../groups/minimax-models-market-place.json | 24 + .../models/groups/paraformer-8k-v1.json | 4 + .../models/groups/paraformer-8k-v2.json | 4 + .../models/groups/paraformer-mtl-v1.json | 4 + .../groups/paraformer-realtime-8k-v1.json | 4 + .../groups/paraformer-realtime-8k-v2.json | 4 + .../models/groups/paraformer-realtime-v1.json | 4 + .../models/groups/paraformer-realtime-v2.json | 4 + .../models/groups/paraformer-v1.json | 4 + .../models/groups/paraformer-v2.json | 4 + .../groups/pixverse-c1-market-place.json | 44 + .../pixverse-capability-market-place.json | 14 + .../models/groups/pixverse-market-place.json | 44 + .../groups/pixverse-v6-market-place.json | 44 + .../models/groups/qvq-max.json | 5 + .../models/groups/qvq-plus.json | 5 + .../groups/qwen-audio-realtime-flash.json | 12 +- .../groups/qwen-audio-realtime-plus.json | 12 +- .../models/groups/qwen-audio-tts.json | 8 + .../models/groups/qwen-coder-plus.json | 5 + .../models/groups/qwen-coder-turbo.json | 5 + .../models/groups/qwen-deep-research.json | 5 + .../models/groups/qwen-doc-turbo.json | 8 + .../models/groups/qwen-embedding.json | 29 + .../models/groups/qwen-flash-character.json | 6 + .../models/groups/qwen-flash.json | 30 + .../models/groups/qwen-image-2.0-pro.json | 4 + .../models/groups/qwen-image-2.0.json | 4 + .../models/groups/qwen-image-edit-max.json | 4 + .../models/groups/qwen-image-edit.json | 8 + .../models/groups/qwen-image-max.json | 4 + .../models/groups/qwen-image-plus.json | 8 + .../models/groups/qwen-long.json | 14 + .../models/groups/qwen-math-plus.json | 20 + .../models/groups/qwen-math-turbo.json | 5 + .../models/groups/qwen-max.json | 8 + .../models/groups/qwen-mt-flash.json | 5 + .../models/groups/qwen-mt-image.json | 4 + .../models/groups/qwen-mt-lite.json | 5 + .../models/groups/qwen-mt-plus.json | 5 + .../models/groups/qwen-mt-turbo.json | 5 + .../groups/qwen-omni-turbo-realtime.json | 18 + .../models/groups/qwen-omni-turbo.json | 27 + .../models/groups/qwen-plus-character.json | 1 + .../models/groups/qwen-plus.json | 94 + .../models/groups/qwen-rerank.json | 13 + .../models/groups/qwen-tts-realtime.json | 10 + .../models/groups/qwen-tts.json | 10 + .../models/groups/qwen-turbo.json | 14 + .../models/groups/qwen-vl-embedding.json | 10 + .../models/groups/qwen-vl-max.json | 9 + .../models/groups/qwen-vl-ocr.json | 19 + .../models/groups/qwen-vl-plus.json | 9 + .../models/groups/qwen-voice-design.json | 4 + .../models/groups/qwen-voice-enrollment.json | 4 + .../models/groups/qwen2.5.json | 9 + .../groups/qwen3-asr-flash-filetrans.json | 4 + .../groups/qwen3-asr-flash-realtime.json | 4 + .../models/groups/qwen3-asr-flash.json | 4 + .../groups/qwen3-coder-30b-a3b-instruct.json | 11 + .../qwen3-coder-480b-a35b-instruct.json | 11 + .../models/groups/qwen3-coder-flash.json | 23 + .../models/groups/qwen3-coder-plus.json | 23 + .../qwen3-livetranslate-flash-realtime.json | 7 + .../groups/qwen3-livetranslate-flash.json | 7 + .../models/groups/qwen3-max.json | 42 + .../groups/qwen3-omni-30b-a3b-captioner.json | 5 + .../groups/qwen3-omni-flash-realtime.json | 9 + .../models/groups/qwen3-omni-flash.json | 14 + .../groups/qwen3-tts-flash-realtime.json | 17 +- .../models/groups/qwen3-tts-flash.json | 4 + .../qwen3-tts-instruct-flash-realtime.json | 4 + .../groups/qwen3-tts-instruct-flash.json | 4 + .../models/groups/qwen3-tts-vc-realtime.json | 4 + .../models/groups/qwen3-tts-vc.json | 4 + .../models/groups/qwen3-tts-vd-realtime.json | 4 + .../models/groups/qwen3-tts-vd.json | 4 + .../models/groups/qwen3-vl-flash.json | 30 + .../models/groups/qwen3-vl-plus.json | 30 + .../models/groups/qwen3.5-flash.json | 27 + .../qwen3.5-livetranslate-flash-realtime.json | 7 + .../models/groups/qwen3.5-ocr.json | 5 + .../groups/qwen3.5-omni-flash-realtime.json | 7 + .../models/groups/qwen3.5-omni-flash.json | 111 - .../groups/qwen3.5-omni-plus-realtime.json | 7 + .../models/groups/qwen3.5-omni-plus.json | 13 + .../models/groups/qwen3.5-plus.json | 27 + .../models/groups/qwen3.5.json | 28 + .../models/groups/qwen3.6-flash.json | 19 + .../models/groups/qwen3.6-max.json | 11 + .../models/groups/qwen3.6-plus.json | 19 + .../models/groups/qwen3.6.json | 10 + .../models/groups/qwen3.7-max.json | 17 + .../models/groups/qwen3.7-plus.json | 21 + .../models/groups/qwen3.json | 119 + .../models/groups/qwq-plus.json | 7 + .../models/groups/sambert.json | 172 + .../models/groups/siliconflow-models.json | 20 + .../groups/stepfun-models-market-place.json | 6 + .../groups/tongyi-intent-detect-v3.json | 5 + .../groups/tongyi-xiaomi-analysis-flash.json | 5 + .../groups/tongyi-xiaomi-analysis-pro.json | 5 + .../groups/tripo-models-market-place.json | 311 ++ .../groups/vanchin-models-market-place.json | 69 +- .../models/groups/video-style-transform.json | 5 + .../models/groups/videoretalk.json | 4 + .../vidu-image-models-market-place.json | 42 +- .../groups/vidu-models-market-place.json | 114 + .../models/groups/wan-image-edit.json | 20 + .../models/groups/wan-image-to-video.json | 73 + .../models/groups/wan-reference-to-video.json | 17 + .../models/groups/wan-text-to-image.json | 32 + .../models/groups/wan-text-to-video.json | 32 + .../models/groups/wan-video-edit.json | 5 + .../groups/wanx-background-generation-v2.json | 4 + .../groups/wanx-sketch-to-image-lite.json | 4 + .../models/groups/wanx-style-repaint-v1.json | 4 + .../models/groups/wanx2.1-vace-plus.json | 4 + .../models/groups/wordart-semantic.json | 4 + .../models/groups/wordart-texture.json | 4 + .../groups/xiaomi-models-market-place.json | 6 + .../models/groups/z-image-turbo.json | 5 + .../groups/zhipu-models-market-place.json | 18 + .../bailian-docs-llm-wiki/models/index.json | 42 +- skills/bailian-docs-llm-wiki/models/index.md | 31 +- .../bailian-docs-llm-wiki/models/models.jsonl | 3 +- ...i-bailian-2023-12-29-batchupdatefiletag.md | 246 ++ ...api-bailian-2023-12-29-creatememorynode.md | 178 -- .../api-bailian-2023-12-29-getmemory.md | 176 -- .../more/how-to-use-search-filters.md | 2 +- .../application-auto-evaluation.md | 242 -- ...dcast-2025-02-28-podcasttaskresultquery.md | 777 ----- ...-aipodcast-2025-02-28-podcasttasksubmit.md | 269 -- .../api-aipodcast-2025-02-28-endpoint.md | 19 - .../api-aipodcast-2025-02-28-overview.md | 35 - .../api-aipodcast-2025-02-28-ram.md | 109 - .../brief-introduction-of-ai-podcast.md | 63 - .../api-aimiaobi-2023-08-01-createdataset.md | 0 .../api-aimiaobi-2023-08-01-deletedataset.md | 0 .../api-aimiaobi-2023-08-01-getdataset.md | 0 .../api-aimiaobi-2023-08-01-listdatasets.md | 0 .../api-aimiaobi-2023-08-01-updatedataset.md | 0 .../voice-clone-design-http-api.md | 8 +- .../voice-clone-python-sdk.md | 8 +- ...ge-generation-and-editing-api-reference.md | 2661 ---------------- .../wanx-sketch-to-image-api-reference.md | 968 ++++++ .../get-temporary-file-url.md | 1185 +++++++ .../more-models/qwen-vl-ocr-api-reference.md | 2762 ----------------- .../omni-realtime-interaction-process.md | 76 + .../model-experience/tts-model.md | 8 +- .../tpm-reservation.md | 1 + .../security-and-compliance/privacy-notice.md | 13 - .../model-user-guide/test-1/model-pricing.md | 8 - .../wiki/api/3d-generation.md | 69 +- .../wiki/api/application-call.md | 84 +- .../application-component-api-reference.md | 97 +- .../wiki/api/file-management-api.md | 49 +- .../wiki/api/frameworks.md | 60 +- .../wiki/api/image-generation.md | 125 +- .../wiki/api/knowledge.md | 44 +- .../wiki/api/long-term-memory-new.md | 111 +- .../wiki/api/managed-agents-api.md | 80 +- .../wiki/api/model-production.md | 46 +- .../wiki/api/more-about-models.md | 68 +- .../wiki/api/more-models.md | 100 +- skills/bailian-docs-llm-wiki/wiki/api/more.md | 52 +- .../wiki/api/omni-realtime-api.md | 117 +- .../wiki/api/preparations.md | 54 +- .../wiki/api/qwen-api-reference.md | 45 +- .../wiki/api/realtime-api-user-guide.md | 113 +- .../wiki/api/toolkits-and-frameworks.md | 132 +- .../wiki/api/vector-and-sort.md | 71 +- .../wiki/api/video-generation-api.md | 118 +- .../application-frameworks-comparison.md | 69 - .../comparisons/application-frameworks.md | 69 - .../application-orchestration-comparison.md | 71 + .../comparisons/generation-api-comparison.md | 66 + .../comparisons/generation-apis-comparison.md | 65 - .../comparisons/image-vs-video-generation.md | 65 - .../knowledge-base-vs-memory-library.md | 63 - .../knowledge-solutions-comparison.md | 73 + .../wiki/comparisons/memory-solutions.md | 62 - ...del-deployment-and-inference-comparison.md | 69 + .../comparisons/model-deployment-options.md | 57 - .../model-evaluation-monitoring.md | 66 - .../comparisons/realtime-api-comparison.md | 109 +- .../tooling-frameworks-comparison.md | 93 + .../wiki/concepts/asynchronous-processing.md | 46 + .../wiki/concepts/function-calling.md | 71 +- .../wiki/concepts/long-term-memory.md | 59 +- .../wiki/concepts/mcp.md | 47 + .../wiki/concepts/model-context-protocol.md | 50 - .../wiki/concepts/model-deployment.md | 68 - .../wiki/concepts/multi-modal.md | 71 - .../wiki/concepts/openai-compatible-api.md | 72 + .../concepts/openai-compatible-interface.md | 68 - .../wiki/concepts/plugin.md | 57 - .../wiki/concepts/prompt-engineering.md | 60 +- .../wiki/concepts/rag.md | 50 +- .../wiki/concepts/security-and-compliance.md | 51 - .../wiki/concepts/streaming-output.md | 69 +- .../wiki/concepts/token.md | 56 +- .../wiki/concepts/tool-integration.md | 65 + .../wiki/guides/application-evaluation.md | 55 +- .../wiki/guides/application-monitoring.md | 70 +- .../application-permission-management.md | 55 +- .../application-publishing-and-sharing.md | 56 +- .../wiki/guides/application-support.md | 41 +- .../wiki/guides/application-use-cases.md | 73 +- .../guides/bailian-application-calling.md | 124 +- .../wiki/guides/data-connection-overview.md | 60 +- .../wiki/guides/fine-tuning.md | 85 +- .../wiki/guides/get-started-with-models.md | 108 +- .../wiki/guides/knowledge-base.md | 50 +- .../wiki/guides/llm-application.md | 69 +- .../wiki/guides/managed-agents.md | 58 +- .../wiki/guides/memory-library-overview.md | 75 +- .../wiki/guides/model-compression.md | 44 +- .../wiki/guides/model-context-protocol.md | 76 +- .../wiki/guides/model-data-overview.md | 68 +- .../wiki/guides/model-deployment-1.md | 58 +- .../guides/model-evaluation-introduction.md | 78 +- .../wiki/guides/model-experience.md | 91 +- .../wiki/guides/model-high-speed-inference.md | 56 +- .../wiki/guides/model-monitoring.md | 121 +- .../wiki/guides/plug-in.md | 46 +- .../wiki/guides/prompt.md | 62 +- .../wiki/guides/release-notes.md | 37 +- .../wiki/guides/security-and-compliance.md | 81 +- .../wiki/guides/skill.md | 41 +- .../wiki/guides/start-using.md | 52 +- .../wiki/guides/support.md | 53 +- .../wiki/guides/test-1.md | 34 +- .../wiki/guides/token-plan-guide.md | 84 +- .../wiki/guides/use-cases.md | 37 +- .../use-chat-client-or-development-tool.md | 55 +- skills/bailian-docs-llm-wiki/wiki/index.md | 33 +- 276 files changed, 8473 insertions(+), 11127 deletions(-) delete mode 100644 skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json create mode 100644 skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json create mode 100644 skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-evaluation/application-auto-evaluation.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-other => api-aimiaobi-2023-08-01-dir-wonderful-search-data-source}/api-aimiaobi-2023-08-01-createdataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-other => api-aimiaobi-2023-08-01-dir-wonderful-search-data-source}/api-aimiaobi-2023-08-01-deletedataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-other => api-aimiaobi-2023-08-01-dir-wonderful-search-data-source}/api-aimiaobi-2023-08-01-getdataset.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-other => api-aimiaobi-2023-08-01-dir-wonderful-search-data-source}/api-aimiaobi-2023-08-01-listdatasets.md (100%) rename skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/{api-aimiaobi-2023-08-01-dir-other => api-aimiaobi-2023-08-01-dir-wonderful-search-data-source}/api-aimiaobi-2023-08-01-updatedataset.md (100%) delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md create mode 100644 skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md delete mode 100644 skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/privacy-notice.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration-comparison.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/generation-api-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-solutions-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-and-inference-comparison.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/comparisons/tooling-frameworks-comparison.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/asynchronous-processing.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/mcp.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-context-protocol.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/model-deployment.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/multi-modal.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-api.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/openai-compatible-interface.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/plugin.md delete mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/security-and-compliance.md create mode 100644 skills/bailian-docs-llm-wiki/wiki/concepts/tool-integration.md diff --git a/skills/bailian-docs-llm-wiki/SKILL.md b/skills/bailian-docs-llm-wiki/SKILL.md index 836eff6a..4c8b7d7e 100644 --- a/skills/bailian-docs-llm-wiki/SKILL.md +++ b/skills/bailian-docs-llm-wiki/SKILL.md @@ -103,6 +103,7 @@ description: >- | `Multimodal-Omni` | 全模态 | | `ME` | 多模态嵌入 | | `TR` | 翻译 | +| `3D-generation` | 3D 生成 | | `Realtime-Chatting` | Realtime-Chatting | 一个模型常常带多个 capability，`index.md` 中按 `capabilities[0]`（主能力）归类， diff --git a/skills/bailian-docs-llm-wiki/llms.txt b/skills/bailian-docs-llm-wiki/llms.txt index 762df236..3a8973b3 100644 --- a/skills/bailian-docs-llm-wiki/llms.txt +++ b/skills/bailian-docs-llm-wiki/llms.txt @@ -11,61 +11,61 @@ - [账单查询与成本管理](raw/model-user-guide/test-1/bill-query-and-cost-management.md) - [模型调用价格](raw/model-user-guide/test-1/model-pricing.md) - **开始使用** - - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [什么是阿里云百炼](raw/model-user-guide/get-started-with-models/what-is-model-studio.md) + - [首次调用千问API](raw/model-user-guide/get-started-with-models/first-api-call-to-qwen.md) - [选择模型](raw/model-user-guide/get-started-with-models/models.md) + - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) - [Base URL总览](raw/model-user-guide/get-started-with-models/base-url.md) - [限流](raw/model-user-guide/get-started-with-models/rate-limit.md) - - [选择地域、服务部署范围和接入域名](raw/model-user-guide/get-started-with-models/regions.md) - **Token Plan** - - **个人版** - - [概述](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) - - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md) - - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md) - **团队版** + - [概述](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-quickstart.md) - [团队管理](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-management.md) - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-faq.md) - - [概述](raw/model-user-guide/token-plan-guide/token-plan-team-edition/token-plan-team-overview.md) + - **个人版** + - [概述](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-overview.md) + - [快速开始](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-quick-start.md) + - [常见问题](raw/model-user-guide/token-plan-guide/token-plan-personal/token-plan-personal-faq.md) - **Coding Plan** - - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) - [常见问题](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan-faq.md) + - [Coding Plan概述](raw/model-user-guide/token-plan-guide/coding-plan-guide/coding-plan.md) - **最佳实践** - - [接入 Harness 工具](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md) - [接入多模态生成模型](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-multimodal-gen.md) + - [接入 Harness 工具](raw/model-user-guide/token-plan-guide/token-plan-best-practice/token-plan-harness-tool.md) - [联网搜索](raw/model-user-guide/token-plan-guide/token-plan-best-practice/web-search-mcp.md) - [添加视觉理解能力](raw/model-user-guide/token-plan-guide/token-plan-best-practice/add-vision-skill.md) - [Token Plan 概述](raw/model-user-guide/token-plan-guide/token-plan-overview.md) +- **模型体验** + - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) + - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) + - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) + - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) + - [语音合成](raw/model-user-guide/model-experience/tts-model.md) + - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) + - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) + - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) + - [语音识别](raw/model-user-guide/model-experience/asr-model.md) + - [全模态](raw/model-user-guide/model-experience/omni.md) + - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) - **接入客户端/开发工具** - [OpenClaw](raw/model-user-guide/use-chat-client-or-development-tool/openclaw.md) - [Hermes Agent](raw/model-user-guide/use-chat-client-or-development-tool/hermes-agent.md) - [Claude Code](raw/model-user-guide/use-chat-client-or-development-tool/claude-code.md) - [Cursor](raw/model-user-guide/use-chat-client-or-development-tool/cursor.md) - [OpenCode](raw/model-user-guide/use-chat-client-or-development-tool/opencode.md) - - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - [Codex](raw/model-user-guide/use-chat-client-or-development-tool/codex.md) + - [QwenPaw](raw/model-user-guide/use-chat-client-or-development-tool/qwenpaw.md) - [Chatbox](raw/model-user-guide/use-chat-client-or-development-tool/chatbox.md) - [Cline](raw/model-user-guide/use-chat-client-or-development-tool/cline.md) - - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) + - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Qoder CN（原 Lingma）](raw/model-user-guide/use-chat-client-or-development-tool/lingma-agent.md) - [Kilo CLI](raw/model-user-guide/use-chat-client-or-development-tool/kilo-cli.md) - [使用Postman或cURL调用图像/视频生成API](raw/model-user-guide/use-chat-client-or-development-tool/first-call-to-image-and-video-api.md) + - [Qoder](raw/model-user-guide/use-chat-client-or-development-tool/qoder-agent.md) - [Dify](raw/model-user-guide/use-chat-client-or-development-tool/dify.md) - [更多工具](raw/model-user-guide/use-chat-client-or-development-tool/more-tools.md) - - [Qwen Code](raw/model-user-guide/use-chat-client-or-development-tool/qwen-code.md) - [Cherry Studio](raw/model-user-guide/use-chat-client-or-development-tool/cherry-studio.md) -- **模型体验** - - [文本生成](raw/model-user-guide/model-experience/text-generation-model.md) - - [视觉理解](raw/model-user-guide/model-experience/vision-model.md) - - [图片生成与编辑](raw/model-user-guide/model-experience/image-model.md) - - [Tripo 3D模型生成](raw/model-user-guide/model-experience/tripo-3d-generation-guide.md) - - [视频生成与编辑](raw/model-user-guide/model-experience/video-generate-edit-model.md) - - [音乐生成](raw/model-user-guide/model-experience/fun-music.md) - - [语音合成](raw/model-user-guide/model-experience/tts-model.md) - - [语音识别](raw/model-user-guide/model-experience/asr-model.md) - - [语音转语音](raw/model-user-guide/model-experience/s2s-model.md) - - [全模态](raw/model-user-guide/model-experience/omni.md) - - [向量与重排序](raw/model-user-guide/model-experience/embedding-rerank-model.md) - **模型推理** - [TPM 预留](raw/model-user-guide/model-high-speed-inference/tpm-reservation.md) - [快速模式](raw/model-user-guide/model-high-speed-inference/fast-mode.md) @@ -77,12 +77,12 @@ - [0 代码强化大模型安全合规能力](raw/model-user-guide/fine-tuning/fine-tune-text-generation-model/enhance-the-security-compliance-of-large-models.md) - **语音合成模型调优** - [CosyVoice模型调优](raw/model-user-guide/fine-tuning/fine-tune-speech-synthesis-model/fine-tune-speech-synthesis-model-by-api.md) - - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - [微调视频生成模型](raw/model-user-guide/fine-tuning/wan-video-generation-finetune-guide.md) + - [微调图像生成模型](raw/model-user-guide/fine-tuning/wan-image-generation-finetune-guide.md) - **模型部署** + - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [预置吞吐长输入与缓存](raw/model-user-guide/model-deployment-1/ptu-long-input-and-cache.md) - [模型导入](raw/model-user-guide/model-deployment-1/model-import.md) - - [模型部署](raw/model-user-guide/model-deployment-1/model-deployment-introduction.md) - [使用 API或命令行进行模型部署](raw/model-user-guide/model-deployment-1/model-deployment-quick-start.md) - **模型评测** - [模型评测](raw/model-user-guide/model-evaluation-introduction/model-evaluation-overview.md) @@ -90,27 +90,22 @@ - **模型压缩** - [模型压缩](raw/model-user-guide/model-compression/model-compression-introduction.md) - **用量统计与性能监控** - - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) - [模型用量](raw/model-user-guide/model-monitoring/model-usage-statistics.md) + - [模型监控](raw/model-user-guide/model-monitoring/model-telemetry.md) - **安全合规** - **传输安全** - - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - [获取RSA的公钥](raw/model-user-guide/security-and-compliance/transmission-security/model-interface-aes-encryption.md) + - [以加密的方式接入模型推理功能](raw/model-user-guide/security-and-compliance/transmission-security/encrypted-access-to-model-inference.md) - [通过终端节点私网访问阿里云百炼模型或应用 API](raw/model-user-guide/security-and-compliance/transmission-security/access-model-studio-through-privatelink.md) - **安全存储** - [配置终端节点并发起连接](raw/model-user-guide/security-and-compliance/secure-storage/configure-an-endpoint-and-initiate-a-connection.md) - [配置可用区IP](raw/model-user-guide/security-and-compliance/secure-storage/configure-zone-ip.md) - [配置私有网络中的资源](raw/model-user-guide/security-and-compliance/secure-storage/configure-resources-in-private-network.md) - [配置MSE云原生网关](raw/model-user-guide/security-and-compliance/secure-storage/configure-mse.md) - - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [权限管理](raw/model-user-guide/security-and-compliance/permission-management-overview.md) - - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) + - [输⼊输出AI安全护栏](raw/model-user-guide/security-and-compliance/content-security.md) - [模型备案信息公示](raw/model-user-guide/security-and-compliance/model-filing-information-publicity.md) - - [合规资质与隐私说明](raw/model-user-guide/security-and-compliance/privacy-notice.md) -- **模型数据** - - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) - - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) - - [日志回流](raw/model-user-guide/model-data-overview/model-log-backflow.md) + - [千问大模型应用上架及合规备案](raw/model-user-guide/security-and-compliance/compliance-and-launch-filing-guide-for-ai-apps-powered-by-the-tongyi-model.md) - **实践教程** - **三方模型调用教程** - [DeepSeek-阿里云](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/deepseek-api.md) @@ -122,24 +117,28 @@ - [GLM-智谱](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/glm-zhipu.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api.md) - [MiniMax](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/minimax-api-by-minimax.md) + - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [MiMo-小米](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/mimo.md) - [Stepfun-阶跃星辰](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/stepfun.md) - - [Vidu视频生成Prompt指南](raw/model-user-guide/use-cases/third-party-model-integration-tutorial/vidu-video-generation-prompt-guide.md) - [HappyHorse 打造一站式影视创作平台](raw/model-user-guide/use-cases/infinite-canvas.md) - [高效搭建 AI 智能体与工作流应用](raw/model-user-guide/use-cases/build-ai-applications-based-on-alibaba-cloud-model-studio.md) - [深度研究：生成你的独家洞察报告](raw/model-user-guide/use-cases/deep-research.md) - [AI 解题 + 批改：推动课程教学智变](raw/model-user-guide/use-cases/ai-homework-helper.md) - [文生文Prompt指南](raw/model-user-guide/use-cases/prompt-engineering-guide.md) - - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - [文生视频/图生视频Prompt指南](raw/model-user-guide/use-cases/text-to-video-prompt.md) - [基于LlamaIndex构建RAG应用](raw/model-user-guide/use-cases/build-rag-applications-based-on-llamaindex.md) + - [文生图Prompt指南](raw/model-user-guide/use-cases/text-to-image-prompt.md) - [自定义模型调优、部署与评测](raw/model-user-guide/use-cases/model-training-best-practices.md) - [借助大模型将文档转换为视频](raw/model-user-guide/use-cases/use-llm-to-convert-document-to-video.md) - [限流应对最佳实践 ](raw/model-user-guide/use-cases/rate-limiting-best-practices.md) - [显式缓存最佳实践](raw/model-user-guide/use-cases/explicit-cache-guide.md) +- **模型数据** + - [数据清洗或增强](raw/model-user-guide/model-data-overview/data-processing.md) + - [训练集与评测集](raw/model-user-guide/model-data-overview/training-set-and-evaluation-set.md) + - [日志回流](raw/model-user-guide/model-data-overview/model-log-backflow.md) - **产品动态** - - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) - [模型平台功能更新](raw/model-user-guide/release-notes/model-release-notes.md) + - [模型上下架与更新](raw/model-user-guide/release-notes/newly-released-models.md) - **服务支持** - [常见问题](raw/model-user-guide/support/faq-about-alibaba-cloud-model-studio.md) - [相关协议](raw/model-user-guide/support/related-agreements.md) @@ -157,46 +156,42 @@ - **Managed Agents** - [概述](raw/application-user-guide/managed-agents/managed-agents-introduction.md) - [快速开始](raw/application-user-guide/managed-agents/managed-agents-quick-start.md) - - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) - - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) - [构建 Agent](raw/application-user-guide/managed-agents/managed-agents-agent.md) + - [配置 Agent 环境](raw/application-user-guide/managed-agents/managed-agents-environment.md) + - [委派任务给 Agent](raw/application-user-guide/managed-agents/managed-agents-session.md) - [Agent 上下文管理](raw/application-user-guide/managed-agents/managed-agents-context.md) - **开始使用** - [0代码构建私有知识问答应用](raw/application-user-guide/start-using/build-knowledge-base-qa-assistant-without-coding.md) - [应用功能动态](raw/application-user-guide/start-using/application-release-notes.md) - **Prompt** - - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) - [Prompt模板概述](raw/application-user-guide/prompt/prompt-template.md) - - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) - [使用Prompt样例库优化模型输出](raw/application-user-guide/prompt/prompt-sample-optimization.md) + - [自定义Prompt模板](raw/application-user-guide/prompt/prompt-custom-template.md) + - [Prompt自动优化](raw/application-user-guide/prompt/optimize-prompt.md) + - [基于大模型输入输出样例的Prompt自动优化](raw/application-user-guide/prompt/prompt-feedback-optimization.md) - **记忆库** - - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) - - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) - [记忆库](raw/application-user-guide/memory-library-overview/memory-library.md) + - [为 OpenClaw 配置长期记忆插件](raw/application-user-guide/memory-library-overview/modelstudio-memory-for-openclaw.md) + - [长期记忆 API](raw/application-user-guide/memory-library-overview/long-term-memory-2-0.md) +- **数据连接** + - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) +- **Skill** + - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) - **知识库（RAG）** + - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) - [RAG效果优化](raw/application-user-guide/knowledge-base/rag-optimization.md) - - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库日志与监控](raw/application-user-guide/knowledge-base/rag-knowledge-base-log-monitoring.md) - - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) + - [知识库API指南](raw/application-user-guide/knowledge-base/rag-knowledge-base-api-guide.md) - [知识库计费说明](raw/application-user-guide/knowledge-base/billing-for-knowledge-base.md) - - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) - [知识问答](raw/application-user-guide/knowledge-base/rag-knowledge-qa.md) - - [知识库](raw/application-user-guide/knowledge-base/rag-knowledge-base.md) -- **数据连接** - - [数据连接](raw/application-user-guide/data-connection-overview/data-connection.md) + - [知识检索](raw/application-user-guide/knowledge-base/rag-knowledge-retrieval.md) + - [知识库配额与限制](raw/application-user-guide/knowledge-base/rag-knowledge-base-specifications.md) - **MCP** - [模型上下文协议（MCP）](raw/application-user-guide/model-context-protocol/mcp-introduction.md) - - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) + - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) - [外部调用](raw/application-user-guide/model-context-protocol/mcp-external-calls.md) - [MCP 常见问题](raw/application-user-guide/model-context-protocol/mcp-faq.md) - - [自定义 MCP 服务](raw/application-user-guide/model-context-protocol/custom-mcp.md) -- **Skill** - - [Skill](raw/application-user-guide/skill/introduction-to-skill.md) -- **插件** - - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) - - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) - - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) + - [官方 MCP 服务](raw/application-user-guide/model-context-protocol/official-and-third-party-mcp.md) - **应用发布与分享** - [分享智能体应用](raw/application-user-guide/application-publishing-and-sharing/share-an-application.md) - [使用智能体或工作流作为组件](raw/application-user-guide/application-publishing-and-sharing/use-agent-or-workflow-as-component.md) @@ -205,13 +200,18 @@ - [调用智能体应用](raw/application-user-guide/bailian-application-calling/call-single-agent-application.md) - [调用工作流应用](raw/application-user-guide/bailian-application-calling/invoke-workflow-application.md) - [应用的自定义参数传递](raw/application-user-guide/bailian-application-calling/pass-through-of-application-parameters.md) +- **插件** + - [插件概述](raw/application-user-guide/plug-in/plug-in-overview.md) + - [官方和第三方插件](raw/application-user-guide/plug-in/plugins.md) + - [自定义插件](raw/application-user-guide/plug-in/custom-plug-ins.md) +- **应用观测** + - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **应用评测** - **新版应用评测** - [新版评测集](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/new-version-of-evaluation-set.md) - [评测任务](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/evaluation-task.md) - [标签管理](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/label-management.md) - [评估器](raw/application-user-guide/application-evaluation/new-version-of-application-evaluation/grader.md) - - [自动评测](raw/application-user-guide/application-evaluation/application-auto-evaluation.md) - [手动评测](raw/application-user-guide/application-evaluation/evaluate-manual-application.md) - [评测集](raw/application-user-guide/application-evaluation/application-evaluation-dataset.md) - **应用广场** @@ -224,93 +224,84 @@ - [服务接入点](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-endpoint.md) - [授权信息](raw/application-user-guide/application-gallery/edu-tutor/api-reference-edututor/api-edututor-2025-07-07-ram.md) - [通义拍照解题辅导产品介绍](raw/application-user-guide/application-gallery/edu-tutor/brief-introduction-of-edu-tutor.md) - - **官方应用-通义音频播客生成** - - **API参考** - - **API目录** - - [PodcastTaskSubmit - 播客任务提交](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md) - - [PodcastTaskResultQuery - 播客任务结果查询](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md) - - [通义音频播客生成产品介绍](raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md) - **官方应用-多模态交互开发套件** - **使用指南** - - [应用创建](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-creation.md) - [应用配置](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-configuration.md) + - [应用创建](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-creation.md) - [应用体验与发布](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-app-experience-and-publishing.md) - [百炼应用推荐模板](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/agent-template.md) - [指令列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/instruction-list.md) - - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - [音色列表](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-timbre-list.md) + - [多语言对话](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multi-language-dialogue.md) - [对话日志接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-chatlog.md) - [三方Agent接入](raw/application-user-guide/application-gallery/multimodal-products/multimodal-guidelines/multimodal-integration-a2a.md) - **SDK安装** - [服务端Java SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-java.md) - - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) - [服务端Python SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-python.md) - [服务端 Go SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/server-go-sdk.md) - - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) - [移动端iOS SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios.md) + - [移动端Android SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android.md) + - [移动端Android Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-android-lite.md) + - [移动端iOS Lite SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-ios-lite.md) - [RTOS C SDK（License模式）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/mmi-rtos-sdk.md) - [Linux C++ SDK](raw/application-user-guide/application-gallery/multimodal-products/multimodal-sdk/multimodal-sdk-linux.md) - **API参考** - [实时多模态交互协议（WebSocket）](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-interaction-protocol.md) - [HTTP协议](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-http-protocol.md) - [调用官方Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/official-agent.md) + - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [调用三方语音模型](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/third-party-voice-integration.md) - [管理热词](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/management-hot-words.md) - - [调用插件](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/call-plugins.md) - [多模态交互套件-错误码](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/multimodal-error-code.md) - - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) - [多模态对话结果 extra_info 说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/extra-info-description.md) + - [长期记忆开放接口](raw/application-user-guide/application-gallery/multimodal-products/multimodal-api-references/long-term-memory-api.md) - **最佳实践** + - **接入图像生成Agent** + - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) + - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) - **接入百炼及三方Agent** - [百炼及三方Agent直连调用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/agent-direct-call.md) - [接入百炼智能体应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-app.md) - [接入百炼工作流应用](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/bailian-and-tripartite-agent/multimodal-call-workflow.md) + - **接入拍照问答Agent** + - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) + - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - **接入听悟智能纪要Agent** - [录音纪要Agent使用教程](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/recording-summary-agent-tutorial.md) - - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) - [快速集成智能纪要Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/fast-integrate-offline-tingwu-meeting-agent.md) - - **接入拍照问答Agent** - - [通过WebSocket协议接入拍照问答Agent和语音合成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-via-websocket-protocol.md) - - [通过HTTP协议接入拍照问答Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/vqa-agent/vqa-agent-through-the-http-protocol.md) - - **接入图像生成Agent** - - [语音请求直通图像生成Agent(websocket)](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/audio-to-generateimgagent.md) - - [通过HTTP协议接入图像生成Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/generateimgagent/image-agent.md) + - [实时转写能力集成](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/fast-integrate-tingwu-meeting-agent/realtime-tingwu-meeting-agent-integration.md) - [接入多模态备忘录Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/multimodal-memo-agent.md) - [接入音乐电台Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/music-agent.md) - [接入视频通话Agent](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/live-api-integration.md) - [动作情绪控制实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/action-emotion-control-practice.md) - - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) - [自定义指令实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-directive.md) - - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) + - [自定义对话角色实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/custom-role.md) - [基于RTOS SDK (License模式) 实现聊天能力](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/chat-capability-based-on-rtos-sdk.md) - - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) - - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) + - [音频采集和播放说明](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/audio-capture-and-playback-instructions.md) + - [声音复刻及声音设计实践](raw/application-user-guide/application-gallery/multimodal-products/multimodal-best-practices/voice-cloning-and-voice-design.md) - [多模态交互开发套件常见问题](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-faq.md) + - [产品计费](raw/application-user-guide/application-gallery/multimodal-products/product-billing.md) + - [产品概述](raw/application-user-guide/application-gallery/multimodal-products/multimodal-products-overview.md) - **官方应用-全妙轻应用系列** - **计费说明（全妙轻应用）** - [电商零售推广文案写作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-retail-promotion-copywriting-billing.md) - [电商文案智能可控生成计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/e-commerce-copy-intelligent-controllable-generation-billing.md) - [影视互娱剧本创作计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-mutual-entertainment-script-creation-billing.md) - - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) + - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - [泛企业VOC挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-voc-mining-billing.md) + - [车机网络热点信息互动问答计费文档](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/car-machine-network-hot-information-interactive-question-and-answer-billing.md) - [泛企业线索挖掘计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/pan-enterprise-lead-mining-billing.md) - [网络内容安全审核计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/network-content-security-audit-billing.md) - [作文批改计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/composition-correction-billing.md) - [视频智能拆条计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/video-smart-strip-billing.md) - - [影视传媒视频理解计费](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-billing-document/film-and-television-media-video-understanding-billing.md) - **使用指南** - [电商文案智能可控生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/intelligent-and-controllable-generation-of-e-commerce-copywriting.md) - [传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-retail-article-style-and-format-learning.md) + - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - [影视互娱剧本创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/film-and-television-script-creation.md) - [车机网络热点信息互动问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/car-machine-content-platform-news-hot-list-interaction.md) - - [影视传媒视频理解](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/media-video-understanding.md) - - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - [泛企业线索挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-clue-mining.md) + - [泛企业VOC挖掘](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/pan-enterprise-voc-mining.md) - [网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/network-content-security-audit.md) - [作文批改助手](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-guidelines-for-use/composition-correction-assistant.md) - **开发文档** @@ -324,73 +315,73 @@ - **API目录** - **传媒/零售文章风格与格式学习** - [RunStyleWriting - 传媒/零售文章风格与格式学习](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-media-retail-article-style-and-format-learning/api-quanmiaolightapp-2024-08-01-runstylewriting.md) - - **影视互娱剧本创作** - - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) - - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) - - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) - - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) + - **电商零售推广文案写作** + - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) + - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) - **影视传媒视频理解** - - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - [SubmitVideoAnalysisTask - 视频理解-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-submitvideoanalysistask.md) + - [GetVideoAnalysisTask - 视频理解-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysistask.md) - [GetVideoAnalysisConfig - 视频理解-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-getvideoanalysisconfig.md) - [UpdateVideoAnalysisConfig - 视频理解-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysisconfig.md) - [RunVideoAnalysis - 视频理解-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-runvideoanalysis.md) - [UpdateVideoAnalysisTasks - 视频理解-批量取消任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistasks.md) - [UpdateVideoAnalysisTask - 视频理解-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-video-understanding/api-quanmiaolightapp-2024-08-01-updatevideoanalysistask.md) + - **影视传媒智能拆条** + - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) + - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) + - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) + - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) + - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) + - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) - **车机网络热点信息互动问答** - [RunHotTopicChat - 播报单（热榜）问答](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicchat.md) - [RunHotTopicSummary - 播报单热点自定义摘要生成](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-car-machine-network-hot-information-interactive-question-and-answer/api-quanmiaolightapp-2024-08-01-runhottopicsummary.md) - - **泛企业VOC挖掘** - - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) - - **网络内容安全审核** - - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) - **泛企业线索挖掘** - - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) - [GenerateOutputFormat - 获取输出格式示例](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-generateoutputformat.md) + - [RunTagMiningAnalysis - 标签挖掘分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-clue-mining/api-quanmiaolightapp-2024-08-01-runtagmininganalysis.md) - **作文批改** - [RunEssayCorrection - 作文批改](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runessaycorrection.md) - - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) - [RunOcrParse - 图片OCR解析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-runocrparse.md) + - [SubmitEssayCorrectionTask - 提交作文批改任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-submitessaycorrectiontask.md) - [GetEssayCorrectionTask - 获取作文批改任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-composition-correction/api-quanmiaolightapp-2024-08-01-getessaycorrectiontask.md) + - **网络内容安全审核** + - [RunNetworkContentAudit - 网络内容安全审核](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-network-content-security-audit/api-quanmiaolightapp-2024-08-01-runnetworkcontentaudit.md) - **其他** - - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) - [GenerateBroadcastNews - 播报单（热榜）热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-generatebroadcastnews.md) - [ListHotTopicSummaries - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listhottopicsummaries.md) - [GetTagMiningAnalysisTask - 获取标签挖掘分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettagmininganalysistask.md) + - [SubmitTagMiningAnalysisTask - 提交标签挖掘分析任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submittagmininganalysistask.md) - [HotNewsRecommend - 新闻热点推荐](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-hotnewsrecommend.md) - [BatchCancelTasks - 批量取消异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchcanceltasks.md) - - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) - [GetFileContent - 获取文件内容](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getfilecontent.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) + - [BatchQueryTaskStatus - 批量查询异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-batchquerytaskstatus.md) - [CancelAsyncTask - 根据任务ID取消异步任务的执行](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-cancelasynctask.md) - [ExportAnalysisTagDetailByTaskId - 根据任务ID导出分析明细](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-exportanalysistagdetailbytaskid.md) - - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC分析任务结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-getenterprisevocanalysistask.md) - [ListAnalysisTagDetailByTaskId - 获取挖掘结果明细列表](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-listanalysistagdetailbytaskid.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC挖掘异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-submitenterprisevocanalysistask.md) - - **影视传媒智能拆条** - - [SubmitVideoDetectShotTask - 智能拆条-提交异步任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-submitvideodetectshottask.md) - - [GetVideoDetectShotTask - 智能拆条-获取异步任务状态和结果](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshottask.md) - - [UpdateVideoDetectShotConfig - 智能拆条-更新配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshotconfig.md) - - [UpdateVideoDetectShotTask - 智能拆条-修改异步任务状态](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-updatevideodetectshottask.md) - - [GetVideoDetectShotConfig - 智能拆条-获取配置](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-getvideodetectshotconfig.md) - - [RunVideoDetectShot - 智能拆条-在线任务](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-media-intelligent-strip/api-quanmiaolightapp-2024-08-01-runvideodetectshot.md) - - **电商零售推广文案写作** - - [RunMarketingInformationWriting - 电商零售推广文案写作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationwriting.md) - - [RunMarketingInformationExtract - 电商零售内容实体抽取](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-e-commerce-retail-promotion-copy-writing/api-quanmiaolightapp-2024-08-01-runmarketinginformationextract.md) + - [GetTaskExecutionStatistics - 查询任务执行情况统计](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-other/api-quanmiaolightapp-2024-08-01-gettaskexecutionstatistics.md) + - **泛企业VOC挖掘** + - [RunEnterpriseVocAnalysis - 在线企业VOC分析](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-pan-enterprise-voc-mining/api-quanmiaolightapp-2024-08-01-runenterprisevocanalysis.md) + - **影视互娱剧本创作** + - [RunScriptRefine - 影视互娱剧本创作-剧本整理](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptrefine.md) + - [RunScriptChat - 影视互娱剧本创作-交互式创作](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptchat.md) + - [RunScriptPlanning - 影视互娱剧本创作-剧本策划](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptplanning.md) + - [RunScriptContinue - 影视互娱剧本创作-剧本续写](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-dir/api-quanmiaolightapp-2024-08-01-dir-film-and-television-mutual-entertainment-script-creation/api-quanmiaolightapp-2024-08-01-runscriptcontinue.md) + - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-overview.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-ram.md) - - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-endpoint.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-light-application-series/development-documentation/api-reference-1/api-quanmiaolightapp-2024-08-01-changeset.md) - [全妙轻应用更新公告](raw/application-user-guide/application-gallery/quanmiao-light-application-series/light-application-update-announcement.md) - [常见问题](raw/application-user-guide/application-gallery/quanmiao-light-application-series/quanmiao-lightapp-faq.md) - **官方应用-伶鹊CCAI-对话分析AIO** - **使用指南** - [如何开通伶鹊CCAI-对话分析AIO](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/product-activation.md) - - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - [如何对应用进行编辑、删除等管理,如何进行API调用、如何查看调用量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/application-management.md) - - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) + - [如何进行基于对话分析Agent方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-dialogue-analysis-agent.md) - [知识库的使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/using-the-knowledge-base.md) + - [热词组配置管理与使用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/hot-phrase-management.md) + - [如何基于自定义方式创建应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/lingque-ccai-aio-user-guide/create-an-application-based-on-a-custom-method.md) - **API参考** - **API目录** - **热词管理** @@ -403,46 +394,33 @@ - [AnalyzeAudioSync - 语音文件实时分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-dir-not-recommended-or-whitelisted-open/api-contactcenterai-2024-06-03-analyzeaudiosync.md) - [RunCompletion - 通过模版ID调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletion.md) - [RunCompletionMessage - 使用原生Prompt调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-runcompletionmessage.md) - - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) - [AnalyzeConversation - 通过任务类型调用通义晓蜜CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeconversation.md) + - [CreateTask - 通过上传离线任务数据进行通义晓蜜CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-createtask.md) - [AnalyzeImage - 图片内容分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-analyzeimage.md) - - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) - [GetTaskResult - 通过任务ID获取离线任务分析结果](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-gettaskresult.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) + - [GeneralAnalyzeImage - 通用图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-dir/api-contactcenterai-2024-06-03-generalanalyzeimage.md) - [服务接入点](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-endpoint.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-overview.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/api-reference-2/api-contactcenterai-2024-06-03-changeset.md) - **最佳实践** - [客服服务质检最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/customer-service-quality-inspection-best-practices.md) - [字段信息抽取最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/best-practices-for-automatic-work-order-generation.md) - [摘要生成（含摘要/标题/关键词）最佳实践](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/best-practices/summary-best-practices.md) - **接口调用示例** - - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - [通过原生Prompt调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-native-prompt-to-call-tongyi-xiaomi-ccai-aio.md) + - [通过模板ID调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-template-id-to-call-tongyi-xiaomi-ccai-aio.md) - [通过任务类型调用伶鹊CCAI-对话分析AIO应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/call-tongyi-xiaomi-ccai-dialogue-analysis-aio-application-through-task-type.md) - - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) + - [通过上传离线任务数据进行伶鹊CCAI-对话分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/tongyi-xiaomi-ccai-dialogue-analysis-by-uploading-offline-task-data.md) - [ROA风格请求体&签名机制](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/roa-style-request-body-signature-mechanism.md) + - [通过伶鹊CCAI-对话分析AIO应用进行图片分析](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/picture-analysis-through-tongyi-xiaomi-ccai-dialogue-analysis-aio-application.md) - [伶鹊CCAI-对话分析RAM子账号使用方式和授权操作](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/call-tyxm-ccai-aio-api/use-and-authorize-ram-users-for-ccai-dialogue-analysis.md) - [通义晓蜜CCAI更新公告](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/tongyi-xiaomi-ccai-update-announcement.md) - [产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/product-overview-1.md) - [伶鹊CCAI-对话分析AIO产品计费](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/billing-description-magpie-ccai-dialogue-analysis-aio.md) - [CCAI如何进行集成；查看技术集成方案](raw/application-user-guide/application-gallery/official-application-lingque-ccai-dialogue-analysis-aio/technology-integration-scheme.md) - - **官方应用-伶鹊CCAI-客服对话Agent** - - **API参考** - - **API目录** - - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) - - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) - - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) - - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) - - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) - **官方应用-伶鹊CCAI-语音对话机器人** - **API参考** - **API目录** - - **MQ消息订阅配置** - - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) - - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) - - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) - **变量管理** - [DeleteVariable - 删除变量](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-deletevariable.md) - [ListVariable - 获取变量列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-variable-management/api-bailianvoicebot-2025-01-01-listvariable.md) @@ -454,42 +432,55 @@ - [ListVoiceAccessProfile - 获取三方语音配置列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-listvoiceaccessprofile.md) - [DeleteVoiceAccessProfile - 删除三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-deletevoiceaccessprofile.md) - [CreateVoiceAccessProfile - 创建三方语音配置](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-three-way-voice-configuration/api-bailianvoicebot-2025-01-01-createvoiceaccessprofile.md) - - **克隆音管理** - - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) - - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) - - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) + - **MQ消息订阅配置** + - [UpdateSubscription - 更新订阅信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-updatesubscription.md) + - [GetSubscription - 获取消息订阅配置信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-getsubscription.md) + - [DisableSubscription - 关闭消息订阅](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-mq-message-subscription-configuration/api-bailianvoicebot-2025-01-01-disablesubscription.md) + - **热词管理** + - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) + - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) + - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) + - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) + - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) + - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) + - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) + - **克隆音管理** + - [ListCloneVoiceModels - 获取克隆音模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoicemodels.md) + - [DeleteCloneVoice - 删除克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-deleteclonevoice.md) + - [ListCloneVoice - 获取克隆音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-listclonevoice.md) - [UpdateCloneVoice - 更新克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-updateclonevoice.md) + - [CreateCloneVoice - 创建克隆音](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-clone-tone-management/api-bailianvoicebot-2025-01-01-createclonevoice.md) - **应用管理** - [DeleteApplication - 删除语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-deleteapplication.md) - [ListNluModels - 获取对话大模型列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listnlumodels.md) - - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) - [PreviewVoice - TTS合成试听](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-previewvoice.md) + - [ListBackgroundMusics - 获取背景音列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listbackgroundmusics.md) - [ListVoices - 获取音色列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listvoices.md) - [ListApplications - 查询语音机器人应用列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-listapplications.md) - - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) - [CreateApplicationVersion - 创建语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplicationversion.md) - [CreateApplication - 创建语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-createapplication.md) - - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) + - [UpdateApplication - 修改语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplication.md) - [UpdateApplicationVersion - 修改语音机器人应用版本](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-updateapplicationversion.md) + - [GetApplication - 获取语音机器人应用](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-getapplication.md) - [PublishApplicationVersion - 发布语音机器人](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-application-management/api-bailianvoicebot-2025-01-01-publishapplicationversion.md) - - **热词管理** - - [ListVocabulary - 获取热词列表](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-listvocabulary.md) - - [UpdateVocabulary - 更新热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-updatevocabulary.md) - - [ImportVocabulary - 导入热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-importvocabulary.md) - - [GetVocabulary - 获取热词信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-getvocabulary.md) - - [ExportVocabulary - 导出热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-exportvocabulary.md) - - [DeleteVocabulary - 删除热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-deletevocabulary.md) - - [CreateVocabulary - 创建热词](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-dir-hot-word-management/api-bailianvoicebot-2025-01-01-createvocabulary.md) - [BridgeWebCall - 软电话测试通话](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-bridgewebcall.md) - [GetDataChannelCredential - 获取数据通道凭证](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-getdatachannelcredential.md) - [GenerateFileUploadParams - 获取文件上传参数](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-dir/api-bailianvoicebot-2025-01-01-generatefileuploadparams.md) - - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - [API概览](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-overview.md) - [版本说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-changeset.md) - - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/api-reference-chat6/api-bailianvoicebot-2025-01-01-ram.md) - [伶鹊CCAI-语音对话机器人产品计费说明](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/billing-information-lingque-ccai-voice-dialogue-robot.md) - [语音对话机器人产品概述](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/product-0verview.md) + - [语音对话机器人操作指南](raw/application-user-guide/application-gallery/official-application-lingque-ccai-voice-dialogue-robot/operation-guide.md) + - **官方应用-伶鹊CCAI-客服对话Agent** + - **API参考** + - **API目录** + - [SseChat - 问答接口](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-dir/api-bailianchatbot-2024-11-05-ssechat.md) + - [API概览](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-overview.md) + - [授权信息](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/api-reference-5/api-bailianchatbot-2024-11-05-ram.md) + - [计费说明（客服对话Agent）](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/billing-description-beebot-agent.md) + - [产品概述](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/product-overview-voicepica-beebot-agent.md) + - [使用指南](raw/application-user-guide/application-gallery/official-application-voicepica-ccai-beebot-agent/guidelines-for-use.md) - **通义点金** - **API参考** - **API目录** @@ -501,49 +492,49 @@ - [GetLibrary - 获取文档库详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getlibrary.md) - [UploadDocument - 上传文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-uploaddocument.md) - [GetDocumentUrl - 获取文档的下载链接](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumenturl.md) - - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [GetFilterDocumentList - 按元信息过滤查询文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getfilterdocumentlist.md) - - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) + - [PreviewDocument - 预览文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-previewdocument.md) - [DeleteDocument - 删除文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletedocument.md) + - [GetDocumentList - 获取文档列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentlist.md) - [UpdateDocument - 更新文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatedocument.md) - - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) - [CreatePredefinedDocument - 创建预定义文档](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-createpredefineddocument.md) + - [GetDocumentChunkList - 获取文档块列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getdocumentchunklist.md) - [RecallDocument - 文档召回](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-recalldocument.md) - [GetParseResult - 获取文档解析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-getparseresult.md) - [ReIndex - 重建索引](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-reindex.md) - - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) - [UpdateLibrary - 更新文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-updatelibrary.md) + - [DeleteLibrary - 删除文档库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-deletelibrary.md) - [RunLibraryChatGeneration - 文档库会话生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-runlibrarychatgeneration.md) - - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - [GetHistoryListByBizType - 根据业务类型获取对话历史记录](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-gethistorylistbybiztype.md) + - [InvokePlugin - 调用插件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-document-library/api-dianjin-2024-06-28-invokeplugin.md) - **平台能力-应用** - - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [EndToEndRealTimeDialog - 语音实时对话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-endtoendrealtimedialog.md) - [RunDialogAnalysis - 会话分析结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rundialoganalysis.md) + - [RunAgent - 运行智能体](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runagent.md) - [CreateDialog - 创建外呼会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialog.md) - [RealTimeDialog - 实时会话](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialog.md) - [RealtimeDialogAssist - 实时会话辅助](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-realtimedialogassist.md) - [GetDialogDetail - 获取会话详情](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialogdetail.md) - [GetDialogLog - 获取对话日志](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoglog.md) - [GetDialogAnalysisResult - 获取会话分析结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getdialoganalysisresult.md) - - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - [CreateDialogAnalysisTask - 创建会话分析任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdialoganalysistask.md) - [EvictTask - 取消任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-evicttask.md) + - [RebuildTask - 重建任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-rebuildtask.md) - [GetTaskStatus - 获取任务状态](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskstatus.md) - [CreateDocsSummaryTask - 创建多文档总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createdocssummarytask.md) - [CreateAnnualDocSummaryTask - 创建按年份总结文档任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createannualdocsummarytask.md) - - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) - [CreatePdfTranslateTask - 创建pdf文档翻译任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createpdftranslatetask.md) + - [GetSummaryTaskResult - 获取财报总结任务结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getsummarytaskresult.md) + - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - [GetTaskResult - 获取结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gettaskresult.md) - - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - [GetQualityCheckTaskResult - 获取质检结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getqualitychecktaskresult.md) + - [CreateQualityCheckTask - 创建质检任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createqualitychecktask.md) - [RecognizeIntention - 意图识别](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-recognizeintention.md) - [UpdateQaLibrary - 更新QA问答库](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-updateqalibrary.md) + - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) - [SubmitChatQuestion - 提交问题列表](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-submitchatquestion.md) - [GetChatQuestionResp - 获取问答结果](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-getchatquestionresp.md) - [RunChatResultGeneration - 对话结果生成](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-runchatresultgeneration.md) - - [GenDocQaResult - 根据文档解析问答QA](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-gendocqaresult.md) - - [CreateFinReportSummaryTask - 创建财报总结任务](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-platform-capabilities-application/api-dianjin-2024-06-28-createfinreportsummarytask.md) - **其他** - [DashscopeAsyncTaskFinishEvent - Dashscope异步任务完成回调事件](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-dir/api-dianjin-2024-06-28-dir-other/api-dianjin-2024-06-28-dashscopeasynctaskfinishevent.md) - [API概览](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-overview.md) @@ -551,6 +542,19 @@ - [授权信息](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-ram.md) - [版本说明](raw/application-user-guide/application-gallery/tongyi-dianjin/api-reference-3/api-dianjin-2024-06-28-changeset.md) - [产品简介](raw/application-user-guide/application-gallery/tongyi-dianjin/tongyi-dianjin-overview.md) + - **官方应用-通义数据挖掘** + - **API参考** + - **API目录** + - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) + - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) + - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) + - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) + - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) + - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) + - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) + - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) + - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) + - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - **官方应用-通义多模态翻译** - **API参考** - **API目录** @@ -561,11 +565,11 @@ - [GetLongTextTranslateTask - 获取长文本翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-getlongtexttranslatetask.md) - [SubmitHtmlTranslateTask - 提交html翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-submithtmltranslatetask.md) - [GetHtmlTranslateTask - 获取html翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-gethtmltranslatetask.md) - - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - [TermQuery - 术语库查询](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termquery.md) + - [TermEdit - 术语库编辑](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-text-translation/api-anytrans-2025-07-07-termedit.md) - **图片翻译** - - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) - [GetImageTranslateTask - 获取图片翻译任务结果](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-getimagetranslatetask.md) + - [SubmitImageTranslateTask - 提交图片翻译任务](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-image-translation/api-anytrans-2025-07-07-submitimagetranslatetask.md) - **文档翻译** - [SubmitDocTranslateTask - 文档翻译任务提交](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-submitdoctranslatetask.md) - [GetDocTranslateTask - 文档翻译结果获取](raw/application-user-guide/application-gallery/official-application-tongyi-translate/tongyi-translate-api-reference/api-anytrans-2025-07-07-dir/api-anytrans-2025-07-07-dir-document-translation/api-anytrans-2025-07-07-getdoctranslatetask.md) @@ -598,54 +602,41 @@ - [错误码-千问联网检索Agent](raw/application-user-guide/application-gallery/web-search-agent/web-search-agent-error-code.md) - **通义 UI Agent** - [通义 UI Agent](raw/application-user-guide/application-gallery/ui-agent/ui-agent-api.md) - - **官方应用-通义数据挖掘** - - **API参考** - - **API目录** - - [文档上传](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-upload.md) - - [信息抽取](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-information-extraction.md) - - [文档内容审核](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-content-audit.md) - - [打标分类](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-tagging.md) - - [摘要生成](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-summary-generation.md) - - [文档删除](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-directory/document-delete.md) - - [API概览](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-api-overview.md) - - [服务接入点](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-service-access-point.md) - - [错误码](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-api-reference/docmining-error-code.md) - - [通义数据挖掘产品介绍](raw/application-user-guide/application-gallery/tongyi-docmining/docmining-product-introduction.md) - **官方应用-全妙解决方案类产品** - **妙笔、妙策和审校** - **使用指南** - **AI妙笔** - **功能界面** - [妙笔-分布生成创作文章](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/step-by-step-generation.md) - - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [直接生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/direct-generation.md) + - [智能配图](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/smart-image-generation.md) - [搜索素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/function-interface/search-materials.md) - [AI妙笔产品概述](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/product-overview-for-amb.md) - [妙笔首页概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/amb-homepage-overview.md) - [AI工具箱](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/ai-toolbox.md) - [素材库](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/material-library.md) - - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [文章风格和格式学习](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/style-imitation.md) + - [系统配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/amb/system-configuration.md) - [AI妙策](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/ai-miaoce.md) - [智能审校](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/article-review.md) - [深度写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/usage-guide/deep-writing.md) - **文本写作指导** - - **政务公文写作指导** - - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) - - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) - - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) - **传媒类文体写作指导** - [快速写一篇传媒稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/quick-media-writing-prompt.md) - [没有思路，要谋篇布局](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/use-amb-to-help-writing.md) - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/media-style-writing-best-practices/generate-titles-summaries-media-text.md) + - **政务公文写作指导** + - [快速写一篇政务稿（prompt一步式撰写）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/quick-gov-writing-prompt.md) + - [分步式撰写政务稿（精准控制章节内容）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/step-by-step-gov-writing.md) + - [用已有文章，生成标题摘要等](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/government-document-writing-best-practices/generate-titles-summaries-gov-text.md) - [常见FAQ](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/document-writing-best-practices/faq-for-using-quanmiao-series-products.md) - **更新公告** - **功能更新** - [2025年2月26日更新-妙笔](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/february-26-2025-update-miaobi.md) - [2025年1月24日更新-全妙解决方案类产品](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/2025-1-24-function-update-announcement-quanmiao-saas.md) - [2024年3月11更新-AI全妙系列 V2.2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-11-ai-quanmiao-v2-2.md) - - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - [2024年2月28更新-AI全妙系列 V2.2](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-02-28-ai-quanmiao-v2.md) + - [2024年3月1更新-AI全妙系列 V2.2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2024-03-01-ai-quanmiao-v2-2.md) - [2023年12月19更新-AI妙笔V2.1](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/update-announcement/miaobi-and-miaoce-function-update/update-2023-12-19-amb-v2.md) - [计费说明（妙笔）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/miaobi-billing.md) - [计费说明（政务公文配套工具）](raw/application-user-guide/application-gallery/quanmiao-solution-products/miaobi-miaoce-shenjiao/government-document-tool-billing.md) @@ -660,63 +651,72 @@ - **API参考** - **数据结构** - [GenerateTraceability](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-generatetraceability.md) - - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) - [HottopicNews](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-hottopicnews.md) + - [OutlineSearchResult](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinesearchresult.md) - [TopicSelection](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-topicselection.md) + - [OutlineWritingArticle](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-outlinewritingarticle.md) - [WritingOutline](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingoutline.md) - - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) - [WritingStyleTemplateField](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatefield.md) + - [WritingStyleTemplateDefine](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-struct-dir/api-aimiaobi-2023-08-01-struct-writingstyletemplatedefine.md) - **API目录** + - **通用接口-文件上传下载** + - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) + - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) - **通用接口** - [CreateToken - 获取授权token](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-createtoken.md) - [ListDialogues - 生成历史列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listdialogues.md) - - [ListVersions - 获取版本信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listversions.md) - [GetProperties - 获取配置信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-getproperties.md) - - **通用接口-文件上传下载** - - [GenerateUploadConfig - 生成上传配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generateuploadconfig.md) - - [GenerateFileUrlByKey - 生成文件URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-file-upload-and-download/api-aimiaobi-2023-08-01-generatefileurlbykey.md) + - [ListVersions - 获取版本信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-universal-interface/api-aimiaobi-2023-08-01-listversions.md) - **通用接口-异步任务管理** - [SubmitAsyncTask - 提交异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-submitasynctask.md) - [CancelAsyncTask - 取消异步任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-cancelasynctask.md) - [QueryAsyncTask - 查询异步任务明细](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-queryasynctask.md) - [ListAsyncTasks - 获取异步任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-asynchronous-task-management/api-aimiaobi-2023-08-01-listasynctasks.md) - - **通用接口-通用配置** - - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) - - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) - - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) - - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) - - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) - **妙笔-创作文章** - - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - [RunAiHelperWriting - AI帮写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runaihelperwriting.md) + - [RunWritingV2 - 智能写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritingv2.md) - [RunWriting - 直接写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwriting.md) - - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) - [RunTranslateGeneration - 中英翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtranslategeneration.md) + - [RunStepByStepWriting - 分步骤写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runstepbystepwriting.md) - [RunTextPolishing - 润色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtextpolishing.md) - - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) - - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) - [RunWriteToneGeneration - 文风改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runwritetonegeneration.md) + - [RunKeywordsExtractionGeneration - 关键词抽取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runkeywordsextractiongeneration.md) + - [RunContinueContent - 内容续写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runcontinuecontent.md) - [RunTitleGeneration - 标题生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runtitlegeneration.md) - [RunSummaryGenerate - 摘要生成](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runsummarygenerate.md) - [RunExpandContent - 内容扩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runexpandcontent.md) - - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) - [RunAbbreviationContent - 内容缩写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runabbreviationcontent.md) - [RunQuickWriting - 快速写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-runquickwriting.md) - - [GenerateImageTask - 生成智能配图任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-generateimagetask.md) + - [SearchNews - 信息检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-searchnews.md) - [ListBuildConfigs - 获取系统自定义预设](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-listbuildconfigs.md) - - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) + - [GenerateImageTask - 生成智能配图任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-generateimagetask.md) - [FeedbackDialogue - 反馈对话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-feedbackdialogue.md) - - **妙笔-视频审校** - - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) - - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) + - [FetchImageTask - 获取图片任务执行结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-creative-articles/api-aimiaobi-2023-08-01-fetchimagetask.md) + - **通用接口-通用配置** + - [ListGeneralConfigs - 通用配置-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-listgeneralconfigs.md) + - [GetGeneralConfig - 通用配置-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-getgeneralconfig.md) + - [CreateGeneralConfig - 通用配置-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-creategeneralconfig.md) + - [DeleteGeneralConfig - 通用配置-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-deletegeneralconfig.md) + - [UpdateGeneralConfig - 通用配置-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-common-interface-common-configuration/api-aimiaobi-2023-08-01-updategeneralconfig.md) - **妙笔-文体仿写** - [ListStyleLearningResult - 获取文体学习分析结果列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-liststylelearningresult.md) - [RunStyleFeatureAnalysis - 内容特点分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-runstylefeatureanalysis.md) - [DeleteStyleLearningResult - 删除自定义文体](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-deletestylelearningresult.md) - - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) - [GetStyleLearningResult - 获取文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-getstylelearningresult.md) + - [SaveStyleLearningResult - 保存文体学习分析结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-savestylelearningresult.md) - [ListWritingStyles - 获取写作文体列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-style-imitation-writing/api-aimiaobi-2023-08-01-listwritingstyles.md) + - **妙笔-视频审校** + - [QueryVideoAuditResult - 查询视频审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-queryvideoauditresult.md) + - [SubmitVideoAudit - 提交视频审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-review/api-aimiaobi-2023-08-01-submitvideoaudit.md) + - **妙笔-文章审校-词库管理** + - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) + - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) + - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) + - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) + - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) + - [FetchExportTermsTask - 获取导出词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchexporttermstask.md) + - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) + - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) - **妙笔-文章审校-规则库管理** - [SubmitAuditNote - 提交自定义规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-submitauditnote.md) - [ConfirmAndPostProcessAuditNote - 确认提交规则库用于审核](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-confirmandpostprocessauditnote.md) @@ -725,99 +725,85 @@ - [GetAuditNotePostProcessingStatus - 获取规则库后处理进度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnotepostprocessingstatus.md) - [GetAuditNoteProcessingStatus - 查询规则库上传状态](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getauditnoteprocessingstatus.md) - [GetAvailableAuditNotes - 查询可用规则库](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-proofreading-rule-library-management/api-aimiaobi-2023-08-01-getavailableauditnotes.md) - - **妙笔-文章审校-词库管理** - - [ListAuditTerms - 获取自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-listauditterms.md) - - [AddAuditTerms - 添加自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-addauditterms.md) - - [DeleteAuditTerms - 删除指定词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-deleteauditterms.md) - - [EditAuditTerms - 编辑自定义词库记录](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-editauditterms.md) - - [SubmitImportTermsTask - 提交导入词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitimporttermstask.md) - - [FetchImportTermsTask - 获取导入词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchimporttermstask.md) - - [SubmitExportTermsTask - 提交导出词库任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-submitexporttermstask.md) - - [FetchExportTermsTask - 获取导出词库任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-thesaurus-management/api-aimiaobi-2023-08-01-fetchexporttermstask.md) - - **妙笔-文章审校-事实性审核** - - [SubmitFactAuditUrl - 提交事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-submitfactauditurl.md) - - [GetFactAuditUrl - 获取事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-getfactauditurl.md) - - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - **妙笔-文章审校** - [SubmitSmartAudit - 提交智能审校任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-submitsmartaudit.md) - [GetSmartAuditResult - 查询智能审校结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-getsmartauditresult.md) - [ListAuditContentErrorTypes - 获取审校维度列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-listauditcontenterrortypes.md) - [ExportAuditContentResult - 导出智能审校报告](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-article-reviser/api-aimiaobi-2023-08-01-exportauditcontentresult.md) + - **妙笔-文章审校-事实性审核** + - [SubmitFactAuditUrl - 提交事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-submitfactauditurl.md) + - [GetFactAuditUrl - 获取事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-getfactauditurl.md) + - [DeleteFactAuditUrl - 删除事实性审核 URL](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaobi-article-review-factual-review/api-aimiaobi-2023-08-01-deletefactauditurl.md) - **妙笔-文档管理** - [GenerateExportWordTask - 生成导出文档任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-generateexportwordtask.md) - [FetchExportWordTask - 获取导出文档任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-fetchexportwordtask.md) - [CreateGeneratedContent - 保存文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-creategeneratedcontent.md) - [DeleteGeneratedContent - 删除文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-deletegeneratedcontent.md) - [UpdateGeneratedContent - 更新文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-updategeneratedcontent.md) - - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) - [GetGeneratedContent - 获取文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-getgeneratedcontent.md) + - [ListGeneratedContents - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-listgeneratedcontents.md) - [ExportGeneratedContent - 导出文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-document-management/api-aimiaobi-2023-08-01-exportgeneratedcontent.md) - **妙笔-素材库** - [SaveMaterialDocument - 保存素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-savematerialdocument.md) - [DeleteMaterialById - 删除素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-deletematerialbyid.md) - [UpdateMaterialDocument - 更新素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-updatematerialdocument.md) - - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - [GetMaterialById - 获取素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-getmaterialbyid.md) + - [ListMaterialDocuments - 获取素材列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library/api-aimiaobi-2023-08-01-listmaterialdocuments.md) - **妙笔-素材库-自定义文本** - [GetCustomText - 获取自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-getcustomtext.md) - [UpdateCustomText - 更新自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-updatecustomtext.md) - - [ListCustomText - 获取自定义文本列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-listcustomtext.md) - [SaveCustomText - 保存自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-savecustomtext.md) + - [ListCustomText - 获取自定义文本列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-listcustomtext.md) - [DeleteCustomText - 删除自定义文本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-deletecustomtext.md) - [DocumentExtraction - 文档提取](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-material-library-custom-text/api-aimiaobi-2023-08-01-documentextraction.md) - **妙笔-视频混剪** - [GetClipsBuildInResource - 获取智能混剪内置资源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getclipsbuildinresource.md) - - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - [AsyncCreateClipsTimeLine - 创建剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstimeline.md) - [AsyncUploadVideo - 异步上传视频剪辑素材](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncuploadvideo.md) - - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) + - [AsyncEditTimeline - 编辑剪辑口播时间线](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asyncedittimeline.md) - [GetAutoClipsTaskInfo - 获得剪辑任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-getautoclipstaskinfo.md) + - [AsyncCreateClipsTask - 创建剪辑任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-asynccreateclipstask.md) - [ListAutoClipsTask - 智能混剪任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-pen-video-mixed-cut/api-aimiaobi-2023-08-01-listautoclipstask.md) - **妙策-自定义数据源** - [SubmitCustomSourceTopicAnalysis - 提交自定义源话题选题分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-submitcustomsourcetopicanalysis.md) - [ExportCustomSourceAnalysisTask - 导出自定义源-话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-exportcustomsourceanalysistask.md) - [GetCustomSourceTopicAnalysisTask - 获取自定义源话题分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-custom-data-source/api-aimiaobi-2023-08-01-getcustomsourcetopicanalysistask.md) + - **妙策-自定义话题** + - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) + - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) + - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) + - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) + - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) + - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) + - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) - **公文库检索** - [ListDocumentRetrieve - 公文库检索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-public-library-retrieval/api-aimiaobi-2023-08-01-listdocumentretrieve.md) - **妙策-选题热点** - - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) - [RunTopicSelectionMerge - 选题热点融合](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-runtopicselectionmerge.md) - - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) + - [ListHotNewsWithType - 获取选题热点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotnewswithtype.md) - [ListHotTopics - 获取热点话题列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhottopics.md) + - [ListHotSources - 获取三方热榜源列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotsources.md) + - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) - [GetTopicById - 获取热点对象](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-gettopicbyid.md) - [ListHotViewPoints - 获取热门视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listhotviewpoints.md) - - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) - - [ListTimedViewAttitude - 获取时效性视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listtimedviewattitude.md) - [ListWebReviewPoints - 获取网友视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listwebreviewpoints.md) - - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) - [ListPlanningProposal - 获取选题策划列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listplanningproposal.md) - - **妙策-自定义话题** - - [DeleteCustomTopicByTopic - 删除自定义热点事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicbytopic.md) - - [ListTopicViewPointRecommendEventList - 获取热点事件推荐观点列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicviewpointrecommendeventlist.md) - - [ListTopicRecommendEventList - 获取热点推荐事件列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listtopicrecommendeventlist.md) - - [RunCustomHotTopicAnalysis - 自定义热点话题分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicanalysis.md) - - [RunCustomHotTopicViewPointAnalysis - 自定义选题视角分析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-runcustomhottopicviewpointanalysis.md) - - [DeleteCustomTopicViewPointById - 删除自定义选题视角](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-deletecustomtopicviewpointbyid.md) - - [ListCustomViewPoints - 获取自定义视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-custom-topic/api-aimiaobi-2023-08-01-listcustomviewpoints.md) - - **妙策-新闻播报** - - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) - - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) - - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) + - [ListFreshViewPoints - 获取新颖视角列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-listfreshviewpoints.md) + - [ExportHotTopicPlanningProposals - 导出选题策划文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaozi-hot-topics/api-aimiaobi-2023-08-01-exporthottopicplanningproposals.md) - **妙策-openapi** - [SubmitDocClusterTask - 提交内容聚合任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitdocclustertask.md) - [GetDocClusterTask - 获取内容聚合任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getdocclustertask.md) - [SubmitTopicSelectionPerspectiveAnalysisTask - 提交选题热点分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submittopicselectionperspectiveanalysistask.md) - - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - [GetTopicSelectionPerspectiveAnalysisTask - 获取选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-gettopicselectionperspectiveanalysistask.md) + - [GetCustomTopicSelectionPerspectiveAnalysisTask - 获取自定义选题视角分析任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-getcustomtopicselectionperspectiveanalysistask.md) - [SubmitCustomTopicSelectionPerspectiveAnalysisTask - 提交自定义热点选题视角分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tips-openapi/api-aimiaobi-2023-08-01-submitcustomtopicselectionperspectiveanalysistask.md) - - **妙搜-智能搜索** - - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) - - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) - - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) - - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) - - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) + - **妙策-新闻播报** + - [GetHotTopicBroadcast - 查询完整播报单（热榜）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-gethottopicbroadcast.md) + - [GetCustomHotTopicBroadcastJob - 获取自定义播报单任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-getcustomhottopicbroadcastjob.md) + - [SubmitCustomHotTopicBroadcastJob - 提交自定义播报单任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-policy-news-broadcast/api-aimiaobi-2023-08-01-submitcustomhottopicbroadcastjob.md) - **妙搜-数据源** - [CreateDataset - 数据源-创建](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md) + - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) - [UpdateDataset - 数据源-修改](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md) - [ListDatasets - 数据源-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md) - [DeleteDataset - 数据源-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md) @@ -825,16 +811,21 @@ - [GetDatasetDocument - 数据源-获取文档详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdatasetdocument.md) - [UpdateDatasetDocument - 数据源-修改文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedatasetdocument.md) - [ListDatasetDocuments - 数据源-文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasetdocuments.md) - - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - - [GetDataset - 数据源-详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md) - [SearchDatasetDocuments - 数据源-搜索文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-searchdatasetdocuments.md) + - [DeleteDatasetDocument - 数据源-删除数据集文档](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedatasetdocument.md) - **妙策-企业VOC挖掘** - [ExportAnalysisTagDetailByTaskId - 导出标签挖掘结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-exportanalysistagdetailbytaskid.md) - [ValidateUploadTemplate - 校验VOC上传模板](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-validateuploadtemplate.md) - [SubmitEnterpriseVocAnalysisTask - 提交企业VOC分析任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-submitenterprisevocanalysistask.md) - - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) - [ListAnalysisTagDetailByTaskId - 根据任务ID获取标签分析明细列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-listanalysistagdetailbytaskid.md) + - [GetEnterpriseVocAnalysisTask - 获取企业VOC挖掘任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getenterprisevocanalysistask.md) - [GetCategoriesByTaskId - 根据任务ID获取分类列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-miaoce-enterprise-voc-mining/api-aimiaobi-2023-08-01-getcategoriesbytaskid.md) + - **妙搜-智能搜索** + - [ListSearchTasks - 查询妙搜搜索生成历史任务列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtasks.md) + - [RunSearchGeneration - 妙搜-智能搜索](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchgeneration.md) + - [ListSearchTaskDialogues - 查询妙搜搜索生成任务详情列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialogues.md) + - [ListSearchTaskDialogueDatas - 查询搜索生成任务对话详情中数据列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-listsearchtaskdialoguedatas.md) + - [RunSearchSimilarArticles - 妙搜-文搜文](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-smart-search/api-aimiaobi-2023-08-01-runsearchsimilararticles.md) - **系统配置-干预配置** - [ListInterveneCnt - 获得所有干预项的数量](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenecnt.md) - [ListIntervenes - 列出干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenes.md) @@ -842,21 +833,21 @@ - [InsertInterveneGlobalReply - 插入干预全局回复项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertinterveneglobalreply.md) - [ImportInterveneFileAsync - 异步导入干预项文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-importintervenefileasync.md) - [GetInterveneTemplateFileUrl - 获得干预导入模版文件地址](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getintervenetemplatefileurl.md) - - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - [GetInterveneGlobalReply - 获得干预全局回复内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneglobalreply.md) - - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) - [ListInterveneRules - 列出干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listintervenerules.md) - [InsertInterveneRule - 插入干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-insertintervenerule.md) - [GetInterveneRuleDetail - 获得干预规则的详情](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneruledetail.md) - [DeleteInterveneRule - 删除干预规则](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-deleteintervenerule.md) - [ExportIntervenes - 导出干预项内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-exportintervenes.md) - [GetInterveneImportTaskInfo - 获得干预项目导入任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-getinterveneimporttaskinfo.md) + - [ListInterveneImportTasks - 列出干预项导入任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-listinterveneimporttasks.md) + - [ClearIntervenes - 清除所有干预项](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-intervention-configuration/api-aimiaobi-2023-08-01-clearintervenes.md) - **系统配置-信源管理** - [SaveDataSourceOrderConfig - 保存信源权重配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-savedatasourceorderconfig.md) - [GetDataSourceOrderConfig - 获取信源配置权重数据](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-system-configuration-source-management/api-aimiaobi-2023-08-01-getdatasourceorderconfig.md) - **妙读-基础操作类** - - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) - [GetDocInfo - 获取文档信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getdocinfo.md) + - [GetFileContentLength - 获取文件长度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-getfilecontentlength.md) - [UploadBook - 书籍上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploadbook.md) - [UploadDoc - 文档上传](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-uploaddoc.md) - [ListDocs - 获取文档列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-basic-operation-class/api-aimiaobi-2023-08-01-listdocs.md) @@ -866,19 +857,19 @@ - [RunDocBrainmap - 全文脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocbrainmap.md) - [RunDocIntroduction - 文档导读](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocintroduction.md) - [RunDocSummary - 文档摘要](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocsummary.md) - - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) - [RunBookIntroduction - 书籍导读（抽取书籍卖点/书籍摘要）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookintroduction.md) + - [RunDocWashing - 改写](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-rundocwashing.md) - [RunCommentGeneration - 客户之声预测](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runcommentgeneration.md) - [RunBookBrainmap - 书籍脑图](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-generate-class/api-aimiaobi-2023-08-01-runbookbrainmap.md) + - **妙读-问答类** + - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) + - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - **妙读-抽取类** - [RunHotword - 抽取关键词](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-extraction-class/api-aimiaobi-2023-08-01-runhotword.md) - **妙读-其他** - - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - [RunDocTranslation - 文档翻译](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundoctranslation.md) + - [RunDocSmartCard - 文档智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-rundocsmartcard.md) - [RunBookSmartCard - 书籍智能卡片](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-other/api-aimiaobi-2023-08-01-runbooksmartcard.md) - - **妙读-问答类** - - [RunGenerateQuestions - 猜你想问](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rungeneratequestions.md) - - [RunDocQa - 文档问答（文章问答/多模态文件问答）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-reading-question-and-answer-class/api-aimiaobi-2023-08-01-rundocqa.md) - **深度写作** - [SubmitDeepWriteTask - 提交深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-submitdeepwritetask.md) - [GetDeepWriteTask - 查询深度写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-getdeepwritetask.md) @@ -887,161 +878,157 @@ - [RunDeepWriting - 查询深度写作事件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-deep-writing/api-aimiaobi-2023-08-01-rundeepwriting.md) - **PPT生成** - [ListEnterprisePptTemplates - 查询企业专属PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listenterpriseppttemplates.md) - - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - [InitiatePptCreationV2 - 初始化PPT创建操作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreationv2.md) - - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) + - [ListPptTemplates - 查询PPT模板列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listppttemplates.md) - [GetPptArtifactExportResult - 查询PPT导出任务的结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifactexportresult.md) + - [GetPptTemplateSelector - 查询PPT模板筛选器](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getppttemplateselector.md) - [ExportPptArtifact - 导出PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-exportpptartifact.md) - [GetPptArtifact - 查询PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptartifact.md) - [RunPptOutlineGeneration - 生成PPT大纲内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-runpptoutlinegeneration.md) - [ListPptArtifacts - 查询PPT作品列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-listpptartifacts.md) - [InitiatePptCreation - 初始化用来创建PPT的会话](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-initiatepptcreation.md) - - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - [GetPptConfig - 获取PPT组件配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-getpptconfig.md) - [BindPptArtifact - 绑定PPT作品信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-bindpptartifact.md) + - [DeletePptArtifact - 删除PPT作品](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-ppt-generation/api-aimiaobi-2023-08-01-deletepptartifact.md) - **标书生成** - - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) - [GetBiddingRemainLimitNum - 获得标书写作剩余额度](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingremainlimitnum.md) - - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) + - [AsyncUploadTenderDoc - 招标文档解析](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncuploadtenderdoc.md) + - [GetBiddingDocInfo - 获得标书写作结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-getbiddingdocinfo.md) - [EditBiddingDoc - 编辑标书内容](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-editbiddingdoc.md) + - [DownloadBiddingDoc - 下载标书文件](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-downloadbiddingdoc.md) - [AsyncWritingBiddingDoc - 标书写作](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-asyncwritingbiddingdoc.md) - [ListBiddingDoc - 列出标书写作任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-tender-generation/api-aimiaobi-2023-08-01-listbiddingdoc.md) - **其他** - - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - - [SubmitSmartClipTask - 提交智能一键成片任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitsmartcliptask.md) - [GetSmartClipTask - 获取智能剪辑任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getsmartcliptask.md) + - [RunVideoScriptGenerate - AI生成视频剪辑脚本](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-runvideoscriptgenerate.md) - [SaveOrUpdateOssConfig - 配置-云存储-参数配置](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-saveorupdateossconfig.md) - - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) + - [SubmitSmartClipTask - 提交智能一键成片任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitsmartcliptask.md) - [CreateDataPermissions - 权限-批量添加](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdatapermissions.md) - [GenerateViewPoint - 生成选题视角（已过时，不推荐使用）](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-generateviewpoint.md) + - [DeleteDataPermissions - 权限-删除](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedatapermissions.md) - [ListDataPermissions - 权限-列表](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatapermissions.md) - - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - [CancelAuditTask - 取消审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-cancelaudittask.md) + - [GetPptInfo - 查询PPT任务信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getpptinfo.md) - [FetchParseDocumentLayoutTask - 获取排版任务结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-fetchparsedocumentlayouttask.md) - - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md) - [SubmitAuditTask - 提交审核任务](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-submitaudittask.md) + - [QueryAuditTask - 查询审核结果](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-queryaudittask.md) - [服务接入点](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-endpoint.md) - [API概览](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-overview.md) - [授权信息](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-ram.md) - [版本说明](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-changeset.md) - - **最佳实践** - - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) - - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) - - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) - - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) - - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) - - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) - **更多** - [全妙服务关联角色](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-slr.md) - [妙笔写作信源对接](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaobi-writing-source-docking.md) - [妙搜数据集管理通过API引入数据源](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/miaosou-introduce-data-source-through-api.md) - [全妙iframe嵌入方案](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/iframe-embedding-scheme.md) - - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) - [全妙云存储（OSS）设置指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/oss-setup-guide.md) - [全妙PaaS AgentKey 获取指南](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/quanmiao-paas-agentkey-get-guide.md) + - [全妙Logo定制规范及部署方式](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/quanmiao-more/logo-customization-specification-and-deployment-method.md) + - **最佳实践** + - [智能审校最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-smart-audit.md) + - [妙策API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaoce-api.md) + - [妙笔API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaobi-api.md) + - [妙搜API最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-miaosou-api.md) + - [视频混剪最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/best-practices-for-video-mixing-and-cutting.md) + - [妙读最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/miaodu-best-practices.md) + - [PPT生成最佳实践](raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/miaobi-and-miaoce-best-practices/ppt-generation-best-practices.md) - [官方应用-通义听悟Agent](raw/application-user-guide/application-gallery/official-application-tingwu-agent.md) - [官方应用-析言GBI](raw/application-user-guide/application-gallery/xiyan-gbi.md) - [通义法睿](raw/application-user-guide/application-gallery/tongyi-farui.md) -- **应用观测** - - [应用观测](raw/application-user-guide/application-monitoring/application-observation.md) - **权限管理** - [权限管理](raw/application-user-guide/application-permission-management/application-permission-management-overview.md) +- **服务支持** + - [常见问题](raw/application-user-guide/application-support/application-faq.md) + - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) + - [阿里云百炼平台售后服务范围说明](raw/application-user-guide/application-support/application-after-sales-service-scope.md) - **实践教程** - - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [在网站上增加一个AI助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-website-in-10-minutes.md) + - [在企业微信中集成一个 AI 助手](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-work-wechat.md) - [10分钟让微信公众号成为智能客服](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-wechat-in-10-minutes.md) - [在钉钉上增加一个AI机器人](raw/application-user-guide/application-use-cases/add-an-ai-assistant-to-your-dingtalk.md) - [基于本地知识库构建RAG应用](raw/application-user-guide/application-use-cases/build-rag-application-based-on-local-retrieval.md) -- **服务支持** - - [常见问题](raw/application-user-guide/application-support/application-faq.md) - - [相关协议](raw/application-user-guide/application-support/application-related-agreements.md) - - [阿里云百炼平台售后服务范围说明](raw/application-user-guide/application-support/application-after-sales-service-scope.md) ## 模型 API 参考 +- **使用 API** + - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) + - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) + - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) + - [错误码](raw/model-api-reference/preparations/error-code.md) - **图像生成** - - **千问** - - [千问-图像生成与编辑3.0 API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) - - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) - - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) - - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - **万相** - [万相-文生图V2版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) - - [万相-图像生成与编辑2.7 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-图像生成与编辑2.6 API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) - [万相-通用图像编辑2.5](raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-通用图像编辑API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-涂鸦作画API参考](raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - [万相-图像局部重绘API参考](raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) + - **千问** + - [千问-图像生成与编辑3.0 API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) + - [千问-文生图API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) + - [千问-图像编辑API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) + - [千问-图像翻译API参考](raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) + - **Z-Image** + - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - **可灵** - [可灵-图像生成API参考](raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) - **Vidu** - [Vidu-图像生成API参考](raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) - - **Z-Image** - - [Z-Image API参考](raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - **创意工具** - [人像风格重绘API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) - - [虚拟模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [图像画面扩展API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) + - [虚拟模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) - - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [创意海报生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) + - [人物实例分割API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [图像擦除补全API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [图像背景生成API参考](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) + - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [AI试衣OutfitAnyone](raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) - - [人物写真生成FaceChain](raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [常见问题](raw/model-api-reference/image-generation/image-faq.md) -- **使用 API** - - [获取API Key](raw/model-api-reference/preparations/get-api-key.md) - - [安装SDK](raw/model-api-reference/preparations/install-sdk.md) - - [错误码](raw/model-api-reference/preparations/error-code.md) - - [使用百炼 CLI](raw/model-api-reference/preparations/use-model-studio-cli.md) - **3D模型生成** - [Tripo-3D模型生成](raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) -- **实时多模态** - - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) - - [实时多模态交互流程](raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) - - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) - - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) -- **更多模型** - - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) - - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) - - [Qwen-OCR API参考](raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) - - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) - **Realtime API** - **快速开始** - [SDK下载](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) - [Token鉴权](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) - [实现接通模型/应用](raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) - **最佳实践** - - [通过WebRTC使用多模态交互套件实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) - [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) + - [通过WebRTC使用多模态交互套件实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) - [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) - **AOQ客户端API** - **AOQ SDK功能介绍** - [连接状态管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) - [音频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) - - [媒体流发送管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) - [自定义音频播放](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) + - [媒体流发送管理](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) - [自定义音频采集](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) - - [视频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) - [自定义视频输入](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) + - [视频常用功能介绍](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) - [AOQ SDK简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) - [Realtime API简介](raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) +- **实时多模态** + - [客户端事件](raw/model-api-reference/omni-realtime-api/client-events.md) + - [Python SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) + - [服务端事件](raw/model-api-reference/omni-realtime-api/server-events.md) + - [Java SDK](raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) + - [实时多模态交互流程](raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) + - [声音复刻API参考](raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) +- **更多模型** + - [意图理解能力](raw/model-api-reference/more-models/intent-detect-capability.md) + - [Qwen-MT API参考](raw/model-api-reference/more-models/qwen-mt-api.md) + - [Qwen-Deep-Research API 参考](raw/model-api-reference/more-models/qwen-deep-research-api.md) + - [GUI-Plus API参考](raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) + - [通义法睿大语言模型](raw/model-api-reference/more-models/tongyi-farui-api.md) - **工具包/框架** - - [OpenAI Chat接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) - - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Responses接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) - - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) + - [completions 接口](raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) + - [OpenAI Chat接口兼容](raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) + - [OpenAI文件接口兼容](raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI兼容-Batch（文件输入）](raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) @@ -1051,30 +1038,30 @@ - [模型调优](raw/model-api-reference/model-production/fine-tuning-jobs-api.md) - [模型部署](raw/model-api-reference/model-production/deployments-api.md) - **更多** - - [生成临时API Key](raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - [异步任务管理 API](raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [通过HTTP回调URL或MQ接收异步任务完成通知](raw/model-api-reference/more-about-models/async-task-api.md) - [子业务空间的模型调用](raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - [上传本地文件获取临时URL](raw/model-api-reference/more-about-models/get-temporary-file-url.md) - [DashScope SDK连接复用配置](raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) + - [生成临时API Key](raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - **视频生成** - **HappyHorse** - [HappyHorse-文生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) - - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) + - [HappyHorse-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) - **万相** - **万相-早期视频模型（2.1-2.6）** - - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-图生视频-基于首帧API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) + - [万相-文生视频API参考（2.1-2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) - - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) + - [万相2.7-文生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-参考生视频API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) - - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - [万相2.7-视频编辑API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) + - [万相-图生动作API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - [万相-视频换人API参考](raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) - [万相-数字人](raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) - **人像驱动** @@ -1084,40 +1071,83 @@ - [视频口型替换-声动人像VideoRetalk](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [图生表情包视频-表情包Emoji](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) - [视频风格重绘API参考](raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) + - **可灵** + - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - **爱诗** - [爱诗-文生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) - - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) + - [爱诗-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-参考生视频API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - [爱诗-视频对口型API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) - - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - [爱诗-视频动作模仿API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) - - **可灵** - - [可灵-视频生成API文档](raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) + - [爱诗-视频超清API参考](raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - **Vidu** - [Vidu-文生视频API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) - - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) - [Vidu-图生视频-基于首帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) + - [Vidu-图生视频-基于首尾帧API参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) + - [Vidu-参考生视频 API 参考](raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - **音频** + - **语音合成** + - **实时语音合成（Qwen-TTS-Realtime）** + - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) + - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) + - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) + - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) + - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) + - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** + - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) + - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) + - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) + - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) + - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) + - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) + - **实时语音合成（Sambert）** + - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) + - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) + - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) + - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) + - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) + - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) + - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) + - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** + - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) + - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) + - **非实时语音合成（MiniMax）** + - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) + - **声音复刻** + - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) + - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) + - [声音复刻Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md) + - [非实时语音合成（Qwen-TTS）API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md) + - [声音设计API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/voice-design-api-references.md) - **语音识别** - - **实时语音识别（Qwen-ASR-Realtime）** - - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) - - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) - - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) - - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) - - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) - **实时语音识别（Fun-ASR）** - - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) - [Fun-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-websocket-api.md) + - [实时语音识别（Fun-ASR）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-client-events.md) - [Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-python-sdk.md) - [实时语音识别（Fun-ASR）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-server-events.md) - [Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/fun-asr-realtime-java-sdk.md) - - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) - [Fun-ASR实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/ios-sdk-for-fun-asr-real-time-service.md) + - [Fun-ASR实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-real-time-speech-recognition-api-reference/android-sdk-for-fun-asr-real-time-service.md) + - **实时语音识别（Qwen-ASR-Realtime）** + - [实时语音识别（Qwen-ASR-Realtime）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-client-events.md) + - [实时语音识别（Qwen-ASR-Realtime）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-server-events.md) + - [Qwen-ASR实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-interaction-process.md) + - [实时语音识别（Qwen-ASR-Realtime）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-java-sdk.md) + - [实时语音识别（Qwen-ASR-Realtime）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-realtime-api/qwen-asr-realtime-python-sdk.md) + - **实时语音识别（Paraformer）** + - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) + - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) + - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) + - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) + - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) + - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) + - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - **非实时语音识别（Fun-ASR）** - - [Fun-ASR非实时语音识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - [Fun-ASR非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/funauidio-asr-recorded-speech-recognition-python-sdk.md) + - [Fun-ASR非实时语音识别HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-http-api.md) - [Fun-ASR非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-java-sdk.md) - [Fun-ASR非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-android-sdk.md) - [Fun-ASR非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/fun-asr-recorded-speech-recognition-api-reference/fun-asr-recorded-speech-recognition-ios-sdk.md) @@ -1126,66 +1156,23 @@ - [Paraformer非实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-python-sdk.md) - [Paraformer非实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-java-sdk.md) - [Paraformer非实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-android-sdk.md) - - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) - [Paraformer非实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-recorded-speech-recognition-ios-sdk.md) + - [最佳实践](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-recorded-speech-recognition-api-reference/paraformer-best-practices.md) - **定制热词** - [定制热词HTTP API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-http-api.md) - [定制热词Java SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-java-sdk.md) - [定制热词Python SDK参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/custom-hot-words/vocabulary-python-sdk.md) - - **实时语音识别（Paraformer）** - - [Paraformer实时语音识别WebSocket API](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/websocket-for-paraformer-real-time-service.md) - - [实时语音识别（Paraformer）客户端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-client-events.md) - - [Paraformer实时语音识别Python SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-python-sdk.md) - - [实时语音识别（Paraformer）服务端事件](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-server-events.md) - - [Paraformer实时语音识别iOS SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/ios-sdk-for-paraformer-real-time-service.md) - - [Paraformer实时语音识别Java SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/paraformer-real-time-speech-recognition-java-sdk.md) - - [Paraformer实时语音识别Android SDK](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/paraformer-real-time-speech-recognition-api-reference/android-sdk-for-paraformer-real-time-service.md) - [非实时语音识别（Fun-ASR-Flash）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-flash.md) - [非实时语音识别（Fun-ASR-Realtime）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/non-real-time-speech-recognition-for-fun-asr-realtime.md) - [非实时语音识别（Qwen-ASR）API参考](raw/model-api-reference/audio-api-references/speech-recognition-api-reference/qwen-asr-api-reference.md) - - **语音合成** - - **实时语音合成（Qwen-Audio-TTS/CosyVoice）** - - [Qwen-Audio-TTS/CosyVoice WebSocket API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-websocket-api.md) - - [Qwen-Audio-TTS/CosyVoice服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-server-events.md) - - [Qwen-Audio-TTS/CosyVoice客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-client-events.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-java-sdk.md) - - [实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-python-sdk.md) - - [语音合成Qwen-Audio-TTS/CosyVoice Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-android-sdk.md) - - [语音合成Qwen-Audio-TTS/CosyVoice iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/cosyvoice-large-model-for-speech-synthesis/cosyvoice-ios-sdk.md) - - **实时语音合成（Qwen-TTS-Realtime）** - - [Qwen-TTS-Realtime WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/interactive-process-of-qwen-tts-realtime-synthesis.md) - - [客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-client-events.md) - - [Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-python-sdk.md) - - [服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-server-events.md) - - [Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-realtime-api-reference/qwen-tts-realtime-java-sdk.md) - - **实时语音合成（Sambert）** - - [Sambert客户端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-client-events.md) - - [Sambert WebSocket API 参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-websocket-api.md) - - [语音合成Sambert Python SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-python-sdk.md) - - [Sambert服务端事件](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-server-events.md) - - [语音合成Sambert Java SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-java-sdk.md) - - [语音合成Sambert Android SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-android-sdk.md) - - [语音合成Sambert iOS SDK](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sambert-speech-synthesis/sambert-ios-sdk.md) - - **非实时语音合成（Qwen-Audio-TTS/CosyVoice）** - - [非实时语音合成Qwen-Audio-TTS/CosyVoice HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-http-api.md) - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-python-sdk.md) - - [非实时语音合成Qwen-Audio-TTS/CosyVoice Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/non-realtime-cosyvoice-api/cosyvoice-tts-java-sdk.md) - - **非实时语音合成（MiniMax）** - - [MiniMax同步语音合成API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/minimax-speech-synthesis/minimax-synchronous-speech-synthesis-api.md) - - **声音复刻** - - [声音复刻HTTP API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md) - - [声音复刻Java SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-java-sdk.md) - - [声音复刻Python SDK参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md) - - [非实时语音合成（Qwen-TTS）API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/qwen-tts-api.md) - - [声音设计API参考](raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/voice-design-api-references.md) - **音乐生成** - [音乐生成Fun-Music API参考](raw/model-api-reference/audio-api-references/music-generation-references/fun-music-api.md) - **语音翻译** - **实时音视频翻译（Qwen-Livetranslate-Realtime）** - [客户端事件](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/live-translator-client-events.md) - [服务端事件](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/live-translator-server-events.md) - - [实时音视频翻译（Qwen-LiveTranslate）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-python-sdk.md) - [实时音视频翻译（Qwen-LiveTranslate）Java SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-java-sdk.md) + - [实时音视频翻译（Qwen-LiveTranslate）Python SDK-API参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/live-translator-api/qwen-livetranslate-python-sdk.md) - [音视频翻译-通义千问 API 参考](raw/model-api-reference/audio-api-references/speech-translation-api-reference/qwen3-livetranslate-flash-api.md) - **语音对话** - **实时语音对话** @@ -1205,32 +1192,30 @@ ## 应用 API 参考 +- **应用调用** + - **DashScope API** + - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) + - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) + - **Responses API** + - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) + - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) + - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) - **Managed Agents** - [API 总览与认证](raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) - - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Agent](raw/application-api-reference/managed-agents-api/agent-api.md) + - [快速开始](raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Session and Event](raw/application-api-reference/managed-agents-api/session-api.md) - - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - [File](raw/application-api-reference/managed-agents-api/files-api.md) + - [Environment](raw/application-api-reference/managed-agents-api/environment-api.md) - [Skill](raw/application-api-reference/managed-agents-api/skills-api.md) -- **应用调用** - - **Responses API** - - [异步调用API参考](raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - - [同步调用 API 参考](raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) - - **DashScope API** - - [新版智能体应用 API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - - [应用 DashScope API 参考](raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) - - [获取APP ID和Workspace ID](raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) -- **框架** - - **Spring AI Alibaba** - - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) - - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) +- **长期记忆** + - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - **应用组件** - **API目录** - **数据连接（原应用数据）** - - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [DeleteCategory - 删除类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) + - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) + - [ListCategory - 类目列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [ApplyFileUploadLease - 申请文件上传租约](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) @@ -1238,64 +1223,64 @@ - [DescribeFile - 查询文件状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) - - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) - - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) + - [GetAvailableParserTypes - 获取文件支持的解析器类型](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ChangeParseSetting - 修改类目解析设置](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) - - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) + - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) + - [DeleteFile - 删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) + - [DeleteFiles - 批量删除文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [AddConnector - 新增连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) + - [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [GetConnector - 获取连接器信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [UpdateConnector - 编辑连接器](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) - - [AddCategory - 新增类目](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) - - [AddTable - 添加表格](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) - **Prompt工程** - [CreatePromptTemplate - 创建Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetPromptTemplate - 获取Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [UpdatePromptTemplate - 更新Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - [DeletePromptTemplate - 删除Prompt模板](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [ListPromptTemplates - 获取Prompt模板列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) + - **知识库** + - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) + - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) + - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) + - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) + - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) + - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) + - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) + - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) + - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) + - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) + - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) + - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) + - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) + - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) + - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - **其他** - **长期记忆（旧）** - [CreateMemory - 创建长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - [UpdateMemory - 更新长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) - - [GetMemory - 获取长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [DeleteMemory - 删除长期记忆体](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) - [ListMemories - 获取长期记忆体列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) - - [CreateMemoryNode - 创建记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) + - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [UpdateMemoryNode - 更新记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) - - [GetMemoryNode - 获取记忆片段](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [GetAlipayTransferStatus - 查询支付宝打赏状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) - [GetAlipayUrl - 获取支付宝打赏URL](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) - [ApplyTempStorageLease - 申请临时文件上传许可](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) - [AddChunk - 新增切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md) - - **知识库** - - [CreateIndex - 创建知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) - - [SubmitIndexAddDocumentsJob - 提交知识库追加任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) - - [GetIndexJobStatus - 查询知识库创建任务状态](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) - - [Retrieve - 检索知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) - - [ListIndexDocuments - 查询知识库下的文件列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) - - [ListIndexFileDetails - 查询知识库下的文件详情](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - - [UpdateIndex - 更新知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) - - [DeleteIndexDocument - 删除知识库下的文件](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - - [DeleteIndex - 删除知识库](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) - - [ListIndices - 查询知识库列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - - [ListChunks - 查询索引下的分片列表](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - - [UpdateChunk - 修改切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - - [DeleteChunk - 删除切片](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - - [GetIndexMonitor - 获取知识库监控数据](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) - - [SubmitIndexJob - 提交知识库创建任务](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) - - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [API概览](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) + - [服务接入点](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) - [版本说明](raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) +- **框架** + - **Spring AI Alibaba** + - [通过Spring AI Alibaba检索阿里云百炼知识库](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) + - [使用Spring AI Alibaba集成阿里云百炼大模型应用](raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) + - [通过LlamaIndex API构建RAG应用](raw/application-api-reference/frameworks/llamaindex.md) - **更多** - [生成临时API Key](raw/application-api-reference/more/application-obtain-temporary-authentication-token.md) - [服务关联角色](raw/application-api-reference/more/bailian-service-linked-role.md) - [知识库SearchFilters](raw/application-api-reference/more/how-to-use-search-filters.md) -- **长期记忆** - - [长期记忆（新）API 参考](raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) - [知识检索与问答](raw/application-api-reference/knowledge.md) diff --git a/skills/bailian-docs-llm-wiki/models/families.jsonl b/skills/bailian-docs-llm-wiki/models/families.jsonl index a312928d..3af3b0ac 100644 --- a/skills/bailian-docs-llm-wiki/models/families.jsonl +++ b/skills/bailian-docs-llm-wiki/models/families.jsonl @@ -59,6 +59,7 @@ {"slug":"qwen-audio-tts","name":"Qwen-Audio-TTS","description":"Qwen-Audio-TTS是一款面向实时交互场景和高质量语音生成场景的语音合成大模型。模型支持多种小语种和中文方言，并具有Free-style 指令遵循能力、Context能力和细粒度标签控制能力，可更灵活地控制情绪、语气、角色、语速、音量等表达方式。","primaryCapability":"Realtime-Text-to-Speech","capabilities":["Realtime-Text-to-Speech"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-audio-3.0-tts-flash","name":"qwen-audio-3.0-tts-flash","capabilities":["Realtime-Text-to-Speech"]},{"model":"qwen-audio-3.0-tts-plus","name":"qwen-audio-3.0-tts-plus","capabilities":["Realtime-Text-to-Speech"]}],"detailPath":"groups/qwen-audio-tts.json"} {"slug":"qwen-coder-plus","name":"Qwen-Coder-Plus","description":"千问系列代码及编程模型是专门用于编程和代码生成的语言模型，性能出色，效果突出。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-plus","name":"Qwen-Coder-Plus","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-plus.json","maxContextWindow":131072} {"slug":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","description":"Qwen-Coder-Turbo模型是专门用于编程和代码生成的语言模型，推理速度快，成本低。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-coder-turbo","name":"Qwen-Coder-Turbo","contextWindow":131072,"capabilities":["TG"]}],"detailPath":"groups/qwen-coder-turbo.json","maxContextWindow":131072} +{"slug":"qwen-deep-research","name":"qwen-deep-research","description":"千问深入研究是一款面向复杂研究任务的高级智能体系统，具备多轮推理与全局规划能力，能够运用互联网搜索等多种工具，对任务进行精细化拆解，开展推理与分析，最终为用户生成可溯源、逻辑严谨的研究型报告。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-deep-research","name":"qwen-deep-research","contextWindow":1000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-deep-research.json","maxContextWindow":1000000} {"slug":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","description":"快速对文档进行精准信息抽取，打标分类，内容审核及摘要总结。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen-doc-turbo","name":"Qwen-Doc-Turbo","contextWindow":262144,"capabilities":["TG"]}],"detailPath":"groups/qwen-doc-turbo.json","maxContextWindow":262144} {"slug":"qwen-embedding","name":"Qwen-Embedding","description":"基于Qwen模型基座训练的多语言文本统一向量模型，文本检索、聚类、分类性能大幅提升，多语言支持，适用于向量检索、向量化等等场景，可搭配检索增强、文档处理场景使用，支持64~2048维用户自定义向量维度。","primaryCapability":"TR","capabilities":["TR"],"providers":["qwen","qwen-domain-model"],"itemCount":7,"items":[{"model":"qwen3.7-text-embedding","name":"Qwen3.7-通用文本向量","contextWindow":131072,"capabilities":["TR"]},{"model":"text-embedding-async-v1","name":"通用文本向量-async-v1","capabilities":["TR"]},{"model":"text-embedding-async-v2","name":"通用文本向量-async-v2","capabilities":["TR"]},{"model":"text-embedding-v1","name":"通用文本向量-v1","capabilities":["TR"]},{"model":"text-embedding-v2","name":"通用文本向量-v2","capabilities":["TR"]},{"model":"text-embedding-v3","name":"通用文本向量-v3","capabilities":["TR"]},{"model":"text-embedding-v4","name":"通用文本向量-v4","capabilities":["TR"]}],"detailPath":"groups/qwen-embedding.json","maxContextWindow":131072} {"slug":"qwen-flash-character","name":"Qwen-Flash-Character","description":"千问系列多语言角色扮演模型，本模型是动态更新版本，模型更新会提前通知，适合拟人化的角色扮演，同时优化了限定人设指令遵循、话题推进、倾听共情等能力，支持个性化角色的深度还原。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-flash-character","name":"Qwen-Flash-Character","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/qwen-flash-character.json","maxContextWindow":8192} @@ -68,6 +69,7 @@ {"slug":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","description":"内容丰实：支持最大 4.5k token 输入，支持图中图密集信息排版，让报纸、分镜、菜单、试卷等复杂版面一次生成。\n细节真实：支持 10px 小字精准渲染，微表情、毛孔、发丝等细节生动还原，逼近真实摄影的质感。\n知识厚实：支持 12 国语言、20+ 字体原生渲染，主流网页、游戏、直播等界面仿真，外部知识全纳入。\nQwen-Image-3.0-Pro 不只是在追求\"好看\"，更在追求**“好用”**——让图像生成真正成为可落地的生产力工具。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-3.0-pro","name":"Qwen-Image-3.0-Pro","capabilities":["IG"]}],"detailPath":"groups/qwen-image-3.0-pro.json"} {"slug":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","description":"千问图像编辑模型Max系列，提供更稳定、更丰富的编辑能力：提升工业设计与几何推理能力；提升角色一致性；减轻偏移问题；集成Lora能力，可以进行更多功能的图像编辑。此版本为2026年1月16日快照。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-edit-max","name":"Qwen-Image-Edit-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit-max.json"} {"slug":"qwen-image-edit","name":"Qwen-Image-Edit-Plus","description":"千问系列图像编辑Plus模型，在首版Edit模型基础上进一步优化了推理性能与系统稳定性，大幅缩短图像生成与编辑的响应时间；支持单次请求返回多张图片，显著提升用户体验。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image-edit","name":"Qwen-Image-Edit","capabilities":["IG"]},{"model":"qwen-image-edit-plus","name":"Qwen-Image-Edit-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-edit.json"} +{"slug":"qwen-image-max","name":"Qwen-Image-Max","description":"千问图像生成模型Max系列，在各类生成任务中表现出色，相较Plus系列大幅度降低生成图片的AI感，提升图像真实性；具备更真实的人物质感、更细腻的自然纹理、更美观的文字渲染。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-image-max","name":"Qwen-Image-Max","capabilities":["IG"]}],"detailPath":"groups/qwen-image-max.json"} {"slug":"qwen-image-plus","name":"Qwen-Image-Plus","description":"千问系列图像生成模型，参数规模200亿。具备卓越的文本渲染能力，在复杂文本渲染、各类生成与编辑任务重表现出色，在多个公开基准测试中获得SOTA，模型性能大幅提升。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen-image","name":"Qwen-Image","capabilities":["IG"]},{"model":"qwen-image-plus","name":"Qwen-Image-Plus","capabilities":["IG"]}],"detailPath":"groups/qwen-image-plus.json"} {"slug":"qwen-long","name":"Qwen-Long","description":"Qwen-Long是在通义实验室针对超长上下文处理场景的大语言模型，支持中文、英文等不同语言输入，支持最长1000万tokens(约1500万字或1.5万页文档)的超长上下文对话。配合同步上线的文档服务，可支持文本文件（ TXT、DOCX、PDF、XLSX、EPUB、MOBI、MD、CSV）和图片文件（BMP、PNG、JPG/JPEG、GIF 以及PDF扫描件）的解析和对话。说明：通过HTTP直接提交请求，支持1M tokens长度，超过此长度建议通过文件方式提交。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":2,"items":[{"model":"qwen-long","name":"Qwen-Long","contextWindow":10000000,"capabilities":["TG"]},{"model":"qwen-long-latest","name":"Qwen-Long-Latest","contextWindow":10000000,"capabilities":["TG"]}],"detailPath":"groups/qwen-long.json","maxContextWindow":10000000} {"slug":"qwen-math-plus","name":"Qwen-Math-Plus","description":"Qwen-Math-Plus模型具有强大的数学解题能力,擅长处理中英文数学题，包括方程、计算、证明等方向。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":4,"items":[{"model":"qwen-math-plus","name":"Qwen-Math-Plus","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0816","name":"Qwen-Math-Plus-2024-08-16","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-0919","name":"Qwen-Math-Plus-2024-09-19","contextWindow":4096,"capabilities":["TG"]},{"model":"qwen-math-plus-latest","name":"Qwen-Math-Plus-Latest","contextWindow":4096,"capabilities":["TG"]}],"detailPath":"groups/qwen-math-plus.json","maxContextWindow":4096} @@ -89,6 +91,7 @@ {"slug":"qwen-vl-embedding","name":"Qwen-VL-Embedding","description":"基于Qwen-VL底座训练的统一多模态向量模型，支持文本、图片、视频单模态/混合模态输入，输出统一表征向量，适用于跨模态检索、图搜、视频检索、图像聚类、复杂多模态信息检索、打标等场景。","primaryCapability":"ME","capabilities":["ME"],"providers":["qwen"],"itemCount":2,"items":[{"model":"qwen2.5-vl-embedding","name":"Qwen2.5-VL-Embedding","capabilities":["ME"]},{"model":"qwen3-vl-embedding","name":"Qwen3-VL-Embedding","capabilities":["ME"]}],"detailPath":"groups/qwen-vl-embedding.json"} {"slug":"qwen-vl-max","name":"Qwen-VL-Max","description":"Qwen-VL-Max，即千问超大规模视觉语言模型。相比增强版，再次提升视觉推理能力和指令遵循能力，提供更高的视觉感知和认知水平。在更多复杂任务上提供最佳的性能。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-vl-max","name":"Qwen-VL-Max","contextWindow":131072,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-max.json","maxContextWindow":131072} {"slug":"qwen-vl-ocr","name":"Qwen-VL-OCR","description":"Qwen-VL-OCR，即基于Qwen-VL训练的OCR识别大模型。通过统一模型的方式聚合多种图文识别、解析、处理类任务，提供强大的图文识别能力。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen-domain-model"],"itemCount":3,"items":[{"model":"qwen-vl-ocr","name":"QwenVL-OCR","contextWindow":38192,"capabilities":["VU"]},{"model":"qwen-vl-ocr-1028","name":"QwenVL-OCR-2024-10-28","contextWindow":34096,"capabilities":["VU"]},{"model":"qwen-vl-ocr-latest","name":"QwenVL-OCR-Latest","contextWindow":38192,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-ocr.json","maxContextWindow":38192} +{"slug":"qwen-vl-plus","name":"Qwen-VL-Plus","description":"Qwen-VL-Plus，即千问大规模视觉语言模型增强版。大幅提升细节识别能力和文字识别能力，支持超百万像素分辨率和任意长宽比规格的图像。在广泛的视觉任务上提供卓越的性能。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-vl-plus","name":"QwenVL-Plus","contextWindow":131072,"capabilities":["VU"]}],"detailPath":"groups/qwen-vl-plus.json","maxContextWindow":131072} {"slug":"qwen-voice-design","name":"Qwen-声音设计","description":"Qwen-Voice-Design模型是千问语音模型的声音设计系列模型，仅需输入简单的文字描述，即可迅速设计出符合要求的相关声音。结合qwen3-tts-vd-realtime模型使用，可设计输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-design","name":"Qwen-Voice-Design","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-design.json"} {"slug":"qwen-voice-enrollment","name":"Qwen-声音复刻","description":"千问voice-enrollment模型是千问语音模型的声音复刻系列模型，仅需5s以上的音频，即可迅速复刻高相似度声音。结合qwen3-tts-vc-realtime模型使用，可将一个人的声音高保真复刻，输出11个语种的语音。且合成音频可以根据文本自适应调节语气，对复杂文本合成也有较好的处理能力。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen-voice-enrollment","name":"Qwen-Voice-Enrollment","capabilities":["TTS"]}],"detailPath":"groups/qwen-voice-enrollment.json"} {"slug":"qwen2.5","name":"Qwen2.5-开源模型","description":"Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen2.5-omni-7b","name":"Qwen2.5-Omni-7B","contextWindow":32768,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen2.5.json","maxContextWindow":32768} @@ -119,7 +122,6 @@ {"slug":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","description":"Qwen3.5-LiveTranslate-Flash的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3.5-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，通义千问3.5-LiveTranslate-Flash 实现了离线和实时两种音视频翻译能力，能听懂60种语言，会说29种语言。","primaryCapability":"Realtime-Audio-Translate","capabilities":["Realtime-Audio-Translate"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","contextWindow":53248,"capabilities":["Realtime-Audio-Translate"]}],"detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json","maxContextWindow":53248} {"slug":"qwen3.5-ocr","name":"Qwen3.5-OCR","description":"Qwen3.5系列OCR模型，在文档解析、文本定位、关键信息提取等方面全面升级，在真实场景的业务卡证（如国内国际身份证、驾驶证等业务场景）抽取效果显著提升。","primaryCapability":"VU","capabilities":["VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","contextWindow":65536,"capabilities":["VU"]}],"detailPath":"groups/qwen3.5-ocr.json","maxContextWindow":65536} {"slug":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash-realtime.json","maxContextWindow":262144} -{"slug":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-flash.json","maxContextWindow":262144} {"slug":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话，WebSearch和复杂FunctionCall的调用，并且具备智能语义打断的交互能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态交互体验。","primaryCapability":"Realtime-Omni","capabilities":["Realtime-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","contextWindow":262144,"capabilities":["Realtime-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus-realtime.json","maxContextWindow":262144} {"slug":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","description":"Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。","primaryCapability":"Multimodal-Omni","capabilities":["Multimodal-Omni"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","contextWindow":262144,"capabilities":["Multimodal-Omni"]}],"detailPath":"groups/qwen3.5-omni-plus.json","maxContextWindow":262144} {"slug":"qwen3.5-plus","name":"Qwen3.5-Plus","description":"Qwen3.5原生视觉语言系列Plus模型，展现出与当前顶尖前沿模型相媲美的卓越性能，模型效果在纯文本与多模态方面相较3系列均实现飞跃式进步。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["qwen"],"itemCount":1,"items":[{"model":"qwen3.5-plus","name":"Qwen3.5-Plus","contextWindow":1000000,"capabilities":["TG","Reasoning","VU"]}],"detailPath":"groups/qwen3.5-plus.json","maxContextWindow":1000000} @@ -140,10 +142,12 @@ {"slug":"tongyi-intent-detect-v3","name":"意图分类模型","description":"意图识别和槽位填充是对话系统中的基础任务。本模型实现了一个基于 API的意图（intent）和槽位参数（slots）联合预测。在一次模型输出中，同时完成多个指令API的返回和槽位参数的填充。返回的结果为标准json格式。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"tongyi-intent-detect-v3","name":"意图分类模型","contextWindow":8192,"capabilities":["TG"]}],"detailPath":"groups/tongyi-intent-detect-v3.json","maxContextWindow":8192} {"slug":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","description":"通义晓蜜-对话分析-flash是专注于日常任务，如对话信息抽取、场景分类等分析类需求的模型，自定义分析标准遵循与对话语义理解能力显著提升，适用于低时延的离线在线分析任务。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-flash.json","maxContextWindow":32768} {"slug":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","description":"通义晓蜜-对话分析-pro是专注于高阶复杂分析，如针对具备复杂业务逻辑的复杂质检规则等分析需求的模型，支持自定义更细粒度的分析标准，具备更强的多轮上下文建模、深层语义理解与推理能力。","primaryCapability":"TG","capabilities":["TG"],"providers":["qwen-domain-model"],"itemCount":1,"items":[{"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","contextWindow":32768,"capabilities":["TG"]}],"detailPath":"groups/tongyi-xiaomi-analysis-pro.json","maxContextWindow":32768} +{"slug":"tripo-models-market-place","name":"Tripo","description":"AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。","primaryCapability":"3D-generation","capabilities":["3D-generation"],"providers":["tripo"],"itemCount":2,"items":[{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","capabilities":["3D-generation"]},{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","capabilities":["3D-generation"]}],"detailPath":"groups/tripo-models-market-place.json"} {"slug":"vanchin-models-market-place","name":"Vanchin DeepSeek","description":"由快手万擎提供的DeepSeek系列模型API服务。","primaryCapability":"TG","capabilities":["TG","Reasoning","VU"],"providers":["deepseek"],"itemCount":6,"items":[{"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","contextWindow":8192,"capabilities":["VU","TG"]},{"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","contextWindow":131072,"capabilities":["TG"]},{"model":"vanchin/deepseek-v3.1-terminus","name":"Vanchin/DeepSeek-V3.1-Terminus","contextWindow":131072,"capabilities":["TG","Reasoning"]},{"model":"vanchin/deepseek-v3.2-think","name":"Vanchin/DeepSeek-V3.2-think","contextWindow":131072,"capabilities":["Reasoning","TG"]},{"model":"vanchin/deepseek-v4-pro","name":"vanchin/deepseek-v4-pro","contextWindow":1048576,"capabilities":["TG"]}],"detailPath":"groups/vanchin-models-market-place.json","maxContextWindow":1048576} {"slug":"video-style-transform","name":"视频风格重绘","description":"视频风格重绘可以将输入的视频帧序列进行多种风格化的重绘/生成，使新视频画面在兼顾原始人物和物体相貌的同时，带来不同风格的绘画效果。当前支持预置重绘风格包括日式漫画、美式漫画、清新漫画、3D卡通、国风卡通。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"video-style-transform","name":"视频风格重绘","capabilities":["VG"]}],"detailPath":"groups/video-style-transform.json"} {"slug":"videoretalk","name":"声动人像VideoRetalk","description":"VideoRetalk是一个人物视频生成模型，可基于人物视频和人声音频，生成人物讲话口型与输入音频相匹配的新视频。","primaryCapability":"VG","capabilities":["VG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"videoretalk","name":"声动人像VideoRetalk","capabilities":["VG"]}],"detailPath":"groups/videoretalk.json"} {"slug":"vidu-image-models-market-place","name":"Vidu AI生图","description":"由生数科技提供Vidu系列图片生成API服务，多图参考，精准还原，高速高质。","primaryCapability":"IG","capabilities":["IG"],"providers":["vidu"],"itemCount":4,"items":[{"model":"vidu/vidu-image_reference2image","name":"Vidu-image_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-fast_reference2image","name":"ViduQ2-fast_reference2image","capabilities":["IG"]},{"model":"vidu/viduq2-pro_reference2image","name":"ViduQ2-Pro_reference2image","capabilities":["IG"]},{"model":"vidu/viduq3-fast_reference2image","name":"ViduQ3-fast_reference2image","capabilities":["IG"]}],"detailPath":"groups/vidu-image-models-market-place.json"} +{"slug":"vidu-models-market-place","name":"Vidu AI生视频","description":"由生数科技提供Vidu系列视频生成API服务，电影级画质、一致性保持、精准可控。","primaryCapability":"VG","capabilities":["VG"],"providers":["vidu"],"itemCount":20,"items":[{"model":"vidu/viduq2_reference2video","name":"ViduQ2_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2_text2video","name":"ViduQ2_text2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_img2video","name":"ViduQ2-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_reference2video","name":"ViduQ2-Pro_reference2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro_start-end2video","name":"ViduQ2-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq2-pro-fast_img2video","name":"ViduQ2-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_img2video","name":"ViduQ2-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq2-turbo_start-end2video","name":"ViduQ2-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3_reference2video","name":"ViduQ3_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-ad_reference2video","name":"ViduQ3-Ad_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-drama_reference2video","name":"ViduQ3-Drama_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-mix_reference2video","name":"ViduQ3-mix_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_img2video","name":"ViduQ3-Pro_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_start-end2video","name":"ViduQ3-Pro_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro_text2video","name":"ViduQ3-Pro_text2video","capabilities":["VG"]},{"model":"vidu/viduq3-pro-fast_img2video","name":"ViduQ3-Pro-fast_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_img2video","name":"ViduQ3-Turbo_img2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_reference2video","name":"ViduQ3-Turbo_reference2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_start-end2video","name":"ViduQ3-Turbo_start-end2video","capabilities":["VG"]},{"model":"vidu/viduq3-turbo_text2video","name":"ViduQ3-Turbo_text2video","capabilities":["VG"]}],"detailPath":"groups/vidu-models-market-place.json"} {"slug":"virtualmodel-v2","name":"虚拟模特V2","description":"虚拟模特可以对上传的真人或者人台实拍商品展示图进行智能生成，将其中的模特和背景替换为心仪的内容，在保持人物姿态不变的情况下，使用虚拟模特对商品进行更加精美、多样的展示。支持各种与模特产生互动的商品，如手持小商品、服装、鞋靴、配饰等。","primaryCapability":"IG","capabilities":["IG"],"providers":["qwen"],"itemCount":1,"items":[{"model":"virtualmodel-v2","name":"虚拟模特V2","capabilities":["IG"]}],"detailPath":"groups/virtualmodel-v2.json"} {"slug":"voice-enrollment","name":"大模型声音复刻及声音设计","description":"大模型声音复刻服务依托先进的大模型技术进行特征提取，无需训练过程就可以完成声音的复刻。仅需提供极短的音频，即可迅速生成高度相似且听感自然的定制声音。\n大模型声音设计使用FunAudioGen-VD模型，支持通过文本Prompt描述，创造声音。无需受限任何音频质量，根据目标场景对音色、语气、语调、语速、情绪等各方面表现力的需求描述，即可生成高质量语音。高度还原专业配音演员的演出水准。","primaryCapability":"TTS","capabilities":["TTS"],"providers":["qwen"],"itemCount":1,"items":[{"model":"voice-enrollment","name":"大模型声音复刻及声音设计","capabilities":["TTS"]}],"detailPath":"groups/voice-enrollment.json"} {"slug":"wan-image-edit","name":"Wan-Image","description":"指令编辑图片内容，轻松实现局部修改、风格变化、一致性保持等","primaryCapability":"IG","capabilities":["IG"],"providers":["wan"],"itemCount":5,"items":[{"model":"wan2.5-i2i-preview","name":"Wan2.5-I2I-Preview","capabilities":["IG"]},{"model":"wan2.6-image","name":"Wan2.6-Image","capabilities":["IG"]},{"model":"wan2.7-image","name":"Wan2.7-Image","capabilities":["IG"]},{"model":"wan2.7-image-pro","name":"Wan2.7-Image-Pro","capabilities":["IG"]},{"model":"wanx2.1-imageedit","name":"Wan2.1-ImageEdit","capabilities":["IG"]}],"detailPath":"groups/wan-image-edit.json"} diff --git a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json index f2a97249..9d280185 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/Kimi-K2.json @@ -28,30 +28,35 @@ { "priceUnit": "每百万tokens", "price": "6.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "27", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "8.125", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.65", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -74,6 +79,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "VU", @@ -130,30 +138,35 @@ { "priceUnit": "每百万tokens", "price": "6.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "27", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "8.125", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.65", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -176,6 +189,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", @@ -231,30 +247,35 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "21", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -277,6 +298,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", @@ -329,18 +353,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -363,6 +390,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -420,18 +450,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -454,6 +487,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json index 9531a790..1297226e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-M2.1.json @@ -23,18 +23,21 @@ { "priceUnit": "每百万tokens", "price": "2.1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.42", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -57,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -109,22 +115,28 @@ { "priceUnit": "每百万tokens", "price": "2.1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.42", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" diff --git a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json index f1d04dde..f52de10f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/MiniMax-speech-market-place.json @@ -20,12 +20,14 @@ { "priceUnit": "每次", "price": "9.9", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" }, { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -48,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -84,12 +89,14 @@ { "priceUnit": "每次", "price": "9.9", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" }, { "priceUnit": "每万字符", "price": "3.5", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -112,6 +119,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -148,12 +158,14 @@ { "priceUnit": "每次", "price": "9.9", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" }, { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -176,6 +188,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -212,12 +227,14 @@ { "priceUnit": "每次", "price": "9.9", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" }, { "priceUnit": "每万字符", "price": "3.5", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -240,6 +257,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json index 4bacc726..ecab22ba 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-parsing-v1.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json index 9bec1d16..f9447f68 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-plus.json @@ -21,10 +21,14 @@ { "priceUnit": "每张", "price": "0.5", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json index 20c35639..cc41aae3 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon-refiner.json @@ -15,6 +15,9 @@ "features": [], "provider": "qwen", "model": "aitryon-refiner", + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -23,6 +26,7 @@ { "priceUnit": "每张", "price": "0.3", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -36,6 +40,7 @@ { "priceUnit": "每张", "price": "0.275", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -49,6 +54,7 @@ { "priceUnit": "每张", "price": "0.25", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -62,6 +68,7 @@ { "priceUnit": "每张", "price": "0.225", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -75,6 +82,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -88,6 +96,7 @@ { "priceUnit": "每张", "price": "0.175", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -101,6 +110,7 @@ { "priceUnit": "每张", "price": "0.15", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json index de21e8c9..aac566d9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/aitryon.json +++ b/skills/bailian-docs-llm-wiki/models/groups/aitryon.json @@ -21,10 +21,14 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json index bbdba13c..a52c8c95 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-detect-gen2.json @@ -17,10 +17,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json index 0242409f..37f73866 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-gen2.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.08", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json index 5d5b1624..d0c3f13b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/animate-anyone-template-gen2.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.08", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json index 33e35cc7..d7129c86 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json +++ b/skills/bailian-docs-llm-wiki/models/groups/cosyvoice.json @@ -20,6 +20,7 @@ { "priceUnit": "每万字符", "price": "0.8", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -72,6 +76,7 @@ { "priceUnit": "每万字符", "price": "1.5", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -88,6 +93,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -124,6 +132,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -140,6 +149,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -173,6 +185,7 @@ { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -189,6 +202,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -223,10 +239,14 @@ { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -259,10 +279,14 @@ { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -299,10 +323,14 @@ { "priceUnit": "每万字符", "price": "2", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json index ef3c42e5..c7e46d56 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/deepseek.json +++ b/skills/bailian-docs-llm-wiki/models/groups/deepseek.json @@ -19,26 +19,26 @@ "web-search" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v4-pro", "prices": [ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -61,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -77,17 +80,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-pro\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-pro',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-pro\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-pro\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-pro\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-pro\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-pro\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -110,26 +113,26 @@ "web-search" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v4-flash", "prices": [ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -152,6 +155,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -169,17 +175,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v4-flash\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v4-flash',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v4-flash\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v4-flash\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v4-flash\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -203,56 +209,61 @@ "batch" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v3.2", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -260,6 +271,7 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -283,6 +295,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -304,17 +319,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -336,20 +351,19 @@ "web-search" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v3.2-exp", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -372,6 +386,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -393,17 +410,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2-exp\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2-exp',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.2-exp\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.2-exp',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2-exp\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2-exp\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.2-exp\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.2-exp\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.2-exp\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -426,26 +443,26 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v3.1", "prices": [ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -468,6 +485,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -489,17 +509,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true,\n \"enable_thinking\": true\n}'", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3.1\", # 您可以按需更换为其它深度思考模型\n messages=messages,\n extra_body={\"enable_thinking\": True},\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化 openai 客户端\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY, // 从环境变量读取\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\nlet isAnswering = false;\nasync function main() {\n try {\n const messages = [{ role: 'user', content: '你是谁' }];\n const stream = await openai.chat.completions.create({\n model: 'deepseek-v3.1',\n messages,\n stream: true,\n enable_thinking: true\n });\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20));\n for await (const chunk of stream) {\n if (!chunk.choices.length) continue;\n const delta = chunk.choices[0].delta;\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n }\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20));\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n }\n }\n } catch (error) {\n console.error('Error:', error);\n }\n}\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3.1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"enable_thinking\": true,\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3.1\",\n messages=messages,\n result_format=\"message\",\n # 开启深度思考\n enable_thinking=True,\n)\n\nif response.status_code == 200:\n # 打印思考过程\n print(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n print(response.output.choices[0].message.reasoning_content)\n \n # 打印回复\n print(\"=\" * 20 + \"完整回复\" + \"=\" * 20)\n print(response.output.choices[0].message.content)\nelse:\n print(f\"HTTP返回码：{response.status_code}\")\n print(f\"错误码：{response.code}\")\n print(f\"错误信息：{response.message}\")", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation();\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3.1\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .enableThinking(true)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(\"====================思考过程====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getReasoningContent());\n System.out.println(\"\\n====================完整回复====================\");\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -522,38 +542,40 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-v3", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -576,6 +598,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -597,17 +622,17 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"deepseek-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-v3\",\n \"messages\": [\n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key=\"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\ncompletion = client.chat.completions.create(\n model=\"deepseek-v3\",\n messages=[\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n ],\n stream=True\n)\nfor chunk in completion:\n print(chunk.choices[0].delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nconst openai = new OpenAI(\n {\n // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: \"sk-xxx\",\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"deepseek-v3\",\n messages: [\n { role: \"system\", content: \"You are a helpful assistant.\" },\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true\n});\nfor await (const chunk of completion) {\n process.stdout.write(chunk.choices[0].delta.content);\n}", "docUrl": "https://help.aliyun.com/document_detail/3016807.html" } }, "dashscope": { "completionsAPI": { - "curl": "curl --location \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"deepseek-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", - "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", + "curl": "curl --location \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header \"Content-Type: application/json\" \\\n--data '{\n \"model\": \"deepseek-v3\",\n \"input\":{\n \"messages\":[ \n {\n \"role\": \"system\",\n \"content\": \"You are a helpful assistant.\"\n },\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\": {\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\n\ndashscope.base_http_api_url = 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1'\n\nmessages = [\n {\"role\": \"system\", \"content\": \"You are a helpful assistant.\"},\n {\"role\": \"user\", \"content\": \"你是谁？\"},\n]\nresponse = Generation.call(\n # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key = \"sk-xxx\",\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-v3\",\n messages=messages,\n result_format=\"message\",\n)\n\nprint(response.output.choices[0].message.content)", + "java": "import java.util.Arrays;\nimport java.lang.System;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport com.alibaba.dashscope.protocol.Protocol;\n\npublic class Main {\n public static GenerationResult callWithMessage() throws ApiException, NoApiKeyException, InputRequiredException {\n Generation gen = new Generation(Protocol.HTTP.getValue(), \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\");\n Message systemMsg = Message.builder()\n .role(Role.SYSTEM.getValue())\n .content(\"You are a helpful assistant.\")\n .build();\n Message userMsg = Message.builder()\n .role(Role.USER.getValue())\n .content(\"你是谁？\")\n .build();\n GenerationParam param = GenerationParam.builder()\n // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-v3\")\n .messages(Arrays.asList(systemMsg, userMsg))\n .resultFormat(GenerationParam.ResultFormat.MESSAGE)\n .build();\n return gen.call(param);\n }\n public static void main(String[] args) {\n try {\n GenerationResult result = callWithMessage();\n System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent());\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n System.err.println(\"错误信息：\"+e.getMessage());\n }\n }\n}", "docUrl": "https://help.aliyun.com/document_detail/3016809.html" } } @@ -629,20 +654,19 @@ "web-search" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1-0528", "prices": [ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -665,6 +689,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], @@ -685,16 +712,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-0528\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-0528',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-0528\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-0528',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-0528\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-0528\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-0528\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-0528\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-0528\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -716,38 +743,40 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1", "prices": [ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -770,6 +799,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], @@ -790,16 +822,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -818,20 +850,19 @@ "model-experience" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1-distill-qwen-7b", "prices": [ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -854,6 +885,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -874,16 +908,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-7b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-7b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-7b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-7b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-7b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-7b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -902,20 +936,19 @@ "model-experience" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1-distill-qwen-32b", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -938,6 +971,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -958,16 +994,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-32b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-32b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-32b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-32b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-32b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-32b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -986,20 +1022,19 @@ "model-experience" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1-distill-qwen-14b", "prices": [ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1022,6 +1057,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1042,16 +1080,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-14b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-14b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-14b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-14b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-14b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-14b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } @@ -1070,9 +1108,6 @@ "model-experience" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "deepseek-r1-distill-qwen-1.5b", "qpmInfo": { "model-default-actual": { @@ -1106,16 +1141,16 @@ "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-1.5b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'deepseek-r1-distill-qwen-1.5b',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-1.5b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"deepseek-r1-distill-qwen-1.5b\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"deepseek-r1-distill-qwen-1.5b\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"deepseek-r1-distill-qwen-1.5b\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/embedding.json b/skills/bailian-docs-llm-wiki/models/groups/embedding.json index fe9ab027..fc01efe1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/embedding.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], @@ -82,12 +87,14 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -110,6 +117,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], @@ -145,12 +155,14 @@ { "priceUnit": "每百万tokens", "price": "0.9", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -173,6 +185,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json index e4b79d5d..acc736db 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-detect-v1.json @@ -17,10 +17,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json index 09b61cd4..5be8940b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emo-v1.json @@ -19,16 +19,21 @@ { "priceUnit": "每秒", "price": "0.08", + "timeBand": "standard", "type": "video_duration_1-1", "priceName": "视频生成（1:1画幅视频）" }, { "priceUnit": "每秒", "price": "0.16", + "timeBand": "standard", "type": "video_duration_3-4", "priceName": "视频生成（3:4画幅视频）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json index e9a48c75..ba35481e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-detect-v1.json @@ -17,10 +17,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json index dd438fae..dec649c9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/emoji-v1.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.08", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json index 7bd76785..6add2e2c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json +++ b/skills/bailian-docs-llm-wiki/models/groups/facechain-generation.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.18", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json index b49111d1..3fa87312 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/farui-plus.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json index 7436007c..9315f829 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-flash.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json index 020604ef..69ca7592 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr-realtime.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.00033", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], @@ -69,6 +73,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -85,6 +90,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json index 14f06761..3cf5006c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-asr.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], @@ -68,6 +72,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -84,6 +89,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json index e6046966..90f25aac 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/fun-music.json +++ b/skills/bailian-docs-llm-wiki/models/groups/fun-music.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.002", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -70,6 +74,7 @@ { "priceUnit": "每秒", "price": "0.005", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -86,6 +91,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json index b8ca49e8..9b7f2c9d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-4.5.json @@ -24,18 +24,21 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -58,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -126,6 +132,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -134,30 +143,35 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -171,30 +185,35 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -270,6 +289,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -278,18 +300,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -303,18 +328,21 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "22", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -389,6 +417,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -397,18 +428,21 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "14", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -422,18 +456,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -512,6 +549,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -520,18 +560,21 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "14", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -545,18 +588,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -634,6 +680,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -642,12 +691,14 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "14", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -661,12 +712,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -743,6 +796,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -751,12 +807,14 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -770,12 +828,14 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json index 853a389d..b96913ed 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json +++ b/skills/bailian-docs-llm-wiki/models/groups/glm-fast.json @@ -20,26 +20,26 @@ "web-search" ], "provider": "zhipu-ai", - "limit": { - "message": "model not exist" - }, "model": "glm-5.2-fast-preview", "prices": [ { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "56", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -62,6 +62,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -78,46 +81,19 @@ "name": "GLM-5.2-Fast-Preview", "docUrl": "https://help.aliyun.com/document_detail/2974045.html", "category": "Third-party", - "predictConfig": [ - { - "name": "system", - "key": "systemMessage", - "tip": "系统人设，例如“你是一个AI助手”。" - }, - { - "name": "top_p", - "key": "top_p", - "default": 0.8, - "tip": "控制核采样方法的概率阈值，取值越大，生成的随机性越高。", - "range": [ - 0.0001, - 1 - ] - }, - { - "name": "temperature", - "key": "temperature", - "default": 0.7, - "tip": "控制生成随机性和多样性，数值越高多样性越强，数值越低一致性越强，范围(0,2)。建议该参数和top_p只设置1个。", - "range": [ - 0, - 1.9999 - ] - } - ], "samples": { "openai": { "default": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", - "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2-fast-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁\"\n }\n ],\n \"stream\": true\n}'", + "python": "from openai import OpenAI\nimport os\nclient = OpenAI(\n # 如果没有配置环境变量，请用阿里云百炼API Key替换：api_key=\"sk-xxx\"\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\nmessages = [{\"role\": \"user\", \"content\": \"你是谁\"}]\ncompletion = client.chat.completions.create(\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n stream=True\n)\nis_answering = False # 是否进入回复阶段\nprint(\"\\n\" + \"=\" * 20 + \"思考过程\" + \"=\" * 20)\nfor chunk in completion:\n if chunk.choices:\n delta = chunk.choices[0].delta\n # 只收集思考内容\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content is not None:\n if not is_answering:\n print(delta.reasoning_content, end=\"\", flush=True)\n # 收到content，开始进行回复\n if hasattr(delta, \"content\") and delta.content:\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(delta.content, end=\"\", flush=True)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\n// 初始化OpenAI客户端\nconst openai = new OpenAI({\n // 如果没有配置环境变量，请用阿里云百炼API Key替换：apiKey: \"sk-xxx\"\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nlet reasoningContent = ''; // 完整思考过程\nlet answerContent = ''; // 完整回复\nlet isAnswering = false; // 是否进入回复阶段\n\nasync function main() {\n const messages = [{ role: 'user', content: '你是谁' }];\n\n const stream = await openai.chat.completions.create({\n model: 'glm-5.2-fast-preview',\n messages,\n stream: true,\n });\n\n console.log('\\n' + '='.repeat(20) + '思考过程' + '='.repeat(20) + '\\n');\n\n for await (const chunk of stream) {\n if (chunk.choices?.length) {\n const delta = chunk.choices[0].delta;\n // 只收集思考内容\n if (delta.reasoning_content !== undefined && delta.reasoning_content !== null) {\n if (!isAnswering) {\n process.stdout.write(delta.reasoning_content);\n }\n reasoningContent += delta.reasoning_content;\n }\n\n // 收到content，开始进行回复\n if (delta.content !== undefined && delta.content) {\n if (!isAnswering) {\n console.log('\\n' + '='.repeat(20) + '完整回复' + '='.repeat(20) + '\\n');\n isAnswering = true;\n }\n process.stdout.write(delta.content);\n answerContent += delta.content;\n }\n }\n }\n}\n\nmain();" } }, "dashscope": { "default": { - "curl": "curl -X POST \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", - "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", - "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2-fast-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" + "curl": "curl -X POST \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/text-generation/generation\" \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"glm-5.2-fast-preview\",\n \"input\":{\n \"messages\":[\n {\n \"role\": \"user\",\n \"content\": \"你是谁？\"\n }\n ]\n },\n \"parameters\":{\n \"result_format\": \"message\"\n }\n}'", + "python": "import os\nfrom dashscope import Generation\nimport dashscope\ndashscope.base_http_api_url = \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/\"\n\nmessages = [{\"role\": \"user\", \"content\": \"你是谁？\"}]\n\ncompletion = Generation.call(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n model=\"glm-5.2-fast-preview\",\n messages=messages,\n result_format=\"message\", \n stream=True,\n incremental_output=True, \n)\n\nis_answering = False\n\nprint(\"=\" * 20 + \"思考过程\" + \"=\" * 20)\n\nfor chunk in completion:\n if (\n chunk.output.choices[0].message.content == \"\"\n and chunk.output.choices[0].message.reasoning_content == \"\"\n ):\n pass\n else:\n if (\n chunk.output.choices[0].message.reasoning_content != \"\"\n and chunk.output.choices[0].message.content == \"\"\n ):\n print(chunk.output.choices[0].message.reasoning_content, end=\"\", flush=True)\n elif chunk.output.choices[0].message.content != \"\":\n if not is_answering:\n print(\"\\n\" + \"=\" * 20 + \"完整回复\" + \"=\" * 20)\n is_answering = True\n print(chunk.output.choices[0].message.content, end=\"\", flush=True)", + "java": "import java.util.Arrays;\nimport org.slf4j.Logger;\nimport org.slf4j.LoggerFactory;\nimport com.alibaba.dashscope.aigc.generation.Generation;\nimport com.alibaba.dashscope.aigc.generation.GenerationParam;\nimport com.alibaba.dashscope.aigc.generation.GenerationResult;\nimport com.alibaba.dashscope.common.Message;\nimport com.alibaba.dashscope.common.Role;\nimport com.alibaba.dashscope.exception.ApiException;\nimport com.alibaba.dashscope.exception.InputRequiredException;\nimport com.alibaba.dashscope.exception.NoApiKeyException;\nimport io.reactivex.Flowable;\nimport java.lang.System;\nimport com.alibaba.dashscope.utils.Constants;\n\npublic class Main {\n static {\n Constants.baseHttpApiUrl=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1\";\n }\n private static final Logger logger = LoggerFactory.getLogger(Main.class);\n private static StringBuilder reasoningContent = new StringBuilder();\n private static StringBuilder finalContent = new StringBuilder();\n private static boolean isFirstPrint = true;\n\n private static void handleGenerationResult(GenerationResult message) {\n String reasoning = message.getOutput().getChoices().get(0).getMessage().getReasoningContent();\n String content = message.getOutput().getChoices().get(0).getMessage().getContent();\n\n if (!reasoning.isEmpty()) {\n reasoningContent.append(reasoning);\n if (isFirstPrint) {\n System.out.println(\"====================思考过程====================\");\n isFirstPrint = false;\n }\n System.out.print(reasoning);\n }\n\n if (!content.isEmpty()) {\n finalContent.append(content);\n if (!isFirstPrint) {\n System.out.println(\"\\n====================完整回复====================\");\n isFirstPrint = true;\n }\n System.out.print(content);\n }\n }\n private static GenerationParam buildGenerationParam(Message userMsg) {\n return GenerationParam.builder()\n // 若没有配置环境变量，请用阿里云百炼API Key将下行替换为：.apiKey(\"sk-xxx\")\n .apiKey(System.getenv(\"DASHSCOPE_API_KEY\"))\n .model(\"glm-5.2-fast-preview\")\n .incrementalOutput(true)\n .resultFormat(\"message\")\n .messages(Arrays.asList(userMsg))\n .build();\n }\n public static void streamCallWithMessage(Generation gen, Message userMsg)\n throws NoApiKeyException, ApiException, InputRequiredException {\n GenerationParam param = buildGenerationParam(userMsg);\n Flowable result = gen.streamCall(param);\n result.blockingForEach(message -> handleGenerationResult(message));\n }\n\n public static void main(String[] args) {\n try {\n Generation gen = new Generation();\n Message userMsg = Message.builder().role(Role.USER.getValue()).content(\"你是谁？\").build();\n streamCallWithMessage(gen, userMsg);\n } catch (ApiException | NoApiKeyException | InputRequiredException e) {\n logger.error(\"An exception occurred: {}\", e.getMessage());\n }\n System.exit(0);\n }\n}" } } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json index 1741f402..6bacfb0c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gui-plus.json @@ -20,12 +20,14 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -48,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json index 4b7651be..25f4e375 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-chat-v1.json @@ -21,6 +21,7 @@ { "priceUnit": "每秒", "price": "0.00015", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json index 284c4893..0b94192d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/gummy-realtime-v1.json @@ -21,6 +21,7 @@ { "priceUnit": "每秒", "price": "0.00015", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json index 16dc8b05..a98be88f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-i2v.json @@ -67,6 +67,7 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], @@ -105,6 +106,7 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_720p", "priceName": "视频生成（720P）" @@ -112,6 +114,7 @@ { "priceUnit": "每秒", "price": "1.6", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" @@ -135,6 +138,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json index bb531658..c1473542 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-r2v.json @@ -52,6 +52,7 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], @@ -90,6 +91,7 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_720p", "priceName": "视频生成（720P）" @@ -97,6 +99,7 @@ { "priceUnit": "每秒", "price": "1.6", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" @@ -120,6 +123,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json index 88af0269..2b6c624a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-t2v.json @@ -51,6 +51,7 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [], "capabilities": [ "VG" ], @@ -88,6 +89,7 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_720p", "priceName": "视频生成（720P）" @@ -95,6 +97,7 @@ { "priceUnit": "每秒", "price": "1.6", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" @@ -116,6 +119,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json index 0e68aaf6..45abf738 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/happyhorse-video-edit.json @@ -21,6 +21,7 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_720p", "priceName": "视频生成（720P）" @@ -28,6 +29,7 @@ { "priceUnit": "每秒", "price": "1.6", + "timeBand": "standard", "discount": 0.8, "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" @@ -51,6 +53,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json index 9a55992a..f5c19ff8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kimi-models-market-place.json @@ -60,6 +60,7 @@ "type": "model-default" } }, + "priceTimeBands": [], "capabilities": [ "TG", "VU", @@ -110,18 +111,21 @@ { "priceUnit": "每百万tokens", "price": "13", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "54", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -144,6 +148,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -193,18 +200,21 @@ { "priceUnit": "每百万tokens", "price": "6.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "27", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -227,6 +237,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "VU", @@ -276,18 +289,21 @@ { "priceUnit": "每百万tokens", "price": "6.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "27", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -310,6 +326,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -361,18 +380,21 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "21", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -395,6 +417,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", diff --git a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json index d2ced42d..74f3a828 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/kling-models-market-place.json @@ -21,24 +21,28 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1.2", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.8", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -61,6 +65,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -98,36 +105,42 @@ { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "type": "720P_no_reference_video", "priceName": "视频生成（720P 无参考视频）" }, { "priceUnit": "每秒", "price": "1.2", + "timeBand": "standard", "type": "1080P_no_reference_video", "priceName": "视频生成（1080P 无参考视频）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "720P_no_audio_no_reference_video", "priceName": "视频生成（720P 无声 无参考视频）" }, { "priceUnit": "每秒", "price": "0.8", + "timeBand": "standard", "type": "1080P_no_audio_no_reference_video", "priceName": "视频生成（1080P 无声 无参考视频）" }, { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "type": "720P_no_audio_reference_video", "priceName": "视频生成（720P 无声 有参考视频）" }, { "priceUnit": "每秒", "price": "1.2", + "timeBand": "standard", "type": "1080P_no_audio_reference_video", "priceName": "视频生成（1080P 无声 有参考视频）" } @@ -150,6 +163,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -186,12 +202,14 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" } @@ -214,6 +232,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -250,18 +271,21 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每秒", "price": "0.4", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -284,6 +308,9 @@ "async_user_concurrency_limit": 10 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json index 97cf97fd..8caed608 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait-detect.json @@ -17,10 +17,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json index 9c6fa97b..cdb9e994 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json +++ b/skills/bailian-docs-llm-wiki/models/groups/liveportrait.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.02", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json index d750913e..ce841311 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/minimax-models-market-place.json @@ -24,18 +24,21 @@ { "priceUnit": "每百万tokens", "price": "4.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.84", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -58,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning", @@ -102,18 +108,21 @@ { "priceUnit": "每百万tokens", "price": "2.1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.42", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -136,6 +145,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -180,18 +192,21 @@ { "priceUnit": "每百万tokens", "price": "2.1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.21", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -214,6 +229,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -257,18 +275,21 @@ { "priceUnit": "每百万tokens", "price": "2.1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.21", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -291,6 +312,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json index 95fb8e84..f6a5cb21 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v1.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00008", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json index c499ffb6..66eeb163 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-8k-v2.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.00008", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json index 66df088d..7e92a47b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-mtl-v1.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00008", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json index ddfe5da2..a973e02c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v1.json @@ -19,10 +19,14 @@ { "priceUnit": "每秒", "price": "0.00024", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json index dd3a1678..33683761 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-8k-v2.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00024", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json index 7c751e45..7b803b38 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v1.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00024", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json index 25bf199e..b30e2e67 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-realtime-v2.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00024", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json index 911d0ca5..095e5d0b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v1.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00008", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json index c7643257..08d76e2c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/paraformer-v2.json @@ -21,10 +21,14 @@ { "priceUnit": "每秒", "price": "0.00008", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json index 6e68510e..73c22ab7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-c1-market-place.json @@ -21,48 +21,56 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.39", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.18", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.56", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -85,6 +93,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -121,48 +132,56 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.39", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.18", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.56", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -185,6 +204,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -222,48 +244,56 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.39", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.18", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.56", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -286,6 +316,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -323,48 +356,56 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.39", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.18", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.56", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -385,6 +426,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json index 8594769d..047a5a2d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-capability-market-place.json @@ -19,6 +19,7 @@ { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "video_generation", "priceName": "视频生成" } @@ -39,6 +40,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -75,18 +79,21 @@ { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -107,6 +114,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -145,6 +155,7 @@ { "priceUnit": "每秒", "price": "0.12", + "timeBand": "standard", "type": "video_generation", "priceName": "视频生成" } @@ -165,6 +176,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json index 424d6161..27c33ec6 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-market-place.json @@ -20,48 +20,56 @@ { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.44", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -82,6 +90,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -118,48 +129,56 @@ { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.44", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -180,6 +199,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -216,48 +238,56 @@ { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.44", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -278,6 +308,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -314,48 +347,56 @@ { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.47", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.44", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -376,6 +417,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json index b44d8299..300923d9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/pixverse-v6-market-place.json @@ -20,48 +20,56 @@ { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.68", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -84,6 +92,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -120,48 +131,56 @@ { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.68", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -184,6 +203,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -220,48 +242,56 @@ { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.68", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -284,6 +314,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -321,48 +354,56 @@ { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "video_ratio_360p", "priceName": "视频生成（360P）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.36", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.68", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "360P_no_audio", "priceName": "视频生成（360P 无声）" }, { "priceUnit": "每秒", "price": "0.21", + "timeBand": "standard", "type": "540P_no_audio", "priceName": "视频生成（540P 无声）" }, { "priceUnit": "每秒", "price": "0.27", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.53", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -385,6 +426,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json index 796918f8..6d822276 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-max.json @@ -24,12 +24,14 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "32", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -52,6 +54,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json index 84557be9..ce0876d0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qvq-plus.json @@ -24,12 +24,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -52,6 +54,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json index 57093657..26689f0d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-flash.json @@ -18,32 +18,33 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-audio-3.0-realtime-flash", "prices": [ { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "audio_text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "audio_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "100", + "timeBand": "standard", "type": "audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -66,6 +67,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Chatting" ], @@ -82,7 +86,7 @@ "samples": { "dashscope": { "default": { - "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-flash\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", + "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-flash\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", "docUrl": "https://help.aliyun.com/document_detail/3041584.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json index ed7d84a1..0b6570bc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-realtime-plus.json @@ -18,32 +18,33 @@ "function-calling" ], "provider": "qwen", - "limit": { - "message": "model not exist" - }, "model": "qwen-audio-3.0-realtime-plus", "prices": [ { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "audio_text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "audio_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "150", + "timeBand": "standard", "type": "audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -66,6 +67,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Chatting" ], @@ -82,7 +86,7 @@ "samples": { "dashscope": { "default": { - "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-plus\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", + "python": "import asyncio\nimport base64\nimport json\nimport os\nimport pyaudio\nimport websockets\n\nAPI_KEY = os.environ[\"DASHSCOPE_API_KEY\"]\nURL = \"wss://[workspace-id].cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime?model=qwen-audio-3.0-realtime-plus\"\n\npya = pyaudio.PyAudio()\nmic = pya.open(format=pyaudio.paInt16, channels=1, rate=16000, input=True)\nspk = pya.open(format=pyaudio.paInt16, channels=1, rate=24000, output=True)\n\nasync def main():\n headers = {\"Authorization\": f\"Bearer {API_KEY}\"}\n async with websockets.connect(URL, additional_headers=headers) as ws:\n await ws.send(json.dumps({\n \"type\": \"session.update\",\n \"session\": {\n \"modalities\": [\"text\", \"audio\"],\n \"voice\": \"longanqian\",\n \"turn_detection\": {\n \"type\": \"server_vad\",\n \"threshold\": 0.5,\n \"silence_duration_ms\": 500\n }\n }\n }))\n\n async def send_audio():\n while True:\n data = await asyncio.to_thread(mic.read, 3200, False)\n await ws.send(json.dumps({\n \"type\": \"input_audio_buffer.append\",\n \"audio\": base64.b64encode(data).decode()\n }))\n await asyncio.sleep(0.02)\n\n async def recv_events():\n async for msg in ws:\n event = json.loads(msg)\n t = event[\"type\"]\n if t == \"response.audio.delta\":\n audio = base64.b64decode(event[\"delta\"])\n await asyncio.to_thread(spk.write, audio)\n elif t == \"conversation.item.input_audio_transcription.completed\":\n print(f\"[You] {event['transcript']}\")\n elif t == \"response.audio_transcript.done\":\n print(f\"[AI] {event['transcript']}\")\n elif t == \"error\":\n print(f\"[Error] {event['error']['message']}\")\n\n await asyncio.gather(send_audio(), recv_events())\n\nif __name__ == \"__main__\":\n try:\n asyncio.run(main())\n except KeyboardInterrupt:\n mic.close()\n spk.close()\n pya.terminate()\n print(\"\\nSession ended.\")", "docUrl": "https://help.aliyun.com/document_detail/3041584.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json index c7029617..017170a2 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-audio-tts.json @@ -19,6 +19,7 @@ { "priceUnit": "每万字符", "price": "1.4", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -71,6 +75,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -87,6 +92,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json index ed3a26b6..f232cd2b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-plus.json @@ -22,12 +22,14 @@ { "priceUnit": "每百万tokens", "price": "3.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -50,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json index 1f7b4cb5..8515fe9f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-coder-turbo.json @@ -22,12 +22,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -50,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json index 18a61bd0..47c7c74c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-deep-research.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "54", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "163", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json index 4f4a37c3..f298bbd1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-doc-turbo.json @@ -21,30 +21,35 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -67,6 +72,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json index a3b4c73d..a1d9bd71 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-embedding.json @@ -44,6 +44,7 @@ "type": "model-default" } }, + "priceTimeBands": [], "capabilities": [ "TR" ], @@ -85,12 +86,14 @@ { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "embedding_token_batch", "priceName": "向量输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -113,6 +116,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -148,12 +154,14 @@ { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "embedding_token_batch", "priceName": "向量输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -176,6 +184,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -209,12 +220,14 @@ { "priceUnit": "每百万tokens", "price": "0.35", + "timeBand": "standard", "type": "embedding_token_batch", "priceName": "向量输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -237,6 +250,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -270,12 +286,14 @@ { "priceUnit": "每百万tokens", "price": "0.35", + "timeBand": "standard", "type": "embedding_token_batch", "priceName": "向量输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -298,6 +316,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -331,10 +352,14 @@ { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -368,10 +393,14 @@ { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json index 9f1cee04..a09a14d8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash-character.json @@ -23,18 +23,21 @@ { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.05", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -57,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json index 554f773f..76b3815f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-flash.json @@ -42,6 +42,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -50,48 +53,56 @@ { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.03", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.075", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.188", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.015", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -99,6 +110,7 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -113,48 +125,56 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -162,6 +182,7 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -176,48 +197,56 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.24", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -225,6 +254,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json index 2a1085ab..d9fc3077 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0-pro.json @@ -23,6 +23,7 @@ { "priceUnit": "每张", "price": "0.5", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -39,6 +40,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json index eb692b65..a11b024f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-2.0.json @@ -23,6 +23,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -39,6 +40,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json index 827da9ab..3632a0f1 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit-max.json @@ -22,6 +22,7 @@ { "priceUnit": "每张", "price": "0.5", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -38,6 +39,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json index fdb26226..9e108b88 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-edit.json @@ -21,6 +21,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -81,6 +85,7 @@ { "priceUnit": "每张", "price": "0.3", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -97,6 +102,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json index 00f7fbb7..e1afb6dd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-max.json @@ -21,6 +21,7 @@ { "priceUnit": "每张", "price": "0.5", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -43,6 +44,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json index ce2053da..631846a7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-image-plus.json @@ -21,6 +21,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -47,6 +48,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -88,6 +92,7 @@ { "priceUnit": "每张", "price": "0.25", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -114,6 +119,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json index 2313f981..ed59eb91 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-long.json @@ -21,24 +21,28 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -61,6 +65,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -116,24 +123,28 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -156,6 +167,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json index f1b5fafe..3b8f37fd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-plus.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -104,12 +109,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -132,6 +139,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -188,12 +198,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -216,6 +228,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -271,12 +286,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -299,6 +316,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json index 95a2dccd..c692a9bb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-math-turbo.json @@ -22,12 +22,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -50,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json index 946f1650..1877fc13 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-max.json @@ -27,30 +27,35 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "9.6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -73,6 +78,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json index 754c5177..525b1c58 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-flash.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.95", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json index 7c7f1c3e..bd1d8856 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-image.json @@ -19,6 +19,7 @@ { "priceUnit": "每张", "price": "0.003", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -39,6 +40,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json index b1e3862a..5494447d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-lite.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json index 627c99cd..030de0bc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-plus.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "5.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json index 1c77270d..b40ee375 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-mt-turbo.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.95", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json index 4b9adfaf..de420aed 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo-realtime.json @@ -23,36 +23,42 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "6.4", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "50", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -75,6 +81,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], @@ -123,36 +132,42 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "6.4", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "50", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -175,6 +190,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json index e3181535..08fd9c4d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-omni-turbo.json @@ -26,90 +26,105 @@ { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "4.5", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "50", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "vision_input_token_cache", "priceName": "输入：图片/视频（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.08", + "timeBand": "standard", "type": "text_input_token_cache", "priceName": "输入：文本（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "audio_input_token_cache", "priceName": "输入：音频（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "text_input_token_batch", "priceName": "输入：文本（Batch File）" }, { "priceUnit": "每百万tokens", "price": "12.5", + "timeBand": "standard", "type": "audio_input_token_batch", "priceName": "输入：音频（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "vision_input_token_batch", "priceName": "输入：图片/视频（Batch File）" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "multi_output_token_batch", "priceName": "输出：文本+音频（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.25", + "timeBand": "standard", "type": "multiin_text_output_token_batch", "priceName": "输出：文本（Batch File，输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "purein_text_output_token_batch", "priceName": "输出：文本（Batch File，输入仅包含文本时）" } @@ -132,6 +147,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], @@ -183,36 +201,42 @@ { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "4.5", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "50", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -235,6 +259,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json index 39d23237..6bf713f9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus-character.json @@ -58,6 +58,7 @@ "type": "model-default" } }, + "priceTimeBands": [], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json index bbb4e97e..5e2a329f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-plus.json @@ -42,6 +42,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -50,90 +53,105 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "0.16", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.16", + "timeBand": "standard", "type": "thinking_input_token_cache", "priceName": "输入（思考模式缓存命中）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.08", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "thinking_input_token_cache_creation_5m", "priceName": "显式缓存创建（思考）" }, { "priceUnit": "每百万tokens", "price": "0.08", + "timeBand": "standard", "type": "thinking_input_token_cache_read", "priceName": "显式缓存命中（思考）" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -141,6 +159,7 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -148,6 +167,7 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "discount": 0.5, "type": "thinking_input_token_batch_chat", "priceName": "输入（思考模式 Batch Chat)" @@ -155,6 +175,7 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "discount": 0.5, "type": "thinking_output_token_batch_chat", "priceName": "输出（思考模式 Batch Chat)" @@ -169,90 +190,105 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "thinking_input_token_cache", "priceName": "输入（思考模式缓存命中）" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.24", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "thinking_input_token_cache_creation_5m", "priceName": "显式缓存创建（思考）" }, { "priceUnit": "每百万tokens", "price": "0.24", + "timeBand": "standard", "type": "thinking_input_token_cache_read", "priceName": "显式缓存命中（思考）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -260,6 +296,7 @@ { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -267,6 +304,7 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "discount": 0.5, "type": "thinking_input_token_batch_chat", "priceName": "输入（思考模式 Batch Chat)" @@ -274,6 +312,7 @@ { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "discount": 0.5, "type": "thinking_output_token_batch_chat", "priceName": "输出（思考模式 Batch Chat)" @@ -288,90 +327,105 @@ { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "48", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "64", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "0.96", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.96", + "timeBand": "standard", "type": "thinking_input_token_cache", "priceName": "输入（思考模式缓存命中）" }, { "priceUnit": "每百万tokens", "price": "32", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "thinking_input_token_cache_creation_5m", "priceName": "显式缓存创建（思考）" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "thinking_input_token_cache_read", "priceName": "显式缓存命中（思考）" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -379,6 +433,7 @@ { "priceUnit": "每百万tokens", "price": "48", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -386,6 +441,7 @@ { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "discount": 0.5, "type": "thinking_input_token_batch_chat", "priceName": "输入（思考模式 Batch Chat)" @@ -393,6 +449,7 @@ { "priceUnit": "每百万tokens", "price": "64", + "timeBand": "standard", "discount": 0.5, "type": "thinking_output_token_batch_chat", "priceName": "输出（思考模式 Batch Chat)" @@ -471,6 +528,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -479,48 +539,56 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" } @@ -534,48 +602,56 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" } @@ -589,48 +665,56 @@ { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "48", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "64", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "32", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" } @@ -688,12 +772,14 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -716,6 +802,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -772,12 +861,14 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -800,6 +891,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json index 20ba1fca..56edeefc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-rerank.json @@ -21,12 +21,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -49,6 +51,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -84,6 +89,7 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -106,6 +112,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], @@ -143,6 +152,7 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -165,6 +175,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json index 26cb0c69..e5bfe739 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts-realtime.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], @@ -90,12 +95,14 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -118,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Text-to-Speech" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json index 386ac268..7d9a5c8e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-tts.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "qwen_tts_multi_output_token", "priceName": "输出：音频" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -90,12 +95,14 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "qwen_tts_multi_output_token", "priceName": "输出：音频" } @@ -118,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json index e7f53f6d..647da1d4 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-turbo.json @@ -25,66 +25,77 @@ { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "thinking_input_token_cache", "priceName": "输入（思考模式缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "thinking_output_token_batch", "priceName": "思考模式输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "thinking_input_token_batch", "priceName": "输入（思考模式 Batch File）" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -107,6 +118,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json index 4a818fa6..5c455f4c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-embedding.json @@ -20,12 +20,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -48,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], @@ -84,12 +89,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "embedding_image_token", "priceName": "图片输入" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "embedding_token", "priceName": "文本输入" } @@ -112,6 +119,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ME" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json index c07de66c..2c8fb683 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-max.json @@ -28,36 +28,42 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.32", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "150", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -80,6 +86,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json index 1239f40f..4e74e9fc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-ocr.json @@ -23,24 +23,28 @@ { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -63,6 +67,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -117,12 +124,14 @@ { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -145,6 +154,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -201,24 +213,28 @@ { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -241,6 +257,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json index fcb69a54..e673bbbb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-vl-plus.json @@ -29,36 +29,42 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.16", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -81,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json index 40311a6c..543cd8ec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-design.json @@ -20,6 +20,7 @@ { "priceUnit": "每次", "price": "0.2", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json index 8e1ec12b..9dd791fe 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen-voice-enrollment.json @@ -19,6 +19,7 @@ { "priceUnit": "次", "price": "0.01", + "timeBand": "standard", "type": "tts_vc_model", "priceName": "声音复刻及声音设计" } @@ -35,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json index 56fc3814..79028dbb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen2.5.json @@ -25,36 +25,42 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "38", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "76", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -77,6 +83,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json index 018c8b38..0f99e445 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-filetrans.json @@ -20,6 +20,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json index 55ce754a..6dc7164d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash-realtime.json @@ -20,6 +20,7 @@ { "priceUnit": "每秒", "price": "0.00033", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json index 7df8bfc9..708242d0 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-asr-flash.json @@ -20,6 +20,7 @@ { "priceUnit": "每秒", "price": "0.00022", + "timeBand": "standard", "type": "content_duration", "priceName": "音频时长" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json index 01975fab..cf19c7ef 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-30b-a3b-instruct.json @@ -36,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -44,12 +47,14 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -63,12 +68,14 @@ { "priceUnit": "每百万tokens", "price": "2.25", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "9", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -82,12 +89,14 @@ { "priceUnit": "每百万tokens", "price": "3.75", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -101,12 +110,14 @@ { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "37.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json index e3c85793..fb3827be 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-480b-a35b-instruct.json @@ -36,6 +36,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -44,12 +47,14 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -63,12 +68,14 @@ { "priceUnit": "每百万tokens", "price": "9", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "36", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -82,12 +89,14 @@ { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "60", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -101,12 +110,14 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json index b61014ad..f4aeedcb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-flash.json @@ -38,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -46,30 +49,35 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.1", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -83,30 +91,35 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.875", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -120,30 +133,35 @@ { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "3.125", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -157,30 +175,35 @@ { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "6.25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json index b8057431..92e082b9 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-coder-plus.json @@ -38,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -46,30 +49,35 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -83,30 +91,35 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -120,30 +133,35 @@ { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "12.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -157,30 +175,35 @@ { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "200", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json index 9359f9c8..161c9a99 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash-realtime.json @@ -22,24 +22,28 @@ { "priceUnit": "每百万tokens", "price": "64", + "timeBand": "standard", "type": "translate_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "translate_vision_input_token", "priceName": "输入：图片" }, { "priceUnit": "每百万tokens", "price": "64", + "timeBand": "standard", "type": "translate_multi_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "240", + "timeBand": "standard", "type": "translate_multi_output_token", "priceName": "输出：音频" } @@ -62,6 +66,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json index 6386eb81..ac5f1cad 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-livetranslate-flash.json @@ -21,24 +21,28 @@ { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "translate_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "translate_vision_input_token", "priceName": "输入：图片" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "translate_multi_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "translate_multi_output_token", "priceName": "输出：音频" } @@ -61,6 +65,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json index bc82e62a..53a805fd 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-max.json @@ -110,6 +110,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -118,48 +121,56 @@ { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.25", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3.125", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -167,6 +178,7 @@ { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -181,48 +193,56 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -230,6 +250,7 @@ { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -244,48 +265,56 @@ { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "3.5", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "14", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "8.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.7", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -293,6 +322,7 @@ { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -396,6 +426,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -404,18 +437,21 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -429,18 +465,21 @@ { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -454,18 +493,21 @@ { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "60", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json index 271a47c9..28679897 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-30b-a3b-captioner.json @@ -20,12 +20,14 @@ { "priceUnit": "每百万tokens", "price": "15.8", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "12.7", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" } @@ -48,6 +50,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "ASR" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json index 263adc45..1cdbea0b 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash-realtime.json @@ -24,36 +24,42 @@ { "priceUnit": "每百万tokens", "price": "2.2", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "18.9", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "3.9", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "8.3", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "15.2", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "75.1", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" } @@ -76,6 +82,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json index 5e54cd14..ad450cbc 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-omni-flash.json @@ -27,66 +27,77 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "text_input_token", "priceName": "输入：文本" }, { "priceUnit": "每百万tokens", "price": "15.8", + "timeBand": "standard", "type": "audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "3.3", + "timeBand": "standard", "type": "vision_input_token", "priceName": "输入：图片/视频" }, { "priceUnit": "每百万tokens", "price": "6.9", + "timeBand": "standard", "type": "purein_text_output_token", "priceName": "输出：文本（输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "12.7", + "timeBand": "standard", "type": "multiin_text_output_token", "priceName": "输出：文本（输入包含图片/音频/视频时）" }, { "priceUnit": "每百万tokens", "price": "62.6", + "timeBand": "standard", "type": "multi_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" }, { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "thinking_text_input_token", "priceName": "输入：文本（思考）" }, { "priceUnit": "每百万tokens", "price": "15.8", + "timeBand": "standard", "type": "thinking_audio_input_token", "priceName": "输入：音频（思考）" }, { "priceUnit": "每百万tokens", "price": "3.3", + "timeBand": "standard", "type": "thinking_vision_input_token", "priceName": "输入：图片/视频（思考）" }, { "priceUnit": "每百万tokens", "price": "6.9", + "timeBand": "standard", "type": "thinking_purein_text_output_token", "priceName": "输出：文本（思考模式下，输入仅包含文本时）" }, { "priceUnit": "每百万tokens", "price": "12.7", + "timeBand": "standard", "type": "thinking_multiin_text_output_token", "priceName": "输出：文本（思考模式下，输入包含图片/音频/视频时）" } @@ -109,6 +120,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json index 34a61955..dd611505 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash-realtime.json @@ -20,24 +20,16 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } ], "qpmInfo": { - "user-spec": { - "count_limit_period": 1, - "start_time": 1762516375, - "count_limit": 20, - "end_time": 253370736000, - "type": "user-spec" - }, "model-default-actual": { "count_limit_period": 1, - "start_time": 1762516375, - "count_limit": 20, - "end_time": 253370736000, - "type": "user-spec" + "count_limit": 3, + "type": "model-default" }, "model-default": { "count_limit_period": 1, @@ -45,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json index 9195d43d..cbdeb533 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-flash.json @@ -22,6 +22,7 @@ { "priceUnit": "每万字符", "price": "0.8", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -38,6 +39,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json index 587cc5ae..8548bb42 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash-realtime.json @@ -20,6 +20,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json index 8c3f0c4c..c44a82a8 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-instruct-flash.json @@ -21,6 +21,7 @@ { "priceUnit": "每万字符", "price": "0.8", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json index 35e45139..4f320d79 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc-realtime.json @@ -20,6 +20,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json index c30acbf4..62c48e27 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vc.json @@ -21,6 +21,7 @@ { "priceUnit": "每万字符", "price": "0.8", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json index 2ab0d68f..dfe9a084 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd-realtime.json @@ -20,6 +20,7 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -36,6 +37,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json index a44db4e3..fd45e733 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-tts-vd.json @@ -21,6 +21,7 @@ { "priceUnit": "每万字符", "price": "0.8", + "timeBand": "standard", "type": "cosy_tts_number", "priceName": "语音合成" } @@ -37,6 +38,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json index aa275109..1a0d0c07 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-flash.json @@ -42,6 +42,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -50,48 +53,56 @@ { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.03", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.075", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.1875", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.015", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -99,6 +110,7 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -113,48 +125,56 @@ { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.375", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.03", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -162,6 +182,7 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -176,48 +197,56 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.06", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -225,6 +254,7 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json index a28289a1..d62b1543 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3-vl-plus.json @@ -43,6 +43,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -51,48 +54,56 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.1", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -100,6 +111,7 @@ { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -114,48 +126,56 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.875", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.15", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -163,6 +183,7 @@ { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -177,48 +198,56 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.3", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -226,6 +255,7 @@ { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json index 42799f67..32e7eb61 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-flash.json @@ -112,6 +112,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -120,42 +123,49 @@ { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "0.25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.02", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -163,6 +173,7 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -177,42 +188,49 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.08", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -220,6 +238,7 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -234,42 +253,49 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -277,6 +303,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json index 56cf0c2d..fefde783 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-livetranslate-flash-realtime.json @@ -22,24 +22,28 @@ { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "translate_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "3.3", + "timeBand": "standard", "type": "translate_vision_input_token", "priceName": "输入：图片" }, { "priceUnit": "每百万tokens", "price": "100", + "timeBand": "standard", "type": "translate_multi_text_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "160", + "timeBand": "standard", "type": "translate_multi_output_token", "priceName": "输出：音频" } @@ -62,6 +66,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Audio-Translate" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json index f6d0ac91..456eef48 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-ocr.json @@ -22,12 +22,14 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -50,6 +52,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json index f7b2ad78..d5f395ce 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash-realtime.json @@ -43,24 +43,28 @@ { "priceUnit": "每百万tokens", "price": "27", + "timeBand": "standard", "type": "omni_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "107", + "timeBand": "standard", "type": "omni_audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" }, { "priceUnit": "每百万tokens", "price": "3.3", + "timeBand": "standard", "type": "omni_no_audio_input_token", "priceName": "输入：文本/图片/视频" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "omni_no_audio_output_token", "priceName": "输出：文本" } @@ -83,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json deleted file mode 100644 index 438c6296..00000000 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-flash.json +++ /dev/null @@ -1,111 +0,0 @@ -{ - "name": "Qwen3.5-Omni-Flash", - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "items": [ - { - "inferenceMetadata": { - "response_modality": [ - "Text", - "Audio" - ], - "request_modality": [ - "Text", - "Image", - "Video", - "Audio" - ] - }, - "builtInToolMultiPrices": [ - { - "supportedApi": "Completions API", - "name": "search_strategy:agent", - "docUrl": "https://help.aliyun.com/document_detail/2867560.html", - "type": "search_strategy:agent", - "prices": [ - { - "priceUnit": "千次调用", - "price": "4", - "currency": "CNY" - } - ] - } - ], - "description": "Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720P（1 FPS）音视频理解与对话，并进一步拓展语言范围，支持60+种语言音频输入，30+语言语音输出，并且具备强大的结构化音视频理解能力，广泛应用于文本创作、语音助手、多媒体分析等场景，提供自然流畅的多模态理解与交互体验。", - "collectionTag": "Qwen3.5", - "features": [ - "web-search" - ], - "provider": "qwen", - "model": "qwen3.5-omni-flash", - "iconUrl": "", - "prices": [ - { - "priceUnit": "每百万tokens", - "price": "18", - "type": "omni_audio_input_token", - "priceName": "输入：音频" - }, - { - "priceUnit": "每百万tokens", - "price": "72", - "type": "omni_audio_output_token", - "priceName": "输出：文本+音频（输出的文本不计费）" - }, - { - "priceUnit": "每百万tokens", - "price": "2.2", - "type": "omni_no_audio_input_token", - "priceName": "输入：文本/图片/视频" - }, - { - "priceUnit": "每百万tokens", - "price": "13.3", - "type": "omni_no_audio_output_token", - "priceName": "输出：文本" - } - ], - "qpmInfo": { - "model-default-actual": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - }, - "model-default": { - "count_limit_period": 60, - "usage_limit": 100000, - "usage_limit_field": "total_tokens", - "count_limit": 60, - "usage_limit_period": 60, - "type": "model-default" - } - }, - "capabilities": [ - "Multimodal-Omni" - ], - "modelAlias": "qwen3.5-omni-flash", - "versionTag": "MAJOR", - "equivalentSnapshot": "qwen3.5-omni-flash-2026-03-15", - "maxOutputTokens": 65536, - "latestOnlineAt": "2026-03-30T03:58:51.000+00:00", - "contextWindow": 262144, - "maxInputTokens": 196608, - "inferenceProvider": "aliyun-bailian", - "name": "Qwen3.5-Omni-Flash", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html", - "category": "Multimodal", - "samples": { - "openai": { - "default": { - "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"qwen3.5-omni-flash\",\n \"messages\": [\n {\n \"role\": \"user\", \n \"content\": \"你是谁？\"\n }\n ],\n \"stream\":true,\n \"stream_options\":{\n \"include_usage\":true\n },\n \"modalities\":[\"text\",\"audio\"],\n \"audio\":{\"voice\":\"Ethan\",\"format\":\"wav\"}\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n # 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"qwen3.5-omni-flash\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n # 设置输出数据的模态，当前支持两种：[\"text\",\"audio\"]、[\"text\"]\n modalities=[\"text\", \"audio\"],\n audio={\"voice\": \"Ethan\", \"format\": \"wav\"},\n # stream 必须设置为 True，否则会报错\n stream=True,\n stream_options={\"include_usage\": True},\n)\n\nfor chunk in completion:\n if chunk.choices:\n print(chunk.choices[0].delta)\n else:\n print(chunk.usage)", - "nodejs": "import OpenAI from \"openai\";\n\nconst openai = new OpenAI(\n {\n // 新加坡和北京地域的API Key不同。获取API Key：https://www.alibabacloud.com/help/zh/model-studio/get-api-key\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: \"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\"\n }\n);\nconst completion = await openai.chat.completions.create({\n model: \"qwen3.5-omni-flash\",\n messages: [\n { role: \"user\", content: \"你是谁？\" }\n ],\n stream: true,\n stream_options: {\n include_usage: true\n },\n modalities: [\"text\", \"audio\"],\n audio: { voice: \"Ethan\", format: \"wav\" }\n});\n\nfor await (const chunk of completion) {\n if (Array.isArray(chunk.choices) && chunk.choices.length > 0) {\n console.log(chunk.choices[0].delta);\n } else {\n console.log(chunk.usage);\n }\n}", - "docUrl": "https://help.aliyun.com/document_detail/2867839.html" - } - } - } - } - ] -} diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json index 2d288870..13196f6e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus-realtime.json @@ -43,24 +43,28 @@ { "priceUnit": "每百万tokens", "price": "80", + "timeBand": "standard", "type": "omni_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "300", + "timeBand": "standard", "type": "omni_audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "omni_no_audio_input_token", "priceName": "输入：文本/图片/视频" }, { "priceUnit": "每百万tokens", "price": "60", + "timeBand": "standard", "type": "omni_no_audio_output_token", "priceName": "输出：文本" } @@ -83,6 +87,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Realtime-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json index 72ace6d3..1796efaa 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-omni-plus.json @@ -44,48 +44,56 @@ { "priceUnit": "每百万tokens", "price": "53", + "timeBand": "standard", "type": "omni_audio_input_token", "priceName": "输入：音频" }, { "priceUnit": "每百万tokens", "price": "213", + "timeBand": "standard", "type": "omni_audio_output_token", "priceName": "输出：文本+音频（输出的文本不计费）" }, { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "type": "omni_no_audio_input_token", "priceName": "输入：文本/图片/视频" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "omni_no_audio_output_token", "priceName": "输出：文本" }, { "priceUnit": "每百万tokens", "price": "26.5", + "timeBand": "standard", "type": "omni_audio_input_token_batch", "priceName": "输入：音频（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3.5", + "timeBand": "standard", "type": "omni_no_audio_input_token_batch", "priceName": "输入：文本/图片/视频（Batch File）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "omni_no_audio_output_token_batch", "priceName": "输出：文本（Batch File）" }, { "priceUnit": "每百万tokens", "price": "53", + "timeBand": "standard", "discount": 0.5, "type": "omni_audio_input_token_batch_chat", "priceName": "输入：音频（Batch Chat）" @@ -93,6 +101,7 @@ { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "discount": 0.5, "type": "omni_no_audio_input_token_batch_chat", "priceName": "输入：文本/图片/视频（Batch Chat）" @@ -100,6 +109,7 @@ { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "discount": 0.5, "type": "omni_no_audio_output_token_batch_chat", "priceName": "输出：文本（Batch Chat）" @@ -123,6 +133,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Multimodal-Omni" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json index 2dae9b2d..024154ec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5-plus.json @@ -112,6 +112,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -120,42 +123,49 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.08", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -163,6 +173,7 @@ { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -177,42 +188,49 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -220,6 +238,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -234,42 +253,49 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -277,6 +303,7 @@ { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json index ce14d6e2..7e4c9192 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.5.json @@ -110,6 +110,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -118,12 +121,14 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "7.2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -137,12 +142,14 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -296,6 +303,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -304,12 +314,14 @@ { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3.2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -323,12 +335,14 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -483,6 +497,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -491,12 +508,14 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -510,12 +529,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "14.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -676,6 +697,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -684,12 +708,14 @@ { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -703,12 +729,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json index b42c7d6c..d1f2011e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-flash.json @@ -111,6 +111,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -119,48 +122,56 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "7.2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "3.6", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.12", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" }, { "priceUnit": "每百万tokens", "price": "7.2", + "timeBand": "standard", "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" } @@ -174,48 +185,56 @@ { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "14.4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.48", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "4.8", + "timeBand": "standard", "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" }, { "priceUnit": "每百万tokens", "price": "28.8", + "timeBand": "standard", "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json index cfbc14e0..2cbc8a4a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-max.json @@ -41,6 +41,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -49,24 +52,28 @@ { "priceUnit": "每百万tokens", "price": "9", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "54", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "11.25", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.9", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } @@ -80,24 +87,28 @@ { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "90", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "18.75", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json index 5b652f21..26a02835 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6-plus.json @@ -111,6 +111,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -119,42 +122,49 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -162,6 +172,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -176,42 +187,49 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "48", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache_read", "priceName": "显式缓存命中" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -219,6 +237,7 @@ { "priceUnit": "每百万tokens", "price": "48", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json index 5f91ea86..b846d044 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.6.json @@ -95,12 +95,14 @@ { "priceUnit": "每百万tokens", "price": "1.8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "10.8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -123,6 +125,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "TG", @@ -188,12 +193,14 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -216,6 +223,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "VU", diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json index 49527a64..069d1a07 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-max.json @@ -68,6 +68,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "input_token", "priceName": "输入" @@ -75,6 +76,7 @@ { "priceUnit": "每百万tokens", "price": "36", + "timeBand": "standard", "discount": 0.5, "type": "output_token", "priceName": "输出" @@ -82,6 +84,7 @@ { "priceUnit": "每百万tokens", "price": "2.4", + "timeBand": "standard", "discount": 0.5, "type": "input_token_cache", "priceName": "输入（缓存命中）" @@ -89,18 +92,21 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "18", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "15", + "timeBand": "standard", "discount": 0.5, "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" @@ -108,6 +114,7 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_cache_read", "priceName": "显式缓存命中" @@ -115,6 +122,7 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -122,6 +130,7 @@ { "priceUnit": "每百万tokens", "price": "36", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -145,6 +154,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -243,12 +255,14 @@ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "36", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -271,6 +285,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json index e7f11c96..b9aa9afa 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.7-plus.json @@ -111,6 +111,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -119,6 +122,7 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.8, "type": "input_token", "priceName": "输入" @@ -126,6 +130,7 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "discount": 0.8, "type": "output_token", "priceName": "输出" @@ -133,6 +138,7 @@ { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache", "priceName": "输入（缓存命中）" @@ -140,18 +146,21 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" @@ -159,6 +168,7 @@ { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache_read", "priceName": "显式缓存命中" @@ -166,6 +176,7 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -173,6 +184,7 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" @@ -187,6 +199,7 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "discount": 0.8, "type": "input_token", "priceName": "输入" @@ -194,6 +207,7 @@ { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "discount": 0.8, "type": "output_token", "priceName": "输出" @@ -201,6 +215,7 @@ { "priceUnit": "每百万tokens", "price": "1.2", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache", "priceName": "输入（缓存命中）" @@ -208,18 +223,21 @@ { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache_creation_5m", "priceName": "显式缓存创建" @@ -227,6 +245,7 @@ { "priceUnit": "每百万tokens", "price": "0.6", + "timeBand": "standard", "discount": 0.8, "type": "input_token_cache_read", "priceName": "显式缓存命中" @@ -234,6 +253,7 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "discount": 0.5, "type": "input_token_batch_chat", "priceName": "输入（Batch Chat)" @@ -241,6 +261,7 @@ { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "discount": 0.5, "type": "output_token_batch_chat", "priceName": "输出（Batch Chat)" diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json index 9b09f528..d54be612 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwen3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwen3.json @@ -26,12 +26,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -54,6 +56,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -111,12 +116,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -139,6 +146,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -198,12 +208,14 @@ { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -226,6 +238,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -286,12 +301,14 @@ { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -314,6 +331,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -374,18 +394,21 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -408,6 +431,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -475,18 +501,21 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -509,6 +538,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -574,12 +606,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -602,6 +636,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "Reasoning" @@ -662,12 +699,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -690,6 +729,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU" ], @@ -742,12 +784,14 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -770,6 +814,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -824,12 +871,14 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -852,6 +901,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -911,12 +963,14 @@ { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -939,6 +993,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1004,12 +1061,14 @@ { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -1032,6 +1091,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1087,12 +1149,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -1115,6 +1179,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1171,12 +1238,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1199,6 +1268,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -1258,24 +1330,28 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -1298,6 +1374,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1358,24 +1437,28 @@ { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.75", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "7.5", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" } @@ -1398,6 +1481,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1459,30 +1545,35 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "20", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "40", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -1505,6 +1596,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1579,30 +1673,35 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "30", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -1625,6 +1724,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1696,30 +1798,35 @@ { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.5", + "timeBand": "standard", "type": "thinking_input_token", "priceName": "输入（思考）" }, { "priceUnit": "每百万tokens", "price": "5", + "timeBand": "standard", "type": "thinking_output_token", "priceName": "输出（思考）" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "ft", "priceName": "调优" } @@ -1742,6 +1849,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -1823,6 +1933,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "multiPrices": [ { "rangeStart": 0, @@ -1831,12 +1944,14 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1850,12 +1965,14 @@ { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -1869,12 +1986,14 @@ { "priceUnit": "每百万tokens", "price": "2.5", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "10", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } diff --git a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json index eb494d3b..a055face 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/qwq-plus.json @@ -25,24 +25,28 @@ { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_batch", "priceName": "输入（Batch File）" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "output_token_batch", "priceName": "输出（Batch File）" } @@ -65,6 +69,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/sambert.json b/skills/bailian-docs-llm-wiki/models/groups/sambert.json index b9233e1d..c4001266 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/sambert.json +++ b/skills/bailian-docs-llm-wiki/models/groups/sambert.json @@ -19,10 +19,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -57,10 +61,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -95,10 +103,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -133,10 +145,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -171,10 +187,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -209,10 +229,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -247,10 +271,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -285,10 +313,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -323,10 +355,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -361,10 +397,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -399,10 +439,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -437,10 +481,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -475,10 +523,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -513,10 +565,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -551,10 +607,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -589,10 +649,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -627,10 +691,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -665,10 +733,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -703,10 +775,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -741,10 +817,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -779,10 +859,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -817,10 +901,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -855,10 +943,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -893,10 +985,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -931,10 +1027,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -969,10 +1069,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1007,10 +1111,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1045,10 +1153,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1083,10 +1195,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1121,10 +1237,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1159,10 +1279,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1197,10 +1321,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1235,10 +1363,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1273,10 +1405,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1311,10 +1447,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1349,10 +1489,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1387,10 +1531,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1425,10 +1573,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1463,10 +1615,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1501,10 +1657,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1539,10 +1699,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1577,10 +1741,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], @@ -1615,10 +1783,14 @@ { "priceUnit": "每万字符", "price": "1", + "timeBand": "standard", "type": "tts_text_number", "priceName": "Sambert 语音合成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TTS" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json index 13c1be04..abc4406d 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json +++ b/skills/bailian-docs-llm-wiki/models/groups/siliconflow-models.json @@ -23,12 +23,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -51,6 +53,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -93,12 +98,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -121,6 +128,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -163,12 +173,14 @@ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -191,6 +203,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -232,12 +247,14 @@ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -260,6 +277,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" diff --git a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json index 0abb94be..104849fb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/stepfun-models-market-place.json @@ -26,18 +26,21 @@ { "priceUnit": "每百万tokens", "price": "1.35", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8.1", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.27", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -60,6 +63,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "VU" diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json index c2e7b9af..bbdb79ad 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-intent-detect-v3.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json index 29c2688e..98ed097a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-flash.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.4", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json index b8f1c727..15aeedae 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json +++ b/skills/bailian-docs-llm-wiki/models/groups/tongyi-xiaomi-analysis-pro.json @@ -19,12 +19,14 @@ { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "2.7", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -47,6 +49,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json new file mode 100644 index 00000000..431ec936 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/models/groups/tripo-models-market-place.json @@ -0,0 +1,311 @@ +{ + "name": "Tripo", + "description": "AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。", + "items": [ + { + "inferenceMetadata": { + "response_modality": [ + "3D-Generation" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Tripo P1.0 是面向实时应用与生产管线的 3D 生成模型，专为需要干净拓扑和引擎可用网格的开发者与创作者设计。模型可在约 2 秒内生成具备专业级拓扑结构的 3D 资产，适用于游戏、Web3D 与各类实时交互场景。针对 UGC 内容生产中对“速度”和“开箱即用”的需求，Tripo P1.0 在保证质量的同时大幅提升生成效率，使资产能够快速接入实时引擎与开发流程。", + "features": [ + "function-calling", + "structured-outputs", + "batch" + ], + "provider": "tripo", + "model": "Tripo/Tripo-P1.0", + "prices": [ + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_no_texture", + "priceName": "文生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_no_texture", + "priceName": "单图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_no_texture", + "priceName": "多图生3D（无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_sd_texture", + "priceName": "文生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_sd_texture", + "priceName": "单图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_sd_texture", + "priceName": "多图生3D（带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_hd_texture", + "priceName": "文生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_hd_texture", + "priceName": "单图生3D（带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_hd_texture", + "priceName": "多图生3D（带高清贴图）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "priceTimeBands": [ + "standard" + ], + "capabilities": [ + "3D-generation" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-27T04:07:42.000+00:00", + "inferenceProvider": "tripo", + "name": "Tripo-P1.0", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-P1.0\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html" + } + } + } + }, + { + "inferenceMetadata": { + "response_modality": [ + "3D-Generation" + ], + "request_modality": [ + "Text" + ] + }, + "description": "Tripo H3.1 是 Tripo 推出的高精度 3D 生成模型，专为需要极致视觉质量与细节表现的创作者设计。模型通过核心算法升级与模块优化，参数规模达 200 亿级，支持十亿体素级三维分辨率与最高 200 万面多边形生成。在保持高精度几何与真实纹理的同时，Tripo H3.1 对输入参考图的还原度与对齐度进一步提升，在角色形体、面部细节与几何文字等复杂结构上实现更稳定、细致的表达，适用于高质量视觉制作与 3D 打印等高精度资产生产场景。", + "features": [ + "function-calling", + "batch", + "structured-outputs" + ], + "provider": "tripo", + "model": "Tripo/Tripo-H3.1", + "prices": [ + { + "priceUnit": "每次", + "price": "0.7", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_no_texture", + "priceName": "文生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_no_texture", + "priceName": "单图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_no_texture", + "priceName": "多图生3D（标准版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "1.4", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_sd_texture", + "priceName": "文生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_sd_texture", + "priceName": "单图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_sd_texture", + "priceName": "多图生3D（标准版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_standard_hd_texture", + "priceName": "文生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_standard_hd_texture", + "priceName": "单图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_standard_hd_texture", + "priceName": "多图生3D（标准版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "2.1", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_no_texture", + "priceName": "文生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_no_texture", + "priceName": "单图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_no_texture", + "priceName": "多图生3D（超清版+无贴图）" + }, + { + "priceUnit": "每次", + "price": "2.8", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_sd_texture", + "priceName": "文生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_sd_texture", + "priceName": "单图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_sd_texture", + "priceName": "多图生3D（超清版+带标清贴图）" + }, + { + "priceUnit": "每次", + "price": "3.5", + "timeBand": "standard", + "type": "generation_3d_text_to_3d_ultra_hd_texture", + "priceName": "文生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_image_to_3d_ultra_hd_texture", + "priceName": "单图生3D（超清版+带高清贴图）" + }, + { + "priceUnit": "每次", + "price": "4.2", + "timeBand": "standard", + "type": "generation_3d_multiview_to_3d_ultra_hd_texture", + "priceName": "多图生3D（超清版+带高清贴图）" + } + ], + "qpmInfo": { + "model-default-actual": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + }, + "model-default": { + "count_limit_period": 60, + "async_user_queue_limit": 500, + "count_limit": 5, + "async_task_timeout": 180, + "type": "model-default", + "async_user_concurrency_limit": 10 + } + }, + "priceTimeBands": [ + "standard" + ], + "capabilities": [ + "3D-generation" + ], + "modelAlias": "", + "versionTag": "MAJOR", + "latestOnlineAt": "2026-04-27T04:08:07.000+00:00", + "inferenceProvider": "tripo", + "name": "Tripo-H3.1", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html", + "samples": { + "dashscope": { + "default": { + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation' \\\n -H 'X-DashScope-Async: enable' \\\n -H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n -H 'Content-Type: application/json' \\\n -d '{\n \"model\": \"Tripo/Tripo-H3.1\",\n \"input\": {\n \"prompt\": \"一只可爱的猫\"\n },\n \"parameters\": {\n \"texture_quality\": \"standard\"\n }\n}'", + "docUrl": "https://help.aliyun.com/document_detail/3030679.html" + } + } + } + } + ] +} diff --git a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json index e3ebd784..8e6333ca 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vanchin-models-market-place.json @@ -18,26 +18,26 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-v4-pro", "prices": [ { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "24", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -60,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -75,7 +78,7 @@ "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v4-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v4-pro\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } @@ -98,27 +101,27 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-v3.2-think", "iconUrl": "", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "3", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -141,6 +144,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -157,7 +163,7 @@ "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.2-think\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.2-think\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } @@ -178,27 +184,27 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-v3.1-terminus", "iconUrl": "", "prices": [ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "12", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -221,6 +227,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -237,7 +246,7 @@ "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3.1-terminus\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } @@ -260,27 +269,27 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-v3", "iconUrl": "", "prices": [ { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "0.8", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -303,6 +312,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -318,7 +330,7 @@ "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-v3\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } @@ -340,27 +352,27 @@ "cache" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-r1", "iconUrl": "", "prices": [ { "priceUnit": "每百万tokens", "price": "4", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "16", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.6", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -383,6 +395,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "Reasoning", "TG" @@ -399,7 +414,7 @@ "samples": { "openai": { "completionsAPI": { - "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-r1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", + "python": "from openai import OpenAI\nimport os\n\nclient = OpenAI(\n api_key=\"sk-xxx\", # 替换为你的百炼API Key\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-r1\",\n messages=[{\"role\": \"user\", \"content\": \"你是谁\"}],\n extra_body={\"enable_thinking\": True},\n stream=True,\n)\n\nfor chunk in completion:\n if not chunk.choices:\n continue\n delta = chunk.choices[0].delta\n if hasattr(delta, \"reasoning_content\") and delta.reasoning_content:\n print(delta.reasoning_content, end=\"\", flush=True)\n if hasattr(delta, \"content\") and delta.content:\n print(delta.content, end=\"\", flush=True)", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } @@ -420,21 +435,20 @@ "structured-outputs" ], "provider": "deepseek", - "limit": { - "message": "model not exist" - }, "model": "vanchin/deepseek-ocr", "iconUrl": "", "prices": [ { "priceUnit": "每百万tokens", "price": "0.216", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "0.216", + "timeBand": "standard", "type": "output_token", "priceName": "输出" } @@ -457,6 +471,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VU", "TG" @@ -472,9 +489,9 @@ "samples": { "openai": { "completionsAPI": { - "curl": "curl -X POST https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"vanchin/deepseek-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\"\n }\n ]\n }\n ]\n}'", - "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\",\n },\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\",\n },\n ],\n }\n ],\n)\n\nprint(completion.choices[0].message.content)", - "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nasync function main() {\n const completion = await openai.chat.completions.create({\n model: 'vanchin/deepseek-ocr',\n messages: [\n {\n role: 'user',\n content: [\n {\n type: 'image_url',\n image_url: {\n url: 'https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg',\n detail: 'high',\n },\n },\n {\n type: 'text',\n text: 'Read all the text in the image.',\n },\n ],\n },\n ],\n });\n\n console.log(completion.choices[0].message.content);\n}\n\nmain();", + "curl": "curl -X POST https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \\\n-H \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n-H \"Content-Type: application/json\" \\\n-d '{\n \"model\": \"vanchin/deepseek-ocr\",\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\"\n }\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\"\n }\n ]\n }\n ]\n}'", + "python": "import os\nfrom openai import OpenAI\n\nclient = OpenAI(\n api_key=os.getenv(\"DASHSCOPE_API_KEY\"),\n base_url=\"https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1\",\n)\n\ncompletion = client.chat.completions.create(\n model=\"vanchin/deepseek-ocr\",\n messages=[\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"type\": \"image_url\",\n \"image_url\": {\n \"url\": \"https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg\",\n \"detail\": \"high\",\n },\n },\n {\n \"type\": \"text\",\n \"text\": \"Read all the text in the image.\",\n },\n ],\n }\n ],\n)\n\nprint(completion.choices[0].message.content)", + "nodejs": "import OpenAI from \"openai\";\nimport process from 'process';\n\nconst openai = new OpenAI({\n apiKey: process.env.DASHSCOPE_API_KEY,\n baseURL: 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/compatible-mode/v1'\n});\n\nasync function main() {\n const completion = await openai.chat.completions.create({\n model: 'vanchin/deepseek-ocr',\n messages: [\n {\n role: 'user',\n content: [\n {\n type: 'image_url',\n image_url: {\n url: 'https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg',\n detail: 'high',\n },\n },\n {\n type: 'text',\n text: 'Read all the text in the image.',\n },\n ],\n },\n ],\n });\n\n console.log(completion.choices[0].message.content);\n}\n\nmain();", "docUrl": "https://help.aliyun.com/document_detail/3027089.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json index def2ebc4..9af6a681 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json +++ b/skills/bailian-docs-llm-wiki/models/groups/video-style-transform.json @@ -19,16 +19,21 @@ { "priceUnit": "每秒", "price": "0.2", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.5", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json index df5f005a..fda316eb 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json +++ b/skills/bailian-docs-llm-wiki/models/groups/videoretalk.json @@ -20,10 +20,14 @@ { "priceUnit": "每秒", "price": "0.08", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json index 53aa900c..0f35b352 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-image-models-market-place.json @@ -15,26 +15,26 @@ "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，对中英文字的精准渲染、UI/图表等设计细节的像素级还原，适合制作海报、信息图等。", "features": [], "provider": "vidu", - "limit": { - "message": "model not exist" - }, "model": "vidu/vidu-image_reference2image", "prices": [ { "priceUnit": "每张", "price": "0.625", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "1", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.46875", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -55,6 +55,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -68,7 +71,7 @@ "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/vidu-image_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/vidu-image_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3045893.html" } } @@ -87,26 +90,26 @@ "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，主打高速高质与低成本，成本比Pro降低约50%。", "features": [], "provider": "vidu", - "limit": { - "message": "model not exist" - }, "model": "vidu/viduq3-fast_reference2image", "prices": [ { "priceUnit": "每张", "price": "0.46875", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.78125", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.09375", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -127,6 +130,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -140,7 +146,7 @@ "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq3-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq3-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3045893.html" } } @@ -159,26 +165,26 @@ "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，擅长处理复杂逻辑，具备超强上下文一致性和工业级稳定性。适合专业设计、漫剧制作等。", "features": [], "provider": "vidu", - "limit": { - "message": "model not exist" - }, "model": "vidu/viduq2-pro_reference2image", "prices": [ { "priceUnit": "每张", "price": "0.9375", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" }, { "priceUnit": "每张", "price": "0.9375", + "timeBand": "standard", "type": "image_type_2k", "priceName": "图片生成（2K）" }, { "priceUnit": "每张", "price": "1.71875", + "timeBand": "standard", "type": "image_type_4k", "priceName": "图片生成（4K）" } @@ -199,6 +205,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -212,7 +221,7 @@ "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-pro_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-pro_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3045893.html" } } @@ -231,14 +240,12 @@ "description": "输入0-14张参考图片或文本描述，支持参考生图、文生图、图片编辑，语义理解能力大幅提升，支持更多风格。", "features": [], "provider": "vidu", - "limit": { - "message": "model not exist" - }, "model": "vidu/viduq2-fast_reference2image", "prices": [ { "priceUnit": "每张", "price": "0.28125", + "timeBand": "standard", "type": "image_type_1k", "priceName": "图片生成（1K）" } @@ -259,6 +266,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -272,7 +282,7 @@ "samples": { "dashscope": { "default": { - "curl": "curl --location 'https://llm-czal8nvvwb8d47ks.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", + "curl": "curl --location 'https://[workspace-id].cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \\\n--header 'X-DashScope-Async: enable' \\\n--header \"Authorization: Bearer $DASHSCOPE_API_KEY\" \\\n--header 'Content-Type: application/json' \\\n--data '{\n \"model\": \"vidu/viduq2-fast_reference2image\",\n \"input\": {\n \"messages\": [\n {\n \"role\": \"user\",\n \"content\": [\n {\n \"text\": \"一间有着精致窗户的花店,漂亮的木质门,摆放着花朵\"\n }\n ]\n }\n ]\n },\n \"parameters\": {\n \"size\":\"1024*1024\"\n }\n}'", "docUrl": "https://help.aliyun.com/document_detail/3045893.html" } } diff --git a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json index fddb6d41..9c6aeb50 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/vidu-models-market-place.json @@ -20,12 +20,14 @@ { "priceUnit": "每秒", "price": "0.75", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.90625", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -46,6 +48,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -83,6 +88,7 @@ { "priceUnit": "每秒", "price": "0.875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -103,6 +109,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -140,12 +149,14 @@ { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -166,6 +177,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -204,18 +218,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -238,6 +255,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -273,18 +293,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -307,6 +330,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -345,18 +371,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -379,6 +408,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -415,18 +447,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -449,6 +484,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -484,18 +522,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -518,6 +559,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -556,18 +600,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.4375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -590,6 +637,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -626,18 +676,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.34375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -660,6 +713,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -696,18 +752,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.34375", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -730,6 +789,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -767,18 +829,21 @@ { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -801,6 +866,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -837,18 +905,21 @@ { "priceUnit": "每秒", "price": "0.0875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -871,6 +942,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -906,18 +980,21 @@ { "priceUnit": "每秒", "price": "0.1125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.21875", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -940,6 +1017,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -978,18 +1058,21 @@ { "priceUnit": "每秒", "price": "0.0875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.46875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1012,6 +1095,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1048,18 +1134,21 @@ { "priceUnit": "每秒", "price": "0.21875", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.28125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.71875", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1082,6 +1171,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1117,18 +1209,21 @@ { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.625", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1151,6 +1246,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1186,18 +1284,21 @@ { "priceUnit": "每秒", "price": "0.15625", + "timeBand": "standard", "type": "video_ratio_540p", "priceName": "视频生成（540P）" }, { "priceUnit": "每秒", "price": "0.3125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.40625", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1220,6 +1321,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1255,12 +1359,14 @@ { "priceUnit": "每秒", "price": "0.78125", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.9375", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1283,6 +1389,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -1318,12 +1427,14 @@ { "priceUnit": "每秒", "price": "0.1", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.2", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -1346,6 +1457,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json index 97ce5685..142946c5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-edit.json @@ -23,6 +23,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -43,6 +44,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -85,6 +89,7 @@ { "priceUnit": "每张", "price": "0.5", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -105,6 +110,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -151,6 +159,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -173,6 +182,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -210,6 +222,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -232,6 +245,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -265,10 +281,14 @@ { "priceUnit": "每张", "price": "0.14", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json index 3c98f6f4..dc5a125c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-image-to-video.json @@ -24,12 +24,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -50,6 +52,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -97,24 +102,28 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.5", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -135,6 +144,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -175,18 +187,21 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "discount": 0.5, "type": "720P_batch", "priceName": "视频生成（720P Batch Chat）" @@ -194,6 +209,7 @@ { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "discount": 0.5, "type": "1080P_batch", "priceName": "视频生成（1080P Batch Chat）" @@ -215,6 +231,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -255,18 +274,21 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -289,6 +311,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -328,12 +353,14 @@ { "priceUnit": "每秒", "price": "0.14", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -354,6 +381,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -389,18 +419,21 @@ { "priceUnit": "每秒", "price": "0.1", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.2", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.48", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -423,6 +456,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -463,12 +499,14 @@ { "priceUnit": "每秒", "price": "0.4", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_pro", "priceName": "视频生成（pro）" } @@ -491,6 +529,9 @@ "async_user_concurrency_limit": 1 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -527,12 +568,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" }, { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "type": "video_ratio_pro", "priceName": "视频生成（pro）" } @@ -555,6 +598,9 @@ "async_user_concurrency_limit": 1 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -587,10 +633,14 @@ { "priceUnit": "每张", "price": "0.004", + "timeBand": "standard", "type": "image_detect_number", "priceName": "图片检测" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -625,16 +675,21 @@ { "priceUnit": "每秒", "price": "0.5", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.9", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -669,18 +724,21 @@ { "priceUnit": "每秒", "price": "0.1", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.2", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.48", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -703,6 +761,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -741,6 +802,7 @@ { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } @@ -761,6 +823,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -794,6 +859,7 @@ { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } @@ -814,6 +880,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -847,10 +916,14 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json index f1e7799b..50aa4cec 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-reference-to-video.json @@ -23,12 +23,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -49,6 +51,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -88,24 +93,28 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "0.5", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.15", + "timeBand": "standard", "type": "720P_no_audio", "priceName": "视频生成（720P 无声）" }, { "priceUnit": "每秒", "price": "0.25", + "timeBand": "standard", "type": "1080P_no_audio", "priceName": "视频生成（1080P 无声）" } @@ -126,6 +135,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -163,12 +175,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -189,6 +203,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json index 852b0911..1e8dda5a 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-image.json @@ -22,6 +22,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -44,6 +45,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -80,6 +84,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -102,6 +107,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -138,6 +146,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -158,6 +167,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -194,6 +206,7 @@ { "priceUnit": "每张", "price": "0.14", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -214,6 +227,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -249,6 +265,7 @@ { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -269,6 +286,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -304,6 +324,7 @@ { "priceUnit": "每张", "price": "0.14", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -324,6 +345,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -359,10 +383,14 @@ { "priceUnit": "每张", "price": "0.16", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], @@ -398,6 +426,7 @@ { "priceUnit": "每张", "price": "0.04", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } @@ -418,6 +447,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json index f4218289..3700e359 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-text-to-video.json @@ -23,12 +23,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -49,6 +51,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -90,18 +95,21 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "discount": 0.5, "type": "720P_batch", "priceName": "视频生成（720P Batch Chat）" @@ -109,6 +117,7 @@ { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "discount": 0.5, "type": "1080P_batch", "priceName": "视频生成（1080P Batch Chat）" @@ -130,6 +139,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -168,18 +180,21 @@ { "priceUnit": "每秒", "price": "0.3", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -202,6 +217,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -236,12 +254,14 @@ { "priceUnit": "每秒", "price": "0.14", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -262,6 +282,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -295,6 +318,7 @@ { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -315,6 +339,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], @@ -348,12 +375,14 @@ { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_480p", "priceName": "视频生成（480P）" }, { "priceUnit": "每秒", "price": "0.24", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" } @@ -374,6 +403,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json index 77815faf..f1126d6c 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wan-video-edit.json @@ -24,12 +24,14 @@ { "priceUnit": "每秒", "price": "0.6", + "timeBand": "standard", "type": "video_ratio_720p", "priceName": "视频生成（720P）" }, { "priceUnit": "每秒", "price": "1", + "timeBand": "standard", "type": "video_ratio_1080p", "priceName": "视频生成（1080P）" } @@ -50,6 +52,9 @@ "async_user_concurrency_limit": 5 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json index d3845269..9e5d1660 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-background-generation-v2.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.08", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json index 3e0b63b6..10bfcb49 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-sketch-to-image-lite.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.06", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json index 2271a317..da9ab951 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx-style-repaint-v1.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.12", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json index 870443af..e0026151 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wanx2.1-vace-plus.json @@ -21,6 +21,7 @@ { "priceUnit": "每秒", "price": "0.7", + "timeBand": "standard", "type": "video_ratio", "priceName": "视频生成（std）" } @@ -41,6 +42,9 @@ "async_user_concurrency_limit": 2 } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "VG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json index 0c1096c5..bde791f5 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-semantic.json @@ -19,10 +19,14 @@ { "priceUnit": "每张", "price": "0.24", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json index 2f4ee929..a9a14eb7 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json +++ b/skills/bailian-docs-llm-wiki/models/groups/wordart-texture.json @@ -20,10 +20,14 @@ { "priceUnit": "每张", "price": "0.08", + "timeBand": "standard", "type": "image_number", "priceName": "图片生成" } ], + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json index b97bce31..74d5362f 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/xiaomi-models-market-place.json @@ -23,18 +23,21 @@ { "priceUnit": "每百万tokens", "price": "7", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "21", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.4", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -57,6 +60,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json index 9a3083df..d22f5a8e 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json +++ b/skills/bailian-docs-llm-wiki/models/groups/z-image-turbo.json @@ -22,12 +22,14 @@ { "priceUnit": "每张", "price": "0.1", + "timeBand": "standard", "type": "image_standard", "priceName": "图片生成（标准）" }, { "priceUnit": "每张", "price": "0.2", + "timeBand": "standard", "type": "image_thinking", "priceName": "图片生成（思考）" } @@ -44,6 +46,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "IG" ], diff --git a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json index f3432ef9..3c10f860 100644 --- a/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json +++ b/skills/bailian-docs-llm-wiki/models/groups/zhipu-models-market-place.json @@ -24,18 +24,21 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -58,6 +61,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" @@ -102,18 +108,21 @@ { "priceUnit": "每百万tokens", "price": "8", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "28", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "2", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -136,6 +145,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG" ], @@ -181,18 +193,21 @@ { "priceUnit": "每百万tokens", "price": "6", + "timeBand": "standard", "type": "input_token", "priceName": "输入" }, { "priceUnit": "每百万tokens", "price": "22", + "timeBand": "standard", "type": "output_token", "priceName": "输出" }, { "priceUnit": "每百万tokens", "price": "1.5", + "timeBand": "standard", "type": "input_token_cache", "priceName": "输入（缓存命中）" } @@ -215,6 +230,9 @@ "type": "model-default" } }, + "priceTimeBands": [ + "standard" + ], "capabilities": [ "TG", "Reasoning" diff --git a/skills/bailian-docs-llm-wiki/models/index.json b/skills/bailian-docs-llm-wiki/models/index.json index 325c6806..d1a34c3e 100644 --- a/skills/bailian-docs-llm-wiki/models/index.json +++ b/skills/bailian-docs-llm-wiki/models/index.json @@ -1,7 +1,7 @@ { "updatedAt": "2026-07-23", "totalFamilies": 170, - "totalModels": 386, + "totalModels": 387, "capabilityDistribution": { "TG": 35, "IG": 31, @@ -11,16 +11,17 @@ "ASR": 12, "VU": 9, "Realtime-ASR": 7, - "Multimodal-Omni": 5, "Realtime-Omni": 4, + "Multimodal-Omni": 4, "Realtime-Audio-Translate": 3, "ME": 2, "Realtime-Chatting": 2, "Realtime-Text-to-Speech": 2, - "TR": 2 + "TR": 2, + "3D-generation": 1 }, "providerDistribution": { - "qwen": 102, + "qwen": 101, "qwen-domain-model": 34, "wan": 13, "happyhorse": 4, @@ -32,6 +33,7 @@ "vidu": 2, "kling": 1, "stepfun": 1, + "tripo": 1, "xiaomi": 1 }, "families": [ @@ -2060,22 +2062,6 @@ ], "maxContextWindow": 262144 }, - { - "slug": "qwen3.5-omni-flash", - "name": "Qwen3.5-Omni-Flash", - "primaryCapability": "Multimodal-Omni", - "capabilities": [ - "Multimodal-Omni" - ], - "providers": [ - "qwen" - ], - "itemCount": 1, - "items": [ - "qwen3.5-omni-flash" - ], - "maxContextWindow": 262144 - }, { "slug": "qwen3.5-omni-plus-realtime", "name": "Qwen3.5-Omni-Plus-Realtime", @@ -2480,6 +2466,22 @@ ], "maxContextWindow": 32768 }, + { + "slug": "tripo-models-market-place", + "name": "Tripo", + "primaryCapability": "3D-generation", + "capabilities": [ + "3D-generation" + ], + "providers": [ + "tripo" + ], + "itemCount": 2, + "items": [ + "Tripo/Tripo-H3.1", + "Tripo/Tripo-P1.0" + ] + }, { "slug": "vanchin-models-market-place", "name": "Vanchin DeepSeek", diff --git a/skills/bailian-docs-llm-wiki/models/index.md b/skills/bailian-docs-llm-wiki/models/index.md index b2ba1b4c..df6d98d0 100644 --- a/skills/bailian-docs-llm-wiki/models/index.md +++ b/skills/bailian-docs-llm-wiki/models/index.md @@ -1,6 +1,6 @@ # 百炼模型市场索引 -> 自动生成 · 共 170 个模型家族 · 386 个主干模型 · 更新于 2026-07-23 +> 自动生成 · 共 170 个模型家族 · 387 个主干模型 · 更新于 2026-07-23 **机器查询走结构化文件**： @@ -335,19 +335,6 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Qwen3-LiveTranslate-Flash](groups/qwen3-livetranslate-flash.json) — Qwen3-LiveTranslate-Flash，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言跨模态对齐和视觉增强等技术，… - 模型：`qwen3-livetranslate-flash` -## 全模态 `Multimodal-Omni` — 5 个家族 - -- [Qwen-Omni-Turbo](groups/qwen-omni-turbo.json) — 千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 - - 模型：`qwen-omni-turbo`, `qwen-omni-turbo-latest` -- [Qwen2.5-开源模型](groups/qwen2.5.json) — Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。 - - 模型：`qwen2.5-omni-7b` -- [Qwen3-Omni-Flash](groups/qwen3-omni-flash.json) — Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互… - - 模型：`qwen3-omni-flash` -- [Qwen3.5-Omni-Flash](groups/qwen3.5-omni-flash.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… - - 模型：`qwen3.5-omni-flash` -- [Qwen3.5-Omni-Plus](groups/qwen3.5-omni-plus.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… - - 模型：`qwen3.5-omni-plus` - ## 实时全模态 `Realtime-Omni` — 4 个家族 - [Qwen-Omni-Turbo-Realtime](groups/qwen-omni-turbo-realtime.json) — 千问全新多模态理解生成大模型实时版，适合实时音频交互场景。支持音频伴随文本、图像、视频混合输入理解，具备语音和文本同时流式生成能力，提供了4种自然对话音色。 @@ -359,6 +346,17 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - [Qwen3.5-Omni-Plus-Realtime](groups/qwen3.5-omni-plus-realtime.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本，支持60+种语言音频输入，30+语言语音输出以及可控语音对话… - 模型：`qwen3.5-omni-plus-realtime` +## 全模态 `Multimodal-Omni` — 4 个家族 + +- [Qwen-Omni-Turbo](groups/qwen-omni-turbo.json) — 千问全新多模态理解生成大模型，支持文本, 图像，语音，视频输入理解和混合输入理解，具备文本和语音同时流式生成能力，多模态内容理解速度显著提升，提供了4种自然对话音色。 + - 模型：`qwen-omni-turbo`, `qwen-omni-turbo-latest` +- [Qwen2.5-开源模型](groups/qwen2.5.json) — Qwen2.5系列开源模型，包含文本生成模型、视觉理解模型、多模态模型等多个领域领先模型。 + - 模型：`qwen2.5-omni-7b` +- [Qwen3-Omni-Flash](groups/qwen3-omni-flash.json) — Qwen3-Omni-Flash多模态大模型，基于Thinker–Talker混合专家（MoE）架构，支持文本、图像、音频、视频的高效理解与语音生成能力，可进行119种语言文本交互和20种语言语音交互… + - 模型：`qwen3-omni-flash` +- [Qwen3.5-Omni-Plus](groups/qwen3.5-omni-plus.json) — Qwen3.5-Omni是Qwen最新一代全模态大模型，支持文本，图片，音频，音视频理解与交互。作为 Qwen3-Omni 的全面进化版本， 支持超过 10 小时的音频理解及超过 400 秒的 720… + - 模型：`qwen3.5-omni-plus` + ## 实时音频翻译 `Realtime-Audio-Translate` — 3 个家族 - [Qwen3-LiveTranslate-Flash-Realtime](groups/qwen3-livetranslate-flash-realtime.json) — Qwen3-LiveTranslate-Flash-Realtime的实时版本，一款高精度、高响应、高鲁棒性的多语言实时音视频同传大模型。依托Qwen3-Omni强大的基座能力、海量多模态数据、跨语言… @@ -395,3 +393,8 @@ join：`models.jsonl[].family == families.jsonl[].slug == index.json.families[]. - 模型：`qwen3.7-text-embedding`, `text-embedding-async-v1`, `text-embedding-async-v2`, `text-embedding-v1`, `text-embedding-v2`, `text-embedding-v3`, `text-embedding-v4` - [Qwen-Rerank](groups/qwen-rerank.json) — 基于Qwen LLM底座训练的文本排序模型，对输入的Query和候选Docs进行相关性排序，支持100+语种和长文本输入，适用于文本检索、RAG等场景，效果对齐Qwen家族开源Rerank系列模型。 - 模型：`gte-rerank-v2`, `qwen3-rerank`, `qwen3-vl-rerank` + +## 3D 生成 `3D-generation` — 1 个家族 + +- [Tripo](groups/tripo-models-market-place.json) — AI驱动的3D通用大模型Tripo，支持文本或图片输入，数秒内一键生成高质量3D模型。 + - 模型：`Tripo/Tripo-H3.1`, `Tripo/Tripo-P1.0` diff --git a/skills/bailian-docs-llm-wiki/models/models.jsonl b/skills/bailian-docs-llm-wiki/models/models.jsonl index 2fb2a1a5..c2dddf53 100644 --- a/skills/bailian-docs-llm-wiki/models/models.jsonl +++ b/skills/bailian-docs-llm-wiki/models/models.jsonl @@ -208,7 +208,6 @@ {"model":"qwen3.5-livetranslate-flash-realtime","name":"Qwen3.5-LiveTranslate-Flash-Realtime","family":"qwen3.5-livetranslate-flash-realtime","familyName":"Qwen3.5-LiveTranslate-Flash-Realtime","provider":"qwen-domain-model","capabilities":["Realtime-Audio-Translate"],"features":[],"contextWindow":53248,"maxInputTokens":49152,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Audio","Text"],"request_modality":["Audio","Image"]},"prices":[{"type":"translate_audio_input_token","unit":"每百万tokens","price":"40"},{"type":"translate_vision_input_token","unit":"每百万tokens","price":"3.3"},{"type":"translate_multi_text_output_token","unit":"每百万tokens","price":"100"},{"type":"translate_multi_output_token","unit":"每百万tokens","price":"160"}],"qpmInfo":{"model-default-actual":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":10,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://www.alibabacloud.com/help/en/document_detail/2983281.html","detailPath":"groups/qwen3.5-livetranslate-flash-realtime.json"} {"model":"qwen3.5-ocr","name":"Qwen3.5-OCR","family":"qwen3.5-ocr","familyName":"Qwen3.5-OCR","provider":"qwen","capabilities":["VU"],"features":["model-experience"],"contextWindow":65536,"maxInputTokens":49152,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.5"},{"type":"output_token","unit":"每百万tokens","price":"2"}],"qpmInfo":{"model-default-actual":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":100,"count_limit_period":1,"usage_limit":3000000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2860683.html","detailPath":"groups/qwen3.5-ocr.json"} {"model":"qwen3.5-omni-flash-realtime","name":"Qwen3.5-Omni-Flash-Realtime","family":"qwen3.5-omni-flash-realtime","familyName":"Qwen3.5-Omni-Flash-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"27"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"107"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"3.3"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"20"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2880812.html","detailPath":"groups/qwen3.5-omni-flash-realtime.json"} -{"model":"qwen3.5-omni-flash","name":"Qwen3.5-Omni-Flash","family":"qwen3.5-omni-flash","familyName":"Qwen3.5-Omni-Flash","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"18"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"72"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"2.2"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"13.3"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-flash.json"} {"model":"qwen3.5-omni-plus-realtime","name":"Qwen3.5-Omni-Plus-Realtime","family":"qwen3.5-omni-plus-realtime","familyName":"Qwen3.5-Omni-Plus-Realtime","provider":"qwen","capabilities":["Realtime-Omni"],"features":["web-search","function-calling"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"80"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"300"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"10"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"60"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus-realtime.json"} {"model":"qwen3.5-omni-plus","name":"Qwen3.5-Omni-Plus","family":"qwen3.5-omni-plus","familyName":"Qwen3.5-Omni-Plus","provider":"qwen","capabilities":["Multimodal-Omni"],"features":["web-search","function-calling","batch"],"contextWindow":262144,"maxInputTokens":196608,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text","Audio"],"request_modality":["Text","Image","Video","Audio"]},"prices":[{"type":"omni_audio_input_token","unit":"每百万tokens","price":"53"},{"type":"omni_audio_output_token","unit":"每百万tokens","price":"213"},{"type":"omni_no_audio_input_token","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token","unit":"每百万tokens","price":"40"},{"type":"omni_audio_input_token_batch","unit":"每百万tokens","price":"26.5"},{"type":"omni_no_audio_input_token_batch","unit":"每百万tokens","price":"3.5"},{"type":"omni_no_audio_output_token_batch","unit":"每百万tokens","price":"20"},{"type":"omni_audio_input_token_batch_chat","unit":"每百万tokens","price":"53"},{"type":"omni_no_audio_input_token_batch_chat","unit":"每百万tokens","price":"7"},{"type":"omni_no_audio_output_token_batch_chat","unit":"每百万tokens","price":"40"}],"qpmInfo":{"model-default-actual":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":60,"count_limit_period":60,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2867839.html","detailPath":"groups/qwen3.5-omni-plus.json"} {"model":"qwen3.5-plus","name":"Qwen3.5-Plus","family":"qwen3.5-plus","familyName":"Qwen3.5-Plus","provider":"qwen","capabilities":["TG","Reasoning","VU"],"features":["model-experience","function-calling","structured-outputs","web-search","prefix-completion","cache","batch"],"contextWindow":1000000,"maxInputTokens":991808,"maxOutputTokens":65536,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image","Video"]},"qpmInfo":{"model-default-actual":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30},"model-default":{"count_limit":500,"count_limit_period":1,"usage_limit":2500000,"usage_limit_field":"total_tokens","usage_limit_period":30}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2712576.html","detailPath":"groups/qwen3.5-plus.json"} @@ -298,6 +297,8 @@ {"model":"tongyi-intent-detect-v3","name":"意图分类模型","family":"tongyi-intent-detect-v3","familyName":"意图分类模型","provider":"qwen","capabilities":["TG"],"features":[],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.4"},{"type":"output_token","unit":"每百万tokens","price":"1"}],"qpmInfo":{"model-default-actual":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6},"model-default":{"count_limit":20,"count_limit_period":1,"usage_limit":100000,"usage_limit_field":"total_tokens","usage_limit_period":6}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/2861138.html","detailPath":"groups/tongyi-intent-detect-v3.json"} {"model":"tongyi-xiaomi-analysis-flash","name":"通义晓蜜-对话分析-flash","family":"tongyi-xiaomi-analysis-flash","familyName":"通义晓蜜-对话分析-flash","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.2"},{"type":"output_token","unit":"每百万tokens","price":"0.4"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-flash.json"} {"model":"tongyi-xiaomi-analysis-pro","name":"通义晓蜜-对话分析-pro","family":"tongyi-xiaomi-analysis-pro","familyName":"通义晓蜜-对话分析-pro","provider":"qwen-domain-model","capabilities":["TG"],"features":[],"contextWindow":32768,"maxInputTokens":28672,"maxOutputTokens":4096,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"1"},{"type":"output_token","unit":"每百万tokens","price":"2.7"}],"qpmInfo":{"model-default-actual":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":600,"count_limit_period":60,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3015075.html","detailPath":"groups/tongyi-xiaomi-analysis-pro.json"} +{"model":"Tripo/Tripo-H3.1","name":"Tripo-H3.1","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","batch","structured-outputs"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_standard_no_texture","unit":"每次","price":"0.7"},{"type":"generation_3d_image_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_multiview_to_3d_standard_no_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_text_to_3d_standard_sd_texture","unit":"每次","price":"1.4"},{"type":"generation_3d_image_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_multiview_to_3d_standard_sd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_text_to_3d_standard_hd_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_standard_hd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_ultra_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_ultra_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_ultra_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_ultra_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_ultra_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} +{"model":"Tripo/Tripo-P1.0","name":"Tripo-P1.0","family":"tripo-models-market-place","familyName":"Tripo","provider":"tripo","capabilities":["3D-generation"],"features":["function-calling","structured-outputs","batch"],"inferenceMetadata":{"response_modality":["3D-Generation"],"request_modality":["Text"]},"prices":[{"type":"generation_3d_text_to_3d_no_texture","unit":"每次","price":"2.1"},{"type":"generation_3d_image_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_multiview_to_3d_no_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_text_to_3d_sd_texture","unit":"每次","price":"2.8"},{"type":"generation_3d_image_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_multiview_to_3d_sd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_text_to_3d_hd_texture","unit":"每次","price":"3.5"},{"type":"generation_3d_image_to_3d_hd_texture","unit":"每次","price":"4.2"},{"type":"generation_3d_multiview_to_3d_hd_texture","unit":"每次","price":"4.2"}],"qpmInfo":{"model-default-actual":{"count_limit":5,"count_limit_period":60},"model-default":{"count_limit":5,"count_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3030679.html","detailPath":"groups/tripo-models-market-place.json"} {"model":"vanchin/deepseek-ocr","name":"Vanchin/DeepSeek-OCR","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["VU","TG"],"features":["structured-outputs"],"contextWindow":8192,"maxInputTokens":8192,"maxOutputTokens":8192,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text","Image"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"0.216"},{"type":"output_token","unit":"每百万tokens","price":"0.216"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-r1","name":"Vanchin/DeepSeek-R1","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["Reasoning","TG"],"features":["function-calling","prefix-completion","cache"],"contextWindow":131072,"maxInputTokens":98304,"maxOutputTokens":32768,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"4"},{"type":"output_token","unit":"每百万tokens","price":"16"},{"type":"input_token_cache","unit":"每百万tokens","price":"1.6"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} {"model":"vanchin/deepseek-v3","name":"Vanchin/DeepSeek-V3","family":"vanchin-models-market-place","familyName":"Vanchin DeepSeek","provider":"deepseek","capabilities":["TG"],"features":["structured-outputs","prefix-completion","function-calling","cache"],"contextWindow":131072,"maxInputTokens":131072,"maxOutputTokens":16384,"inferenceMetadata":{"response_modality":["Text"],"request_modality":["Text"]},"prices":[{"type":"input_token","unit":"每百万tokens","price":"2"},{"type":"output_token","unit":"每百万tokens","price":"8"},{"type":"input_token_cache","unit":"每百万tokens","price":"0.8"}],"qpmInfo":{"model-default-actual":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60},"model-default":{"count_limit":50,"count_limit_period":6,"usage_limit":1000000,"usage_limit_field":"total_tokens","usage_limit_period":60}},"versionTag":"MAJOR","docUrl":"https://help.aliyun.com/document_detail/3027089.html","detailPath":"groups/vanchin-models-market-place.json"} diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md new file mode 100644 index 00000000..07acc02b --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md @@ -0,0 +1,246 @@ +# BatchUpdateFileTag - 批量更新文档标签 + +该接口用于批量更新数据连接中的文档标签。 + +## 调试 + +[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/BatchUpdateFileTag) + + [![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png) 调试](https://api.aliyun.com/api/bailian/2023-12-29/BatchUpdateFileTag) + +## **授权信息** + +当前API暂无授权信息透出。 + +## 请求语法 + +``` +PUT /{WorkspaceId}/datacenter/batchupdatetag HTTP/1.1 +``` + +## 路径参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +WorkspaceId + +string + +是 + +业务空间 ID。在百炼的[控制台首页](https://bailian.console.aliyun.com/knowledge-base#/home)，单击页面左上角业务空间详情图标获取。 + +llm-3shx2gu255oqxxxx + +## 请求参数 + +**名称** + +**类型** + +**必填** + +**描述** + +**示例值** + +FileInfos + +array + +是 + +需要更新的文档列表 + +object + +是 + +FileId + +string + +是 + +数据中心的文件 ID，您可以在[应用数据](https://bailian.console.aliyun.com/?tab=app#/data-center)页面，单击文件名称旁的 ID 图标获取。 + +file\_3d5319366e2c46309f4c11cfbeacd5fd\_10045951 + +tags + +array + +是 + +- 文件关联的标签列表。最多传入 100 个标签，所有标签字符长度总和不能超过 700。 + + +string + +否 + +标签值，每个标签最多 32 个字符，支持 Unicode 中 letter 分类下的字符（其中包括英文、中文和数字等），下划线\_，中划线-，标签中不能包含空格。 + +TagA + +UpdateMode + +string + +否 + +更新模式，仅支持 APPEND（追加）和 OVERWRITE（覆盖） + +OVERWRITE + +## **返回参数** + +**名称** + +**类型** + +**描述** + +**示例值** + +object + +Schema of Response + +Code + +string + +错误状态码 + +Success + +Data + +object + +接口返回的业务字段。 + +UpdateFileTagResultList + +array + +标签更新的结果列表 + +object + +FileId + +string + +文件 ID。 + +file\_f40f2a32205d44b4a93b11617113da15\_10045951 + +Success + +boolean + +接口调用是否成功，可能值为： + +- true：成功。 + +- false：失败。 + + +true + +ErrorCode + +string + +返回错误码，仅当 Success 为 false 时返回。 + +NoPermission + +ErrorMessage + +string + +错误描述信息，仅当 Success 为 false 时返回。 + +FileId not exists. + +Message + +string + +错误信息 + +Required parameter(FileId) missing or invalid, please check the request parameters. + +RequestId + +string + +Id of the request + +17204B98-xxxx-4F9A-8464-2446A84821CA + +Status + +string + +接口返回的状态码。 + +200 + +Success + +boolean + +接口调用是否成功，可能值： + +- true：成功。 + +- false：失败。 + + +true + +## 示例 + +正常返回示例 + +`JSON`格式 + +``` +{ + "Code": "Success", + "Data": { + "UpdateFileTagResultList": [ + { + "FileId": "file_f40f2a32205d44b4a93b11617113da15_10045951", + "Success": true, + "ErrorCode": "NoPermission", + "ErrorMessage": "FileId not exists." + } + ] + }, + "Message": "Required parameter(FileId) missing or invalid, please check the request parameters.", + "RequestId": "17204B98-xxxx-4F9A-8464-2446A84821CA", + "Status": "200", + "Success": true +} +``` + +## 错误码 + +访问[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)查看更多错误码。 + +## **变更历史** + +更多信息，参考[变更详情](https://api.aliyun.com/document/bailian/2023-12-29/BatchUpdateFileTag#workbench-doc-change-demo)。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md deleted file mode 100644 index ae0dfe52..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md +++ /dev/null @@ -1,178 +0,0 @@ -# CreateMemoryNode - 创建记忆片段 - -创建记忆片段。 - -## 调试 - -[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/CreateMemoryNode) - -[![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png)调试](https://api.aliyun.com/api/bailian/2023-12-29/CreateMemoryNode) - -## 授权信息 - -下表是API对应的授权信息，可以在RAM权限策略语句的`Action`元素中使用，用来给RAM用户或RAM角色授予调用此API的权限。具体说明如下： - -- 操作：是指具体的权限点。 -- 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 -- 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 - - 对于不支持资源级授权的操作，用`全部资源`表示。 -- 条件关键字：是指云产品自身定义的条件关键字。 -- 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 - -操作 - -访问级别 - -资源类型 - -条件关键字 - -关联操作 - -sfm:CreateMemoryNode - -create - -\*全部资源 - -`*` - -无 - -无 - -## 请求语法 - -``` -POST /{workspaceId}/memories/{memoryId}/memoryNodes HTTP/1.1 -``` - -## 请求参数 - -名称 - -类型 - -必填 - -描述 - -示例值 - -workspaceId - -string - -否 - -长期记忆体所属的业务空间 ID。获取方式请参见[如何使用业务空间](https://help.aliyun.com/zh/model-studio/use-workspace)。 - -llm-us9hjmt32nysdm5v - -memoryId - -string - -是 - -长期记忆体 ID。即 **CreateMemory** 接口返回的`memoryId`。 - -6bff4f317a14442fbc9f73d29dbd5fc3 - -content - -string - -是 - -记忆片段内容。长度为 1~200 个字符，支持中文、英文、数字、下划线（\_）、短划线（-）、半角句号（.）和半角冒号（:）。 - -用户喜欢吃西红柿炒鸡蛋 - -## 返回参数 - -名称 - -类型 - -描述 - -示例值 - -object - -Schema of Response - -memoryNodeId - -string - -记忆片段 ID。 - -68de06c95368463a8be4a84efcxxxxxx - -requestId - -string - -请求 ID。 - -8C56C7AF-xxxx-19CE-B018-E05E1EDCF4C5 - -## 示例 - -正常返回示例 - -`JSON`格式 - -``` -{ - "memoryNodeId": "68de06c95368463a8be4a84efcxxxxxx", - "requestId": "8C56C7AF-xxxx-19CE-B018-E05E1EDCF4C5" -} -``` - -## 错误码 - -HTTP status code - -错误码 - -错误信息 - -描述 - -400 - -Memory.MemoryNodeContentInvalid - -Memory node content is invalid. - -长期记忆节点的内容无效 - -404 - -Memory.MemoryIdNotFound - -Memory Id not exist or is not authorized. - -memoryId 未找到 - -404 - -Memory.MemoryNodeNotFound - -MemoryNode not found. - -长期记忆节点未找到 - -500 - -Memory.InternalError - -Memory service inner exception. - -长期记忆服务内部异常。 - -访问[错误中心](< https://api.aliyun.com/document/bailian/2023-12-29/errorCode>)查看更多错误码。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md deleted file mode 100644 index c0601157..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md +++ /dev/null @@ -1,176 +0,0 @@ -# GetMemory - 获取长期记忆体 - -获取指定长期记忆体的描述信息。 - -## 接口说明 - -- 本接口具有幂等性。 - -**限流说明：** 请确保两次请求间隔至少 1 秒，否则可能触发系统限流。如遇限流，请稍后重试。 - -## 调试 - -[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/bailian/2023-12-29/GetMemory) - -[![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png)调试](https://api.aliyun.com/api/bailian/2023-12-29/GetMemory) - -## 授权信息 - -下表是API对应的授权信息，可以在RAM权限策略语句的`Action`元素中使用，用来给RAM用户或RAM角色授予调用此API的权限。具体说明如下： - -- 操作：是指具体的权限点。 -- 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 -- 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 - - 对于不支持资源级授权的操作，用`全部资源`表示。 -- 条件关键字：是指云产品自身定义的条件关键字。 -- 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 - -操作 - -访问级别 - -资源类型 - -条件关键字 - -关联操作 - -sfm:GetMemory - -get - -\*全部资源 - -`*` - -无 - -无 - -## 请求语法 - -``` -GET /{workspaceId}/memories/{memoryId} HTTP/1.1 -``` - -## 请求参数 - -名称 - -类型 - -必填 - -描述 - -示例值 - -workspaceId - -string - -是 - -长期记忆体所属的业务空间 ID。获取方式请参见[如何使用业务空间](https://help.aliyun.com/zh/model-studio/use-workspace)。 - -llm-3z7uw7fwz0vexxxx - -memoryId - -string - -是 - -长期记忆体 ID，对应 [CreateMemory](https://help.aliyun.com/zh/model-studio/developer-reference/api-bailian-2023-12-29-creatememory) 接口返回的`memoryId`。 - -6bff4f317a14442fbc9f73d29dbxxxx - -## 返回参数 - -名称 - -类型 - -描述 - -示例值 - -object - -Schema of Response - -description - -string - -长期记忆体的描述信息。 - -我的大模型应用$APP\_ID关于A用户的长期记忆体 - -memoryId - -string - -长期记忆体 ID。 - -6bff4f317a14442fbc9f73d29dbdxxxx - -requestId - -string - -请求 ID。 - -6a71f2d9-f1c9-913b-818b-11402910xxxx - -workspaceId - -string - -长期记忆体所属的业务空间 ID。 - -llm-3z7uw7fwz0vexxxx - -## 示例 - -正常返回示例 - -`JSON`格式 - -``` -{ - "description": "我的大模型应用$APP_ID关于A用户的长期记忆体", - "memoryId": "6bff4f317a14442fbc9f73d29dbdxxxx", - "requestId": "6a71f2d9-f1c9-913b-818b-11402910xxxx", - "workspaceId": "llm-3z7uw7fwz0vexxxx" -} -``` - -## 错误码 - -HTTP status code - -错误码 - -错误信息 - -描述 - -404 - -Memory.MemoryIdNotFound - -Memory Id not exist or is not authorized. - -memoryId 未找到 - -500 - -Memory.InternalError - -Memory service inner exception. - -长期记忆服务内部异常。 - -访问[错误中心](< https://api.aliyun.com/document/bailian/2023-12-29/errorCode>)查看更多错误码。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md index 24c1ae25..3f7490cf 100644 --- a/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md +++ b/skills/bailian-docs-llm-wiki/raw/application-api-reference/more/how-to-use-search-filters.md @@ -2565,4 +2565,4 @@ RAM用户（子账号）请先获取阿里云百炼的数据权限再调用[Retr ## 错误码 -如果调用失败并收到报错信息，请参见[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)进行解决。 \ No newline at end of file +如果调用失败并收到报错信息，请参见[错误中心](https://api.aliyun.com/document/bailian/2023-12-29/errorCode)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-evaluation/application-auto-evaluation.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-evaluation/application-auto-evaluation.md deleted file mode 100644 index b06a5907..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-evaluation/application-auto-evaluation.md +++ /dev/null @@ -1,242 +0,0 @@ -# 自动评测 - -人工评测智能体应用需要手动构建评测集，耗时费力，同时评测结果依赖领域专家的判断，虽保证了专业性，但过程难以量化且可能带入个人主观偏好。阿里云百炼提供了自动评测功能，利用大模型、基于应用的知识库自动生成评测集，评估智能体的回答并生成评测报告与调优建议。自动评测支持两种模式： - -- **单应用评测**：深度评估单个智能体应用的表现，生成包含评分、错误分析和优化建议的详细报告，用于快速发现问题并针对性优化。 - -- **多应用横向评测**：在同一评测基准下，对比评估多个应用（或同一应用的不同版本）的各项指标，用于选型决策或版本迭代效果验证。 - - -## 前提条件 - -1. 自动评测仅面向已发布的[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)。应用发布请参考[应用分享](https://help.aliyun.com/zh/model-studio/share-an-application)。 - -2. 自动评测将基于知识库自动生成评测集，请确保应用已配置知识库。详情可参考[创建和使用知识库](https://help.aliyun.com/zh/model-studio/rag-knowledge-base)。 - -3. 智能体应用评测依赖于智能体推理的过程数据，请确保已开通`应用观测`功能，并将需要评测的应用添加到观测列表中。详情可参考[用量监控与性能分析](https://help.aliyun.com/zh/model-studio/application-observation)。 - -4. 子账号（RAM用户）需获取`管理员`或`应用评测-操作`权限，才能够使用自动评测功能。详情可参考[页面权限](https://help.aliyun.com/zh/model-studio/member-management#febd776ce5lbx)。 - - -## 操作流程 - -完成一次完整的自动评测需要经历四个阶段：创建评测任务、设置评测集、配置评测规则和执行评测。 - -### **创建评测任务** - -1. 进入阿里云百炼控制台[自动评测](https://bailian.console.aliyun.com/?&tab=app&scm=20140722.S_%E7%99%BE%E7%82%BCprompt._.RL_%E7%99%BE%E7%82%BCprompt-LOC_aillm-OR_chat-V_3-RC_llm#/efm/app_evaluate/tabs)界面，单击**创建评测任务**。 - - 若尚未开通应用观测，请在弹出窗口单击**立即前往**完成开通。 - - -2. 选择需要评测的智能体应用。可以选择 1 个应用进行评测，或选择**最多 8 个**应用进行横向评测。 - - **说明** - - - 仅支持选择**已发布**且**配置知识库**的应用。 - - - 同一智能体的不同版本将被视为不同的、独立的应用。 - - - 进行多应用横向评测时，所有被选应用必须都已关联了至少一个相同的知识库。 - - - 在**选择智能体**页面，左栏显示智能体列表，勾选目标智能体后，右栏自动展示该智能体的可用版本列表，勾选需要评测的版本，然后单击**下一步**进入设置评测集步骤。 - -3. 选择用于评测的知识库。单应用评测时，可从该应用关联的所有知识库中选择一个或多个用于生成评测集；多应用横向评测时，系统会列出所有被选应用的公共知识库，选择一个或多个用于生成评测集。 - - 在**创建评测任务**页面的第一步**选择评测应用**中，先在**已选智能体**区域选择需要评测的应用，再在下方**选择知识库范围**区域勾选目标知识库，完成后单击**下一步**。 - -4. 确认选择无误后，单击**下一步**。若所选应用没有开通应用观测，请在弹出窗口单机**一键开通并进入下一步**。 - - -### **设置评测集** - -评测集选择支持**生成评测集**和**选择已有评测集**两种方式： - -- 生成评测集：基于上一步选中的知识库，由大模型自动生成评测集。所有生成的评测集均可在[评测集](https://bailian.console.aliyun.com/?&tab=app&scm=20140722.S_%E7%99%BE%E7%82%BCprompt._.RL_%E7%99%BE%E7%82%BCprompt-LOC_aillm-OR_chat-V_3-RC_llm#/efm/app_evaluate/tabs?tab=group)页面查看。 - -- 选择已有评测集：复用已有评测集，需确保所选评测集内各问题的参考答案，均能在当前指定的知识库中找到。否则将导致评测结果不准确。评测集格式请参考[评测集](https://help.aliyun.com/zh/model-studio/application-evaluation-dataset)。 - - -以下以生成评测集为例。 - -1. 输入评测集名称。 - -2. 选择任务类型。生成评测集时，需选择 2 至 8 种任务类型。系统默认提供“事实型”、“分析型”、“比较型”、“教程型”四种任务类型。此外也支持自定义新类型，单击**增加任务类型**，输入任务类型、描述和示例即可。 - - 页面提供**生成评测集**和**选择已有评测集**两种方式。选择生成评测集后，需填写**评测集名称**并确认**已选知识库**，完成任务类型配置后单击**生成评测集**按钮提交。 - -3. 选择用于生成评测任务的模型。为确保生成质量，目前仅支持使用`qwen-max`和`qwen-plus`模型。各模型能力请参考[选择模型](https://help.aliyun.com/zh/model-studio/models)，计费规则请参考[模型调用计费](https://help.aliyun.com/zh/model-studio/model-pricing)。 - -4. 模型选择完毕后，页面下方将显示模型 Token 的预估平均消耗和预估最大消耗。评测完成后，可以在[自动评测](https://bailian.console.aliyun.com/tab=app&scm=20140722.S_%E7%99%BE%E7%82%BCprompt._.RL_%E7%99%BE%E7%82%BCprompt-LOC_aillm-OR_chat-V_3-RC_llm#/efm/app_evaluate/tabs)页面查看实际 Token 消耗明细。 - - **说明** - - - **预估平均消耗**是**参考值**，最终用量请以实际账单为准。 - - - **预估最大消耗**是为防止意外的超长输出而设置的成本硬性上限，实际消耗通常远低于此值。 - - -5. 单击**生成评测集**，在弹出页面确认配置信息，然后单击**继续生成**。 - - 弹出的确认对话框中显示 **已选知识库**、**任务类型**（事实型、教程型、比较型、分析型）和 **评测模型选择** 三项配置，并提示评测集生成发起后不支持修改且会消耗较多tokens。 - -6. 等待评测集生成，生成状态可在[评测集](https://bailian.console.aliyun.com/?&tab=app&scm=20140722.S_%E7%99%BE%E7%82%BCprompt._.RL_%E7%99%BE%E7%82%BCprompt-LOC_aillm-OR_chat-V_3-RC_llm#/efm/app_evaluate/tabs?tab=group)管理页面查看。评测集生成的时间开销，主要取决于所选知识库的数量与文档总量。生成完毕后，单击**下一步**。 - - 进入**查看评测集**页面，页面显示评测集名称及版本信息，可单击**编辑评测集**进行修改。评测集以表格形式展示，包含**任务类型**、**用户query**、**参考答案**、**粗粒度关键词**和**细粒度关键词**列。确认评测集内容无误后，单击**下一步**进入评测规则设置。 - - -### **配置评测规则** - -1. 选择分类采样数。用于设置每个任务类型需要采样的问题数量。系统将从每个类型下随机抽取指定数量的问题用于最终评测。 - - 分类采样数包含**事实型(Factual)**、**教程型(Tutorial)**、**比较型(Comparative)**和**分析型(Analytical)**四个类型，可通过滑块分别设置采样数量。本示例中将**事实型**和**分析型**各设为1，其余设为0，**评测总数**为2。 - -2. 选择评测模型。为确保结果准确，目前仅支持使用`qwen-max`和`qwen-plus`模型，模型能力和计费规则请参考[选择模型](https://help.aliyun.com/zh/model-studio/models)和[模型调用计费](https://help.aliyun.com/zh/model-studio/model-pricing)。 - -3. 模型选择完毕后，下方将显示预估平均消耗和预估最大消耗。采样的评测数据越多，消耗的 Token 数量越大。评测完成后，可在[自动评测](https://bailian.console.aliyun.com/tab=app&scm=20140722.S_%E7%99%BE%E7%82%BCprompt._.RL_%E7%99%BE%E7%82%BCprompt-LOC_aillm-OR_chat-V_3-RC_llm#/efm/app_evaluate/tabs)页面查看实际 Token 消耗明细。 - - **说明** - - - **预估平均消耗**是**参考值**，最终用量请以实际账单为准。 - - - **预估最大消耗**是为防止意外的超长输出而设置的成本硬性上限，实际消耗通常远低于此值。 - - - 完成配置后，单击**试运行**可先验证评测流程，确认无误后单击**发起评测任务**开始正式评测。 - -4. 在正式评测之前，可以选择试运行以预览评测效果，**试运行仅支持单个应用的评测结果预览**。试运行将随机抽取一道题执行完整评测，此过程会消耗少量 Token。单击**试运行**，在弹窗中选择需要试运行评测的应用，再次单击**试运行**。 - - 查看试运行结果。 - - 试运行结果弹窗中展示每条评测用例的**用户query**、**参考答案**、**实际运行结果**、**任务类型**（如教程型、分析型、事实型）以及**大模型打分**（星级评分）。 - - -### **执行评测** - -1. 确认评测集与评测规则配置无误后，单击**发起评测任务**。 - - 在**设置评测规则**页面，通过滑块调整各评测维度的权重值（范围0~1），在**评测模型**下拉框中选择模型（如 `qwen-max`），页面将展示预估平均和最大token消耗。可单击**试运行**预览评测效果。 - -2. 在弹出窗口确认评测配置和预计消耗后，单击**开始评测**。 - - 等待评测完成。自动评测的时间开销，主要取决于评测样本的总规模。 - - 评测任务开始后，任务列表中该任务的评测状态显示为**评测中(X%)**，如需中途停止，可在操作列单击**终止**。 - -3. 评测完成后，可在任务列表中单击**追加评测**，为本次任务加入新的应用进行对比。此操作适用于单应用评测和多应用评测。追加后的应用总数不能超过 8 个。 - - -## 评测报告分析 - -### 评分机制 - -系统使用大模型对每个回答进行评分（1-5 分），评分时会对比智能体的输出和评测集中的参考答案，评估答案的准确性、完整性和相关性。评分规则如下： - -- **5 分：**答案正确，质量优秀**。** - -- **4 分**：答案正确，质量良好。 - -- **4 分以下**：答案错误，系统将自动进行归因分析。 - - -### 结果分析 - -1. **总正确率**：展示应用的整体表现评估，计算公式：总正确率 = 得分不低于4分的回答数量 / 总回答数量 × 100%。多应用评测会以图表形式展示各应用的对比。 - - 单应用： - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3639694571/p994661.png) - - 多应用： - - ![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3639694571/p993712.png) - -2. **BadCase 分析**：BadCase 列表默认按分数从低到高展示 Top-5 的错误评测条目，点击**查看全部数据**可以查看全部错误评测条目。 - - > 若无 BadCase，BadCase 分析列表将为空。如需查看全部结果，可单击页面右上角**下载评测结果**。 - - BadCase分析页面左侧为汇总区域，包含**大模型打分**和**问题分类**两个Tab页签，通过环形图展示不同分数段（2-4分、低于2分）的BadCase数量分布。右侧BadCase列表包含**任务类型**、**用户输入**、**应用输出**、**打分/问题分类**等列。 - -3. **调优建议**：系统会根据归因分析结果，提供针对Prompt、检索配置或知识库切片的具体优化建议。 - - 例如，针对**切片不完整**问题（通常由于创建知识库时选择的切分策略不合适，导致关键信息被切分到多个切片），系统建议：1. 根据参考答案长度调整最大分段长度和分段重叠长度；2. 选择自定义切分方式，自由调整切分策略（如按页码切分、按标题切分、按正则切分等）；3. 找到被错误切分的切片，手动调整切片信息，使得相关性强的内容在同一个切片内。 - -4. **RAG 智能体评价**：展示各问题类型（如事实型、分析型）的单项得分。 - - 评分标准分为表现优秀（>=4）、表现良好（2-4）和待提升（<2）三档。事实型（Factual）和比较型（Comparative）表现优秀占比均为100.0%；分析型（Analytical）表现优秀占比80.0%，表现良好占比20.0%；教程型（Tutorial）各项占比均为0.0%。 - - -### 归因分析 - -对于得分低于 4 分的 BadCase，系统会自动进行归因分析，定位问题出在 RAG 流程的哪个环节。各项归因类型的含义及优化建议如下： - -1. **模型理解有误**：已获取正确知识，但应用配置的提示词不明确或模型推理能力不足，导致答案错误。需补充更清晰的回答要求或切换更强的模型。 - -2. **重排不佳**：正确切片已被召回，但排序靠后未被包含在最终传递给大模型的上下文中。需调整重排配置或增加传递给模型的切片数量。 - -3. **检索无效**：通常是由于设置的检索策略不合适，导致召回了过多或过少的切片。需根据数据特点调整检索方式。 - -4. **切片不完整**：通常是由于创建知识库时，切分粒度过细导致一个完整的语义单元被分割到多个不同的切片中。需增大切片长度或启用语义切分。 - -5. **未获取知识**：可能是由于知识库召回无结果或缺失与问题相关的内容。需向知识库中补充相应知识。 - - -## 最佳实践 - -单次评测只能反映应用在特定时间点的表现。要持续保障智能体应用的质量，需要将自动评测融入日常的开发和运维流程。 - -### 建立持续评测机制 - -以下场景建议触发一次评测： - -- **知识库更新后**：新增、修改或删除知识内容可能影响检索和回答质量。 - -- **调整 Prompt 后**：提示词的变化直接影响模型的输出行为。 - -- **更换或升级模型后**：不同模型的理解和生成能力存在差异。 - -- **调整检索/重排策略后**：这些配置直接影响 RAG 流程的召回质量。 - -- **定期回归（每周或每月）**：即使没有主动变更，也建议定期评测以监控潜在的质量波动。 - - -### 建立优化闭环 - -建议按以下流程进行迭代： - -1. **识别 BadCase**：在评测报告中，重点关注得分低于 4 分的回答。 - -2. **分析归因，定位问题**：根据系统给出的归因类型，参考[归因分析](#c686b552abp3n)中的说明，快速定位每个 BadCase 的产生原因。 - -3. **实施针对性优化**：根据归因结果，修改对应的配置。 - -4. **发布新版本，再次评测**：将优化后的应用发布为新版本，使用同一评测集进行评测。 - -5. **对比结果，确认改进**：对比新旧版本的评测报告，验证优化效果。若优化效果未达预期或发现新问题，则返回第一步，开始新一轮优化循环。 - - -## 常见问题 - -### **模型 Token 的预估平均消耗和预估最大消耗有什么区别？为什么实际消耗与预估消耗不符？** - -1. **预估平均消耗**是**参考值**，最终用量请以实际账单为准。 - -2. **预估最大消耗**是**理论上限**，基于模型最大输入输出 Token 长度计算，实际消耗不会超过这个值。 - - -### **为什么评测集生成和应用评测的进度长时间保持在0%？** - -评测集生成和应用评测均为离线任务，需在后台排队执行，排队期间进度将保持0%。任务开始执行后，进度会自动更新。 - -### **评测任务运行时，关闭应用观测会有什么影响？** - -为确保评测任务正常运行，请勿在评测期间关闭应用观测，否则可能导致评测任务失败、数据丢失或最终评测报告不准确。 - -### 为什么评测报告中显示的用例数量与设置的不符？ - -自动评测可能会失败。评测报告只显示成功完成评测的任务用例，失败的任务不计入最终正确率的计算。 - -### **为什么评测任务失败了，还会消耗 Token？** - -评测任务是分步执行的。每个成功完成的步骤都会消耗 Token 并计费。如果任务在后续步骤失败，此前已消耗的 Token 仍然会计入用量。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md deleted file mode 100644 index c7bc8cab..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttaskresultquery.md +++ /dev/null @@ -1,777 +0,0 @@ -# PodcastTaskResultQuery - 播客任务结果查询 - -ai播客生成任务结果查询。 - -## 调试 - -[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/AIPodcast/2025-02-28/PodcastTaskResultQuery) - -[![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png)调试](https://api.aliyun.com/api/AIPodcast/2025-02-28/PodcastTaskResultQuery) - -## 授权信息 - -下表是API对应的授权信息，可以在RAM权限策略语句的`Action`元素中使用，用来给RAM用户或RAM角色授予调用此API的权限。具体说明如下： - -- 操作：是指具体的权限点。 -- 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 -- 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 - - 对于不支持资源级授权的操作，用`全部资源`表示。 -- 条件关键字：是指云产品自身定义的条件关键字。 -- 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 - -操作 - -访问级别 - -资源类型 - -条件关键字 - -关联操作 - -aipodcast:PodcastTaskResultQuery - -none - -\*全部资源 - -`*` - -无 - -无 - -## 请求语法 - -``` -POST /podcast/task HTTP/1.1 -``` - -## 请求参数 - -名称 - -类型 - -必填 - -描述 - -示例值 - -workspaceId - -string - -是 - -当前请求所使用的百炼业务空间 id - -llm-ep8ba0dr6seiddri - -taskId - -string - -是 - -任务唯一标识 - -63c4e0eaab3b4c0db208ecafa990e8d1 - -## 返回参数 - -名称 - -类型 - -描述 - -示例值 - -object - -Schema of Response - -code - -string - -响应状态码。 - -"success" - -message - -string - -响应消息。 - -"success" - -requestId - -string - -请求 id，用于追溯 API 调用链路。 - -C38F034D-7F36-531C-95AC-0C752F80E840 - -success - -boolean - -是否成功：true 成功，false 失败 - -True - -httpStatusCode - -string - -HTTP 状态码 - -200 - -data - -object - -响应数据。 - -taskId - -string - -任务唯一标识 - -63c4e0eaab3b4c0db208ecafa990e8d1 - -taskStatus - -string - -任务状态。 - -- PENDING：待执行 -- RUNNING：执行中 -- SUCCEEDED：成功 -- INVALID：失效 -- FAILED：失败 -- UNKNOWN： 未知 - -SUCCEEDED - -script - -string - -播客文字稿(未交互增强) - -"\[{\\"text\\": \\"听众朋友们，晚上好！今天咱们聊聊最近大家都很关心的一个话题——甲流来袭，我们该怎么应对？\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"嗯，这个话题确实挺重要。甲流听起来有点吓人，但其实只要科学防护，就不用太担心。\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"没错，先给大家科普一下，甲流全名叫甲型流感，是由甲型流感病毒感染引起的流行性感冒。跟普通感冒比起来，它的症状更重，传播也更快。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"哦？那具体都有哪些症状呢？我记得好像会发烧吧。\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"对的，高烧是甲流最常见的症状之一，体温可能高达39到40度。除了发烧，还会有头痛、全身肌肉和关节疼痛、乏力、食欲不振等症状。有些人甚至会出现恶心、呕吐的情况。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"听起来确实不太好受。不过婴儿的症状是不是会稍微不一样？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"是的，婴儿可能会表现为高烧、烦躁、哭闹增加，还有吃奶减少等。所以家长要特别注意观察孩子的状态。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"嗯，说到这，我突然想到一个问题：甲流是怎么传播的呢？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"主要通过呼吸道传播和接触传播。比如近距离接触甲流病人或者高度疑似病人后，就可能被感染。另外，患者的分泌物也可能携带病毒，直接或间接接触这些分泌物也会有风险。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"哦，原来如此。那易感人群有哪些呢？是不是老年人和小孩更容易中招？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"没错，老人、孕妇、小孩都是易感人群。此外，患有慢性疾病、肥胖或者免疫功能低下的人群也要格外小心。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"听你这么一说，感觉预防真的很重要啊。那有没有什么有效的预防措施呢？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"当然有！最有效的方法就是接种疫苗。每年9到10月份接种流感疫苗，保护效力可以持续到冬春流感高发季节。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"嗯，除了打疫苗，还有什么其他办法吗？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"保持良好的卫生习惯也很重要。比如勤洗手，尽量避免用手接触眼、鼻、口。同时，避免在拥挤的场所逗留，减少人际接触。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"嗯，这点我特别赞同。现在很多人都习惯了戴口罩，这其实也是个很好的防护措施。\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"没错，戴口罩能有效阻断飞沫传播。另外，增强免疫力也很关键。保持充足的睡眠、均衡的饮食和适量的运动都能帮助身体更好地抵抗病毒。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"听你这么一说，感觉生活中的小细节真的很重要。比如咳嗽或打喷嚏时，用纸巾、手肘遮住口鼻，就能减少病毒传播。\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"对，还有别忘了保持环境清洁和通风。家里、教室每天勤开窗通风，必要时进行消毒。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"看来这些看似简单的小事，其实都能起到大作用。那万一真的不幸中招了，该怎么办呢？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"如果确诊是甲型流感，一定要隔离、卧床休息，并遵医嘱用药。比如磷酸奥司他韦胶囊就是一种常用的抗病毒药物。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"哦，原来还有专门针对甲流的药。不过如果症状比较严重，比如持续高烧或者呼吸困难，那就得赶紧去医院了。\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"没错，千万别拖延病情。医院可以排查是否有并发症，比如肺炎之类的。总之，早发现、早治疗很重要。\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"总结一下，甲流虽然可怕，但只要我们做好防护，及时就医，就没啥好怕的。大家记住了吗？\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"记住了！接种疫苗、勤洗手、戴口罩、增强免疫力，还有必要时及时就医。希望大家都能健健康康地度过这个季节！\\", \\"speaker\\": \\"speaker-1\\"}, {\\"text\\": \\"好了，今天的节目就到这里。感谢大家收听，我们下期再见！\\", \\"speaker\\": \\"speaker-2\\"}, {\\"text\\": \\"再见！\\", \\"speaker\\": \\"speaker-1\\"}\]" - -resultUrl - -any - -以 URL 形式返回的解析结果，链接有效期为一小时。 - -{"audio":"http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/audio.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748853849&Signature=e1KlRgmjAjuUkPVWIEhoRbn4X0w%3D","script":"http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/script.txt?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748853511&Signature=th5sI%2BB1FZuQ6tRLg4qGGX1fevI%3D"} - -extraResult - -any - -播客分段音频等详细信息 - -extraResult 详细结构描述如下 - -``` -- extraResult.segment(list[float]):分段音频时长坐标 -- extraResult.segmentDetails.index(int):分段音频下标 -- extraResult.segmentDetails.time(list[float]):分段音频位置 -- extraResult.segmentDetails.audioSegmentUrl(str):分段音频可下载链接，有效期一小时 -- extraResult.segmentDetails.script.speaker(str):分段音频主理人 -- extraResult.segmentDetails.script.text(str):分段音频文字稿 -``` - -{ "segment": \[ \[ 0.0, 6370.0 \], \[ 6370.0, 15020.0 \], \[ 15020.0, 26071.791 \], \[ 26071.791, 31001.791 \], \[ 31001.791, 45646.582 \], \[ 45646.582, 50816.582 \], \[ 50816.582, 59501.418 \], \[ 59501.418, 63631.418 \], \[ 63631.418, 77161.25 \], \[ 77161.25, 84691.25 \], \[ 84691.25, 93623.086 \], \[ 93623.086, 99233.086 \], \[ 99233.086, 108894.875 \], \[ 108894.875, 111864.875 \], \[ 111864.875, 121678.664 \], \[ 121678.664, 128448.664 \], \[ 128448.664, 139774.45 \], \[ 139774.45, 149824.45 \], \[ 149824.45, 157966.25 \], \[ 157966.25, 164576.25 \], \[ 164576.25, 175432.08 \], \[ 175432.08, 184482.08 \], \[ 184482.08, 192451.88 \], \[ 192451.88, 199781.88 \], \[ 199781.88, 208151.88 \], \[ 208151.88, 212601.88 \], \[ 212601.88, 213131.88 \] \], "segmentDetails": \[ { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_0.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=cWk2X%2B7MtbIEm%2Fu1mwVts59hpwU%3D", "index": 0, "time": \[ 0.0, 6370.0 \], "script": { "speaker": "speaker-1", "text": "听众朋友们，晚上好！今天咱们聊聊最近大家都很关心的一个话题——甲流来袭，我们该怎么应对？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_1.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=CFGULg%2FJ8I3htf0YvVL5mz7FLXg%3D", "index": 1, "time": \[ 6370.0, 15020.0 \], "script": { "speaker": "speaker-2", "text": "嗯，这个话题确实挺重要。甲流听起来有点吓人，但其实只要科学防护，就不用太担心。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_2.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=Q2Ck9DLRqsvJ1Bt6aqMh0rnPPDQ%3D", "index": 2, "time": \[ 15020.0, 26071.791 \], "script": { "speaker": "speaker-1", "text": "没错，先给大家科普一下，甲流全名叫甲型流感，是由甲型流感病毒感染引起的流行性感冒。（对）跟普通感冒比起来，它的症状更重，传播也更快。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_3.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=vPwoJQnhKv2Ti4aovV6v3lP4V5w%3D", "index": 3, "time": \[ 26071.791, 31001.791 \], "script": { "speaker": "speaker-2", "text": "哦？那具体都有哪些症状呢？我记得好像会发烧吧。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_4.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=fUh04j4iLyYG4GV5wQ1f5GkFvJE%3D", "index": 4, "time": \[ 31001.791, 45646.582 \], "script": { "speaker": "speaker-1", "text": "对的，高烧是甲流最常见的症状之一，体温可能高达39到40度。（嗯）除了发烧，还会有头痛、全身肌肉和关节疼痛、乏力、食欲不振等症状。有些人甚至会出现恶心、呕吐的情况。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_5.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=v5c0d%2Bk4OmjmRRFVCKZ5JhDOq5o%3D", "index": 5, "time": \[ 45646.582, 50816.582 \], "script": { "speaker": "speaker-2", "text": "听起来确实不太好受。不过婴儿的症状是不是会稍微不一样？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_6.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=ZeyAxNn8wbRFXK4qXPJMKKGhIl0%3D", "index": 6, "time": \[ 50816.582, 59501.418 \], "script": { "speaker": "speaker-1", "text": "是的，婴儿可能会表现为高烧、烦躁、哭闹增加，还有吃奶减少等。（确实）所以家长要特别注意观察孩子的状态。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_7.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854445&Signature=UEyc%2BbAqgIa1C3SS4EjtNegTDTg%3D", "index": 7, "time": \[ 59501.418, 63631.418 \], "script": { "speaker": "speaker-2", "text": "嗯，说到这，我突然想到一个问题：甲流是怎么传播的呢？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_8.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=JYlAsSnLJNMFDEv9Bm5hSf4SrNY%3D", "index": 8, "time": \[ 63631.418, 77161.25 \], "script": { "speaker": "speaker-1", "text": "主要通过呼吸道传播和接触传播。（没错）比如近距离接触甲流病人或者高度疑似病人后，就可能被感染。另外，患者的分泌物也可能携带病毒，直接或间接接触这些分泌物也会有风险。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_9.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=eHG1QMbfsSSqfHwAD79Xs531kFw%3D", "index": 9, "time": \[ 77161.25, 84691.25 \], "script": { "speaker": "speaker-2", "text": "哦，原来如此。那易感人群有哪些呢？是不是老年人和小孩更容易中招？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_10.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=rElcCYlOaChj%2BJWT94c7%2F%2FMlTUo%3D", "index": 10, "time": \[ 84691.25, 93623.086 \], "script": { "speaker": "speaker-1", "text": "没错，老人、孕妇、小孩都是易感人群。（真的）此外，患有慢性疾病、肥胖或者免疫功能低下的人群也要格外小心。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_11.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=dBjOIrNqHSiF3RCrbsWyLtrE6Ow%3D", "index": 11, "time": \[ 93623.086, 99233.086 \], "script": { "speaker": "speaker-2", "text": "听你这么一说，感觉预防真的很重要啊。那有没有什么有效的预防措施呢？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_12.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=EJ0qCENMhqIUEu6UZ0RTP0cdibQ%3D", "index": 12, "time": \[ 99233.086, 108894.875 \], "script": { "speaker": "speaker-1", "text": "当然有！最有效的方法就是接种疫苗。（对）每年9到10月份接种流感疫苗，保护效力可以持续到冬春流感高发季节。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_13.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=nH02zWlwtrLK2fcskFbRvuMTD3s%3D", "index": 13, "time": \[ 108894.875, 111864.875 \], "script": { "speaker": "speaker-2", "text": "嗯，除了打疫苗，还有什么其他办法吗？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_14.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=9y3D2UGy6zg77gFWP4qFtHkF5S4%3D", "index": 14, "time": \[ 111864.875, 121678.664 \], "script": { "speaker": "speaker-1", "text": "保持良好的卫生习惯也很重要。比如勤洗手，尽量避免用手接触眼、鼻、口。（没错）同时，避免在拥挤的场所逗留，减少人际接触。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_15.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854446&Signature=EbwEURP7u9HpwrsrPC2Thwd4XUg%3D", "index": 15, "time": \[ 121678.664, 128448.664 \], "script": { "speaker": "speaker-2", "text": "嗯，这点我特别赞同。现在很多人都习惯了戴口罩，这其实也是个很好的防护措施。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_16.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=4rj5XQD0SaspVCj7QmGFrIGyACA%3D", "index": 16, "time": \[ 128448.664, 139774.45 \], "script": { "speaker": "speaker-1", "text": "没错，戴口罩能有效阻断飞沫传播。另外，增强免疫力也很关键。（嗯）保持充足的睡眠、均衡的饮食和适量的运动都能帮助身体更好地抵抗病毒。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_17.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=1aGlxcVI8r3ac7kY%2FHAt2%2Bg2Jss%3D", "index": 17, "time": \[ 139774.45, 149824.45 \], "script": { "speaker": "speaker-2", "text": "听你这么一说，感觉生活中的小细节真的很重要。比如咳嗽或打喷嚏时，用纸巾、手肘遮住口鼻，就能减少病毒传播。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_18.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=41GmphAH1k2hwtKm5xG5HU8sZYU%3D", "index": 18, "time": \[ 149824.45, 157966.25 \], "script": { "speaker": "speaker-1", "text": "对，还有别忘了保持环境清洁和通风。（确实）家里、教室每天勤开窗通风，必要时进行消毒。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_19.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=1yD7PvQwmJCFtDVZxiBSCgxFQWE%3D", "index": 19, "time": \[ 157966.25, 164576.25 \], "script": { "speaker": "speaker-2", "text": "看来这些看似简单的小事，其实都能起到大作用。那万一真的不幸中招了，该怎么办呢？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_20.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=D%2BVKscQXjtyzbNx5NJhx%2Fyny1qI%3D", "index": 20, "time": \[ 164576.25, 175432.08 \], "script": { "speaker": "speaker-1", "text": "如果确诊是甲型流感，一定要隔离、卧床休息，并遵医嘱用药。（没错）比如磷酸奥司他韦胶囊就是一种常用的抗病毒药物。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_21.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=Ugt6%2FUFdc7ACT%2BFxJm4Tk8VOX84%3D", "index": 21, "time": \[ 175432.08, 184482.08 \], "script": { "speaker": "speaker-2", "text": "哦，原来还有专门针对甲流的药。不过如果症状比较严重，比如持续高烧或者呼吸困难，那就得赶紧去医院了。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_22.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=YJy3rQ2cpGwjtW%2BSzehQscSFicg%3D", "index": 22, "time": \[ 184482.08, 192451.88 \], "script": { "speaker": "speaker-1", "text": "没错，千万别拖延病情。医院可以排查是否有并发症，比如肺炎之类的。（对）总之，早发现、早治疗很重要。" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_23.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854447&Signature=5WvLaj3cCmK5m%2FAS8EiIbaAgxYs%3D", "index": 23, "time": \[ 192451.88, 199781.88 \], "script": { "speaker": "speaker-2", "text": "总结一下，甲流虽然可怕，但只要我们做好防护，及时就医，就没啥好怕的。大家记住了吗？" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_24.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854448&Signature=H4gi%2B%2BUJ%2BJ36rjZeUecTsqFjTqY%3D", "index": 24, "time": \[ 199781.88, 208151.88 \], "script": { "speaker": "speaker-1", "text": "记住了！接种疫苗、勤洗手、戴口罩、增强免疫力，还有必要时及时就医。希望大家都能健健康康地度过这个季节！" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_25.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854448&Signature=oGTU2Ki3FkmV2hvL3sPFVM7ZPGI%3D", "index": 25, "time": \[ 208151.88, 212601.88 \], "script": { "speaker": "speaker-2", "text": "好了，今天的节目就到这里。感谢大家收听，我们下期再见！" } }, { "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment\_26.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1748854448&Signature=N4lXSxOmRSjsRFounST8hBM3FnY%3D", "index": 26, "time": \[ 212601.88, 213131.88 \], "script": { "speaker": "speaker-1", "text": "再见！" } } \] } - -resultUrl 中 script 返回的内容为交互增强后的播客稿内容与 script 参数内容不一致。 - -## 示例 - -正常返回示例 - -`JSON`格式 - -``` -{ - "code": "success", - "message": "success", - "requestId": "C38F034D-7F36-531C-95AC-0C752F80E840", - "success": true, - "httpStatusCode": 200, - "data": { - "taskId": "63c4e0eaab3b4c0db208ecafa990e8d1\n", - "taskStatus": "SUCCEEDED", - "script": [ - { - "text": "听众朋友们，晚上好！今天咱们聊聊最近大家都很关心的一个话题——甲流来袭，我们该怎么应对？", - "speaker": "speaker-1" - }, - { - "text": "嗯，这个话题确实挺重要。甲流听起来有点吓人，但其实只要科学防护，就不用太担心。", - "speaker": "speaker-2" - }, - { - "text": "没错，先给大家科普一下，甲流全名叫甲型流感，是由甲型流感病毒感染引起的流行性感冒。跟普通感冒比起来，它的症状更重，传播也更快。", - "speaker": "speaker-1" - }, - { - "text": "哦？那具体都有哪些症状呢？我记得好像会发烧吧。", - "speaker": "speaker-2" - }, - { - "text": "对的，高烧是甲流最常见的症状之一，体温可能高达39到40度。除了发烧，还会有头痛、全身肌肉和关节疼痛、乏力、食欲不振等症状。有些人甚至会出现恶心、呕吐的情况。", - "speaker": "speaker-1" - }, - { - "text": "听起来确实不太好受。不过婴儿的症状是不是会稍微不一样？", - "speaker": "speaker-2" - }, - { - "text": "是的，婴儿可能会表现为高烧、烦躁、哭闹增加，还有吃奶减少等。所以家长要特别注意观察孩子的状态。", - "speaker": "speaker-1" - }, - { - "text": "嗯，说到这，我突然想到一个问题：甲流是怎么传播的呢？", - "speaker": "speaker-2" - }, - { - "text": "主要通过呼吸道传播和接触传播。比如近距离接触甲流病人或者高度疑似病人后，就可能被感染。另外，患者的分泌物也可能携带病毒，直接或间接接触这些分泌物也会有风险。", - "speaker": "speaker-1" - }, - { - "text": "哦，原来如此。那易感人群有哪些呢？是不是老年人和小孩更容易中招？", - "speaker": "speaker-2" - }, - { - "text": "没错，老人、孕妇、小孩都是易感人群。此外，患有慢性疾病、肥胖或者免疫功能低下的人群也要格外小心。", - "speaker": "speaker-1" - }, - { - "text": "听你这么一说，感觉预防真的很重要啊。那有没有什么有效的预防措施呢？", - "speaker": "speaker-2" - }, - { - "text": "当然有！最有效的方法就是接种疫苗。每年9到10月份接种流感疫苗，保护效力可以持续到冬春流感高发季节。", - "speaker": "speaker-1" - }, - { - "text": "嗯，除了打疫苗，还有什么其他办法吗？", - "speaker": "speaker-2" - }, - { - "text": "保持良好的卫生习惯也很重要。比如勤洗手，尽量避免用手接触眼、鼻、口。同时，避免在拥挤的场所逗留，减少人际接触。", - "speaker": "speaker-1" - }, - { - "text": "嗯，这点我特别赞同。现在很多人都习惯了戴口罩，这其实也是个很好的防护措施。", - "speaker": "speaker-2" - }, - { - "text": "没错，戴口罩能有效阻断飞沫传播。另外，增强免疫力也很关键。保持充足的睡眠、均衡的饮食和适量的运动都能帮助身体更好地抵抗病毒。", - "speaker": "speaker-1" - }, - { - "text": "听你这么一说，感觉生活中的小细节真的很重要。比如咳嗽或打喷嚏时，用纸巾、手肘遮住口鼻，就能减少病毒传播。", - "speaker": "speaker-2" - }, - { - "text": "对，还有别忘了保持环境清洁和通风。家里、教室每天勤开窗通风，必要时进行消毒。", - "speaker": "speaker-1" - }, - { - "text": "看来这些看似简单的小事，其实都能起到大作用。那万一真的不幸中招了，该怎么办呢？", - "speaker": "speaker-2" - }, - { - "text": "如果确诊是甲型流感，一定要隔离、卧床休息，并遵医嘱用药。比如磷酸奥司他韦胶囊就是一种常用的抗病毒药物。", - "speaker": "speaker-1" - }, - { - "text": "哦，原来还有专门针对甲流的药。不过如果症状比较严重，比如持续高烧或者呼吸困难，那就得赶紧去医院了。", - "speaker": "speaker-2" - }, - { - "text": "没错，千万别拖延病情。医院可以排查是否有并发症，比如肺炎之类的。总之，早发现、早治疗很重要。", - "speaker": "speaker-1" - }, - { - "text": "总结一下，甲流虽然可怕，但只要我们做好防护，及时就医，就没啥好怕的。大家记住了吗？", - "speaker": "speaker-2" - }, - { - "text": "记住了！接种疫苗、勤洗手、戴口罩、增强免疫力，还有必要时及时就医。希望大家都能健健康康地度过这个季节！", - "speaker": "speaker-1" - }, - { - "text": "好了，今天的节目就到这里。感谢大家收听，我们下期再见！", - "speaker": "speaker-2" - }, - { - "text": "再见！", - "speaker": "speaker-1" - } - ], - "resultUrl": { - "audio": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/audio.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748853849&Signature=e1KlRgmjAjuUkPVWIEhoRbn4X0w%3D", - "script": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/script.txt?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748853511&Signature=th5sI%2BB1FZuQ6tRLg4qGGX1fevI%3D" - }, - "extraResult": { - "segment": [ - [ - 0, - 6370 - ], - [ - 6370, - 15020 - ], - [ - 15020, - 26071.791 - ], - [ - 26071.791, - 31001.791 - ], - [ - 31001.791, - 45646.582 - ], - [ - 45646.582, - 50816.582 - ], - [ - 50816.582, - 59501.418 - ], - [ - 59501.418, - 63631.418 - ], - [ - 63631.418, - 77161.25 - ], - [ - 77161.25, - 84691.25 - ], - [ - 84691.25, - 93623.086 - ], - [ - 93623.086, - 99233.086 - ], - [ - 99233.086, - 108894.875 - ], - [ - 108894.875, - 111864.875 - ], - [ - 111864.875, - 121678.664 - ], - [ - 121678.664, - 128448.664 - ], - [ - 128448.664, - 139774.45 - ], - [ - 139774.45, - 149824.45 - ], - [ - 149824.45, - 157966.25 - ], - [ - 157966.25, - 164576.25 - ], - [ - 164576.25, - 175432.08 - ], - [ - 175432.08, - 184482.08 - ], - [ - 184482.08, - 192451.88 - ], - [ - 192451.88, - 199781.88 - ], - [ - 199781.88, - 208151.88 - ], - [ - 208151.88, - 212601.88 - ], - [ - 212601.88, - 213131.88 - ] - ], - "segmentDetails": [ - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_0.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=cWk2X%2B7MtbIEm%2Fu1mwVts59hpwU%3D", - "index": 0, - "time": [ - 0, - 6370 - ], - "script": { - "speaker": "speaker-1", - "text": "听众朋友们，晚上好！今天咱们聊聊最近大家都很关心的一个话题——甲流来袭，我们该怎么应对？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_1.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=CFGULg%2FJ8I3htf0YvVL5mz7FLXg%3D", - "index": 1, - "time": [ - 6370, - 15020 - ], - "script": { - "speaker": "speaker-2", - "text": "嗯，这个话题确实挺重要。甲流听起来有点吓人，但其实只要科学防护，就不用太担心。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_2.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=Q2Ck9DLRqsvJ1Bt6aqMh0rnPPDQ%3D", - "index": 2, - "time": [ - 15020, - 26071.791 - ], - "script": { - "speaker": "speaker-1", - "text": "没错，先给大家科普一下，甲流全名叫甲型流感，是由甲型流感病毒感染引起的流行性感冒。（对）跟普通感冒比起来，它的症状更重，传播也更快。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_3.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=vPwoJQnhKv2Ti4aovV6v3lP4V5w%3D", - "index": 3, - "time": [ - 26071.791, - 31001.791 - ], - "script": { - "speaker": "speaker-2", - "text": "哦？那具体都有哪些症状呢？我记得好像会发烧吧。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_4.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=fUh04j4iLyYG4GV5wQ1f5GkFvJE%3D", - "index": 4, - "time": [ - 31001.791, - 45646.582 - ], - "script": { - "speaker": "speaker-1", - "text": "对的，高烧是甲流最常见的症状之一，体温可能高达39到40度。（嗯）除了发烧，还会有头痛、全身肌肉和关节疼痛、乏力、食欲不振等症状。有些人甚至会出现恶心、呕吐的情况。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_5.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=v5c0d%2Bk4OmjmRRFVCKZ5JhDOq5o%3D", - "index": 5, - "time": [ - 45646.582, - 50816.582 - ], - "script": { - "speaker": "speaker-2", - "text": "听起来确实不太好受。不过婴儿的症状是不是会稍微不一样？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_6.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=ZeyAxNn8wbRFXK4qXPJMKKGhIl0%3D", - "index": 6, - "time": [ - 50816.582, - 59501.418 - ], - "script": { - "speaker": "speaker-1", - "text": "是的，婴儿可能会表现为高烧、烦躁、哭闹增加，还有吃奶减少等。（确实）所以家长要特别注意观察孩子的状态。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_7.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854445&Signature=UEyc%2BbAqgIa1C3SS4EjtNegTDTg%3D", - "index": 7, - "time": [ - 59501.418, - 63631.418 - ], - "script": { - "speaker": "speaker-2", - "text": "嗯，说到这，我突然想到一个问题：甲流是怎么传播的呢？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_8.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=JYlAsSnLJNMFDEv9Bm5hSf4SrNY%3D", - "index": 8, - "time": [ - 63631.418, - 77161.25 - ], - "script": { - "speaker": "speaker-1", - "text": "主要通过呼吸道传播和接触传播。（没错）比如近距离接触甲流病人或者高度疑似病人后，就可能被感染。另外，患者的分泌物也可能携带病毒，直接或间接接触这些分泌物也会有风险。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_9.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=eHG1QMbfsSSqfHwAD79Xs531kFw%3D", - "index": 9, - "time": [ - 77161.25, - 84691.25 - ], - "script": { - "speaker": "speaker-2", - "text": "哦，原来如此。那易感人群有哪些呢？是不是老年人和小孩更容易中招？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_10.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=rElcCYlOaChj%2BJWT94c7%2F%2FMlTUo%3D", - "index": 10, - "time": [ - 84691.25, - 93623.086 - ], - "script": { - "speaker": "speaker-1", - "text": "没错，老人、孕妇、小孩都是易感人群。（真的）此外，患有慢性疾病、肥胖或者免疫功能低下的人群也要格外小心。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_11.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=dBjOIrNqHSiF3RCrbsWyLtrE6Ow%3D", - "index": 11, - "time": [ - 93623.086, - 99233.086 - ], - "script": { - "speaker": "speaker-2", - "text": "听你这么一说，感觉预防真的很重要啊。那有没有什么有效的预防措施呢？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_12.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=EJ0qCENMhqIUEu6UZ0RTP0cdibQ%3D", - "index": 12, - "time": [ - 99233.086, - 108894.875 - ], - "script": { - "speaker": "speaker-1", - "text": "当然有！最有效的方法就是接种疫苗。（对）每年9到10月份接种流感疫苗，保护效力可以持续到冬春流感高发季节。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_13.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=nH02zWlwtrLK2fcskFbRvuMTD3s%3D", - "index": 13, - "time": [ - 108894.875, - 111864.875 - ], - "script": { - "speaker": "speaker-2", - "text": "嗯，除了打疫苗，还有什么其他办法吗？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_14.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=9y3D2UGy6zg77gFWP4qFtHkF5S4%3D", - "index": 14, - "time": [ - 111864.875, - 121678.664 - ], - "script": { - "speaker": "speaker-1", - "text": "保持良好的卫生习惯也很重要。比如勤洗手，尽量避免用手接触眼、鼻、口。（没错）同时，避免在拥挤的场所逗留，减少人际接触。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_15.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854446&Signature=EbwEURP7u9HpwrsrPC2Thwd4XUg%3D", - "index": 15, - "time": [ - 121678.664, - 128448.664 - ], - "script": { - "speaker": "speaker-2", - "text": "嗯，这点我特别赞同。现在很多人都习惯了戴口罩，这其实也是个很好的防护措施。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_16.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=4rj5XQD0SaspVCj7QmGFrIGyACA%3D", - "index": 16, - "time": [ - 128448.664, - 139774.45 - ], - "script": { - "speaker": "speaker-1", - "text": "没错，戴口罩能有效阻断飞沫传播。另外，增强免疫力也很关键。（嗯）保持充足的睡眠、均衡的饮食和适量的运动都能帮助身体更好地抵抗病毒。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_17.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=1aGlxcVI8r3ac7kY%2FHAt2%2Bg2Jss%3D", - "index": 17, - "time": [ - 139774.45, - 149824.45 - ], - "script": { - "speaker": "speaker-2", - "text": "听你这么一说，感觉生活中的小细节真的很重要。比如咳嗽或打喷嚏时，用纸巾、手肘遮住口鼻，就能减少病毒传播。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_18.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=41GmphAH1k2hwtKm5xG5HU8sZYU%3D", - "index": 18, - "time": [ - 149824.45, - 157966.25 - ], - "script": { - "speaker": "speaker-1", - "text": "对，还有别忘了保持环境清洁和通风。（确实）家里、教室每天勤开窗通风，必要时进行消毒。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_19.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=1yD7PvQwmJCFtDVZxiBSCgxFQWE%3D", - "index": 19, - "time": [ - 157966.25, - 164576.25 - ], - "script": { - "speaker": "speaker-2", - "text": "看来这些看似简单的小事，其实都能起到大作用。那万一真的不幸中招了，该怎么办呢？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_20.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=D%2BVKscQXjtyzbNx5NJhx%2Fyny1qI%3D", - "index": 20, - "time": [ - 164576.25, - 175432.08 - ], - "script": { - "speaker": "speaker-1", - "text": "如果确诊是甲型流感，一定要隔离、卧床休息，并遵医嘱用药。（没错）比如磷酸奥司他韦胶囊就是一种常用的抗病毒药物。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_21.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=Ugt6%2FUFdc7ACT%2BFxJm4Tk8VOX84%3D", - "index": 21, - "time": [ - 175432.08, - 184482.08 - ], - "script": { - "speaker": "speaker-2", - "text": "哦，原来还有专门针对甲流的药。不过如果症状比较严重，比如持续高烧或者呼吸困难，那就得赶紧去医院了。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_22.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=YJy3rQ2cpGwjtW%2BSzehQscSFicg%3D", - "index": 22, - "time": [ - 184482.08, - 192451.88 - ], - "script": { - "speaker": "speaker-1", - "text": "没错，千万别拖延病情。医院可以排查是否有并发症，比如肺炎之类的。（对）总之，早发现、早治疗很重要。" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_23.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854447&Signature=5WvLaj3cCmK5m%2FAS8EiIbaAgxYs%3D", - "index": 23, - "time": [ - 192451.88, - 199781.88 - ], - "script": { - "speaker": "speaker-2", - "text": "总结一下，甲流虽然可怕，但只要我们做好防护，及时就医，就没啥好怕的。大家记住了吗？" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_24.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854448&Signature=H4gi%2B%2BUJ%2BJ36rjZeUecTsqFjTqY%3D", - "index": 24, - "time": [ - 199781.88, - 208151.88 - ], - "script": { - "speaker": "speaker-1", - "text": "记住了！接种疫苗、勤洗手、戴口罩、增强免疫力，还有必要时及时就医。希望大家都能健健康康地度过这个季节！" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_25.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854448&Signature=oGTU2Ki3FkmV2hvL3sPFVM7ZPGI%3D", - "index": 25, - "time": [ - 208151.88, - 212601.88 - ], - "script": { - "speaker": "speaker-2", - "text": "好了，今天的节目就到这里。感谢大家收听，我们下期再见！" - } - }, - { - "audioSegmentUrl": "http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202506021536413361/segment_26.mp3?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ****&Expires=1748854448&Signature=N4lXSxOmRSjsRFounST8hBM3FnY%3D", - "index": 26, - "time": [ - 212601.88, - 213131.88 - ], - "script": { - "speaker": "speaker-1", - "text": "再见！" - } - } - ] - } - } -} -``` - -## 错误码 - -访问[错误中心](< https://api.aliyun.com/document/AIPodcast/2025-02-28/errorCode>)查看更多错误码。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md deleted file mode 100644 index 2a79a172..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-dir/api-aipodcast-2025-02-28-podcasttasksubmit.md +++ /dev/null @@ -1,269 +0,0 @@ -# PodcastTaskSubmit - 播客任务提交 - -ai播客生成任务提交。 - -## 调试 - -[您可以在OpenAPI Explorer中直接运行该接口，免去您计算签名的困扰。运行成功后，OpenAPI Explorer可以自动生成SDK代码示例。](https://api.aliyun.com/api/AIPodcast/2025-02-28/PodcastTaskSubmit) - -[![](https://img.alicdn.com/tfs/TB16JcyXHr1gK0jSZR0XXbP8XXa-24-26.png)调试](https://api.aliyun.com/api/AIPodcast/2025-02-28/PodcastTaskSubmit) - -## 授权信息 - -下表是API对应的授权信息，可以在RAM权限策略语句的`Action`元素中使用，用来给RAM用户或RAM角色授予调用此API的权限。具体说明如下： - -- 操作：是指具体的权限点。 -- 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 -- 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 - - 对于不支持资源级授权的操作，用`全部资源`表示。 -- 条件关键字：是指云产品自身定义的条件关键字。 -- 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 - -操作 - -访问级别 - -资源类型 - -条件关键字 - -关联操作 - -aipodcast:PodcastTaskSubmit - -none - -\*全部资源 - -`*` - -无 - -无 - -## 请求语法 - -``` -POST /podcast/task/submit HTTP/1.1 -``` - -## 请求参数 - -名称 - -类型 - -必填 - -描述 - -示例值 - -workspaceId - -string - -是 - -当前请求所使用的百炼业务空间 id - -llm-ep8ba0dr6seiddxx - -fileUrls - -array - -否 - -用播客任务生成的文件可访问链接 - -string - -否 - -文件可访问链接 - -http://xxx-ai-file.oss-cn-beijing.aliyuncs.com/202503241702148295/script.txt?OSSAccessKeyId=LTAI5tPLWJfJHNkZbfnQ\*\*\*\*&Expires=1742810622&Signature=TBBdikHzOWW3YqDw3sNMTXiMo6A%3D - -topic - -string - -否 - -播客内容生成的主题 - -甲流来袭，我们该如何应对？ - -counts - -integer - -否 - -对话人数 - -枚举值： - -- 1：单人。 -- 2：双人。 - -2 - -voices - -array - -否 - -指定播客生成的人声音色，数组中的起始音色，为第一个对话人的音色 - -枚举值： - -- Dylan：北京话-男声。 -- en\_female：女声。 -- Cherry：女声。 -- Ethan：男声。 -- news\_male：男声。 -- Jada：吴语-女声。 -- Chelsie：女声。 -- en\_male：男声。 -- news\_female：女声。 -- Serena：女声。 -- Sunny：四川话-女声。 - -string - -否 - -人声音色 - -news\_female - -text - -string - -否 - -文案 - -最近，甲型流感似乎成了大家热议的话题。但你真的了解甲流吗？它和普通感冒有什么区别？为什么症状看起来如此严重？更重要的是，我们应该如何预防和治疗？本期对话中，我们将深入探讨甲流的传播方式、易感人群以及科学的防护措施，帮助你在流感高发季节保护自己和家人的健康。 - -sourceLang - -string - -否 - -播客内容生成的结果（包含播客稿、播客音频）语种选型，目前仅支持中文及英文，默认语种为中文。 - -zh - -## 返回参数 - -名称 - -类型 - -描述 - -示例值 - -object - -Schema of Response - -code - -string - -响应状态码。 - -"success" - -message - -string - -响应消息。 - -"success" - -requestId - -string - -请求 id，用于追溯 API 调用链路。 - -9CE5B91A-6E6B-55FB-A1AF-037DF01C84B3 - -success - -boolean - -true 接口调用成功，false 接口调用失败 - -True - -httpStatusCode - -string - -HTTP 状态码。 - -200 - -data - -object - -返回数据 - -taskId - -string - -任务唯一标识 - -63c4e0eaab3b4c0db208ecafa990e8d1 - -taskStatus - -string - -任务状态。 - -- PENDING：待执行 -- RUNNING：执行中 -- SUCCEEDED：成功 -- INVALID：失效 -- FAILED：失败 -- UNKNOWN： 未知 - -SUCCEEDED - -## 示例 - -正常返回示例 - -`JSON`格式 - -``` -{ - "code": "success", - "message": "success", - "requestId": "9CE5B91A-6E6B-55FB-A1AF-037DF01C84B3", - "success": true, - "httpStatusCode": 200, - "data": { - "taskId": "63c4e0eaab3b4c0db208ecafa990e8d1", - "taskStatus": "SUCCEEDED" - } -} -``` - -## 错误码 - -访问[错误中心](< https://api.aliyun.com/document/AIPodcast/2025-02-28/errorCode>)查看更多错误码。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md deleted file mode 100644 index 260e77d6..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-endpoint.md +++ /dev/null @@ -1,19 +0,0 @@ -# 服务接入点 - -## 亚太 - -地域名称 - -地域ID - -公网接入地址 - -VPC接入地址 - -华北2（北京） - -cn-beijing - -aipodcast.cn-beijing.aliyuncs.com - -aipodcast-vpc.cn-beijing.aliyuncs.com diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md deleted file mode 100644 index 245f1ba4..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-overview.md +++ /dev/null @@ -1,35 +0,0 @@ -# API概览 - -## **API标准及多语言预置SDK** - -本产品（`AIPodcast/2025-02-28`）的OpenAPI采用[ROA](https://help.aliyun.com/zh/sdk/product-overview/roa-mechanism)签名风格。我们已经为开发者封装了常见编程语言的SDK，开发者可通过[下载SDK](https://api.aliyun.com/api-tools/sdk/AIPodcast?version=2025-02-28)直接调用本产品OpenAPI而无需关心技术细节。如果现有SDK不能满足使用需求，可通过签名机制进行自签名对接。由于自签名细节非常复杂，需花费 5个工作日左右。因此建议加入我们的服务钉钉群（147535001692），在专家指导下进行签名对接。 - -在使用API前，您需要准备好身份账号及访问密钥（AccessKey），才能有效通过客户端工具（SDK、CLI等）访问API。细节请参见[获取AccessKey](https://help.aliyun.com/zh/ram/user-guide/create-an-accesskey-pair)。 - -## **自定义签名场景** - -若您的业务场景有特殊需求，需通过自签名方式对接 API，建议优先咨询我们的技术支持团队（服务钉钉群：147535001692），获取专业指导以确保高效接入。 - -## **账号与安全准备** - -阿里云账号具备对所有资源的完全管理权限。一旦 AccessKey 泄露，所有相关资源都将面临未经授权访问的风险。为确保安全，建议创建一个仅具备 API 访问权限的[RAM用户](https://help.aliyun.com/zh/ram/user-guide/create-a-ram-user)并配置其 AccessKey，同时基于最小权限原则 (PoLP) 配置 RAM 策略。仅在明确需要阿里云账号权限的特定场景下，才使用阿里云账号。 - -## API目录 - -API - -标题 - -API概述 - -[PodcastTaskResultQuery](https://help.aliyun.com/zh/model-studio/api-aipodcast-2025-02-28-podcasttaskresultquery) - -播客任务结果查询 - -ai播客生成任务结果查询。 - -[PodcastTaskSubmit](https://help.aliyun.com/zh/model-studio/api-aipodcast-2025-02-28-podcasttasksubmit) - -播客任务提交 - -ai播客生成任务提交。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md deleted file mode 100644 index 92bc6236..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/api-reference-aipodcast/api-aipodcast-2025-02-28-ram.md +++ /dev/null @@ -1,109 +0,0 @@ -# 授权信息 - -访问控制（RAM）是阿里云提供的管理用户身份与资源访问权限的服务。使用RAM可以让您避免与其他用户共享阿里云账号密钥，并可按需为用户授予最小权限。RAM中使用权限策略描述授权的具体内容。 - -本文为您介绍大模型服务平台百炼（AIPodcast）为RAM权限策略定义的操作（Action）、资源（Resource）和条件（Condition）。大模型服务平台百炼（AIPodcast）的RAM代码（RamCode）为 aipodcast，支持的授权粒度为操作级。 - -## 权限策略通用结构 - -权限策略支持JSON格式，其通用结构如下： - -``` -{ - "Version": "1", - "Statement": [ - { - "Effect": "", - "Action": "", - "Resource": "", - "Condition": { - "": { - "": [ - "" - ] - } - } - } - ] -} -``` - -各字段含义如下： - -- Effect：权限策略效果。取值：Allow（允许）、Deny（拒绝）。 -- Action：授予允许或拒绝权限的具体操作。具体信息，请参见[操作（Action）](#title-auth-detail-2)。 -- Resource：受操作影响的具体对象，您可以使用资源ARN来描述指定资源。具体信息，请参见[资源（Resource）](#title-auth-detail-3)。 -- Condition：指授权生效的条件。可选字段。具体信息，请参见[条件（Condition）](#title-auth-detail-4)。 - - Condition\_operator：条件运算符，不同类型的条件对应不同的条件运算符。具体信息，请参见[权限策略基本元素](https://help.aliyun.com/zh/ram/policy-elements)。 - - Condition\_key：条件关键字。 - - Condition\_value：条件关键字对应的值。 - -## 操作（Action） - -下表是大模型服务平台百炼（AIPodcast）定义的操作，这些操作可以在RAM权限策略语句的`Action`元素中使用，用来授予执行该操作的权限。下面对表中的具体项提供说明： - -- 操作：是指具体的权限点。 -- API：是指操作对应的API接口。 -- 访问级别：是指每个操作的访问级别，取值为写入（Write）、读取（Read）或列出（List）。 -- 资源类型：是指操作中支持授权的资源类型。具体说明如下： - - 对于必选的资源类型，用前面加 \* 表示。 - - 对于不支持资源级授权的操作，用`全部资源`表示。 -- 条件关键字：是指云产品自身定义的条件关键字。该列不体现适用于任何操作的[通用条件关键字](https://help.aliyun.com/zh/ram/policy-elements)。 -- 关联操作：是指成功执行操作所需要的其他权限。操作者必须同时具备关联操作的权限，操作才能成功。 - -操作 - -API - -访问级别 - -资源类型 - -条件关键字 - -关联操作 - -aipodcast:PodcastTaskResultQuery - -[PodcastTaskResultQuery](https://help.aliyun.com/zh/model-studio/api-aipodcast-2025-02-28-podcasttaskresultquery) - -none - -\*全部资源 - -`*` - -无 - -无 - -aipodcast:PodcastTaskSubmit - -[PodcastTaskSubmit](https://help.aliyun.com/zh/model-studio/api-aipodcast-2025-02-28-podcasttasksubmit) - -none - -\*全部资源 - -`*` - -无 - -无 - -## 资源（Resource） - -大模型服务平台百炼（AIPodcast）不支持在RAM权限策略语句的`Resource`中指定资源ARN。如果要允许对大模型服务平台百炼（AIPodcast）的访问权限，请在策略语句中指定`"Resource": "*"`。 - -## 条件（Condition） - -大模型服务平台百炼（AIPodcast）未定义产品级别的条件关键字。如需查看适用于所有云产品的通用条件关键字，请参见[通用条件关键字](https://help.aliyun.com/zh/ram/policy-elements)。 - -## 相关操作 - -您可以创建自定义权限策略，并将权限策略授予RAM用户、RAM用户组或RAM角色。具体操作如下： - -- [创建自定义权限策略](https://help.aliyun.com/zh/ram/create-a-custom-policy) -- [为RAM用户授权](https://help.aliyun.com/zh/ram/user-guide/grant-permissions-to-the-ram-user) -- [为RAM用户组授权](https://help.aliyun.com/zh/ram/user-guide/grant-permissions-to-a-ram-user-group) -- [为RAM角色授权](https://help.aliyun.com/zh/ram/user-guide/grant-permissions-to-a-ram-role) diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md deleted file mode 100644 index 36ddf65b..00000000 --- a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/official-application-aipodcast/brief-introduction-of-ai-podcast.md +++ /dev/null @@ -1,63 +0,0 @@ -# 通义音频播客生成产品介绍 - -## **产品概述** - -播客音频生成是以通义千问大模型为基座的音频内容创作应用，通过大模型技术将文档内容转换成一段AI解读的播客节目，由两位AI主持人以对话的形式生动地对谈。 - -## **功能介绍** - -**功能点** - -**说明** - -文档类型 - -支持用户上传各种文档资料（word、pdf、txt等），并能根据这些内容生成音频解读概述。 - -声音复刻 - -可以复刻输入的样例声音作为生成音频的音色。 - -话题延伸 - -不仅针对文档的原始内容，还将进行适当的话题延展，以使对话更加发人深省。 - -主题指定 - -指定想了解的特定主题，作为播客音频的重点包含内容。 - -语种支持 - -支持生成中文或英文的音频内容。 - -## **应用场景** - -面向追求高效学习、需要情感陪伴以及希望解放视觉注意力的用户群体。长文本内容，都可以通过转换为音频的方式来提升触达效率。 - -- 媒体创作：智能生成新闻访谈音频，提高多场景内容生产效率。 - -- 企业培训：培训资料转音频课程，助力员工碎片化时间高效提升。 - -- 教育教学：课程音频智能转化，支持课外按需回放学习。 - -- 电商带货：产品文案转语音导购，搭载数字人实现全时商品讲解。 - - -## **计量计费** - -### **计费规则** - -- 播客音频生成接口按照使用次数后付费，如果仅开通服务，不会产生费用，费用发生以实际调用为准。 - -- 按使用次数进行计费，如果您账户的可用额度（含阿里云账户余额和代金券）小于待结算的账单，会收到余额不足的短信或邮件提醒。 - -- 调用限制：对于每个开通账号，限流为5QPM（每分钟不超过5次请求）。 - -- 计费逻辑：按照成功生成音频的次数\*单价进行计费。 - -- 页面使用次数与接口使用次数共享，会占用免费额度以及调用后收费。 - - -### **计费单价** - -开通成功赠送100次免费体验额度，超过免费额度后每次成功生成播客音频计费0.6元。 diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-createdataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-createdataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-deletedataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-deletedataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-getdataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-getdataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-listdatasets.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-listdatasets.md diff --git a/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md b/skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md similarity index 100% rename from skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-other/api-aimiaobi-2023-08-01-updatedataset.md rename to skills/bailian-docs-llm-wiki/raw/application-user-guide/application-gallery/quanmiao-solution-products/ai-quan-miao-development-document/amb-api-reference/api-aimiaobi-2023-08-01-dir/api-aimiaobi-2023-08-01-dir-wonderful-search-data-source/api-aimiaobi-2023-08-01-updatedataset.md diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md index 5fbf90b8..a2402160 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-design-http-api.md @@ -181,7 +181,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -189,7 +189,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 @@ -303,7 +303,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -313,7 +313,7 @@ curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/ **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md index c85a081b..42410c27 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/audio-api-references/speech-synthesis-api-reference/sound-reengraving/voice-clone-python-sdk.md @@ -112,7 +112,7 @@ List\[str\] **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash、v3-plus和v3-flash模型支持。 辅助模型识别样本音频的语种，从而更准确地提取音色特征，提升复刻效果。若设置的语种与实际音频语种不符（例如为中文音频设置 `en`），系统将忽略该设置并自动检测语种。 @@ -120,7 +120,7 @@ List\[str\] 取值范围（因模型而异）： -- qwen-audio-3.0-tts-flash： +- qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash： - zh：中文 @@ -205,7 +205,7 @@ float **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 音频预处理后用于声音复刻的参考音频最大时长（秒）。取值范围：\[3.0, 30.0\]。时间越长效果越好。 @@ -219,7 +219,7 @@ bool **重要** -仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 +仅适用于Qwen-Audio-TTS/CosyVoice声音复刻（model为`voice-enrollment`时），且仅qwen-audio-3.0-tts-plus、qwen-audio-3.0-tts-flash、cosyvoice-v3.5-plus、v3.5-flash和v3-flash模型支持。 是否开启音频预处理（降噪、音频增强、音量规整）。有背景噪音时建议开启；安静环境建议关闭以最大程度还原音色。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md deleted file mode 100644 index d0e365f7..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md +++ /dev/null @@ -1,2661 +0,0 @@ -# 万相-图像生成与编辑2.7 API参考 - -万相2.7图像生成与编辑模型支持文生图、文生组图、图生组图、图像编辑和多图参考生成。 - -## 模型概览 - -**模型名称** - -**模型简介** - -**输出图像规格** - -wan2.7-image-pro - -万相2.7 image专业版，文生图（非组图生成）支持4K高清输出 - -图片格式：PNG。 - -图像分辨率和尺寸请参见[size参数](https://help.aliyun.com/zh/model-studio/wan-image-generation-api-reference#wan27-param-size-section)。 - -wan2.7-image - -万相2.7 image，生成速度更快 - -**说明** - -调用前，请查阅各地域支持的[模型列表与价格](https://help.aliyun.com/zh/model-studio/model-pricing#e2540d71a2utl)。 - -## 前提条件 - -您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 - -**重要** - -华北2（北京）和新加坡地域拥有独立的 **API Key** 与**请求地址**，不可混用，跨地域调用将导致鉴权失败或服务报错。 - -**重要** - -阿里云百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - -- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - -- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - -其中 `{WorkspaceId}` 为您的业务空间 ID，可在阿里云百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 - -## **HTTP同步调用** - -一次请求即可获得结果，流程简单，推荐大多数场景使用。 - -## **北京** - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -## **新加坡** - -`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -#### 请求参数 - -## 文生图 - -> wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"text": "一间有着精致窗户的花店，漂亮的木质门，摆放着花朵"} - ] - } - ] - }, - "parameters": { - "size": "2K", - "n": 1, - "watermark": false, - "thinking_mode": true - } -}' -``` - -## **图像编辑** - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp"}, - {"image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp"}, - {"text": "把图2的涂鸦喷绘在图1的汽车上"} - ] - } - ] - }, - "parameters": { - "size": "2K", - "n": 1, - "watermark": false - } -}' -``` - -## **交互式编辑** - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"image": "https://img.alicdn.com/imgextra/i3/O1CN0157XGE51l6iL9441yX_!!6000000004770-49-tps-1104-1472.webp"}, - {"image": "https://img.alicdn.com/imgextra/i3/O1CN01SfG4J41UYn9WNt4X1_!!6000000002530-49-tps-1696-960.webp"}, - {"text": "把图1的闹钟放在图2的框选的位置，保持场景和光线融合自然"} - ] - } - ] - }, - "parameters": { - "bbox_list": [[],[[989, 515, 1138, 681]]], - "size": "2K", - "n": 1, - "watermark": false - } -}' -``` - -## **组图生成** - -> wan2.7-image-pro组图生成仅支持最高2K分辨率。 - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"text": "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。"} - ] - } - ] - }, - "parameters": { - "enable_sequential": true, - "n": 4, - "size": "2K" - } -}' -``` - -##### 请求头（Headers） - -**Content-Type** `_string_` **（必选）** - -请求内容类型。此参数必须设置为`application/json`。 - -**Authorization** `_string_`**（必选）** - -请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 - -##### 请求体（Request Body） - -**model** `_string_` **（必选）** - -模型名称。可选值：`wan2.7-image-pro`、`wan2.7-image`。 - -**input** `_object_` **（必选）** - -输入的基本信息。 - -**属性** - -**messages** `_array_` **（必选）** - -请求内容数组。当前**仅支持单轮对话**，即传入一组role、content参数，不支持多轮对话。 - -**属性** - -**role** `_string_` **（必选）** - -消息的角色。此参数固定设置为`user`。 - -**content** `_array_` **（必选）** - -消息内容数组。 - -**属性** - -**text** `_string_` - -用户输入提示词。支持中英文，长度不超过5000个字符，每个汉字、字母、数字或符号计为一个字符，超过部分会自动截断。 - -**image** `_string_` - -输入图像的URL或Base64编码字符串。 - -图像限制： - -- 图像格式：JPEG、JPG、PNG（不支持透明通道）、BMP、WEBP。 - -- 图像分辨率：图像的宽高范围均为\[240, 8000\]像素，宽高比范围\[1:8, 8:1\]。 - -- 文件大小：不超过20MB。 - - -图像数量限制： - -- 可传入0-9张图片。 - -- 当输入多张图像时，需在`content`数组中传入多个`image`对象，并按照数组顺序定义图像顺序。 - - -支持的输入格式： - -1. 使用公网可访问URL - - - 支持HTTP或HTTPS协议。 - - - 示例值：`http://wanx.alicdn.com/material/xxx.jpeg`。 - -2. 传入 Base64 编码图像后的字符串 - - - 格式：data:{MIME\_type};base64,{base64\_data} - - - 示例：data:image/jpeg;base64,GDU7MtCZzEbTbmRZ...（仅示意，实际需传入完整字符串） - - - Base64 编码规范请参见[图像传入方式](https://help.aliyun.com/zh/model-studio/wan-image-edit#8db0e2215frua)。 - - -**parameters** `_object_` （可选） - -模型参数配置。 - -**属性** - -**bbox\_list** `_array[array[array[integer]]]_` （可选） - -交互式编辑框选区域。 - -- 对应关系：列表长度必须与输入图片数量一致。若某张图片无需编辑，请在对应位置传入空列表 `[]`。 - -- 坐标格式：`[x1, y1, x2, y2]`（左上角 x, 左上角 y, 右下角 x, 右下角 y），使用原图绝对像素坐标，左上角坐标为（0，0）。 - -- 限制条件：单张图片最多支持 2 个边界框。 - - -示例：输入 3 张图片，其中第 2 张无框选，第 1 张有两个框选： - -``` -[ - [[0, 0, 12, 12], [25, 25, 100, 100]], # 图 1 (2个框) - [], # 图 2 (无框) - [[10, 10, 50, 50]] # 图 3 (1个框) -] -``` - -**enable\_sequential** `_boolean_` （可选） - -控制生图模式： - -- false：默认值。 - -- true ：启用组图输出模式。 - - -**size** `_string_` （可选） - -关于输出图片分辨率参数，支持以下两种方式，不可混用： - -**模型：wan2.7-image-pro** - -- **方式一：指定输出图片的分辨率（推荐）** - - - 支持 1K、2K（默认）、4K 三种规格 - - - **适用范围**： - - - 文生图（无图片输入，非组图生成）：支持1K、2K、4K。 - - - 其他场景：支持1K、2K。 - - - **各规格总像素**：1K：1024\*1024、2K：2048\*2048、4K：4096\*4096 - - - **图像比例**： - - - 当有图片输入时：输出宽高比与输入图像（多图输入时为最后一张）一致，并缩放到选定分辨率。 - - - 当没有图片输入时：输出为正方形。 - -- **方式二：指定生成图像的宽高像素值** - - - 文生图：总像素在 \[768\*768, 4096\*4096\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - - 其他场景：总像素在 \[768\*768, 2048\*2048\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - -**模型：wan2.7-image** - -- **方式一：指定输出图片的分辨率（推荐）** - - - 支持1K、2K（默认）两种规格，不支持4K。 - -- **方式二：指定生成图像的宽高像素值** - - - 所有场景下，总像素在 \[768\*768, 2048\*2048\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - -> 输出图片的像素值可能和指定像素值存在微小差异。 - -**n** `_integer_` （可选） - -**重要** - -n直接影响费用。费用 = 单价 × 成功生成的图片张数，请在调用前确认[模型价格](https://help.aliyun.com/zh/model-studio/model-pricing#e2540d71a2utl)。 - -- 关闭组图模式时，该数值代表生成图像数量，取值范围 1-4，**默认为 1**； - -- 开启组图模式时，该数值代表最大生成图像数量，取值范围 1-12，**默认为 12**。实际数量由模型决定且不超过 n。 - - -**thinking\_mode** `_boolean_` （可选） - -是否开启思考模式，默认为`true`（开启）。仅在关闭组图模式且无图片输入时生效。开启时，模型将增强推理能力以提升出图质量，但会增加生成耗时。 - -**color\_palette** `_array_` （可选） - -自定义颜色主题，一个包含颜色（hex）和占比（ratio）的对象数组，需要包含 3 至 10 种颜色，推荐设置为 8 种。 - -仅当关闭组图模式（`enable_sequential=false`）时可用。 - -**属性** - -**hex** `_string_` **（必选）** - -十六进制（HEX）格式的色值。 - -**ratio** `_string_` **（必选）** - -颜色所占的百分比，需精确到小数点后两位（如`"25.00%"`）。所有 ratio 值相加**总和必须为 100.00%**。 - -**点击查看输入示例** - -``` -"color_palette": [ - { - "hex": "#C2D1E6", - "ratio": "23.51%" - }, - { - "hex": "#CDD8E9", - "ratio": "20.13%" - }, - { - "hex": "#B5C8DB", - "ratio": "15.88%" - }, - { - "hex": "#C0B5B4", - "ratio": "13.27%" - }, - { - "hex": "#DAE0EC", - "ratio": "10.11%" - }, - { - "hex": "#636574", - "ratio": "8.93%" - }, - { - "hex": "#CACAD2", - "ratio": "5.55%" - }, - { - "hex": "#CBD4E4", - "ratio": "2.62%" - } -] -``` - -**watermark** `_bool_` （可选） - -是否添加水印标识，水印位于图片右下角，文案固定为“AI生成”。 - -- false：默认值，不添加水印。 - -- true：添加水印。 - - -**seed** `_integer_` （可选） - -随机数种子，取值范围`[0,2147483647]`。 - -使用相同的`seed`参数值可使生成内容保持相对稳定。若不提供，算法将自动使用随机数种子。 - -**注意**：模型生成过程具有概率性，即使使用相同的`seed`，也不能保证每次生成结果完全一致。 - -#### 响应参数 - -## 任务执行成功 - -任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 - -``` -{ - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "content": [ - { - "image": "https://dashscope-xxx.oss-xxx.aliyuncs.com/xxx.png?Expires=xxx", - "type": "image" - } - ], - "role": "assistant" - } - } - ], - "finished": true - }, - "usage": { - "image_count": 1, - "input_tokens": 10867, - "output_tokens": 2, - "size": "1488*704", - "total_tokens": 10869 - }, - "request_id": "71dfc3c6-f796-9972-97e4-bc4efc4faxxx" -} -``` - -## 任务执行异常 - -如果因为某种原因导致任务执行失败，将返回相关信息，可以通过code和message字段明确指示错误原因。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "request_id": "a4d78a5f-655f-9639-8437-xxxxxx", - "code": "InvalidParameter", - "message": "num_images_per_prompt must be 1" -} -``` - -**output** `_object_` - -任务输出信息。 - -**属性** - -**choices** `_array_` - -模型生成的输出内容。 - -**属性** - -**finish\_reason** `_string_` - -任务停止原因。自然停止时为`stop`。 - -**message** `_object_` - -模型返回的消息。 - -**属性** - -**role** `_string_` - -消息的角色，固定为`assistant`。 - -**content** `_array_` - -**属性** - -**type** `_string_` - -输出的类型，固定为image。 - -**image** `_string_` - -生成图像的 URL，图像格式为PNG。 - -**链接有效期为24小时**，请及时下载并保存图像。 - -**finished** `_boolean_` - -任务是否结束。 - -- true：已结束。 - -- false：未结束。 - - -**usage** `_object_` - -输出信息统计。只对成功的结果计数。 - -**属性** - -**image\_count** `_integer_` - -生成图像的张数。 - -**size** `_string_` - -生成的图像分辨率。示例值：1376\*768。 - -**input\_tokens** `_integer_` - -输入token数量（不计费）。按图片张数计费。 - -**output\_tokens** `_integer_` - -输出token数量（不计费）。按图片张数计费。 - -**total\_tokens** `_integer_` - -总token数量（不计费）。按图片张数计费。 - -**request\_id** `_string_` - -请求唯一标识。可用于请求明细溯源和问题排查。 - -**code** `_string_` - -请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**message** `_string_` - -请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -## **HTTP异步调用** - -适用于耗时较长的任务，支持查询任务状态和结果。 - -### 步骤1：创建任务获取任务ID - -## **北京** - -`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -## **新加坡** - -`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation` - -调用时请将`{WorkspaceId}`替换为真实的[业务空间ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -#### 请求参数 - -## **文生图** - -> wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header "X-DashScope-Async: enable" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"text": "一间有着精致窗户的花店，漂亮的木质门，摆放着花朵"} - ] - } - ] - }, - "parameters": { - "size": "2K", - "n": 1, - "watermark": false, - "thinking_mode": true - } -}' -``` - -## **图像编辑** - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header "X-DashScope-Async: enable" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp"}, - {"image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp"}, - {"text": "把图2的涂鸦喷绘在图1的汽车上"} - ] - } - ] - }, - "parameters": { - "size": "2K", - "n": 1, - "watermark": false - } -}' -``` - -## **交互式编辑** - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header "X-DashScope-Async: enable" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"image": "https://img.alicdn.com/imgextra/i3/O1CN0157XGE51l6iL9441yX_!!6000000004770-49-tps-1104-1472.webp"}, - {"image": "https://img.alicdn.com/imgextra/i3/O1CN01SfG4J41UYn9WNt4X1_!!6000000002530-49-tps-1696-960.webp"}, - {"text": "把图1的闹钟放在图2的框选的位置，保持场景和光线融合自然"} - ] - } - ] - }, - "parameters": { - "bbox_list": [[],[[989, 515, 1138, 681]]], - "size": "2K", - "n": 1, - "watermark": false - } -}' -``` - -## **组图生成** - -> wan2.7-image-pro组图生成仅支持最高2K分辨率。 - -``` -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image-generation/generation' \ ---header 'Content-Type: application/json' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header "X-DashScope-Async: enable" \ ---data '{ - "model": "wan2.7-image-pro", - "input": { - "messages": [ - { - "role": "user", - "content": [ - {"text": "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。"} - ] - } - ] - }, - "parameters": { - "enable_sequential": true, - "n": 4, - "size": "2K" - } -}' -``` - -##### 请求头（Headers） - -**Content-Type** `_string_` **（必选）** - -请求内容类型。此参数必须设置为`application/json`。 - -**Authorization** `_string_`**（必选）** - -请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 - -**X-DashScope-Async** `_string_` **（必选）** - -异步处理配置参数。HTTP请求只支持异步，**必须设置为**`**enable**`。 - -**重要** - -缺少此请求头将报错：“current user api does not support synchronous calls”。 - -##### 请求体（Request Body） - -**model** `_string_` **（必选）** - -模型名称。可选值：`wan2.7-image-pro`、`wan2.7-image`。 - -**input** `_object_` **（必选）** - -输入的基本信息。 - -**属性** - -**messages** `_array_` **（必选）** - -请求内容数组。当前**仅支持单轮对话**，即传入一组role、content参数，不支持多轮对话。 - -**属性** - -**role** `_string_` **（必选）** - -消息的角色。此参数固定设置为`user`。 - -**content** `_array_` **（必选）** - -消息内容数组。 - -**属性** - -**text** `_string_` - -用户输入提示词。支持中英文，长度不超过5000个字符，每个汉字、字母、数字或符号计为一个字符，超过部分会自动截断。 - -**image** `_string_` - -输入图像的URL或Base64编码字符串。 - -图像限制： - -- 图像格式：JPEG、JPG、PNG（不支持透明通道）、BMP、WEBP。 - -- 图像分辨率：图像的宽高范围均为\[240, 8000\]像素，宽高比范围\[1:8, 8:1\]。 - -- 文件大小：不超过20MB。 - - -图像数量限制： - -- 可传入0-9张图片。 - -- 当输入多张图像时，需在`content`数组中传入多个`image`对象，并按照数组顺序定义图像顺序。 - - -支持的输入格式： - -1. 使用公网可访问URL - - - 支持HTTP或HTTPS协议。 - - - 示例值：`http://wanx.alicdn.com/material/xxx.jpeg`。 - -2. 传入 Base64 编码图像后的字符串 - - - 格式：data:{MIME\_type};base64,{base64\_data} - - - 示例：data:image/jpeg;base64,GDU7MtCZzEbTbmRZ...（仅示意，实际需传入完整字符串） - - - Base64 编码规范请参见[图像传入方式](https://help.aliyun.com/zh/model-studio/wan-image-edit#8db0e2215frua)。 - - -**parameters** `_object_` （可选） - -模型参数配置。 - -**属性** - -**bbox\_list** `_array[array[array[integer]]]_` （可选） - -交互式编辑框选区域。 - -- 对应关系：列表长度必须与输入图片数量一致。若某张图片无需编辑，请在对应位置传入空列表 `[]`。 - -- 坐标格式：`[x1, y1, x2, y2]`（左上角 x, 左上角 y, 右下角 x, 右下角 y），使用原图绝对像素坐标，左上角坐标为（0，0）。 - -- 限制条件：单张图片最多支持 2 个边界框。 - - -示例：输入 3 张图片，其中第 2 张无框选，第 1 张有两个框选： - -``` -[ - [[0, 0, 12, 12], [25, 25, 100, 100]], # 图 1 (2个框) - [], # 图 2 (无框) - [[10, 10, 50, 50]] # 图 3 (1个框) -] -``` - -**enable\_sequential** `_boolean_` （可选） - -控制生图模式： - -- false：默认值。 - -- true ：启用组图输出模式。 - - -**size** `_string_` （可选） - -关于输出图片分辨率参数，支持以下两种方式，不可混用： - -**模型：wan2.7-image-pro** - -- **方式一：指定输出图片的分辨率（推荐）** - - - 支持 1K、2K（默认）、4K 三种规格 - - - **适用范围**： - - - 文生图（无图片输入，非组图生成）：支持1K、2K、4K。 - - - 其他场景：支持1K、2K。 - - - **各规格总像素**：1K：1024\*1024、2K：2048\*2048、4K：4096\*4096 - - - **图像比例**： - - - 当有图片输入时：输出宽高比与输入图像（多图输入时为最后一张）一致，并缩放到选定分辨率。 - - - 当没有图片输入时：输出为正方形。 - -- **方式二：指定生成图像的宽高像素值** - - - 文生图：总像素在 \[768\*768, 4096\*4096\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - - 其他场景：总像素在 \[768\*768, 2048\*2048\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - -**模型：wan2.7-image** - -- **方式一：指定输出图片的分辨率（推荐）** - - - 支持1K、2K（默认）两种规格，不支持4K。 - -- **方式二：指定生成图像的宽高像素值** - - - 所有场景下，总像素在 \[768\*768, 2048\*2048\] 之间，宽高比范围为 \[1:8, 8:1\]。 - - -> 输出图片的像素值可能和指定像素值存在微小差异。 - -**n** `_integer_` （可选） - -**重要** - -n直接影响费用。费用 = 单价 × 成功生成的图片张数，请在调用前确认[模型价格](https://help.aliyun.com/zh/model-studio/model-pricing#e2540d71a2utl)。 - -- 关闭组图模式时，该数值代表生成图像数量，取值范围 1-4，**默认为 1**； - -- 开启组图模式时，该数值代表最大生成图像数量，取值范围 1-12，**默认为 12**。实际数量由模型决定且不超过 n。 - - -**thinking\_mode** `_boolean_` （可选） - -是否开启思考模式，默认为`true`（开启）。仅在关闭组图模式且无图片输入时生效。开启时，模型将增强推理能力以提升出图质量，但会增加生成耗时。 - -**color\_palette** `_array_` （可选） - -自定义颜色主题，一个包含颜色（hex）和占比（ratio）的对象数组，需要包含 3 至 10 种颜色，推荐设置为 8 种。 - -仅当关闭组图模式（`enable_sequential=false`）时可用。 - -**属性** - -**hex** `_string_` **（必选）** - -十六进制（HEX）格式的色值。 - -**ratio** `_string_` **（必选）** - -颜色所占的百分比，需精确到小数点后两位（如`"25.00%"`）。所有 ratio 值相加**总和必须为 100.00%**。 - -**点击查看输入示例** - -``` -"color_palette": [ - { - "hex": "#C2D1E6", - "ratio": "23.51%" - }, - { - "hex": "#CDD8E9", - "ratio": "20.13%" - }, - { - "hex": "#B5C8DB", - "ratio": "15.88%" - }, - { - "hex": "#C0B5B4", - "ratio": "13.27%" - }, - { - "hex": "#DAE0EC", - "ratio": "10.11%" - }, - { - "hex": "#636574", - "ratio": "8.93%" - }, - { - "hex": "#CACAD2", - "ratio": "5.55%" - }, - { - "hex": "#CBD4E4", - "ratio": "2.62%" - } -] -``` - -**watermark** `_bool_` （可选） - -是否添加水印标识，水印位于图片右下角，文案固定为“AI生成”。 - -- false：默认值，不添加水印。 - -- true：添加水印。 - - -**seed** `_integer_` （可选） - -随机数种子，取值范围`[0,2147483647]`。 - -使用相同的`seed`参数值可使生成内容保持相对稳定。若不提供，算法将自动使用随机数种子。 - -**注意**：模型生成过程具有概率性，即使使用相同的`seed`，也不能保证每次生成结果完全一致。 - -#### 响应参数 - -#### 成功响应 - -请保存 task\_id，用于查询任务状态与结果。 - -``` -{ - "output": { - "task_status": "PENDING", - "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx" - }, - "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx" -} -``` - -#### 异常响应 - -创建任务失败，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "code": "InvalidApiKey", - "message": "No API-key provided.", - "request_id": "7438d53d-6eb8-4596-8835-xxxxxx" -} -``` - -**output** `_object_` - -任务输出信息。 - -**属性** - -**task\_id** `_string_` - -任务ID。查询有效期24小时。 - -**task\_status** `_string_` - -任务状态。 - -**枚举值** - -- PENDING：任务排队中 - -- RUNNING：任务处理中 - -- SUCCEEDED：任务执行成功 - -- FAILED：任务执行失败 - -- CANCELED：任务已取消 - -- UNKNOWN：任务不存在或状态未知 - - -**request\_id** `_string_` - -请求唯一标识。可用于请求明细溯源和问题排查。 - -**code** `_string_` - -请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**message** `_string_` - -请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -### 步骤2：根据任务ID查询结果 - -## **北京** - -`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` - -## **新加坡** - -`GET https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/{task_id}` - -#### 请求参数 - -## 查询任务结果 - -将`{task_id}`完整替换为上一步接口返回的`task_id`的值。`task_id`查询有效期为24小时。 - -``` -curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id} \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" -``` - -##### **请求头（Headers）** - -**Authorization** `_string_`**（必选）** - -请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 - -##### **URL路径参数（Path parameters）** - -**task\_id** `_string_`**（必选）** - -任务ID。 - -#### 响应参数 - -## 任务执行成功 - -任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 - -``` -{ - "request_id": "810fa5f5-334c-91f3-aaa4-ed89cf0caxxx", - "output": { - "task_id": "a81ee7cb-014c-473d-b842-76e98311cxxx", - "task_status": "SUCCEEDED", - "submit_time": "2026-03-26 17:16:01.663", - "scheduled_time": "2026-03-26 17:16:01.716", - "end_time": "2026-03-26 17:16:22.961", - "finished": true, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-xxx.oss-xxx.aliyuncs.com/xxx.png?Expires=xxx", - "type": "image" - } - ] - } - } - ] - }, - "usage": { - "size": "2976*1408", - "total_tokens": 11017, - "image_count": 1, - "output_tokens": 2, - "input_tokens": 11015 - } -} -``` - -## 任务执行异常 - -如果因为某种原因导致任务执行失败，将返回相关信息，可以通过code和message字段明确指示错误原因。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 - -``` -{ - "request_id": "a4d78a5f-655f-9639-8437-xxxxxx", - "code": "InvalidParameter", - "message": "num_images_per_prompt must be 1" -} -``` - -**output** `_object_` - -任务输出信息。 - -**属性** - -**task\_id** `_string_` - -任务ID。查询有效期24小时。 - -**task\_status** `_string_` - -任务状态。 - -**枚举值** - -- PENDING：任务排队中 - -- RUNNING：任务处理中 - -- SUCCEEDED：任务执行成功 - -- FAILED：任务执行失败 - -- CANCELED：任务已取消 - -- UNKNOWN：任务不存在或状态未知 - - -**轮询过程中的状态流转：** - -- PENDING（排队中） → RUNNING（处理中）→ SUCCEEDED（成功）/ FAILED（失败）。 - -- 初次查询状态通常为 PENDING（排队中）或 RUNNING（处理中）。 - -- 当状态变为 SUCCEEDED 时，响应中将包含生成的图像URL。 - -- 若状态为 FAILED，请检查错误信息并重试。 - - -**submit\_time** `_string_` - -任务提交时间。格式为 YYYY-MM-DD HH:mm:ss.SSS。 - -**scheduled\_time** `_string_` - -任务执行时间。格式为 YYYY-MM-DD HH:mm:ss.SSS。 - -**end\_time** `_string_` - -任务完成时间。格式为 YYYY-MM-DD HH:mm:ss.SSS。 - -**finished** `_boolean_` - -任务是否结束。 - -- true：已结束。 - -- false：未结束。 - - -**choices** `_array_` - -模型生成的输出内容。 - -**属性** - -**finish\_reason** `_string_` - -任务停止原因，自然停止时为`stop`。 - -**message** `_object_` - -模型返回的消息。 - -**属性** - -**role** `_string_` - -消息的角色，固定为`assistant`。 - -**content** `_array_` - -**属性** - -**type** `_string_` - -输出的类型，枚举值为text、image。 - -**text** `_string_` - -生成的文字。 - -**image** `_string_` - -生成图像的 URL，图像格式为PNG。 - -**链接有效期为24小时**，请及时下载并保存图像。 - -**usage** `_object_` - -输出信息统计。只对成功的结果计数。 - -**属性** - -**image\_count** `_integer_` - -生成图像的张数。 - -**size** `_string_` - -生成的图像分辨率。示例值：1376\*768。 - -**input\_tokens** `_integer_` - -输入token数量（不计费）。按图片张数计费。 - -**output\_tokens** `_integer_` - -输出token数量（不计费）。按图片张数计费。 - -**total\_tokens** `_integer_` - -总token数量（不计费）。按图片张数计费。 - -**request\_id** `_string_` - -请求唯一标识。可用于请求明细溯源和问题排查。 - -**code** `_string_` - -请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -**message** `_string_` - -请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 - -## **Python SDK调用** - -SDK 参数命名与HTTP接口基本一致。 - -任务可能耗时较长，SDK 已封装HTTP异步调用流程，同时支持同步和异步调用。 - -> 具体耗时受限于排队任务数和服务执行情况，请在获取结果时耐心等待。 - -**重要** - -请确保 DashScope Python SDK版本**不低于 1.25.15**，再运行以下代码。更新请参考[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 - -各地域的`base_url`和API Key 不通用，以下示例以北京地域为例进行调用： - -### 华北2（北京） - -`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1` - -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -### 新加坡 - -`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1` - -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -### **图像编辑** - -## **同步调用** - -##### **请求示例** - -``` -import os -import base64 -import mimetypes -import urllib.request -import dashscope -from dashscope.aigc.image_generation import ImageGeneration -from dashscope.api_entities.dashscope_response import Message - -# 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -api_key = os.getenv("DASHSCOPE_API_KEY") - -# --- Base64编码函数 --- -# base64编码格式为 data:{MIME_type};base64,{base64_data} -def encode_file(file_path): - mime_type, _ = mimetypes.guess_type(file_path) - if not mime_type or not mime_type.startswith("image/"): - raise ValueError("不支持或无法识别的图像格式") - with open(file_path, "rb") as image_file: - encoded_string = base64.b64encode(image_file.read()).decode("utf-8") - return f"data:{mime_type};base64,{encoded_string}" - -""" -图像输入方式说明： -以下提供了三种图片输入方式，三选一即可 -1. 使用公网URL - 适合已有公开可访问的图片 -2. 使用本地文件 - 适合本地开发测试 -3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 -""" -# 【方式一】使用公网图片 URL -image_1 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp" -image_2 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp" - -# 【方式二】使用本地文件（支持绝对路径和相对路径） -# image_1 = "file:///path/to/your/car.png" -# image_2 = "file:///path/to/your/paint.png" - -# 【方式三】使用Base64编码的图片 -# image_1 = encode_file("/path/to/your/car.png") -# image_2 = encode_file("/path/to/your/paint.png") - -message = Message( - role="user", - content=[ - {"text": "把图2的涂鸦喷绘在图1的汽车上"}, - {"image": image_1}, - {"image": image_2}, - ], -) -print("----sync call, please wait a moment----") -rsp = ImageGeneration.call( - model="wan2.7-image-pro", - api_key=api_key, - messages=[message], - watermark=False, - n=1, - size="2K", # wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 -) - -# 提取结果图片URL并保存到本地 -if rsp.status_code == 200: - for i, choice in enumerate(rsp.output.choices): - for j, content in enumerate(choice["message"]["content"]): - if content.get("type") == "image": - image_url = content["image"] - file_name = f"output_{i}_{j}.png" - # 结果URL有效期为24小时，请及时下载 - urllib.request.urlretrieve(image_url, file_name) - print(f"Image saved to {file_name}") -else: - print(f"Failed: status_code={rsp.status_code}, message={rsp.message}") -``` - -##### 响应示例 - -> URL 有效期24小时，请及时下载图像。 - -``` -{ - "status_code": 200, - "request_id": "81d868c6-6ce1-92d8-a90d-d2ee71xxxxxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "audio": null, - "finished": true - }, - "usage": { - "input_tokens": 18790, - "output_tokens": 2, - "characters": 0, - "image_count": 1, - "size": "2985*1405", - "total_tokens": 18792 - } -} -``` - -## **异步调用** - -##### **请求示例** - -``` -import os -import base64 -import mimetypes -import urllib.request -import dashscope -from dashscope.aigc.image_generation import ImageGeneration -from dashscope.api_entities.dashscope_response import Message -from http import HTTPStatus - -# 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -api_key = os.getenv("DASHSCOPE_API_KEY") - -# --- Base64编码函数 --- -# base64编码格式为 data:{MIME_type};base64,{base64_data} -def encode_file(file_path): - mime_type, _ = mimetypes.guess_type(file_path) - if not mime_type or not mime_type.startswith("image/"): - raise ValueError("不支持或无法识别的图像格式") - with open(file_path, "rb") as image_file: - encoded_string = base64.b64encode(image_file.read()).decode("utf-8") - return f"data:{mime_type};base64,{encoded_string}" - -""" -图像输入方式说明： -以下提供了三种图片输入方式，三选一即可 -1. 使用公网URL - 适合已有公开可访问的图片 -2. 使用本地文件 - 适合本地开发测试 -3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 -""" -# 【方式一】使用公网图片 URL -image_1 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp" -image_2 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp" - -# 【方式二】使用本地文件（支持绝对路径和相对路径） -# image_1 = "file:///path/to/your/car.png" -# image_2 = "file:///path/to/your/paint.png" - -# 【方式三】使用Base64编码的图片 -# image_1 = encode_file("/path/to/your/car.png") -# image_2 = encode_file("/path/to/your/paint.png") - -# 创建异步任务 -def create_async_task(): - print("Creating async task...") - message = Message( - role="user", - content=[ - {"text": "把图2的涂鸦喷绘在图1的汽车上"}, - {"image": image_1}, - {"image": image_2}, - ], - ) - response = ImageGeneration.async_call( - model="wan2.7-image-pro", - api_key=api_key, - messages=[message], - watermark=False, - n=1, - size="2K", # wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - ) - - if response.status_code == 200: - print("Task created successfully:", response) - return response - else: - raise Exception(f"Failed to create task: {response.code} - {response.message}") - -# 等待任务完成 -def wait_for_completion(task_response): - print("Waiting for task completion...") - status = ImageGeneration.wait(task=task_response, api_key=api_key) - - if status.output.task_status == "SUCCEEDED": - print("Task succeeded!") - # 提取结果图片URL并保存到本地 - for i, choice in enumerate(status.output.choices): - for j, content in enumerate(choice["message"]["content"]): - if content.get("type") == "image": - image_url = content["image"] - file_name = f"output_{i}_{j}.png" - # 结果URL有效期为24小时，请及时下载 - urllib.request.urlretrieve(image_url, file_name) - print(f"Image saved to {file_name}") - else: - raise Exception(f"Task failed with status: {status.output.task_status}") - -# 获取异步任务信息 -def fetch_task_status(task): - print("Fetching task status...") - status = ImageGeneration.fetch(task=task, api_key=api_key) - - if status.status_code == HTTPStatus.OK: - print("Task status:", status.output.task_status) - print("Response details:", status) - else: - print(f"Failed to fetch status: {status.code} - {status.message}") - -# 取消异步任务 -def cancel_task(task): - print("Canceling task...") - response = ImageGeneration.cancel(task=task, api_key=api_key) - - if response.status_code == HTTPStatus.OK: - print("Task canceled successfully:", response.output.task_status) - else: - print(f"Failed to cancel task: {response.code} - {response.message}") - -# 主执行流程 -if __name__ == "__main__": - task = create_async_task() - wait_for_completion(task) -``` - -##### 响应示例 - -1、创建任务的响应示例 - -``` -{ - "status_code": 200, - "request_id": "4fb3050f-de57-4a24-84ff-e37ee5xxxxxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": null, - "audio": null, - "task_id": "127ec645-118f-4884-955d-0eba8dxxxxxx", - "task_status": "PENDING" - }, - "usage": { - "input_tokens": 0, - "output_tokens": 0, - "characters": 0 - } -} -``` - -2、查询任务结果的响应示例 - -> URL 有效期24小时，请及时下载图像。 - -``` -{ - "status_code": 200, - "request_id": "3b99aae5-d26f-9059-8dd0-ee9ca4804xxx", - "code": null, - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "audio": null, - "task_id": "127ec645-118f-4884-955d-0eba8dxxxxxx", - "task_status": "SUCCEEDED", - "submit_time": "2026-03-31 22:58:47.646", - "scheduled_time": "2026-03-31 22:58:47.683", - "end_time": "2026-03-31 22:58:59.642", - "finished": true - }, - "usage": { - "input_tokens": 18711, - "output_tokens": 2, - "characters": 0, - "size": "2985*1405", - "total_tokens": 18713, - "image_count": 1 - } -} -``` - -### **组图生成** - -## **同步调用** - -##### **请求示例** - -``` -import os -import base64 -import mimetypes -import urllib.request -import dashscope -from dashscope.aigc.image_generation import ImageGeneration -from dashscope.api_entities.dashscope_response import Message - -# 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -api_key = os.getenv("DASHSCOPE_API_KEY") - -# --- Base64编码函数 --- -# base64编码格式为 data:{MIME_type};base64,{base64_data} -def encode_file(file_path): - mime_type, _ = mimetypes.guess_type(file_path) - if not mime_type or not mime_type.startswith("image/"): - raise ValueError("不支持或无法识别的图像格式") - with open(file_path, "rb") as image_file: - encoded_string = base64.b64encode(image_file.read()).decode("utf-8") - return f"data:{mime_type};base64,{encoded_string}" - -""" -图像输入方式说明（图生组图场景）： -以下提供了三种图片输入方式，三选一即可 -1. 使用公网URL - 适合已有公开可访问的图片 -2. 使用本地文件 - 适合本地开发测试 -3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 -""" -# 【方式一】使用公网图片 URL -# image_1 = "https://img.alicdn.com/imgextra/i4/O1CN01IM44WN23dq5uY1yla_!!6000000007279-49-tps-1024-1024.webp" - -# 【方式二】使用本地文件（支持绝对路径和相对路径） -# image_1 = "file:///path/to/your/image.png" - -# 【方式三】使用Base64编码的图片 -# image_1 = encode_file("/path/to/your/image.png") - -message = Message( - role="user", - content=[ - { - "text": "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。" - } - # 图生组图场景：取消以下注释并注释掉上方纯文本 - # {"text": "参考图片风格生成四季组图"}, - # {"image": image_1} - ], -) - -print("----sync call, please wait a moment----") -rsp = ImageGeneration.call( - model="wan2.7-image-pro", - api_key=api_key, - messages=[message], - enable_sequential=True, - n=4, - size="2K", # wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 -) - -# 提取结果图片URL并保存到本地 -if rsp.status_code == 200: - for i, choice in enumerate(rsp.output.choices): - for j, content in enumerate(choice["message"]["content"]): - if content.get("type") == "image": - image_url = content["image"] - file_name = f"output_{i}_{j}.png" - # 结果URL有效期为24小时，请及时下载 - urllib.request.urlretrieve(image_url, file_name) - print(f"Image saved to {file_name}") -else: - print(f"Failed: status_code={rsp.status_code}, message={rsp.message}") -``` - -##### 响应示例 - -> URL 有效期24小时，请及时下载图像。 - -``` -{ - "status_code": 200, - "request_id": "56e318fd-ed60-99e8-8ca1-cdef25ca4xxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "audio": null, - "finished": true - }, - "usage": { - "input_tokens": 720, - "output_tokens": 11, - "characters": 0, - "image_count": 4, - "size": "2048*2048", - "total_tokens": 731 - } -} -``` - -## **异步调用** - -##### **请求示例** - -``` -import os -import base64 -import mimetypes -import urllib.request -import dashscope -from dashscope.aigc.image_generation import ImageGeneration -from dashscope.api_entities.dashscope_response import Message - -# 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -api_key = os.getenv("DASHSCOPE_API_KEY") - -# --- Base64编码函数 --- -# base64编码格式为 data:{MIME_type};base64,{base64_data} -def encode_file(file_path): - mime_type, _ = mimetypes.guess_type(file_path) - if not mime_type or not mime_type.startswith("image/"): - raise ValueError("不支持或无法识别的图像格式") - with open(file_path, "rb") as image_file: - encoded_string = base64.b64encode(image_file.read()).decode("utf-8") - return f"data:{mime_type};base64,{encoded_string}" - -""" -图像输入方式说明（图生组图场景）： -以下提供了三种图片输入方式，三选一即可 -1. 使用公网URL - 适合已有公开可访问的图片 -2. 使用本地文件 - 适合本地开发测试 -3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 -""" -# 【方式一】使用公网图片 URL -# image_1 = "https://img.alicdn.com/imgextra/i4/O1CN01IM44WN23dq5uY1yla_!!6000000007279-49-tps-1024-1024.webp" - -# 【方式二】使用本地文件（支持绝对路径和相对路径） -# image_1 = "file:///path/to/your/image.png" - -# 【方式三】使用Base64编码的图片 -# image_1 = encode_file("/path/to/your/image.png") - -def main(): - message = Message( - role="user", - content=[ - { - "text": "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。" - } - # 图生组图场景：取消以下注释并注释掉上方纯文本 - # {"text": "参考图片风格生成四季组图"}, - # {"image": image_1} - ], - ) - - # 提交异步任务 - print("提交异步任务...") - response = ImageGeneration.async_call( - model="wan2.7-image-pro", - api_key=api_key, - messages=[message], - enable_sequential=True, - n=4, - size="2K", # wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - ) - - if response.status_code == 200: - print(f"任务提交成功，任务ID: {response.output.task_id}") - - # 等待任务完成 - status = ImageGeneration.wait(task=response, api_key=api_key) - - if status.output.task_status == "SUCCEEDED": - print("任务完成!") - # 提取结果图片URL并保存到本地 - for i, choice in enumerate(status.output.choices): - for j, content in enumerate(choice["message"]["content"]): - if content.get("type") == "image": - image_url = content["image"] - file_name = f"output_{i}_{j}.png" - # 结果URL有效期为24小时，请及时下载 - urllib.request.urlretrieve(image_url, file_name) - print(f"Image saved to {file_name}") - else: - print(f"任务失败，状态: {status.output.task_status}") - else: - print(f"任务创建失败: {response.code} - {response.message}") - -if __name__ == "__main__": - try: - main() - except Exception as e: - print(f"错误: {e}") -``` - -##### 响应示例 - -1、创建任务的响应示例 - -``` -{ - "status_code": 200, - "request_id": "4fb3050f-de57-4a24-84ff-e37ee5xxxxxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": null, - "audio": null, - "task_id": "77093787-a217-4c29-9cd4-ca7b5ac86xxx", - "task_status": "PENDING" - }, - "usage": { - "input_tokens": 0, - "output_tokens": 0, - "characters": 0 - } -} -``` - -2、查询任务结果的响应示例 - -> URL 有效期24小时，请及时下载图像。 - -``` -{ - "status_code": 200, - "request_id": "56e318fd-ed60-99e8-8ca1-cdef25ca4xxx", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "audio": null, - "task_id": "77093787-a217-4c29-9cd4-ca7b5ac86xxx", - "task_status": "SUCCEEDED", - "submit_time": "2026-03-31 23:04:46.166", - "scheduled_time": "2026-03-31 23:04:46.208", - "end_time": "2026-03-31 23:05:11.664", - "finished": true - }, - "usage": { - "input_tokens": 720, - "output_tokens": 11, - "characters": 0, - "size": "2048*2048", - "total_tokens": 731, - "image_count": 4 - } -} -``` - -## **Java SDK调用** - -SDK 参数命名与HTTP接口基本一致。 - -任务可能耗时较长，SDK 已封装HTTP异步调用流程，同时支持同步和异步调用。 - -**重要** - -请确保 DashScope Java SDK版本不低于 `2.22.13`，否则可能不支持本文所用的部分参数。 - -### 华北2（北京） - -`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1` - -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -### 新加坡 - -`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1` - -调用时请将`WorkspaceId`替换为真实的[Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 - -### **图像编辑** - -## **同步调用** - -##### 请求示例 - -``` -import com.alibaba.dashscope.aigc.imagegeneration.*; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.io.IOException; -import java.io.InputStream; -import java.net.URL; -import java.nio.file.Files; -import java.nio.file.Paths; -import java.nio.file.StandardCopyOption; -import java.util.Arrays; -import java.util.Base64; -import java.util.Collections; -import java.util.List; -import java.util.Map; - -/** - * wan2.7-image-pro 图像编辑 - 同步调用示例 - */ -public class Main { - - static { - // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - } - - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - static String apiKey = System.getenv("DASHSCOPE_API_KEY"); - - // --- Base64编码函数 --- - // base64编码格式为 data:{MIME_type};base64,{base64_data} - public static String encodeFile(String filePath) throws IOException { - byte[] fileContent = Files.readAllBytes(Paths.get(filePath)); - String base64String = Base64.getEncoder().encodeToString(fileContent); - String mimeType = Files.probeContentType(Paths.get(filePath)); - return "data:" + mimeType + ";base64," + base64String; - } - - public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException, IOException { - /* - * 图像输入方式说明： - * 以下提供了三种图片输入方式，三选一即可 - * 1. 使用公网URL - 适合已有公开可访问的图片 - * 2. 使用本地文件 - 适合本地开发测试 - * 3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 - */ - // 【方式一】使用公网图片 URL - String image1 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp"; - String image2 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp"; - - // 【方式二】使用本地文件（支持绝对路径和相对路径） - // 格式要求：file:// + 文件路径 - // String image1 = "file:///path/to/your/car.png"; - // String image2 = "file:///path/to/your/paint.png"; - - // 【方式三】使用Base64编码的图片 - // String image1 = encodeFile("/path/to/your/car.png"); - // String image2 = encodeFile("/path/to/your/paint.png"); - - // 构建多图输入消息 - ImageGenerationMessage message = ImageGenerationMessage.builder() - .role("user") - .content(Arrays.asList( - // 支持多图输入，可以提供多张参考图片 - Collections.singletonMap("text", "把图2的涂鸦喷绘在图1的汽车上"), - Collections.singletonMap("image", image1), - Collections.singletonMap("image", image2) - )).build(); - - ImageGenerationParam param = ImageGenerationParam.builder() - .apiKey(apiKey) - .model("wan2.7-image-pro") - .messages(Collections.singletonList(message)) - .n(1) - .size("2K") // wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - .build(); - - ImageGeneration imageGeneration = new ImageGeneration(); - ImageGenerationResult result = null; - try { - System.out.println("---sync call for image editing, please wait a moment----"); - result = imageGeneration.call(param); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - throw new RuntimeException(e.getMessage()); - } - // 提取结果图片URL并保存到本地 - for (int i = 0; i < result.getOutput().getChoices().size(); i++) { - List> contents = result.getOutput().getChoices().get(i) - .getMessage().getContent(); - for (int j = 0; j < contents.size(); j++) { - if ("image".equals(contents.get(j).get("type"))) { - String imageUrl = (String) contents.get(j).get("image"); - String fileName = "output_" + i + "_" + j + ".png"; - // 结果URL有效期为24小时，请及时下载 - try (InputStream in = new URL(imageUrl).openStream()) { - Files.copy(in, Paths.get(fileName), StandardCopyOption.REPLACE_EXISTING); - } - System.out.println("Image saved to " + fileName); - } - } - } - } - - public static void main(String[] args) throws ApiException, NoApiKeyException, UploadFileException, IOException { - basicCall(); - } -} -``` - -##### 响应示例 - -> URL 有效期24小时，请及时下载并保存图像。 - -``` -{ - "requestId": "1bf6173a-e8de-9f75-94d3-5e618f875xxx", - "usage": { - "input_tokens": 18790, - "output_tokens": 2, - "total_tokens": 18792, - "image_count": 1, - "size": "2985*1405" - }, - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "finished": true - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -## **异步调用** - -##### 请求示例 - -``` -import com.alibaba.dashscope.aigc.imagegeneration.*; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.io.IOException; -import java.io.InputStream; -import java.net.URL; -import java.nio.file.Files; -import java.nio.file.Paths; -import java.nio.file.StandardCopyOption; -import java.util.Arrays; -import java.util.Base64; -import java.util.Collections; -import java.util.List; -import java.util.Map; - -/** - * wan2.7-image-pro 图像编辑 - 异步调用示例 - */ -public class Main { - - static { - // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - } - - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - static String apiKey = System.getenv("DASHSCOPE_API_KEY"); - - // --- Base64编码函数 --- - // base64编码格式为 data:{MIME_type};base64,{base64_data} - public static String encodeFile(String filePath) throws IOException { - byte[] fileContent = Files.readAllBytes(Paths.get(filePath)); - String base64String = Base64.getEncoder().encodeToString(fileContent); - String mimeType = Files.probeContentType(Paths.get(filePath)); - return "data:" + mimeType + ";base64," + base64String; - } - - public static void asyncCall() throws ApiException, NoApiKeyException, UploadFileException, IOException { - /* - * 图像输入方式说明： - * 以下提供了三种图片输入方式，三选一即可 - * 1. 使用公网URL - 适合已有公开可访问的图片 - * 2. 使用本地文件 - 适合本地开发测试 - * 3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 - */ - // 【方式一】使用公网图片 URL - String image1 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/pjeqdf/car.webp"; - String image2 = "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20251229/xsunlm/paint.webp"; - - // 【方式二】使用本地文件（支持绝对路径和相对路径） - // 格式要求：file:// + 文件路径 - // String image1 = "file:///path/to/your/car.png"; - // String image2 = "file:///path/to/your/paint.png"; - - // 【方式三】使用Base64编码的图片 - // String image1 = encodeFile("/path/to/your/car.png"); - // String image2 = encodeFile("/path/to/your/paint.png"); - - // 构建多图输入消息 - ImageGenerationMessage message = ImageGenerationMessage.builder() - .role("user") - .content(Arrays.asList( - // 支持多图输入，可以提供多张参考图片 - Collections.singletonMap("text", "把图2的涂鸦喷绘在图1的汽车上"), - Collections.singletonMap("image", image1), - Collections.singletonMap("image", image2) - )).build(); - - ImageGenerationParam param = ImageGenerationParam.builder() - .apiKey(apiKey) - .model("wan2.7-image-pro") - .n(1) - .size("2K") // wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - .messages(Arrays.asList(message)) - .build(); - - ImageGeneration imageGeneration = new ImageGeneration(); - ImageGenerationResult result = null; - try { - System.out.println("---async call for image editing, creating task----"); - result = imageGeneration.asyncCall(param); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - throw new RuntimeException(e.getMessage()); - } - System.out.println("任务创建结果:"); - System.out.println(JsonUtils.toJson(result)); - - String taskId = result.getOutput().getTaskId(); - // 等待任务完成 - waitTask(taskId); - } - - public static void waitTask(String taskId) throws ApiException, NoApiKeyException, IOException { - ImageGeneration imageGeneration = new ImageGeneration(); - System.out.println("\n---waiting for task completion----"); - ImageGenerationResult result = imageGeneration.wait(taskId, apiKey); - // 提取结果图片URL并保存到本地 - for (int i = 0; i < result.getOutput().getChoices().size(); i++) { - List> contents = result.getOutput().getChoices().get(i) - .getMessage().getContent(); - for (int j = 0; j < contents.size(); j++) { - if ("image".equals(contents.get(j).get("type"))) { - String imageUrl = (String) contents.get(j).get("image"); - String fileName = "output_" + i + "_" + j + ".png"; - // 结果URL有效期为24小时，请及时下载 - try (InputStream in = new URL(imageUrl).openStream()) { - Files.copy(in, Paths.get(fileName), StandardCopyOption.REPLACE_EXISTING); - } - System.out.println("Image saved to " + fileName); - } - } - } - } - - public static void main(String[] args) throws ApiException, NoApiKeyException, UploadFileException, IOException { - asyncCall(); - } -} -``` - -##### 响应示例 - -1、创建任务的响应示例 - -``` -{ - "requestId": "ccf4b2f4-bf30-9e13-9461-3a28c6a7bxxx", - "output": { - "task_id": "8811b4a4-00ac-4aa2-a2fd-017d3b90cxxx", - "task_status": "PENDING" - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -2、查询任务结果的响应示例 - -> URL 有效期24小时，请及时下载并保存图像。 - -``` -{ - "requestId": "60a08540-f1c1-9e76-8cd3-d5949db8cxxx", - "usage": { - "input_tokens": 18711, - "output_tokens": 2, - "total_tokens": 18713, - "image_count": 1, - "size": "2985*1405" - }, - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "task_id": "8811b4a4-00ac-4aa2-a2fd-017d3b90cxxx", - "task_status": "SUCCEEDED", - "finished": true, - "submit_time": "2026-03-31 19:57:58.840", - "scheduled_time": "2026-03-31 19:57:58.877", - "end_time": "2026-03-31 19:58:11.563" - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -### **组图生成** - -## **同步调用** - -##### 请求示例 - -``` -import com.alibaba.dashscope.aigc.imagegeneration.*; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.io.IOException; -import java.io.InputStream; -import java.net.URL; -import java.nio.file.Files; -import java.nio.file.Paths; -import java.nio.file.StandardCopyOption; -import java.util.Arrays; -import java.util.Base64; -import java.util.Collections; -import java.util.List; -import java.util.Map; - -/** - * wan2.7-image-pro 组图生成 - 同步调用示例（北京地域） - */ -public class Main { - - static { - // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - } - - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - static String apiKey = System.getenv("DASHSCOPE_API_KEY"); - - // --- Base64编码函数 --- - // base64编码格式为 data:{MIME_type};base64,{base64_data} - public static String encodeFile(String filePath) throws IOException { - byte[] fileContent = Files.readAllBytes(Paths.get(filePath)); - String base64String = Base64.getEncoder().encodeToString(fileContent); - String mimeType = Files.probeContentType(Paths.get(filePath)); - return "data:" + mimeType + ";base64," + base64String; - } - - public static void basicCall() throws ApiException, NoApiKeyException, UploadFileException, IOException { - /* - * 图像输入方式说明（图生组图场景）： - * 以下提供了三种图片输入方式，三选一即可 - * 1. 使用公网URL - 适合已有公开可访问的图片 - * 2. 使用本地文件 - 适合本地开发测试 - * 3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 - */ - // 【方式一】使用公网图片 URL - // String image1 = "https://img.alicdn.com/imgextra/i4/O1CN01IM44WN23dq5uY1yla_!!6000000007279-49-tps-1024-1024.webp"; - - // 【方式二】使用本地文件（支持绝对路径和相对路径） - // 格式要求：file:// + 文件路径 - // String image1 = "file:///path/to/your/image.png"; - - // 【方式三】使用Base64编码的图片 - // String image1 = encodeFile("/path/to/your/image.png"); - - // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例） - ImageGenerationMessage message = ImageGenerationMessage.builder() - .role("user") - .content(Collections.singletonList( - Collections.singletonMap("text", "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。") - )).build(); - // 图生组图场景：取消以下注释并注释掉上方纯文本构建 - // ImageGenerationMessage message = ImageGenerationMessage.builder() - // .role("user") - // .content(Arrays.asList( - // Collections.singletonMap("text", "参考图片风格生成四季组图"), - // Collections.singletonMap("image", image1) - // )).build(); - - ImageGenerationParam param = ImageGenerationParam.builder() - .apiKey(apiKey) - .model("wan2.7-image-pro") - .messages(Collections.singletonList(message)) - .enableSequential(true) - .n(4) - .size("2K") // wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - .build(); - - ImageGeneration imageGeneration = new ImageGeneration(); - ImageGenerationResult result = null; - try { - System.out.println("----sync call, please wait a moment----"); - result = imageGeneration.call(param); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - throw new RuntimeException(e.getMessage()); - } - // 提取结果图片URL并保存到本地 - for (int i = 0; i < result.getOutput().getChoices().size(); i++) { - List> contents = result.getOutput().getChoices().get(i) - .getMessage().getContent(); - for (int j = 0; j < contents.size(); j++) { - if ("image".equals(contents.get(j).get("type"))) { - String imageUrl = (String) contents.get(j).get("image"); - String fileName = "output_" + i + "_" + j + ".png"; - // 结果URL有效期为24小时，请及时下载 - try (InputStream in = new URL(imageUrl).openStream()) { - Files.copy(in, Paths.get(fileName), StandardCopyOption.REPLACE_EXISTING); - } - System.out.println("Image saved to " + fileName); - } - } - } - } - - public static void main(String[] args) throws ApiException, NoApiKeyException, UploadFileException, IOException { - basicCall(); - } -} -``` - -##### 响应示例 - -> URL 有效期24小时，请及时下载并保存图像。 - -``` -{ - "requestId": "4678c314-b37a-91c9-a2ae-2d3cd54bbxxx", - "usage": { - "input_tokens": 720, - "output_tokens": 11, - "total_tokens": 731, - "image_count": 4, - "size": "2048*2048" - }, - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "finished": true - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -## **异步调用** - -##### 请求示例 - -``` -import com.alibaba.dashscope.aigc.imagegeneration.*; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; -import com.alibaba.dashscope.utils.JsonUtils; - -import java.io.IOException; -import java.io.InputStream; -import java.net.URL; -import java.nio.file.Files; -import java.nio.file.Paths; -import java.nio.file.StandardCopyOption; -import java.util.Arrays; -import java.util.Base64; -import java.util.Collections; -import java.util.List; -import java.util.Map; - -/** - * wan2.7-image-pro 组图生成 - 异步调用示例（北京地域） - */ -public class Main { - - static { - // 以下为华北2（北京）地域的URL，调用时请将{WorkspaceId}替换为真实的业务空间ID，各地域的URL不同。 - Constants.baseHttpApiUrl = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1"; - } - - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey="sk-xxx" - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - static String apiKey = System.getenv("DASHSCOPE_API_KEY"); - - // --- Base64编码函数 --- - // base64编码格式为 data:{MIME_type};base64,{base64_data} - public static String encodeFile(String filePath) throws IOException { - byte[] fileContent = Files.readAllBytes(Paths.get(filePath)); - String base64String = Base64.getEncoder().encodeToString(fileContent); - String mimeType = Files.probeContentType(Paths.get(filePath)); - return "data:" + mimeType + ";base64," + base64String; - } - - public static ImageGenerationResult waitTask(String taskId) - throws ApiException, NoApiKeyException { - ImageGeneration imageGeneration = new ImageGeneration(); - return imageGeneration.wait(taskId, apiKey); - } - - public static void asyncCall() throws ApiException, NoApiKeyException, UploadFileException, IOException { - /* - * 图像输入方式说明（图生组图场景）： - * 以下提供了三种图片输入方式，三选一即可 - * 1. 使用公网URL - 适合已有公开可访问的图片 - * 2. 使用本地文件 - 适合本地开发测试 - * 3. 使用Base64编码 - 适合私有图片或需要加密传输的场景 - */ - // 【方式一】使用公网图片 URL - // String image1 = "https://img.alicdn.com/imgextra/i4/O1CN01IM44WN23dq5uY1yla_!!6000000007279-49-tps-1024-1024.webp"; - - // 【方式二】使用本地文件（支持绝对路径和相对路径） - // 格式要求：file:// + 文件路径 - // String image1 = "file:///path/to/your/image.png"; - - // 【方式三】使用Base64编码的图片 - // String image1 = encodeFile("/path/to/your/image.png"); - - // 构建文本输入消息（支持文生组图以及图生组图，此处以文生组图为例） - ImageGenerationMessage message = ImageGenerationMessage.builder() - .role("user") - .content(Collections.singletonList( - Collections.singletonMap("text", "电影感组图，记录同一只流浪橘猫，特征必须前后一致。第一张：春天，橘猫穿梭在盛开的樱花树下；第二张：夏天，橘猫在老街的树荫下乘凉避暑；第三张：秋天，橘猫踩在满地的金色落叶上；第四张：冬天，橘猫在雪地上走留下足迹。") - )).build(); - // 图生组图场景：取消以下注释并注释掉上方纯文本构建 - // ImageGenerationMessage message = ImageGenerationMessage.builder() - // .role("user") - // .content(Arrays.asList( - // Collections.singletonMap("text", "参考图片风格生成四季组图"), - // Collections.singletonMap("image", image1) - // )).build(); - - ImageGenerationParam param = ImageGenerationParam.builder() - .apiKey(apiKey) - .model("wan2.7-image-pro") - .messages(Collections.singletonList(message)) - .enableSequential(true) - .n(4) - .size("2K") // wan2.7-image-pro仅文生图场景支持4K分辨率，图像编辑和组图生成支持最高2K分辨率 - .build(); - - ImageGeneration imageGeneration = new ImageGeneration(); - ImageGenerationResult taskResult = null; - try { - System.out.println("----async call, creating task----"); - taskResult = imageGeneration.asyncCall(param); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - throw new RuntimeException(e.getMessage()); - } - System.out.println("Task created: " + JsonUtils.toJson(taskResult)); - - // 等待任务完成 - String taskId = taskResult.getOutput().getTaskId(); - ImageGenerationResult result = waitTask(taskId); - // 提取结果图片URL并保存到本地 - for (int i = 0; i < result.getOutput().getChoices().size(); i++) { - List> contents = result.getOutput().getChoices().get(i) - .getMessage().getContent(); - for (int j = 0; j < contents.size(); j++) { - if ("image".equals(contents.get(j).get("type"))) { - String imageUrl = (String) contents.get(j).get("image"); - String fileName = "output_" + i + "_" + j + ".png"; - // 结果URL有效期为24小时，请及时下载 - try (InputStream in = new URL(imageUrl).openStream()) { - Files.copy(in, Paths.get(fileName), StandardCopyOption.REPLACE_EXISTING); - } - System.out.println("Image saved to " + fileName); - } - } - } - } - - public static void main(String[] args) throws ApiException, NoApiKeyException, UploadFileException, IOException { - asyncCall(); - } -} -``` - -##### 响应示例 - -1、创建任务的响应示例 - -``` -{ - "requestId": "7d026dc1-e8c9-9caa-84ac-e82e2da97xxx", - "output": { - "task_id": "2de18c56-c151-4b80-8105-1d164733exxx", - "task_status": "PENDING" - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -2、查询任务结果的响应示例 - -> URL 有效期24小时，请及时下载并保存图像。 - -``` -{ - "requestId": "daea7295-4ce0-928a-9a11-4d2bea058xxx", - "usage": { - "input_tokens": 720, - "output_tokens": 11, - "total_tokens": 731, - "image_count": 4, - "size": "2048*2048" - }, - "output": { - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - }, - { - "image": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/xxxxxx.png?Expires=xxxxxx", - "type": "image" - } - ] - } - } - ], - "task_id": "2de18c56-c151-4b80-8105-1d164733exxx", - "task_status": "SUCCEEDED", - "finished": true, - "submit_time": "2026-03-31 19:49:53.124", - "scheduled_time": "2026-03-31 19:49:53.175", - "end_time": "2026-03-31 19:50:53.160" - }, - "status_code": 200, - "code": "", - "message": "" -} -``` - -## **计费与限流** - -- 模型免费额度和计费单价请参见[模型价格](https://help.aliyun.com/zh/model-studio/model-pricing#e2540d71a2utl)。 - -- 模型限流请参见[万相](https://help.aliyun.com/zh/model-studio/rate-limit#513e0a3df24v7)。 - -- 计费说明：按成功生成的 **图像张数** 计费。模型调用失败或处理错误不产生任何费用，也不消耗[新人免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota)。 - - -## **错误码** - -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md new file mode 100644 index 00000000..4042ba36 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md @@ -0,0 +1,968 @@ +# 万相-涂鸦作画API参考 + +本文介绍万相-涂鸦作画模型的API输入输出参数。 + +**相关指南**：[涂鸦作画](https://help.aliyun.com/zh/model-studio/sketch-to-image) + +**重要** + +本文档仅适用于华北2（北京）地域，且必须使用该地域的[API Key](https://bailian.console.aliyun.com/?tab=model#/api-key)。 + +**重要** + +百炼为华北2（北京）地域推出了业务空间专属域名 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`，**能够为推理请求提供卓越的性能和更高的稳定性**，建议从 `https://dashscope.aliyuncs.com` 迁移至新域名。 + +其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 + +## **模型概览** + +**模型效果示意** + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/2400704371/p883780.png) + +**模型简介** + +**模型名称** + +**模型简介** + +wanx-sketch-to-image-lite + +万相-涂鸦作画通过手绘图案和文字描述，生成精美的涂鸦绘画作品。 + +**模型说明** + +**模型名称** + +**计费单价** + +**限流（主账号与RAM子账号共用）** + +**免费额度**[（查看）](https://help.aliyun.com/zh/model-studio/new-free-quota) + +**任务下发接口QPS限制** + +**同时处理中任务数量** + +wanx-sketch-to-image-lite + +0.06元/张 + +2 + +1 + +500张 + +更多说明请参见[模型计费及限流](#b8457b7223zhp)。 + +## **前提条件** + +涂鸦作画API支持通过HTTP和DashScope SDK进行调用。 + +在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +如需通过SDK进行调用，请[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。目前，该SDK已支持Python和Java。 + +## HTTP调用 + +图像模型处理时间较长，为了避免请求超时，HTTP调用仅支持异步获取模型结果。您需要发起两个请求： + +1. **创建任务获取任务ID**：首先发起创建任务请求，该请求会返回任务ID（task\_id）。 + +2. **根据任务ID查询结果**：使用上一步获得的任务ID，查询任务状态及结果。任务成功执行时将返回图像URL，有效期24小时。 + + +**说明** + +创建任务后，该任务将被加入到排队队列，等待调度执行。后续需要调用“根据任务ID查询结果接口”获取任务状态及结果。 + +### **步骤1：创建任务获取任务ID** + +`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis/` + +#### 请求参数 + +## curl + +``` +curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis' \ +--header 'X-DashScope-Async: enable' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--header 'Content-Type: application/json' \ +--data '{ + "model": "wanx-sketch-to-image-lite", + "input": { + "sketch_image_url": "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg", + "prompt": "一棵参天大树" + }, + "parameters": { + "size": "768*768", + "n": 2, + "sketch_weight": 3, + "style": "" + } +}' +``` + +##### **请求头（Headers）** + +**Content-Type** `_string_` **（必选）** + +请求内容类型。此参数必须设置为`application/json`。 + +**Authorization** `_string_`**（必选）** + +请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 + +**X-DashScope-Async** `_string_` **（必选）** + +异步处理配置参数。HTTP请求只支持异步，**必须设置为**`**enable**`。 + +**重要** + +缺少此请求头将报错：“current user api does not support synchronous calls”。 + +**X-DashScope-WorkSpace** `_string_` （可选） + +阿里云百炼业务空间ID。示例值：llm-xxxx。 + +您可以在此[获取Workspace ID](https://help.aliyun.com/zh/model-studio/obtain-the-app-id-and-workspace-id#d3eb3cd37b7fu)。 + +**详细说明** + +此参数根据阿里云百炼API Key进行填写。 + +- 若为主账号API Key，可不填。不填则使用主账号权限，填写则使用对应的业务空间权限。 + +- 若为RAM子账号API Key，则必填。RAM子账号一定归属于某个业务空间。 + + +业务空间必须具备访问模型的权限，才能调用API。若无权限，请参考[授权子业务空间模型调用、训练和部署](https://help.aliyun.com/zh/model-studio/use-workspace#f2e68d7ba7ubk)。 + +> 关于如何区分阿里云百炼主账号和RAM子账号，请参考[主账号管理](https://help.aliyun.com/zh/model-studio/business-space-management)。 + +##### **请求体（Request Body）** + +**model** `_string_` **（必选）** + +调用模型。 + +**input** `_object_` **（必选）** + +输入的基本信息，比如提示词、图像URL地址。 + +**属性** + +**prompt** `_string_` **（必选）** + +提示词，用来描述生成图像中期望包含的元素和视觉特点。 + +支持中英文，长度不超过75个字符，超过部分会自动截断。 + +示例值：一棵参天大树。 + +**sketch\_image\_url** `_string_` **（必选）** + +输入草图的URL地址。输入草图需要与输出图像的分辨率比例保持一致，否则会导致图片拉伸变形，建议使用白色背景图。 + +URL 需为公网可访问的地址，并支持 HTTP 或 HTTPS 协议。您也可在此[获取临时公网URL](https://help.aliyun.com/zh/model-studio/get-temporary-file-url)。 + +图像限制： + +- 图像格式：JPG、JPEG、PNG、TIFF、WEBP。 + +- 图像分辨率：不小于256×256像素且不超过2048×2048像素。 + +- 图像大小：不超过10 MB。 + +- URL地址中不能包含中文字符。 + + +草图示例： + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/9289386271/p850798.png) + +**parameters** `_object_` （可选） + +图像处理参数。 + +**属性** + +**style** `_string_` （可选） + +输出图像的风格，目前支持以下风格取值： + +- ：默认值，由模型随机输出图像风格。 + +- <3d cartoon>：3D卡通。 + +- ：二次元。 + +- ：油画。 + +- ：水彩。 + +- ：素描。 + +- ：中国画。 + +- ：扁平插画。 + + +**size** `_string_` （可选） + +输出图像的分辨率。目前仅支持一种图像分辨率：768\*768，且为默认值。 + +**n** `_integer_` （可选） + +生成图片的数量。取值范围为1~4张，默认为4。 + +**sketch\_weight** `_integer_` （可选） + +输入草图对输出图像的约束程度。 + +取值范围为0-10，取值间隔为1， 默认值为10。取值越大表示输出图像跟输入草图越相似。 + +**sketch\_extraction** `_boolean_` （可选） + +如果上传图片是RGB图片，而非草图（sketch线稿），此参数可控制是否对输入图片进行sketch边缘提取。 + +默认值为False，表示不进行提取。设置为True时，表示进行提取，此时，`sketch_color`字段失效。 + +**sketch\_color** `_array_` （可选） + +此字段在`sketch_extraction=false`时生效，所包含数值均被视为画笔色，其余数值均会视为背景色。模型会基于一种或多种画笔色描绘的区域生成新的画作。默认值为\[\]。 + +当sketch\_image\_url线稿中的线条不是黑色，而是包含其他一种或多种颜色时，可以指定一个或多个RGB颜色数值作为画笔色。 + +示例值：\[\[134, 134, 134\], \[0, 0, 0\]\] + +#### **响应参数** + +#### 成功响应 + +请保存 task\_id，用于查询任务状态与结果。 + +``` +{ + "output": { + "task_status": "PENDING", + "task_id": "0385dc79-5ff8-4d82-bcb6-xxxxxx" + }, + "request_id": "4909100c-7b5a-9f92-bfe5-xxxxxx" +} +``` + +#### 异常响应 + +创建任务失败，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +``` +{ + "code": "InvalidApiKey", + "message": "No API-key provided.", + "request_id": "7438d53d-6eb8-4596-8835-xxxxxx" +} +``` + +**output** `_object_` + +任务输出信息。 + +**属性** + +**task\_id** `_string_` + +任务ID。查询有效期24小时。 + +**task\_status** `_string_` + +任务状态。 + +**枚举值** + +- PENDING：任务排队中 + +- RUNNING：任务处理中 + +- SUCCEEDED：任务执行成功 + +- FAILED：任务执行失败 + +- CANCELED：任务已取消 + +- UNKNOWN：任务不存在或状态未知 + + +**request\_id** `_string_` + +请求唯一标识。可用于请求明细溯源和问题排查。 + +**code** `_string_` + +请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +**message** `_string_` + +请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +### 步骤2：根据任务ID查询结果 + +`GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + +#### 请求参数 + +#### 查询任务结果 + +请将`86ecf553-d340-4e21-xxxxxxxxx`替换为真实的task\_id。 + +> 若使用新加坡地域的模型，需将base\_url替换为https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx，其中WorkspaceId需替换为真实的业务空间ID。 + +``` +curl -X GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/86ecf553-d340-4e21-xxxxxxxxx \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" +``` + +#### **请求头（Headers）** + +**Authorization** `_string_`**（必选）** + +请求身份认证。接口使用阿里云百炼API Key进行身份认证。示例值：Bearer sk-xxxx。 + +#### **URL路径参数（Path parameters）** + +**task\_id** `_string_`**（必选）** + +任务ID。 + +#### **响应参数** + +#### 任务执行成功 + +任务数据（如任务状态、图像URL等）仅保留24小时，超时后会被自动清除。请您务必及时保存生成的图像。 + +``` +{ + "request_id": "85eaba38-0185-99d7-8d16-4d9135238846", + "output": { + "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", + "task_status": "SUCCEEDED", + "results": [ + { + "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/a1.png" + }, + { + "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/b2.png" + } + ], + "task_metrics": { + "TOTAL": 2, + "SUCCEEDED": 2, + "FAILED": 0 + } + }, + "usage": { + "image_count": 2 + } +} +``` + +#### 任务执行失败 + +若任务执行失败，task\_status将置为 FAILED，并提供错误码和信息。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +``` +{ + "request_id": "e5d70b02-ebd3-98ce-9fe8-759d7d7b107d", + "output": { + "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", + "task_status": "FAILED", + "code": "InvalidParameter", + "message": "The size is not match the allowed size ['1024*1024', '720*1280', '1280*720']", + "task_metrics": { + "TOTAL": 4, + "SUCCEEDED": 0, + "FAILED": 4 + } + } +} +``` + +#### 任务部分失败 + +模型可以在一次任务中生成多张图片。只要有一张图片生成成功，任务状态将标记为`SUCCEEDED`，并且返回相应的图像URL。对于生成失败的图片，结果中会返回相应的失败原因。同时在usage统计中，只会对成功的结果计数。请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +``` +{ + "request_id": "85eaba38-0185-99d7-8d16-4d9135238846", + "output": { + "task_id": "86ecf553-d340-4e21-af6e-a0c6a421c010", + "task_status": "SUCCEEDED", + "results": [ + { + "url": "https://dashscope-result-bj.oss-cn-beijing.aliyuncs.com/123/a1.png" + }, + { + "code": "InternalError.Timeout", + "message": "An internal timeout error has occured during execution, please try again later or contact service support." + } + ], + "task_metrics": { + "TOTAL": 2, + "SUCCEEDED": 1, + "FAILED": 1 + } + }, + "usage": { + "image_count": 1 + } +} +``` + +**output** `_object_` + +任务输出信息。 + +**属性** + +**task\_id** `_string_` + +任务ID。查询有效期24小时。 + +**task\_status** `_string_` + +任务状态。 + +**枚举值** + +- PENDING：任务排队中 + +- RUNNING：任务处理中 + +- SUCCEEDED：任务执行成功 + +- FAILED：任务执行失败 + +- CANCELED：任务已取消 + +- UNKNOWN：任务不存在或状态未知 + + +**task\_metrics** `_object_` + +任务结果统计。 + +**属性** + +**TOTAL** `_integer_` + +总的任务数。 + +**SUCCEEDED** `_integer_` + +任务状态为成功的任务数。 + +**FAILED** `_integer_` + +任务状态为失败的任务数。 + +**results** `_array of object_` + +任务结果列表，包括图像URL、部分任务执行失败报错信息等。 + +**数据结构** + +``` +{ + "results": [ + { + "url": "" + }, + { + "code": "", + "message": "" + } + ] +} +``` + +**code** `_string_` + +请求失败的错误码。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +**message** `_string_` + +请求失败的详细信息。请求成功时不会返回此参数，详情请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + +**usage** `_object_` + +输出信息统计。只对成功的结果计数。 + +**属性** + +**image\_count** `_integer_` + +模型成功生成图片的数量。计费公式：费用 = 图片数量 × 单价。 + +**request\_id** `_string_` + +请求唯一标识。可用于请求明细溯源和问题排查。 + +## DashScope SDK调用 + +请先确认已安装最新版DashScope SDK：[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 + +DashScope SDK目前已支持Python和Java。 + +SDK与HTTP接口的参数名基本一致，参数结构根据不同语言的SDK封装而定。参数说明可参考[HTTP调用](https://help.aliyun.com/zh/model-studio/text-to-image-api-reference#42703589880ts)。 + +由于图像模型处理时间较长，底层服务采用异步方式提供。SDK在上层进行了封装，支持同步、异步两种调用方式。 + +### Python SDK调用 + +## 同步调用 + +##### **请求示例** + +``` +from http import HTTPStatus +from urllib.parse import urlparse, unquote +from pathlib import PurePosixPath +import requests +import dashscope +from dashscope import ImageSynthesis +import os + +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' + +prompt = "一棵参天大树" +sketch_image_url = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg" +model = "wanx-sketch-to-image-lite" +task = "image2image" + +print('----sync call, please wait a moment----') +rsp = ImageSynthesis.call(api_key=os.getenv("DASHSCOPE_API_KEY"), + model=model, + prompt=prompt, + n=1, + style='', + size='768*768', + sketch_image_url=sketch_image_url, + task=task) +print('response: %s' % rsp) +if rsp.status_code == HTTPStatus.OK: + print(rsp.output) + # save file to current directory + for result in rsp.output.results: + file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1] + with open('./%s' % file_name, 'wb+') as f: + f.write(requests.get(result.url).content) +else: + print('sync_call Failed, status_code: %s, code: %s, message: %s' % + (rsp.status_code, rsp.code, rsp.message)) +``` + +##### **响应示例** + +``` +{ + "status_code": 200, + "request_id": "4126d9dd-e037-9f32-8d56-6d29ab3f9a06", + "code": null, + "message": "", + "output": { + "task_id": "b476bc4e-35c1-4c4e-a4d9-xxxxxxx", + "task_status": "SUCCEEDED", + "results": [{ + "url": "https://dashscope-result-sh.oss-cn-shanghai.aliyuncs.com/xxxx.png" + }], + "submit_time": "2024-11-01 09:50:56.081", + "scheduled_time": "2024-11-01 09:50:56.104", + "end_time": "2024-11-01 09:51:22.740", + "task_metrics": { + "TOTAL": 1, + "SUCCEEDED": 1, + "FAILED": 0 + } + }, + "usage": { + "image_count": 1 + } +} +``` + +## 异步调用 + +##### **请求示例** + +``` +from http import HTTPStatus +from urllib.parse import urlparse, unquote +from pathlib import PurePosixPath +import requests +import dashscope +from dashscope import ImageSynthesis +import os + +dashscope.base_http_api_url = 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1' + +prompt = "一棵参天大树" +sketch_image_url = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg" +model = "wanx-sketch-to-image-lite" +task = "image2image" + +# 异步调用 +def async_call(): + print('----create task----') + task_info = create_async_task() + print('----wait task done then save image----') + wait_async_task(task_info) + +# 创建异步任务 +def create_async_task(): + rsp = ImageSynthesis.async_call(api_key=os.getenv("DASHSCOPE_API_KEY"), + model=model, + prompt=prompt, + n=1, + style='', + size='768*768', + sketch_image_url=sketch_image_url, + task=task) + print(rsp) + if rsp.status_code == HTTPStatus.OK: + print(rsp.output) + else: + print('create_async_task Failed, status_code: %s, code: %s, message: %s' % + (rsp.status_code, rsp.code, rsp.message)) + return rsp + +# 等待异步任务结束 +def wait_async_task(task): + rsp = ImageSynthesis.wait(task) + print(rsp) + if rsp.status_code == HTTPStatus.OK: + print(rsp.output.task_status) + # save file to current directory + for result in rsp.output.results: + file_name = PurePosixPath(unquote(urlparse(result.url).path)).parts[-1] + with open('./%s' % file_name, 'wb+') as f: + f.write(requests.get(result.url).content) + else: + print('Failed, status_code: %s, code: %s, message: %s' % + (rsp.status_code, rsp.code, rsp.message)) + +if __name__ == '__main__': + async_call() +``` + +##### **响应示例** + +**1、创建任务的响应示例** + +``` +{ + "status_code": 200, + "request_id": "31b04171-011c-96bd-ac00-f0383b669cc7", + "code": "", + "message": "", + "output": { + "task_id": "4f90cf14-a34e-4eae-xxxxxxxx", + "task_status": "PENDING", + "results": [] + }, + "usage": null +} +``` + +**2、查询任务结果的响应示例** + +``` +{ + "status_code": 200, + "request_id": "d861d3ba-4b29-9491-abad-266ef4fb2f08", + "code": null, + "message": "", + "output": { + "task_id": "4f90cf14-a34e-4eae-xxxxxxxx", + "task_status": "SUCCEEDED", + "results": [{ + "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" + }], + "submit_time": "2024-10-31 20:40:35.631", + "scheduled_time": "2024-10-31 20:40:35.684", + "end_time": "2024-10-31 20:41:02.700", + "task_metrics": { + "TOTAL": 1, + "SUCCEEDED": 1, + "FAILED": 0 + } + }, + "usage": { + "image_count": 1 + } +} +``` + +### Java SDK调用 + +## 同步调用 + +##### 请求示例 + +``` +// Copyright (c) Alibaba, Inc. and its affiliates. + +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis; +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam; +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult; +import com.alibaba.dashscope.exception.ApiException; +import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; + +public class Main { + static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} + + public void syncCall() { + String prompt = "一棵参天大树"; + String sketchImageUrl = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg"; + String model = "wanx-sketch-to-image-lite"; + ImageSynthesisParam param = ImageSynthesisParam.builder() + .model(model) + .prompt(prompt) + .n(1) + .size("768*768") + .sketchImageUrl(sketchImageUrl) + .style("") + .build(); + + String task = "image2image"; + ImageSynthesis imageSynthesis = new ImageSynthesis(task); + ImageSynthesisResult result = null; + try { + System.out.println("---sync call, please wait a moment----"); + result = imageSynthesis.call(param); + } catch (ApiException | NoApiKeyException e){ + throw new RuntimeException(e.getMessage()); + } + System.out.println(JsonUtils.toJson(result)); + } + + public static void main(String[] args){ + Main text2Image = new Main(); + text2Image.syncCall(); + } + +} +``` + +##### **响应示例** + +``` +{ + "request_id": "150edcda-05d5-9ffe-8803-84626d1db623", + "output": { + "task_id": "f2098ff0-146e-404c-bb25-xxxxxxxx", + "task_status": "SUCCEEDED", + "results": [{ + "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" + }], + "task_metrics": { + "TOTAL": 1, + "SUCCEEDED": 1, + "FAILED": 0 + } + }, + "usage": { + "image_count": 1 + } +} +``` + +## 异步调用 + +##### **请求示例** + +``` +// Copyright (c) Alibaba, Inc. and its affiliates. + +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesis; +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisParam; +import com.alibaba.dashscope.aigc.imagesynthesis.ImageSynthesisResult; +import com.alibaba.dashscope.exception.ApiException; +import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.utils.JsonUtils; +import com.alibaba.dashscope.utils.Constants; + +public class Main { + static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} + + public void asyncCall() { + System.out.println("---create task----"); + String taskId = this.createAsyncTask(); + System.out.println("---wait task done then return image url----"); + this.waitAsyncTask(taskId); + } + + /** + * 创建异步任务 + * @return taskId + */ + public String createAsyncTask() { + String prompt = "一棵参天大树"; + String sketchImageUrl = "https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/6609471071/p743851.jpg"; + String model = "wanx-sketch-to-image-lite"; + ImageSynthesisParam param = ImageSynthesisParam.builder() + .model(model) + .prompt(prompt) + .n(1) + .size("768*768") + .sketchImageUrl(sketchImageUrl) + .style("") + .build(); + + String task = "image2image"; + ImageSynthesis imageSynthesis = new ImageSynthesis(task); + ImageSynthesisResult result = null; + try { + result = imageSynthesis.asyncCall(param); + } catch (Exception e){ + throw new RuntimeException(e.getMessage()); + } + String taskId = result.getOutput().getTaskId(); + System.out.println("taskId=" + taskId); + return taskId; + } + + /** + * 等待异步任务结束 + * @param taskId 任务id + * */ + public void waitAsyncTask(String taskId) { + ImageSynthesis imageSynthesis = new ImageSynthesis(); + ImageSynthesisResult result = null; + try { + // If you have set the DASHSCOPE_API_KEY in the system environment variable, the apiKey can be null. + result = imageSynthesis.wait(taskId, null); + } catch (ApiException | NoApiKeyException e){ + throw new RuntimeException(e.getMessage()); + } + + System.out.println(JsonUtils.toJson(result.getOutput())); + System.out.println(JsonUtils.toJson(result.getUsage())); + } + + public static void main(String[] args){ + Main text2Image = new Main(); + text2Image.asyncCall(); + } + +} +``` + +##### **响应示例** + +**1、步骤1：创建任务获取任务ID的响应示例** + +``` +{ + "request_id": "5dbf9dc5-4f4c-9605-85ea-542f97709ba8", + "output": { + "task_id": "7277e20e-aa01-4709-xxxxxxxx", + "task_status": "PENDING" + } +} +``` + +**2、步骤2：根据任务ID查询结果的响应示例** + +``` +{ + "request_id": "c44213ba-7aa3-91e4-97c1-c527ade82597", + "output": { + "task_id": "7277e20e-aa01-4709-xxxxxxxx", + "task_status": "SUCCEEDED", + "results": [{ + "url": "https://dashscope-result-hz.oss-cn-hangzhou.aliyuncs.com/xxxx.png" + }], + "task_metrics": { + "TOTAL": 1, + "SUCCEEDED": 1, + "FAILED": 0 + } + }, + "usage": { + "image_count": 1 + } +} +``` + +## 错误码 + +如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +此API还有特定状态码，具体如下所示。 + +**HTTP状态码** + +**接口错误码（code）** + +**接口错误信息（message）** + +**含义说明** + +400 + +InvalidParameter.DataInspection + +Unable to download the media resource during the data inspection process. + +输入图片无法下载，请检查URL地址是否正确且可访问。 + +400 + +InvalidParameter + +Value error, format of image {url} is not valid : payload.input.sketch + +输入图片格式不合法，请确认图片格式为JPG、JPEG、PNG、TIFF或WEBP。 + +## **常见问题** + +### 模型计费及限流 + +**免费额度** + +- 额度说明：免费额度是指模型成功生成的输出图片数量。输入图片及模型处理失败的情况不占用免费额度。 + +- 领取方式：开通阿里云百炼大模型服务后自动发放，有效期90天。 + +- 使用账号：阿里云主账号与其RAM子账号共享免费额度。 + +- 更多详情请参见[新人免费额度](https://help.aliyun.com/zh/model-studio/new-free-quota)。 + + +**限时免费** + +- 当计费为限时免费时，表示该模型处于公测阶段，免费额度用尽后不可使用。 + + +**计费说明** + +- 当计费有明确单价时，如0.2元/秒，表示该模型已商业化，免费额度用尽或过期后需付费使用。 + +- 计费项：只对模型成功生成的输出图片进行收费，其余情况暂不计费。 + +- 付费方式：由阿里云主账号统一付费。RAM子账号不能独立计量计费，必须由所属的主账号付费。如果您需要查询账单信息，请前往阿里云控制台[账单概览](https://billing-cost.console.aliyun.com/finance/month-bill/account)。 + +- 充值途径：您可以在阿里云控制台[费用与成本](https://billing-cost.console.aliyun.com/home?spm=a2c4g.11186623.0.0.2d543048F4KRQP)页面进行充值。 + +- 模型调用情况：您可以前往阿里云百炼的[模型观测](https://bailian.console.aliyun.com/#/model-telemetry)查看模型调用量及调用次数。 + +- 更多计费问题请参见[计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio)。 + + +**限流** + +- 限流说明：阿里云主账号与其RAM子账号共享限流限制。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md new file mode 100644 index 00000000..4c3b266a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-about-models/get-temporary-file-url.md @@ -0,0 +1,1185 @@ +# 上传本地文件获取临时URL + +在调用多模态、图像、视频或音频模型时，通常需要传入文件的 URL。为此，阿里云百炼提供了**免费**临时存储空间，您可将本地文件上传至该空间并获得 URL（**有效期为 48 小时**）。 + +## **使用限制** + +- **文件与模型绑定**：文件上传时必须指定模型名称，且该模型须与后续调用的**模型一致**，不同模型无法共享文件。 + +- **文件大小限制**：接口上传文件大小不得超过**1GB**，超出限制将导致上传失败。此外，不同模型对输入文件大小有不同限制，超出限制将导致模型调用失败。 + +- **文件与主账号绑定**：文件上传与模型调用所使用的 API Key 必须**属于同一个阿里云主账号**，且上传的文件仅限该主账号及其对应模型使用，无法被其他主账号或其他模型共享。 + +- **文件有效期限制**：文件上传后**有效期48小时**，超时后文件将被自动清理，请确保在有效期内完成模型调用。 + +- **文件使用限制**：文件一旦上传，不可查询、修改或下载，仅能**通过URL参数在模型调用时使用**。 + +- **文件上传限流**：文件上传凭证接口的调用限流按照“阿里云主账号+模型”维度为**100QPS**，**超出限流将导致请求失败**。 + + +**重要** + +- 临时 URL 有效期48小时，过期后无法使用，**请勿用于生产环境。** + +- 文件上传凭证接口限流为 100 QPS 且不支持扩容，**请勿用于生产环境、高并发及压测场景。** + +- 生产环境建议使用[阿里云OSS](https://help.aliyun.com/zh/oss/user-guide/what-is-oss) 等稳定存储，确保文件长期可用并规避限流问题。 + + +## **使用方式** + +1. 获取文件 URL：请先通过[步骤一](#a363e01e741gu)上传文件（图片、视频或音频），获取以`oss://` 为前缀的临时 URL。 + +2. 调用模型：**请务必根据**[**步骤二**](#1c60469225ufa)**使用临时 URL 进行调用**。该步骤不能跳过，否则接口将报错。 + + +## **步骤一：获取临时URL** + +### **方式一：通过代码上传文件** + +本文提供 Python 和 Java 示例代码，简化上传文件操作。您只需**指定模型和待上传的文件**，即可获取临时URL。 + +**前提条件** + +在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)，再[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +#### **示例代码** + +## Python + +**环境配置** + +- 推荐使用Python 3.8及以上版本。 + +- 请安装必要的依赖包。 + + +``` +pip install -U requests +``` + +**输入参数** + +- api\_key：阿里云百炼API KEY。 + +- model\_name：指定文件将要用于哪个模型，如`qwen-vl-plus`。 + +- file\_path：待上传的本地文件路径（图片、视频等）。 + + +``` +import os +import requests +from pathlib import Path +from datetime import datetime, timedelta + +def get_upload_policy(api_key, model_name): + """获取文件上传凭证""" + url = "https://dashscope.aliyuncs.com/api/v1/uploads" + headers = { + "Authorization": f"Bearer {api_key}", + "Content-Type": "application/json" + } + params = { + "action": "getPolicy", + "model": model_name + } + + response = requests.get(url, headers=headers, params=params) + if response.status_code != 200: + raise Exception(f"Failed to get upload policy: {response.text}") + + return response.json()['data'] + +def upload_file_to_oss(policy_data, file_path): + """将文件上传到临时存储OSS""" + file_name = Path(file_path).name + key = f"{policy_data['upload_dir']}/{file_name}" + + with open(file_path, 'rb') as file: + files = { + 'OSSAccessKeyId': (None, policy_data['oss_access_key_id']), + 'Signature': (None, policy_data['signature']), + 'policy': (None, policy_data['policy']), + 'x-oss-object-acl': (None, policy_data['x_oss_object_acl']), + 'x-oss-forbid-overwrite': (None, policy_data['x_oss_forbid_overwrite']), + 'key': (None, key), + 'success_action_status': (None, '200'), + 'file': (file_name, file) + } + + response = requests.post(policy_data['upload_host'], files=files) + if response.status_code != 200: + raise Exception(f"Failed to upload file: {response.text}") + + return f"oss://{key}" + +def upload_file_and_get_url(api_key, model_name, file_path): + """上传文件并获取URL""" + # 1. 获取上传凭证，上传凭证接口有限流，超出限流将导致请求失败 + policy_data = get_upload_policy(api_key, model_name) + # 2. 上传文件到OSS + oss_url = upload_file_to_oss(policy_data, file_path) + + return oss_url + +# 使用示例 +if __name__ == "__main__": + # 从环境变量中获取API Key 或者 在代码中设置 api_key = "your_api_key" + api_key = os.getenv("DASHSCOPE_API_KEY") + if not api_key: + raise Exception("请设置DASHSCOPE_API_KEY环境变量") + + # 设置model名称 + model_name="qwen-vl-plus" + + # 待上传的文件路径 + file_path = "/tmp/cat.png" # 替换为实际文件路径 + + try: + public_url = upload_file_and_get_url(api_key, model_name, file_path) + expire_time = datetime.now() + timedelta(hours=48) + print(f"文件上传成功，有效期为48小时，过期时间: {expire_time.strftime('%Y-%m-%d %H:%M:%S')}") + print(f"临时URL: {public_url}") + print("注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call") + + except Exception as e: + print(f"Error: {str(e)}") +``` + +**输出示例** + +``` +文件上传成功，有效期为48小时，过期时间: 2024-07-18 17:36:15 +临时URL: oss://dashscope-instant/xxx/2024-07-18/xxx/cat.png +注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call +``` + +**重要** + +获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 + +## Java + +**环境配置** + +- 推荐使用JDK 1.8及以上版本。 + +- 请在Maven项目的`pom.xml`文件中导入以下依赖。 + + +``` + + + org.json + json + 20230618 + + + org.apache.httpcomponents + httpclient + 4.5.13 + + + org.apache.httpcomponents + httpmime + 4.5.13 + + +``` + +**输入参数** + +- apiKey：阿里云百炼API KEY。 + +- modelName：指定文件将要用于哪个模型，如`qwen-vl-plus`。 + +- filePath：待上传的本地文件路径（图片、视频等）。 + + +``` +import org.apache.http.client.methods.CloseableHttpResponse; +import org.apache.http.client.methods.HttpGet; +import org.apache.http.client.methods.HttpPost; +import org.apache.http.entity.mime.MultipartEntityBuilder; +import org.apache.http.entity.ContentType; +import org.apache.http.impl.client.CloseableHttpClient; +import org.apache.http.impl.client.HttpClients; +import org.apache.http.HttpStatus; +import org.apache.http.util.EntityUtils; +import org.json.JSONObject; +import java.io.File; +import java.io.IOException; +import java.nio.file.Files; +import java.nio.file.Path; +import java.nio.file.Paths; +import java.time.LocalDateTime; +import java.time.format.DateTimeFormatter; + +public class PublicUrlHandler { + + private static final String API_URL = "https://dashscope.aliyuncs.com/api/v1/uploads"; + + public static JSONObject getUploadPolicy(String apiKey, String modelName) throws IOException { + try (CloseableHttpClient httpClient = HttpClients.createDefault()) { + HttpGet httpGet = new HttpGet(API_URL); + httpGet.addHeader("Authorization", "Bearer " + apiKey); + httpGet.addHeader("Content-Type", "application/json"); + + String query = String.format("action=getPolicy&model=%s", modelName); + httpGet.setURI(httpGet.getURI().resolve(httpGet.getURI() + "?" + query)); + + try (CloseableHttpResponse response = httpClient.execute(httpGet)) { + if (response.getStatusLine().getStatusCode() != 200) { + throw new IOException("Failed to get upload policy: " + + EntityUtils.toString(response.getEntity())); + } + String responseBody = EntityUtils.toString(response.getEntity()); + return new JSONObject(responseBody).getJSONObject("data"); + } + } + } + + public static String uploadFileToOSS(JSONObject policyData, String filePath) throws IOException { + Path path = Paths.get(filePath); + String fileName = path.getFileName().toString(); + String key = policyData.getString("upload_dir") + "/" + fileName; + + HttpPost httpPost = new HttpPost(policyData.getString("upload_host")); + MultipartEntityBuilder builder = MultipartEntityBuilder.create(); + + builder.addTextBody("OSSAccessKeyId", policyData.getString("oss_access_key_id")); + builder.addTextBody("Signature", policyData.getString("signature")); + builder.addTextBody("policy", policyData.getString("policy")); + builder.addTextBody("x-oss-object-acl", policyData.getString("x_oss_object_acl")); + builder.addTextBody("x-oss-forbid-overwrite", policyData.getString("x_oss_forbid_overwrite")); + builder.addTextBody("key", key); + builder.addTextBody("success_action_status", "200"); + byte[] fileContent = Files.readAllBytes(path); + builder.addBinaryBody("file", fileContent, ContentType.DEFAULT_BINARY, fileName); + + httpPost.setEntity(builder.build()); + + try (CloseableHttpClient httpClient = HttpClients.createDefault(); + CloseableHttpResponse response = httpClient.execute(httpPost)) { + if (response.getStatusLine().getStatusCode() != HttpStatus.SC_OK) { + throw new IOException("Failed to upload file: " + + EntityUtils.toString(response.getEntity())); + } + return "oss://" + key; + } + } + + public static String uploadFileAndGetUrl(String apiKey, String modelName, String filePath) throws IOException { + JSONObject policyData = getUploadPolicy(apiKey, modelName); + return uploadFileToOSS(policyData, filePath); + } + + public static void main(String[] args) { + // 获取环境变量中的API密钥 + String apiKey = System.getenv("DASHSCOPE_API_KEY"); + if (apiKey == null || apiKey.isEmpty()) { + System.err.println("请设置DASHSCOPE_API_KEY环境变量"); + System.exit(1); + } + // 模型名称 + String modelName = "qwen-vl-plus"; + //替换为实际文件路径 + String filePath = "src/main/resources/tmp/cat.png"; + + try { + // 检查文件是否存在 + File file = new File(filePath); + if (!file.exists()) { + System.err.println("文件不存在: " + filePath); + System.exit(1); + } + + String publicUrl = uploadFileAndGetUrl(apiKey, modelName, filePath); + LocalDateTime expireTime = LocalDateTime.now().plusHours(48); + DateTimeFormatter formatter = DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss"); + + System.out.println("文件上传成功，有效期为48小时，过期时间: " + expireTime.format(formatter)); + System.out.println("临时URL: " + publicUrl); + System.out.println("注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call"); + } catch (IOException e) { + System.err.println("Error: " + e.getMessage()); + } + } +} +``` + +**输出示例** + +``` +文件上传成功，有效期为48小时，过期时间: 2024-07-18 17:36:15 +临时URL: oss://dashscope-instant/xxx/2024-07-18/xxx/cat.png +注意：使用oss://形式的临时URL时，必须在HTTP请求头（Header）中显式添加参数：X-DashScope-OssResourceResolve: enable，具体请参考：https://help.aliyun.com/zh/model-studio/get-temporary-file-url#http-call +``` + +**重要** + +获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 + +### **方式二：通过命令行工具上传文件** + +对于熟悉命令行的开发者，可使用DashScope提供的命令行工具来上传文件。**执行命令后，即可获取临时URL**。 + +#### **前提条件** + +1. 环境准备：推荐使用 Python 3.8 及以上版本。 + +2. 获取API-KEY：在调用前，您需要[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 + +3. 安装SDK：请确保[DashScope Python SDK](https://help.aliyun.com/zh/model-studio/install-sdk) 版本不低于 `1.24.0`。执行以下命令进行安装或升级： + + +``` +pip install -U dashscope +``` + +#### **方法1：使用环境变量（推荐）** + +此方法更安全，可以避免API-KEY在命令历史或脚本中明文暴露。 + +前提条件：请确保已[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +执行上传命令： + +``` +dashscope oss.upload --model qwen-vl-plus --file cat.png +``` + +输出示例： + +``` +Start oss.upload: model=qwen-vl-plus, file=cat.png, api_key=None +Uploaded oss url: oss://dashscope-instant/xxxx/2025-08-01/xxxx/cat.png +``` + +**重要** + +获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 + +#### **方法2：通过命令行参数指定API-KEY（临时使用）** + +执行上传命令： + +``` +dashscope oss.upload --model qwen-vl-plus --file cat.png --api_key sk-xxxxxxx +``` + +输出示例： + +``` +Start oss.upload: model=qwen-vl-plus, file=cat.png, api_key=sk-xxxxxxx +Uploaded oss url: oss://dashscope-instant/xxx/2025-08-01/xxx/cat.png +``` + +**重要** + +获取临时 URL 后，调用时**必须**在 HTTP 请求头（Header）中显式添加参数：`**X-DashScope-OssResourceResolve: enable**`，具体请参见[通过HTTP调用](#d6a1cb0f01h5k)。 + +#### **命令行参数说明** + +**参数** + +**是否必须** + +**说明** + +**示例** + +oss.upload + +是 + +dashscope的子命令，用于执行文件上传操作。 + +oss.upload + +\--model + +是 + +指定文件将要用于哪个模型。 + +qwen-vl-plus + +\--file + +是 + +本地文件的路径。可以是相对路径或绝对路径。 + +cat.png，/data/img.jpg + +\--api\_key + +否 + +阿里云百炼API-KEY。如已配置环境变量，无需填写此参数。 + +sk-xxxx + +## **步骤二：使用临时URL调用模型** + +#### **使用限制** + +- **文件格式**：临时URL须通过上述方式生成，且以 `oss://`为前缀的URL字符串。 + +- **文件未过期**：文件URL仍在上传后的48小时有效期内。 + +- **模型一致**：模型调用所使用的模型必须与文件上传时指定的模型完全一致。 + +- **账号一致**：模型调用的API KEY必须与文件上传时使用的API KEY同属一个阿里云主账号。 + + +#### **方式一：通过HTTP调用** + +通过curl、Postman或任何其他HTTP客户端直接调用API，则**必须遵循以下规则**： + +**重要** + +- 使用临时URL，**必须**在请求的**Header**中添加参数：`**X-DashScope-OssResourceResolve: enable**`。 + +- 若缺失此Header，系统将无法解析`oss://`链接，请求将失败，报错信息请参考[错误码](#3b9b15a6a8qkl)。 + + +## **请求示例** + +本示例为调用 qwen-vl-plus 模型识别图片内容。 + +**说明** + +请将 `oss://...`替换为真实的临时 URL，否则请求将失败。 + +``` +curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/chat/completions \ +-H "Authorization: Bearer $DASHSCOPE_API_KEY" \ +-H 'Content-Type: application/json' \ +-H 'X-DashScope-OssResourceResolve: enable' \ +-d '{ + "model": "qwen-vl-plus", + "messages": [{ + "role": "user", + "content": + [{"type": "text","text": "这是什么"}, + {"type": "image_url","image_url": {"url": "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"}}] + }] +}' +``` + +## 响应示例 + +``` +{ + "choices": [ + { + "message": { + "content": "这是一张描绘一只白色猫咪在草地上奔跑的图片。这只猫有蓝色的眼睛，看起来非常可爱和活泼。背景是模糊化的自然景色，强调了主体——那只向前冲跑的小猫。这种摄影技巧称为浅景深（或大光圈效果），它使得前景中的小猫变得清晰而锐利，同时使背景虚化以突出主题并营造出一种梦幻般的效果。整体上这张照片给人一种轻松愉快的感觉，并且很好地捕捉到了动物的行为瞬间。", + "role": "assistant" + }, + "finish_reason": "stop", + "index": 0, + "logprobs": null + } + ], + "object": "chat.completion", + "usage": { + "prompt_tokens": 1253, + "completion_tokens": 104, + "total_tokens": 1357 + }, + "created": 1739349052, + "system_fingerprint": null, + "model": "qwen-vl-plus", + "id": "chatcmpl-cfc4f2aa-22a8-9a94-8243-44c5bd9899bc" +} +``` + +## 上传的本地图片示例 + +![image](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/5231249371/p915804.png) + +#### **方式二：通过DashScope SDK调用** + +您也可以使用阿里云百炼提供的 Python 或 Java SDK。 + +- **直接传入 URL**：调用模型 SDK 时，直接将以`oss://`为前缀的URL字符串作为文件参数传入。 + +- **无需关心 Header**：SDK 会自动添加必需的请求头，无需额外操作。 + + +**注意**：并非所有模型都支持 SDK 调用，请以模型 API 文档为准。 + +> 不支持 OpenAI SDK。 + +## Python + +**前提条件** + +请[安装DashScope Python SDK](https://help.aliyun.com/zh/model-studio/install-sdk)，且DashScope Python SDK版本号 >=`1.24.0`。 + +**示例代码** + +本示例为调用 qwen-vl-plus 模型识别图片内容。此代码示例仅适用于 qwen-vl 和 omni 系列模型。 + +## 请求示例 + +**说明** + +请将 image 参数中的 `oss://...`替换为真实的临时 URL，否则请求将失败。 + +``` +import os +import dashscope + +messages = [ + { + "role": "system", + "content": [{"text": "You are a helpful assistant."}] + }, + { + "role": "user", + "content": [ + {"image": "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"}, + {"text": "这是什么"}] + }] + +# 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" +api_key = os.getenv('DASHSCOPE_API_KEY') + +response = dashscope.MultiModalConversation.call( + api_key=api_key, + model='qwen-vl-plus', + messages=messages +) + +print(response) +``` + +## 响应示例 + +``` +{ + "status_code": 200, + "request_id": "ccd9dcfb-98f0-92bc-xxxxxx", + "code": "", + "message": "", + "output": { + "text": null, + "finish_reason": null, + "choices": [ + { + "finish_reason": "stop", + "message": { + "role": "assistant", + "content": [ + { + "text": "这是一张一只猫在草地上奔跑的照片。猫的毛色主要是白色，带有浅棕色的斑点，眼睛是蓝色的，显得非常可爱。背景是一个模糊的绿色草地和一些树木，阳光照射下来，给整个画面增添了一种温暖的感觉。猫的姿态显示出它正在快速移动，可能是在追逐什么或只是在享受户外活动的乐趣。整体来看，这是一幅充满活力和生机的图片。" + } + ] + } + } + ] + }, + "usage": { + "input_tokens": 1112, + "output_tokens": 91, + "input_tokens_details": { + "text_tokens": 21, + "image_tokens": 1091 + }, + "prompt_tokens_details": { + "cached_tokens": 0 + }, + "total_tokens": 1203, + "output_tokens_details": { + "text_tokens": 91 + }, + "image_tokens": 1091 + } +} +``` + +## Java + +**前提条件** + +请[安装DashScope Java SDK](https://help.aliyun.com/zh/model-studio/install-sdk)，且DashScope Java SDK版本号 >= `2.21.0`。 + +**示例代码** + +本示例为调用 qwen-vl-plus 模型识别图片内容。此代码示例仅适用于 qwen-vl 和 omni 系列模型。 + +## 请求示例 + +**说明** + +请将 `oss://...`替换为真实的临时 URL，否则请求将失败。 + +``` +import com.alibaba.dashscope.aigc.multimodalconversation.*; +import com.alibaba.dashscope.common.Role; +import com.alibaba.dashscope.exception.ApiException; +import com.alibaba.dashscope.exception.NoApiKeyException; +import com.alibaba.dashscope.exception.UploadFileException; +import com.alibaba.dashscope.utils.JsonUtils; + +import java.util.Arrays; + +public class MultiModalConversationUsage { + + private static final String modelName = "qwen-vl-plus"; + + // 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" + public static String apiKey = System.getenv("DASHSCOPE_API_KEY"); + + public static void simpleMultiModalConversationCall() throws ApiException, NoApiKeyException, UploadFileException { + MultiModalConversation conv = new MultiModalConversation(); + MultiModalMessageItemText systemText = new MultiModalMessageItemText("You are a helpful assistant."); + MultiModalConversationMessage systemMessage = MultiModalConversationMessage.builder() + .role(Role.SYSTEM.getValue()).content(Arrays.asList(systemText)).build(); + MultiModalMessageItemImage userImage = new MultiModalMessageItemImage( + "oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png"); + MultiModalMessageItemText userText = new MultiModalMessageItemText("这是什么"); + MultiModalConversationMessage userMessage = + MultiModalConversationMessage.builder().role(Role.USER.getValue()) + .content(Arrays.asList(userImage, userText)).build(); + MultiModalConversationParam param = MultiModalConversationParam.builder() + .model(MultiModalConversationUsage.modelName) + .apiKey(apiKey) + .message(systemMessage) + .vlHighResolutionImages(true) + .vlEnableImageHwOutput(true) +// .incrementalOutput(true) + .message(userMessage).build(); + MultiModalConversationResult result = conv.call(param); + System.out.print(JsonUtils.toJson(result)); + + } + + public static void main(String[] args) { + try { + simpleMultiModalConversationCall(); + } catch (ApiException | NoApiKeyException | UploadFileException /*| IOException*/ e) { + System.out.println(e.getMessage()); + } + System.exit(0); + } + +} +``` + +## 响应示例 + +``` +{ + "requestId": "b6d60f91-4a7f-9257-xxxxxx", + "usage": { + "input_tokens": 1112, + "output_tokens": 91, + "total_tokens": 1203, + "image_tokens": 1091, + "input_tokens_details": { + "text_tokens": 21, + "image_tokens": 1091 + }, + "output_tokens_details": { + "text_tokens": 91 + } + }, + "output": { + "choices": [ + { + "finish_reason": "stop", + "message": { + "role": "assistant", + "content": [ + { + "text": "这是一张一只猫在草地上奔跑的照片。猫的毛色主要是白色，带有浅棕色的斑点，眼睛是蓝色的，显得非常可爱。背景是一个模糊的绿色草地和一些树木，阳光照射下来，给整个画面增添了一种温暖的感觉。猫的姿态显示出它正在快速移动，可能是在追逐什么或只是在享受户外活动的乐趣。整体来看，这是一幅充满活力和生机的图片。" + }, + { + "image_hw": [ + [ + "924", + "924" + ] + ] + } + ] + } + } + ] + } +} +``` + +## 附接口说明 + +在上述[获取临时URL](#a363e01e741gu)的两种方式中，代码调用和命令行工具已集成以下三个步骤，简化文件上传操作。以下是各步骤的接口说明。 + +#### **步骤1：获取文件上传凭证** + +##### **前提条件** + +您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。 + +##### **请求接口** + +``` +GET https://dashscope.aliyuncs.com/api/v1/uploads +``` + +**重要** + +文件上传凭证接口限流为 100 QPS（按“阿里云主账号+模型”维度），且临时存储不可扩容。生产环境或高并发场景请使用[阿里云OSS](https://help.aliyun.com/zh/oss/user-guide/what-is-oss)等存储服务。 + +##### **入参描述** + +**传参方式** + +**字段** + +**类型** + +**必选** + +**描述** + +**示例值** + +Header + +Content-Type + +_string_ + +是 + +请求类型：application/json 。 + +application/json + +Authorization + +_string_ + +是 + +阿里云百炼API Key，例如：Bearer sk-xxx。 + +Bearer sk-xxx + +Params + +action + +_string_ + +是 + +操作类型，当前场景为`getPolicy`。 + +getPolicy + +model + +_string_ + +是 + +需要调用的模型名称。 + +qwen-vl-plus + +##### **出参描述** + +**字段** + +**类型** + +**描述** + +**示例值** + +request\_id + +_string_ + +本次请求的系统唯一码。 + +7574ee8f-...-11c33ab46e51 + +data + +_object_ + +\- + +\- + +data.policy + +_string_ + +上传凭证。 + +eyJl...1ZSJ9XX0= + +data.signature + +_string_ + +上传凭证的签名。 + +g5K...d40= + +data.upload\_dir + +_string_ + +上传文件的目录。 + +dashscope-instant/xxx/2024-07-18/xxxx + +data.upload\_host + +_string_ + +上传的host地址。 + +https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com + +data.expire\_in\_seconds + +_string_ + +凭证有效期（单位：秒）。 + +**说明** + +过期后，重新调用本接口获取新的凭证。 + +300 + +data.max\_file\_size\_mb + +_string_ + +本次允许上传的最大文件的大小（单位：MB）。 + +该值与需要访问的模型相关。 + +100 + +data.capacity\_limit\_mb + +_string_ + +同一个主账号每天上传容量限制（单位：MB）。 + +999999999 + +data.oss\_access\_key\_id + +_string_ + +用于上传的access key。 + +LTAxxx + +data.x\_oss\_object\_acl + +_string_ + +上传文件的访问权限，`private`表示私有。 + +private + +data.x\_oss\_forbid\_overwrite + +_string_ + +文件同名时是否可以覆盖，`true`表示不可覆盖。 + +true + +##### **请求示例** + +``` +curl --location 'https://dashscope.aliyuncs.com/api/v1/uploads?action=getPolicy&model=qwen-vl-plus' \ +--header "Authorization: Bearer $DASHSCOPE_API_KEY" \ +--header 'Content-Type: application/json' +``` + +**说明** + +若未配置阿里云百炼API Key到环境变量，请将`$DASHSCOPE_API_KEY`替换为实际API Key，例如：`--header "Authorization: Bearer sk-xxx"`。 + +#### **响应示例** + +``` +{ + "request_id": "52f4383a-c67d-9f8c-xxxxxx", + "data": { + "policy": "eyJl...1ZSJ=", + "signature": "eWy...=", + "upload_dir": "dashscope-instant/xxx/2024-07-18/xxx", + "upload_host": "https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com", + "expire_in_seconds": 300, + "max_file_size_mb": 100, + "capacity_limit_mb": 999999999, + "oss_access_key_id": "LTA...", + "x_oss_object_acl": "private", + "x_oss_forbid_overwrite": "true" + } +} +``` + +#### **步骤2：上传文件至临时存储空间** + +#### **前提条件** + +- 已获取文件上传凭证。 + +- 确保文件上传凭证在有效期内，若凭证过期，请重新调用步骤1的接口获取新的凭证。 + + > 查看文件上传凭证有效期：步骤1的输出参数`data.expire_in_seconds`为凭证有效期，单位为秒。 + + +#### **请求接口** + +``` +POST {data.upload_host} +``` + +**说明** + +请将{data.upload\_host}替换为步骤1的输出参数`data.upload_host`对应的值。 + +#### **入参描述** + +**传参方式** + +**字段** + +**类型** + +**必选** + +**描述** + +**示例值** + +Header + +Content-Type + +_string_ + +否 + +提交表单必须为`multipart/form-data`。 + +在提交表单时，Content-Type会以`multipart/form-data;boundary=xxxxxx`的形式展示。 + +> boundary 是自动生成的随机字符串，无需手动指定。若使用 SDK 拼接表单，SDK 也会自动生成该随机值。 + +multipart/form-data; boundary=9431149156168 + +form-data + +OSSAccessKeyId + +_text_ + +是 + +文件上传凭证接口的输出参数 `data.oss_access_key_id` 的值。 + +LTAm5xxx + +policy + +_text_ + +是 + +文件上传凭证接口的输出参数 `data.policy` 的值。 + +g5K...d40= + +Signature + +_text_ + +是 + +文件上传凭证接口的输出参数 `data.signature` 的值。 + +Sm/tv7DcZuTZftFVvt5yOoSETsc= + +key + +_text_ + +是 + +文件上传凭证接口的输出参数 `data.upload_dir` 的值拼接上`/_文件名_`。 + +例如，`upload_dir` 为 `dashscope-instant/xxx/2024-07-18/xxx`，需要上传的文件名为 `cat.png`，拼接后的完整路径为： + +`dashscope-instant/xxx/2024-07-18/xxx/cat.png` + +x-oss-object-acl + +_text_ + +是 + +文件上传凭证接口的输出参数 `data.x_oss_object_acl` 的值。 + +private + +x-oss-forbid-overwrite + +_text_ + +是 + +文件上传凭证接口的输出参数中`data.x_oss_forbid_overwrite` 的值。 + +true + +success\_action\_status + +_text_ + +否 + +通常取值为 200，上传完成后接口返回 HTTP code 200，表示操作成功。 + +200 + +file + +_text_ + +是 + +文件或文本内容。 + +**说明** + +- 一次只支持上传一个文件。 + +- file必须为最后一个表单域，除file以外的其他表单域并无顺序要求。 + + +例如，待上传文件`cat.png`在Linux系统中的存储路径为`/tmp`，则此处应为`file=@"/tmp/cat.png"`。 + +#### **出参描述** + +调用成功时，本接口无任何参数输出。 + +#### **请求示例** + +``` +curl --location 'https://dashscope-file-xxx.oss-cn-beijing.aliyuncs.com' \ +--form 'OSSAccessKeyId="LTAm5xxx"' \ +--form 'Signature="Sm/tv7DcZuTZftFVvt5yOoSETsc="' \ +--form 'policy="eyJleHBpcmF0aW9 ... ... ... dHJ1ZSJ9XX0="' \ +--form 'x-oss-object-acl="private"' \ +--form 'x-oss-forbid-overwrite="true"' \ +--form 'key="dashscope-instant/xxx/2024-07-18/xxx/cat.png"' \ +--form 'success_action_status="200"' \ +--form 'file=@"/tmp/cat.png"' +``` + +#### **步骤3：生成文件URL** + +文件URL拼接逻辑：`**oss://**` + `**key**` （步骤2的入参`key`）。该URL有效期为 48 小时。 + +``` +oss://dashscope-instant/xxx/2024-07-18/xxxx/cat.png +``` + +## **错误码** + +如果接口调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 + +本文的API还有特定状态码，具体如下所示。 + +**HTTP状态码** + +**接口错误码（code）** + +**接口错误信息（message）** + +**含义说明** + +400 + +invalid\_parameter\_error + +InternalError.Algo.InvalidParameter: The provided URL does not appear to be valid. Ensure it is correctly formatted. + +无效URL，请检查URL是否填写正确。 + +> 若使用临时文件URL，需确保请求的 Header 中添加了参数 `X-DashScope-OssResourceResolve: enable`。 + +400 + +InvalidParameter.DataInspection + +The media format is not supported or incorrect for the data inspection. + +可能的原因有： + +- 请求Header 缺少必要参数，请设置 `X-DashScope-OssResourceResolve: enable`**。** + +- 上传的图片格式不符合模型要求，更多信息请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)。 + + +403 + +AccessDenied + +Invalid according to Policy: Policy expired. + +文件上传凭证已经过期。 + +请重新调用[文件上传凭证接口](#32db94982cllx)生成新凭证。 + +429 + +Throttling.RateQuota + +Requests rate limit exceeded, please try again later. + +调用频次触发限流。 + +[文件上传凭证接口](#32db94982cllx)限流为 100 QPS（按阿里云主账号 + 模型维度）。触发限流后，建议降低请求频率，或迁移至 OSS 等自有存储服务以规避限制。 + +## **常见问题** + +#### **Q：使用** `**oss://**` **前缀的 URL 调用时报错，该如何处理？** + +A：请按以下步骤排查： + +1. **检查请求头（Header）**： + 若您通过 HTTP（如 Postman、curl）直接调用，**必须在** `**Header**` **中添加参数** `**X-DashScope-OssResourceResolve: enable**`。未添加该参数会导致服务端无法识别 OSS 内部协议。关于请求头配置，请参见[通过HTTP调用](#d6a1cb0f01h5k)。 + +2. **检查 URL 有效性**： + `oss://` 链接为临时 URL，请确保该链接是48小时内生成的。如果链接已过期，请重新上传文件获取新的 URL。 + + +#### **Q：文件上传与模型调用使用的API KEY可以不一样吗？** + +A：文件存储和访问权限基于阿里云主账号管理，API Key 仅为主账号的访问凭证。 + +因此，同一阿里云主账号下的不同 API Key 可正常使用，不同主账号的 API Key因账号隔离，模型调用无法跨账号读取文件。 + +请确保文件上传与模型调用使用的 API Key 属于同一阿里云主账号。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md deleted file mode 100644 index 54954450..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md +++ /dev/null @@ -1,2762 +0,0 @@ -# Qwen-OCR API参考 - -本文介绍通过 OpenAI 兼容接口 或 DashScope API 调用通义千问OCR 模型的输入与输出参数。 - -> 相关文档：[文字提取（Qwen-OCR）](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr) - -## OpenAI 兼容 - -## 华北2（北京）地域 - -SDK 调用配置的`base_url`为：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1` - -HTTP 调用配置的`endpoint`：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions` - -## 新加坡地域 - -SDK 调用配置的`base_url`为：`https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1` - -HTTP 调用配置的`endpoint`：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions` - -## 美国（弗吉尼亚）地域 - -SDK 调用配置的`base_url`为：`https://dashscope-us.aliyuncs.com/compatible-mode/v1` - -HTTP 调用配置的`endpoint`：`POST https://dashscope-us.aliyuncs.com/compatible-mode/v1/chat/completions` - -**重要** - -百炼为华北2（北京）、新加坡地域推出了业务空间专属域名，**能够为推理请求提供卓越的性能和更高的稳定性**，建议迁移至新域名： - -- 华北2（北京）地域：从 `https://dashscope.aliyuncs.com` 迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` - -- 新加坡地域：从 `https://dashscope-intl.aliyuncs.com` 迁移至 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com` - - -其中 `{WorkspaceId}` 为您的业务空间 ID，可在百炼控制台的**业务空间详情**页面查看。现有域名仍可正常使用。 - -> 您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。若通过OpenAI SDK进行调用，需要[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)。 - -### 请求体 - -## 非流式输出 - -## Python - -``` -from openai import OpenAI -import os - -PROMPT_TICKET_EXTRACTION = """ -请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。 -要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。 -返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'}, -""" - -try: - client = OpenAI( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下为北京地域的 base_url，若使用弗吉尼亚地域模型，需要将base_url换成https://dashscope-us.aliyuncs.com/compatible-mode/v1 - # 若使用新加坡地域的模型，需将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", - ) - completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[ - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": {"url":"https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg"}, - # 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192 - }, - # 模型支持在以下text字段中传入Prompt，若未传入，则会使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - {"type": "text", - "text": PROMPT_TICKET_EXTRACTION} - ] - } - ]) - print(completion.choices[0].message.content) -except Exception as e: - print(f"错误信息: {e}") -``` - -## Node.js - -``` -import OpenAI from 'openai'; - -// 定义提取车票信息的Prompt -const PROMPT_TICKET_EXTRACTION = ` -请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。 -要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。 -返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'} -`; - -// 初始化OpenAI客户端 -const client = new OpenAI({ - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: "sk-xxx", - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - apiKey: process.env.DASHSCOPE_API_KEY, - // 以下为北京地域的 base_url，若使用弗吉尼亚地域模型，需要将base_url换成https://dashscope-us.aliyuncs.com/compatible-mode/v1 - // 若使用新加坡地域的模型，需将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 - baseURL: "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -}); - -async function main() { - try { - // 创建聊天完成请求 - const completion = await client.chat.completions.create({ - model: "qwen3.5-ocr", - messages: [ - { - role: "user", - content: [ - // 模型支持在text字段中传入Prompt，若未传入，则会使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - { - type: "image_url", - image_url: { - url: "https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg", - }, - // 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - min_pixels: 32 * 32 * 3, - // 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - max_pixels: 32 * 32 * 8192 - }, - {type: "text", - text: PROMPT_TICKET_EXTRACTION} - ] - } - ] - }); - - // 输出结果 - console.log(completion.choices[0].message.content); - } catch (error) { - console.log(`错误信息: ${error}`); - } -} - -main(); -``` - -## curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下是北京地域base_url，若使用弗吉尼亚地域模型，需要将base_url换成https://dashscope-us.aliyuncs.com/compatible-mode/v1/chat/completions -# 如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions -# === 执行时请删除该注释 === - -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" \ --H "Content-Type: application/json" \ --d '{ - "model": "qwen3.5-ocr", - "messages": [ - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": {"url":"https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg"}, - "min_pixels": 3072, - "max_pixels": 8388608 - }, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'"} - ] - } - ] -}' -``` - -## 流式输出 - -## Python - -``` -import os -from openai import OpenAI - -PROMPT_TICKET_EXTRACTION = """ -请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。 -要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。 -返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'}, -""" - -client = OpenAI( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv("DASHSCOPE_API_KEY"), - # 以下是北京地域base-url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1", -) -completion = client.chat.completions.create( - model="qwen3.5-ocr", - messages=[ - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": {"url":"https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg"}, - # 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192 - }, - # 模型支持在text字段中传入Prompt，若未传入，则会使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - {"type": "text","text": PROMPT_TICKET_EXTRACTION} - - ] - } - ], - stream=True, - stream_options={"include_usage": True} -) -for chunk in completion: - print(chunk.model_dump_json()) -``` - -## Node.js - -``` -import OpenAI from 'openai'; - -// 定义提取车票信息的Prompt -const PROMPT_TICKET_EXTRACTION = ` -请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。 -要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。 -返回数据格式以json方式输出，格式为：{'发票号码': 'xxx','起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'} -`; - -const openai = new OpenAI({ - // 若没有配置环境变量，请用百炼API Key将下行替换为：apiKey: "sk-xxx", - // 以下是北京地域base-url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 - apiKey: process.env.DASHSCOPE_API_KEY, - // 以下是北京地域base-url，如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1 - baseURL: 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1', -}); - -async function main() { - const response = await openai.chat.completions.create({ - model: 'qwen3.5-ocr', - messages: [ - { - role: 'user', - content: [ - // 模型支持在text字段中传入Prompt，若未传入，则会使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - { type: 'text', text: PROMPT_TICKET_EXTRACTION}, - { - type: 'image_url', - image_url: { - url: 'https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg', - }, - // 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - // 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192 - } - ] - } - ], - stream: true, - stream_options:{"include_usage": true} - }); -let fullContent = "" - console.log("流式输出内容为：") - for await (const chunk of response) { - if (chunk.choices[0] && chunk.choices[0].delta.content != null) { - fullContent += chunk.choices[0].delta.content; - console.log(chunk.choices[0].delta.content); - } -} - console.log(`完整输出内容为：${fullContent}`) -} - -main(); -``` - -## curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下是北京地域base_url，若使用弗吉尼亚地域模型，需要将base_url换成https://dashscope-us.aliyuncs.com/compatible-mode/v1/chat/completions -# 如果使用新加坡地域的模型，需要将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/compatible-mode/v1/chat/completions -# === 执行时请删除该注释 === - -curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ --H "Authorization: Bearer $DASHSCOPE_API_KEY" \ --H "Content-Type: application/json" \ --d '{ - "model": "qwen3.5-ocr", - "messages": [ - { - "role": "user", - "content": [ - { - "type": "image_url", - "image_url": {"url":"https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg"}, - "min_pixels": 3072, - "max_pixels": 8388608 - }, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'"} - ] - } - ], - "stream": true, - "stream_options": {"include_usage": true} -}' -``` - -**model** `_string_` **（必选）** - -模型名称。支持的模型可参见`[选择模型](https://help.aliyun.com/zh/model-studio/models#55c81ba3ccgct)`。 - -**messages** `_array_` **（必选）** - -传递给大模型的上下文，按对话顺序排列。 - -**消息类型** - -User Message `_object_` **（必选）** - -用户消息，用于向模型传递指令和待识别的图像。 - -**属性** - -**content** `_array_`**（必选）** - -消息内容。 - -**属性** - -**type** `_string_` **（必选）** - -可选值： - -- `text` - - 输入文本时需设为`text`。 - -- `image_url` - - 输入图片时需设为`image_url`。 - - -**text** `_string_` **（可选）** - -输入的文本。 - -默认值为：`Please output only the text content from the image without any additional descriptions or formatting.` ，即模型默认提取图像中的全部文本。 - -**image\_url** `_object_` - -输入的图片信息。当`type`为`image_url`时是必选参数。 - -**属性** - -**url** `_string_`**（必选）** - -图片的 URL或 Base64 Data URL。传入本地文件请参考[文字提取](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#ea4e1d92dbry2)。 - -**min\_pixels** `_integer_` （可选） - -用于设定输入图像的最小像素阈值，单位为像素。 - -当输入图像像素小于`min_pixels`时，会将图像进行放大，直到总像素高于`min_pixels`。 - -**图像Token与像素的转换关系** - -不同模型，每个图像 Token 对应的像素不同： - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：每 Token 对应像素为`32*32`。 - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：每 Token 对应像素为`28*28`。 - - -**min\_pixels 取值范围** - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：默认值和最小值均为3072（即`3×32×32`） - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：默认值和最小值均为 `3136` （即`4×28×28`）。 - - -示例值：`{"type": "image_url","image_url": {"url":"https://xxxx.jpg"},"min_pixels": 3072}` - -**max\_pixels** `_integer_` （可选） - -用于设定输入图像的最大像素阈值，单位为像素。 - -当输入图像像素在`[min_pixels, max_pixels]`区间内时，模型会按原图进行识别。当输入图像像素大于`max_pixels`时，会将图像进行缩小，直到总像素低于`max_pixels`。 - -**图像Token与像素的转换关系** - -不同模型，每个图像 Token 对应的像素不同： - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：每 Token 对应像素为`32*32`。 - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：每 Token 对应像素为`28*28`。 - - -**max\_pixels 取值范围** - -- `qwen3.5-ocr、qwen-vl-ocr-latest、qwen-vl-ocr-2025-11-20` - - - 默认值：8388608 （即`8192x32x32`） - - - 最大值：30720000（即`30000x32x32`） - -- `qwen-vl-ocr、qwen-vl-ocr-2025-08-28`及之前更新的模型 - - - 默认值：6422528（即`8192x28x28`） - - - 最大值：23520000（即`30000x28x28`） - - -示例值：`{"type": "image_url","image_url": {"url":"https://xxxx.jpg"},"max_pixels": 8388608}` - -**role** `_string_` **（必选）** - -用户消息的角色，固定为`user`。 - -**stream** `_boolean_` （可选） 默认值为 `false` - -是否以流式方式输出回复。 - -可选值： - -- `false`：等待模型生成完整回复后一次性返回。 - -- `true`：模型边生成边返回数据块。客户端需逐块读取，以还原完整回复。 - - -**stream\_options** `_object_` （可选） - -流式输出的配置项，仅在 `stream` 为 `true` 时生效。 - -**属性** - -**include\_usage** `_boolean_` （可选）默认值为 `false` - -是否在**最后一个数据块**包含Token消耗信息。 - -可选值： - -- `true`：包含； - -- `false`：不包含。 - - -**max\_tokens** `_integer_` （可选） - -用于限制模型输出的最大 Token 数。若生成内容超过此值，响应将被截断。 - -- `qwen3.5-ocr`：默认值与最大值为32768。 - -- `qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`、`qwen-vl-ocr-2024-10-28`默认值与最大值均为模型的最大输出长度，请参见[模型选型](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#f4299b0a1ace4)。 - -- `qwen-vl-ocr、qwen-vl-ocr-2025-04-13、qwen-vl-ocr-2025-08-28`，默认值和最大值为4096。 - - > 如需提高该参数值（4097~8192范围），请联系商务经理进行申请，并提供以下信息：主账号ID、图像类型（如文档图、电商图、合同等）、模型名称、预计 QPS 和每日请求总数，以及模型输出长度超过4096的请求占比。 - - -**logprobs** `_boolean_` （可选）默认值为 `false` - -是否返回输出 Token 的对数概率，可选值： - -- `true` - - 返回 - -- `false` - - 不返回 - - -**top\_logprobs** `_integer_` （可选）默认值为0 - -指定在每一步生成时，返回模型最大概率的候选 Token 个数。 - -取值范围：\[0,5\] - -仅当 `logprobs` 为 `true` 时生效。 - -**temperature** `_float_` （可选） 默认值为0.01 - -采样温度，控制模型生成文本的多样性。 - -temperature越高，生成的文本更多样，反之，生成的文本更确定。 - -取值范围： \[0, 2) - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> 建议设置为默认值即可。 - -**top\_p** `_float_` （可选）默认值为0.001 - -核采样的概率阈值，控制模型生成文本的多样性。 - -top\_p越高，生成的文本更多样。反之，生成的文本更确定。 - -取值范围：（0,1.0\] - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> 建议设置为默认值即可。 - -**top\_k** `_integer_` （可选）默认值为1 - -生成过程中采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个Token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。取值为None或当top\_k大于100时，表示不启用top\_k策略，此时仅有top\_p策略生效。 - -取值需要大于或等于0。 - -该参数非OpenAI标准参数。通过 Python SDK调用时，请放入 **extra\_body** 对象中，配置方式为：`extra_body={"top_k": xxx}`；通过 Node.js SDK 或 HTTP 方式调用时，请作为顶层参数传递。 - -> 建议设置为默认值即可。 - -**repetition\_penalty** `_float_` （可选）默认值为1.0 - -模型生成时连续序列中的重复度。提高repetition\_penalty时可以降低模型生成的重复度，1.0表示不做惩罚。该参数对模型效果影响较大，建议保持默认值。 - -> 建议设置为默认值即可。 - -**presence\_penalty** `_float_` （可选）默认值为0.0 - -控制模型生成文本时的内容重复度。 - -取值范围：\[-2.0, 2.0\]。正值降低重复度，负值增加重复度。 - -在创意写作或头脑风暴等需要多样性、趣味性或创造力的场景中，建议调高该值；在技术文档或正式文本等强调一致性与术语准确性的场景中，建议调低该值。 - -**原理介绍** - -如果参数值是正数，模型将对目前文本中已存在的Token施加一个惩罚值（惩罚值与文本出现的次数无关），减少这些Token重复出现的几率，从而减少内容重复度，增加用词多样性。 - -> 建议设置为默认值即可。 - -**seed** `_integer_` （可选） - -随机数种子。用于确保在相同输入和参数下生成结果可复现。若调用时传入相同的 `seed` 且其他参数不变，模型将尽可能返回相同结果。 - -取值范围：`[0,231−1]`。 - -> 建议设置为默认值即可。 - -**stop** `_string 或 array_` （可选） - -用于指定停止词。当模型生成的文本中出现`stop` 指定的字符串或`token_id`时，生成将立即终止。 - -可传入敏感词以控制模型的输出。 - -> stop为数组时，不可将`token_id`和字符串同时作为元素输入，比如不可以指定为`["你好",104307]`。 - -### chat响应对象（非流式输出） - -``` -{ - "id": "chatcmpl-ba21fa91-dcd6-4dad-90cc-6d49c3c39094", - "choices": [ - { - "finish_reason": "stop", - "index": 0, - "logprobs": null, - "message": { - "content": "```json\n{\n \"销售方名称\": \"null\",\n \"购买方名称\": \"蔡应时\",\n \"不含税价\": \"230769.23\",\n \"组织机构代码\": \"null\",\n \"发票代码\": \"142011726001\"\n}\n```", - "refusal": null, - "role": "assistant", - "annotations": null, - "audio": null, - "function_call": null, - "tool_calls": null - } - } - ], - "created": 1763283287, - "model": "qwen3.5-ocr", - "object": "chat.completion", - "service_tier": null, - "system_fingerprint": null, - "usage": { - "completion_tokens": 72, - "prompt_tokens": 1185, - "total_tokens": 1257, - "completion_tokens_details": { - "accepted_prediction_tokens": null, - "audio_tokens": null, - "reasoning_tokens": null, - "rejected_prediction_tokens": null, - "text_tokens": 72 - }, - "prompt_tokens_details": { - "audio_tokens": null, - "cached_tokens": null, - "image_tokens": 1001, - "text_tokens": 184 - } - } -} -``` - -**id** `_string_` - -本次请求的唯一标识符。 - -**choices** `_array_` - -模型生成内容的数组。 - -**属性** - -**finish\_reason** `_string_` - -模型停止生成的原因。 - -有两种情况： - -- 自然停止输出时为`stop`； - -- 生成长度过长而结束为`length`。 - - -**index** `_integer_` - -当前对象在`choices`数组中的索引。 - -**message** `_object_` - -模型输出的消息。 - -**属性** - -**content** `_string_` - -大模型的返回结果。 - -**processed\_text** `_string_` - -对模型原始输出进行后处理的结果，自动删除重复片段等。当模型输出存在重复内容时，该字段提供清洗后的文本。 - -> 仅通过 DashScope SDK 和 curl 调用时返回，OpenAI 兼容 SDK 不返回该字段。 - -**refusal** `_string_` - -该参数当前固定为`null`。 - -**role** `_string_` - -消息的角色，固定为`assistant`。 - -**audio** `_object_` - -该参数当前固定为`null`。 - -**function\_call** `_object_` - -该参数当前固定为`null`。 - -**tool\_calls** `_array_` - -该参数当前固定为`null`。 - -**created** `_integer_` - -本次请求被创建时的时间戳。 - -**model** `_string_` - -本次请求使用的模型。 - -**object** `_string_` - -始终为`chat.completion`。 - -**service\_tier** `_string_` - -该参数当前固定为`null`。 - -**system\_fingerprint** `_string_` - -该参数当前固定为`null`。 - -**usage** `_object_` - -本次请求的 Token 消耗信息。 - -**属性** - -**completion\_tokens** `_integer_` - -模型输出的 Token 数。 - -**prompt\_tokens** `_integer_` - -输入的 Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`prompt_tokens`与`completion_tokens`的总和。 - -**completion\_tokens\_details** `_object_` - -模型输出Token的细粒度分类。 - -**属性** - -**accepted\_prediction\_tokens**`_integer_` - -该参数当前固定为`null`。 - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**reasoning\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输出文本对应的 Token 数。 - -**rejected\_prediction\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**prompt\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**cached\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -### chat响应chunk对象（流式输出） - -``` -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"","function_call":null,"refusal":null,"role":"assistant","tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"```","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"json","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"\n","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"{\n","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":" ","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -...... -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"```","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[{"delta":{"content":"","function_call":null,"refusal":null,"role":null,"tool_calls":null},"finish_reason":"stop","index":0,"logprobs":null}],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":null} -{"id":"chatcmpl-f6fbdc0d-78d6-418f-856f-f099c2e4859b","choices":[],"created":1764139204,"model":"qwen3.5-ocr","object":"chat.completion.chunk","service_tier":null,"system_fingerprint":null,"usage":{"completion_tokens":141,"prompt_tokens":513,"total_tokens":654,"completion_tokens_details":{"accepted_prediction_tokens":null,"audio_tokens":null,"reasoning_tokens":null,"rejected_prediction_tokens":null,"text_tokens":141},"prompt_tokens_details":{"audio_tokens":null,"cached_tokens":null,"image_tokens":332,"text_tokens":181}}} -``` - -**id** `_string_` - -本次调用的唯一标识符。每个chunk对象有相同的 id。 - -**choices** `_array_` - -模型生成内容的数组。若设置`include_usage`参数为`true`，则在最后一个chunk中为空。 - -**属性** - -**delta** `_object_` - -流式返回的输出内容。 - -**属性** - -**content** `_string_` - -大模型的返回结果。 - -**function\_call** `_object_` - -该参数当前固定为`null`。 - -**refusal** `_object_` - -该参数当前固定为`null`。 - -**role** `_string_` - -消息对象的角色，只在第一个chunk中有值。 - -**finish\_reason** `_string_` - -模型停止生成的原因。有三种情况： - -- 自然停止输出时为`stop`； - -- 生成未结束时为`null`； - -- 生成长度过长而结束为`length`。 - - -**index** `_integer_` - -当前响应在`choices`数组中的索引。 - -**created** `_integer_` - -本次请求被创建时的时间戳。每个chunk有相同的时间戳。 - -**model** `_string_` - -本次请求使用的模型。 - -**object** `_string_` - -始终为`chat.completion.chunk`。 - -**service\_tier** `_string_` - -该参数当前固定为`null`。 - -**system\_fingerprint**`_string_` - -该参数当前固定为`null`。 - -**usage** `_object_` - -本次请求消耗的Token。只在`include_usage`为`true`时，在最后一个chunk返回。 - -**属性** - -**completion\_tokens** `_integer_` - -模型输出的 Token 数。 - -**prompt\_tokens** `_integer_` - -输入的 Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`prompt_tokens`与`completion_tokens`的总和。 - -**completion\_tokens\_details** `_object_` - -模型输出Token的细粒度分类。 - -**属性** - -**accepted\_prediction\_tokens**`_integer_` - -该参数当前固定为`null`。 - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**reasoning\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输出文本对应的 Token 数。 - -**rejected\_prediction\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**prompt\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**audio\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**cached\_tokens** `_integer_` - -该参数当前固定为`null`。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -## DashScope - -## 华北2（北京）地域 - -HTTP 调用配置的`endpoint`：`POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -SDK 调用无需配置 `base_url`。 - -## 新加坡地域 - -HTTP 调用配置的`endpoint`：`POST https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -SDK调用配置的`base_url`： - -## **Python代码** - -``` -dashscope.base_http_api_url = 'https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1' -``` - -## **Java代码** - -- **方式一：** - - ``` - import com.alibaba.dashscope.protocol.Protocol; - MultiModalConversation conv = new MultiModalConversation(Protocol.HTTP.getValue(), "https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"); - ``` - -- **方式二：** - - ``` - import com.alibaba.dashscope.utils.Constants; - Constants.baseHttpApiUrl="https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1"; - ``` - - -## 美国（弗吉尼亚）地域 - -HTTP 调用配置的`endpoint`：`POST https://dashscope-us.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` - -SDK调用配置的`base_url`： - -## **Python代码** - -``` -dashscope.base_http_api_url = 'https://dashscope-us.aliyuncs.com/api/v1' -``` - -## **Java代码** - -- **方式一：** - - ``` - import com.alibaba.dashscope.protocol.Protocol; - MultiModalConversation conv = new MultiModalConversation(Protocol.HTTP.getValue(), "https://dashscope-us.aliyuncs.com/api/v1"); - ``` - -- **方式二：** - - ``` - import com.alibaba.dashscope.utils.Constants; - Constants.baseHttpApiUrl="https://dashscope-us.aliyuncs.com/api/v1"; - ``` - - -> 您需要已[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)并[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)。若通过DashScope SDK进行调用，需要[安装DashScope SDK](https://help.aliyun.com/zh/model-studio/install-sdk#f3e80b21069aa)。 - -### 请求体 - -## 高精识别 - -> 以下为调用高精识别内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False}] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为高精识别 - ocr_options={"task": "advanced_recognition"} -) -# 高精识别任务以纯文本返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -// dashscope SDK的版本 >= 2.21.8 -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.ADVANCED_RECOGNITION) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data ' -{ - "model": "qwen3.5-ocr", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] - }, - "parameters": { - "ocr_options": { - "task": "advanced_recognition" - } - } -} -' -``` - -## 信息抽取 - -> 以下为调用信息抽取内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -# use [pip install -U dashscope] to update sdk - -import os -import dashscope -from dashscope import MultiModalConversation - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [ - { - "role":"user", - "content":[ - { - "image":"http://duguang-labelling.oss-cn-shanghai.aliyuncs.com/demo_ocr/receipt_zh_demo.jpg", - "min_pixels": 32 * 32 * 3, - "max_pixels": 32 * 32 *8192, - "enable_rotate": False - } - ] - } - ] - -# 指定抽取字段 -params = { - "ocr_options":{ - "task": "key_information_extraction", - "task_config": { - "result_schema": { - "乘车日期": "对应图中乘车日期时间，格式为年-月-日，比如2025-03-05", - "发票代码": "提取图中的发票代码，通常为一组数字或字母组合", - "发票号码": "提取发票上的号码，通常由纯数字组成。" - } - } - } -} - -response = MultiModalConversation.call( - model='qwen3.5-ocr', - messages=messages, - **params, - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv('DASHSCOPE_API_KEY')) - -print(response.output.choices[0].message.content[0]["ocr_result"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.google.gson.JsonObject; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "http://duguang-labelling.oss-cn-shanghai.aliyuncs.com/demo_ocr/receipt_zh_demo.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels",3072); - // 开启图像自动转正功能 - map.put("enable_rotate", false); - - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - - // 创建主JSON对象 - JsonObject resultSchema = new JsonObject(); - resultSchema.addProperty("乘车日期", "对应图中乘车日期时间，格式为年-月-日，比如2025-03-05"); - resultSchema.addProperty("发票代码", "提取图中的发票代码，通常为一组数字或字母组合"); - resultSchema.addProperty("发票号码", "提取发票上的号码，通常由纯数字组成。"); - - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.KEY_INFORMATION_EXTRACTION) - .taskConfig(OcrOptions.TaskConfig.builder() - .resultSchema(resultSchema) - .build()) - .build(); - - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("ocr_result")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data ' -{ - "model": "qwen3.5-ocr", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "image": "http://duguang-labelling.oss-cn-shanghai.aliyuncs.com/demo_ocr/receipt_zh_demo.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] - }, - "parameters": { - "ocr_options": { - "task": "key_information_extraction", - "task_config": { - "result_schema": { - "乘车日期": "对应图中乘车日期时间，格式为年-月-日，比如2025-03-05", - "发票代码": "提取图中的发票代码，通常为一组数字或字母组合", - "发票号码": "提取发票上的号码，通常由纯数字组成。" - } - } - } - } -} -' -``` - -## 表格解析 - -> 以下为调用表格解析内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "http://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/doc_parsing/tables/photo/eng/17.jpg", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False}] - }] -response = dashscope.MultiModalConversation.call( - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为表格解析 - ocr_options= {"task": "table_parsing"} -) -# 表格解析任务以HTML格式返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/doc_parsing/tables/photo/eng/17.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.TABLE_PARSING) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data ' -{ - "model": "qwen3.5-ocr", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "image": "http://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/doc_parsing/tables/photo/eng/17.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] - }, - "parameters": { - "ocr_options": { - "task": "table_parsing" - } - } -} -' -``` - -## **文档解析** - -> 以下为调用文档解析内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "https://img.alicdn.com/imgextra/i1/O1CN01ukECva1cisjyK6ZDK_!!6000000003635-0-tps-1500-1734.jpg", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False}] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为文档解析 - ocr_options= {"task": "document_parsing"} -) -# 文档解析任务以LaTeX格式返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://img.alicdn.com/imgextra/i1/O1CN01ukECva1cisjyK6ZDK_!!6000000003635-0-tps-1500-1734.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.DOCUMENT_PARSING) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation'\ - --header "Authorization: Bearer $DASHSCOPE_API_KEY"\ - --header 'Content-Type: application/json'\ - --data '{ -"model": "qwen3.5-ocr", -"input": { - "messages": [ - { - "role": "user", - "content": [{ - "image": "https://img.alicdn.com/imgextra/i1/O1CN01ukECva1cisjyK6ZDK_!!6000000003635-0-tps-1500-1734.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] -}, -"parameters": { - "ocr_options": { - "task": "document_parsing" - } -} -} -' -``` - -## 公式识别 - -> 以下为调用公式识别内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "http://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/formula_handwriting/test/inline_5_4.jpg", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False }] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx" - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为公式识别 - ocr_options= {"task": "formula_recognition"} -) - -# 公式识别任务以LaTeX格式返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "http://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/formula_handwriting/test/inline_5_4.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.FORMULA_RECOGNITION) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data ' -{ - "model": "qwen3.5-ocr", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "image": "http://duguang-llm.oss-cn-hangzhou.aliyuncs.com/llm_data_keeper/data/formula_handwriting/test/inline_5_4.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] - }, - "parameters": { - "ocr_options": { - "task": "formula_recognition" - } - } -} -' -``` - -## **通用文字识别** - -> 以下为调用通用文字识别内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False}] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为通用文字识别 - ocr_options= {"task": "text_recognition"} -) -# 通用文字识别任务以纯文本格式返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - - // 配置内置任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.TEXT_RECOGNITION) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation'\ - --header "Authorization: Bearer $DASHSCOPE_API_KEY"\ - --header 'Content-Type: application/json'\ - --data '{ -"model": "qwen3.5-ocr", -"input": { - "messages": [ - { - "role": "user", - "content": [{ - "image": "https://help-static-aliyun-doc.aliyuncs.com/file-manage-files/zh-CN/20241108/ctdzex/biaozhun.jpg", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] -}, -"parameters": { - "ocr_options": { - "task": "text_recognition" - } -} -}' -``` - -## 多语言识别 - -> 以下为调用通用多语言识别内置任务的代码示例，详情请参见[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -Python - -``` -import os -import dashscope - -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -messages = [{ - "role": "user", - "content": [{ - "image": "https://img.alicdn.com/imgextra/i2/O1CN01VvUMNP1yq8YvkSDFY_!!6000000006629-2-tps-6000-3000.png", - # 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 是否开启图像自动转正功能 - "enable_rotate": False }] - }] - -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv('DASHSCOPE_API_KEY'), - model='qwen3.5-ocr', - messages=messages, - # 设置内置任务为多语言识别 - ocr_options={"task": "multi_lan"} -) -# 多语言识别任务以纯文本的形式返回结果 -print(response["output"]["choices"][0]["message"].content[0]["text"]) -``` - -Java - -``` -import java.util.Arrays; -import java.util.Collections; -import java.util.Map; -import java.util.HashMap; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.aigc.multimodalconversation.OcrOptions; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - // 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://img.alicdn.com/imgextra/i2/O1CN01VvUMNP1yq8YvkSDFY_!!6000000006629-2-tps-6000-3000.png"); - // 输入图像的最大像素阈值，超过该值图像会进行缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会进行放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 是否开启图像自动转正功能 - map.put("enable_rotate", false); - // 配置内置的OCR任务 - OcrOptions ocrOptions = OcrOptions.builder() - .task(OcrOptions.Task.MULTI_LAN) - .build(); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map - )).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .ocrOptions(ocrOptions) - .build(); - MultiModalConversationResult result = conv.call(param); - System.out.println(result.getOutput().getChoices().get(0).getMessage().getContent().get(0).get("text")); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# 以下为华北2（北京）地域的URL，调用时请将WorkspaceId替换为真实的业务空间ID，各地域的URL不同。 -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ ---data ' -{ - "model": "qwen3.5-ocr", - "input": { - "messages": [ - { - "role": "user", - "content": [ - { - "image": "https://img.alicdn.com/imgextra/i2/O1CN01VvUMNP1yq8YvkSDFY_!!6000000006629-2-tps-6000-3000.png", - "min_pixels": 3072, - "max_pixels": 8388608, - "enable_rotate": false - } - ] - } - ] - }, - "parameters": { - "ocr_options": { - "task": "multi_lan" - } - } -} -' -``` - -## 流式输出 - -## Python - -``` -import os -import dashscope - -# 以下为北京地域base_url，若使用弗吉尼亚地域模型，需要将base_url换成 https://dashscope-us.aliyuncs.com/api/v1 -# 若使用新加坡地域的模型，需将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1 -dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - -PROMPT_TICKET_EXTRACTION = """ -请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。 -要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。 -返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'}, -""" - -messages = [ - { - "role": "user", - "content": [ - { - "image": "https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg", - # 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - "min_pixels": 32 * 32 * 3, - # 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - "max_pixels": 32 * 32 * 8192, - # 开启图像自动转正功能 - "enable_rotate": False, - }, - # 未设置内置任务时，支持在text字段中传入Prompt，若未传入则使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - { - "type": "text", - "text": PROMPT_TICKET_EXTRACTION, - }, - ], - } -] -response = dashscope.MultiModalConversation.call( - # 若没有配置环境变量，请用百炼API Key将下行替换为：api_key="sk-xxx", - # 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - api_key=os.getenv("DASHSCOPE_API_KEY"), - model="qwen3.5-ocr", - messages=messages, - stream=True, - incremental_output=True, -) - -full_content = "" -print("流式输出内容为：") -for response in response: - try: - print(response["output"]["choices"][0]["message"].content[0]["text"]) - full_content += response["output"]["choices"][0]["message"].content[0]["text"] - except: - pass -print(f"完整内容为：{full_content}") -``` - -## Java - -``` -import java.util.*; - -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversation; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationParam; -import com.alibaba.dashscope.aigc.multimodalconversation.MultiModalConversationResult; -import com.alibaba.dashscope.common.MultiModalMessage; -import com.alibaba.dashscope.common.Role; -import com.alibaba.dashscope.exception.ApiException; -import com.alibaba.dashscope.exception.NoApiKeyException; -import com.alibaba.dashscope.exception.UploadFileException; -import io.reactivex.Flowable; -import com.alibaba.dashscope.utils.Constants; - -public class Main { - - // 以下为北京地域 base_url，若使用弗吉尼亚地域模型，需要将base_url换成 https://dashscope-us.aliyuncs.com/api/v1 - // 若使用新加坡地域的模型，需将base_url替换为：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1 - static {Constants.baseHttpApiUrl="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1";} - - public static void simpleMultiModalConversationCall() - throws ApiException, NoApiKeyException, UploadFileException { - MultiModalConversation conv = new MultiModalConversation(); - Map map = new HashMap<>(); - map.put("image", "https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg"); - // 输入图像的最大像素阈值，超过该值图像会缩小，直到总像素低于max_pixels - map.put("max_pixels", 8388608); - // 输入图像的最小像素阈值，小于该值图像会放大，直到总像素大于min_pixels - map.put("min_pixels", 3072); - // 开启图像自动转正功能 - map.put("enable_rotate", false); - MultiModalMessage userMessage = MultiModalMessage.builder().role(Role.USER.getValue()) - .content(Arrays.asList( - map, - // 模型未设置内置任务时，支持在text字段中传入Prompt，若未传入则使用默认的Prompt：Please output only the text content from the image without any additional descriptions or formatting. - Collections.singletonMap("text", "请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'"))).build(); - MultiModalConversationParam param = MultiModalConversationParam.builder() - // 若没有配置环境变量，请用百炼API Key将下行替换为：.apiKey("sk-xxx") - // 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key - .apiKey(System.getenv("DASHSCOPE_API_KEY")) - .model("qwen3.5-ocr") - .message(userMessage) - .incrementalOutput(true) - .build(); - Flowable result = conv.streamCall(param); - result.blockingForEach(item -> { - try { - List> contentList = item.getOutput().getChoices().get(0).getMessage().getContent(); - if (!contentList.isEmpty()){ - System.out.println(contentList.get(0).get("text")); - }// - } catch (Exception e){ - System.exit(0); - } - }); - } - - public static void main(String[] args) { - try { - simpleMultiModalConversationCall(); - } catch (ApiException | NoApiKeyException | UploadFileException e) { - System.out.println(e.getMessage()); - } - System.exit(0); - } -} -``` - -## curl - -``` -# ======= 重要提示 ======= -# 各地域的API Key不同。获取API Key：https://help.aliyun.com/zh/model-studio/get-api-key -# 以下为北京地域base_url，若使用弗吉尼亚地域模型，需要将base_url换成：https://dashscope-us.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation -# 若使用新加坡地域的模型，需要将base_url换成：https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation -# === 执行时请删除该注释 === - -curl --location 'https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation' \ ---header "Authorization: Bearer $DASHSCOPE_API_KEY" \ ---header 'Content-Type: application/json' \ --H 'X-DashScope-SSE: enable' \ ---data '{ - "model": "qwen3.5-ocr", - "input":{ - "messages":[ - { - "role": "user", - "content": [ - { - "image": "https://img.alicdn.com/imgextra/i2/O1CN01ktT8451iQutqReELT_!!6000000004408-0-tps-689-487.jpg", - "min_pixels": 3072, - "max_pixels": 8388608 - }, - {"type": "text", "text": "请提取车票图像中的发票号码、车次、起始站、终点站、发车日期和时间点、座位号、席别类型、票价、身份证号码、购票人姓名。要求准确无误的提取上述关键信息、不要遗漏和捏造虚假信息，模糊或者强光遮挡的单个文字可以用英文问号?代替。返回数据格式以json方式输出，格式为：{'发票号码': 'xxx', '起始站': 'xxx', '终点站': 'xxx', '发车日期和时间点':'xxx', '座位号': 'xxx','票价':'xxx', '身份证号码': 'xxx', '购票人姓名': 'xxx'"} - ] - } - ] - }, - "parameters": { - "incremental_output": true - } -}' -``` - -**model** `_string_` **（必选）** - -模型名称。支持的模型可参见`[选择模型](https://help.aliyun.com/zh/model-studio/models#55c81ba3ccgct)`。 - -**messages** `_array_` **（必选）** - -传递给大模型的上下文，按对话顺序排列。 - -> 通过HTTP调用时，请将**messages** 放入 **input** 对象中。 - -**消息类型** - -User Message `_object_`**（必选）** - -用户消息，用于向模型传递问题、指令或上下文等。 - -**属性** - -**content** `_string 或 array_`**（必选）** - -消息内容。若输入只有文本，则为 string 类型；若输入包含图像数据，则为 array 类型。 - -**属性** - -**text** `_string_` **（可选）** - -输入的文本。 - -默认值为：`Please output only the text content from the image without any additional descriptions or formatting.` ，即模型默认提取图像中的全部文本。 - -**image** `_string_`（可选） - -图片的URL、 Base64 Data URL、或本地路径。传入本地文件请参见[传入本地文件](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#ea4e1d92dbry2)。 - -示例值：`{"image":"https://xxxx.jpeg"}` - -**enable\_rotate** `_boolean_` （可选）默认值为`false` - -是否对倾斜的图像进行校正处理。 - -可选值： - -- `true`：自动校正 - -- `false`：不进行校正 - - -示例值：`{"image":"https://xxxx.jpeg","enable_rotate": True}` - -**min\_pixels** `_integer_` （可选） - -用于设定输入图像的最小像素阈值，单位为像素。 - -当输入图像像素小于`min_pixels`时，会将图像进行放大，直到总像素高于`min_pixels`。 - -**图像Token与像素的转换关系** - -不同模型，每个图像 Token 对应的像素不同： - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：每 Token 对应像素为`32*32`。 - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：每 Token 对应像素为`28*28`。 - - -**min\_pixels 取值范围** - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：默认值和最小值均为3072（即`3×32×32`） - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：默认值和最小值均为 `3136` （即`4×28×28`）。 - - -示例值：`{"image":"https://xxxx.jpeg","min_pixels": 3072}` - -**max\_pixels** `_integer_` （可选） - -用于设定输入图像的最大像素阈值，单位为像素。 - -当输入图像像素在`[min_pixels, max_pixels]`区间内时，模型会按原图进行识别。当输入图像像素大于`max_pixels`时，会将图像进行缩小，直到总像素低于`max_pixels`。 - -**图像Token与像素的转换关系** - -不同模型，每个图像 Token 对应的像素不同： - -- `qwen3.5-ocr`、`qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`：每 Token 对应像素为`32*32`。 - -- `qwen-vl-ocr`、`qwen-vl-ocr-2025-08-28`及之前更新的模型：每 Token 对应像素为`28*28`。 - - -**max\_pixels 取值范围** - -- `qwen3.5-ocr、qwen-vl-ocr-latest、qwen-vl-ocr-2025-11-20` - - - 默认值：8388608 （即`8192x32x32`） - - - 最大值：30720000（即`30000x32x32`） - -- `qwen-vl-ocr、qwen-vl-ocr-2025-08-28`及之前更新的模型 - - - 默认值：6422528（即`8192x28x28`） - - - 最大值：23520000（即`30000x28x28`） - - -示例值：`{"image":"https://xxxx.jpeg","max_pixels": 8388608}` - -**role** `_string_` **（必选）** - -用户消息的角色，固定为`user`。 - -**max\_tokens** `_integer_` （可选） - -用于限制模型输出的最大 Token 数。若生成内容超过此值，响应将被截断。 - -- `qwen3.5-ocr`：默认值与最大值为32768。 - -- `qwen-vl-ocr-latest`、`qwen-vl-ocr-2025-11-20`、`qwen-vl-ocr-2024-10-28`默认值与最大值均为模型的最大输出长度，请参见[模型选型](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#f4299b0a1ace4)。 - -- `qwen-vl-ocr、qwen-vl-ocr-2025-04-13、qwen-vl-ocr-2025-08-28`，默认值和最大值为4096。 - - > 如需提高该参数值（4097~8192范围），请联系商务经理进行申请，并提供以下信息：主账号ID、图像类型（如文档图、电商图、合同等）、模型名称、预计 QPS 和每日请求总数，以及模型输出长度超过4096的请求占比。 - - -> Java SDK中为**maxTokens**_。_通过HTTP调用时，请将 **max\_tokens** 放入 **parameters** 对象中。 - -**ocr\_options** _object_ （可选） - -使用通义千问OCR模型[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)时需要配置的参数。调用内置任务时，无需传入`User Message`，模型内部会采用对应任务的`Prompt`。相关章节：[调用内置任务](https://help.aliyun.com/zh/model-studio/qwen-vl-ocr#1aae916a0br7o)。 - -**属性** - -**task** `_string_` （必选） - -内置任务的名称，可选值如下： - -- `text_recognition`：通用文字识别 - -- `key_information_extraction`：信息抽取 - -- `document_parsing`：文档解析 - -- `table_parsing`：表格解析 - -- `formula_recognition`：公式识别 - -- `multi_lan`：多语言识别 - -- `advanced_recognition`：高精识别 - - -**task\_config** `_object_` （可选） - -当`task`的取值为`key_information_extraction`（信息抽取）时，此参数用于指定需抽取的特定字段。如未指定 `task_config`，模型将默认提取图像中的所有字段。 - -**属性** - -**result\_schema** `_object_` （可选） - -表示需要模型抽取的字段，应为JSON对象结构，最多可嵌套3层JSON 对象。 - -在JSON对象的键（`key`）中指定待抽取字段的名称，对应的值（`value`）可为空，建议在值中提供字段描述或格式要求，可提高信息提取的准确率。 - -示例值： - -``` -"result_schema": { - "发票号码": "发票的唯一识别编号，通常为数字和字母的组合。", - "开票日期": "发票开具的日期，请以YYYY-MM-DD格式提取，例如2023-10-26。", - "销售方名称": "发票上显示的销售方公司全称。", - "总金额": "发票中包含税费的总计金额，要求提取数值并保留两位小数，例如123.45。" -} -``` - -> Java SDK为**OcrOptions**，DashScope Python SDK 最低版本为1.22.2， Java SDK 最低版本为2.18.4。 - -> 通过HTTP调用时，请将 **ocr\_options** 放入 **parameters** 对象中。 - -**seed** `_integer_` （可选） - -随机数种子。用于确保在相同输入和参数下生成结果可复现。若调用时传入相同的 `seed` 且其他参数不变，模型将尽可能返回相同结果。 - -取值范围：`[0,231−1]`。 - -> 建议设置为默认值即可。 - -> 通过HTTP调用时，请将 **seed** 放入 **parameters** 对象中。 - -**temperature** `_float_` （可选） 默认值为0.01 - -采样温度，控制模型生成文本的多样性。 - -temperature越高，生成的文本更多样，反之，生成的文本更确定。 - -取值范围： \[0, 2) - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> 建议设置为默认值即可。 - -> 通过HTTP调用时，请将 **temperature** 放入 **parameters** 对象中。 - -**top\_p** `_float_` （可选）默认值为0.001 - -核采样的概率阈值，控制模型生成文本的多样性。 - -top\_p越高，生成的文本更多样。反之，生成的文本更确定。 - -取值范围：（0,1.0\] - -temperature与top\_p均可以控制生成文本的多样性，建议只设置其中一个值。 - -> 建议设置为默认值即可。 - -> Java SDK中为**topP**_。_通过HTTP调用时，请将 **top\_p** 放入 **parameters** 对象中。 - -**top\_k** `_integer_` （可选）默认值为1 - -生成过程中采样候选集的大小。例如，取值为50时，仅将单次生成中得分最高的50个Token组成随机采样的候选集。取值越大，生成的随机性越高；取值越小，生成的确定性越高。取值为None或当top\_k大于100时，表示不启用top\_k策略，此时仅有top\_p策略生效。 - -取值需要大于或等于0。 - -该参数非OpenAI标准参数。通过 Python SDK调用时，请放入 **extra\_body** 对象中，配置方式为：`extra_body={"top_k": xxx}`；通过 Node.js SDK 或 HTTP 方式调用时，请作为顶层参数传递。 - -> 建议设置为默认值即可。 - -**repetition\_penalty** `_float_` （可选）默认值为1.0 - -模型生成时连续序列中的重复度。提高repetition\_penalty时可以降低模型生成的重复度，1.0表示不做惩罚。该参数对模型效果影响较大，建议保持默认值。 - -> 建议设置为默认值即可。 - -> Java SDK中为**repetitionPenalty**_。_通过HTTP调用时，请将 **repetition\_penalty** 放入 **parameters** 对象中。 - -**presence\_penalty** `_float_` （可选）默认值为0.0 - -控制模型生成文本时的内容重复度。 - -取值范围：\[-2.0, 2.0\]。正值降低重复度，负值增加重复度。 - -在创意写作或头脑风暴等需要多样性、趣味性或创造力的场景中，建议调高该值；在技术文档或正式文本等强调一致性与术语准确性的场景中，建议调低该值。 - -**原理介绍** - -如果参数值是正数，模型将对目前文本中已存在的Token施加一个惩罚值（惩罚值与文本出现的次数无关），减少这些Token重复出现的几率，从而减少内容重复度，增加用词多样性。 - -> 建议设置为默认值即可。 - -**stream** `_boolean_` （可选） 默认值为`false` - -是否流式输出回复。参数值： - -- false：模型生成完所有内容后一次性返回结果。 - -- true：边生成边输出，即每生成一部分内容就立即输出一个片段（chunk）。 - - -> 该参数仅支持Python SDK。通过Java SDK实现流式输出请通过`streamCall`接口调用；通过HTTP实现流式输出请在Header中指定`X-DashScope-SSE`为`enable`。 - -**incremental\_output** `_boolean_` （可选）默认为`false` - -在流式输出模式下是否开启增量输出。推荐您优先设置为`true`。 - -参数值： - -- false：每次输出为当前已经生成的整个序列，最后一次输出为生成的完整结果。 - - ``` - I - I like - I like apple - I like apple. - ``` - -- true（推荐）：增量输出，即后续输出内容不包含已输出的内容。您需要实时地逐个读取这些片段以获得完整的结果。 - - ``` - I - like - apple - . - ``` - - -> Java SDK中为**incrementalOutput**_。_通过HTTP调用时，请将 **incremental\_output** 放入 **parameters** 对象中。 - -**stop** `_string 或 array_` （可选） - -用于指定停止词。当模型生成的文本中出现`stop` 指定的字符串或`token_id`时，生成将立即终止。 - -可传入敏感词以控制模型的输出。 - -> stop为数组时，不可将`token_id`和字符串同时作为元素输入，比如不可以指定为`["你好",104307]`。 - -**logprobs** `_boolean_` （可选）默认值为 `false` - -是否返回输出 Token 的对数概率，可选值： - -- `true` - - 返回 - -- `false` - - 不返回 - - -支持的模型：qwen-vl-ocr-2025-04-13及之后更新的模型 - -> 通过HTTP调用时，请将 **logprobs** 放入 **parameters** 对象中。 - -**top\_logprobs** `_integer_` （可选）默认值为0 - -指定在每一步生成时，返回模型最大概率的候选 Token 个数。仅当 `logprobs` 为 `true` 时生效。 - -取值范围：\[0,5\] - -> Java SDK中为**topLogprobs**_。_通过HTTP调用时，请将 **top\_logprobs** 放入 **parameters** 对象中。 - -### chat响应对象（流式与非流式输出格式一致） - -``` -{"status_code": 200, - "request_id": "8f8c0f6e-6805-4056-bb65-d26d66080a41", - "code": "", - "message": "", - "output": { - "text": null, - "finish_reason": null, - "choices": [ - { - "finish_reason": "stop", - "message": { - "role": "assistant", - "content": [ - { - "ocr_result": { - "kv_result": { - "不含税价": "230769.23", - "发票代码": "142011726001", - "组织机构代码": "null", - "购买方名称": "蔡应时", - "销售方名称": "null" - } - }, - "text": "```json\n{\n \"不含税价\": \"230769.23\",\n \"发票代码\": \"142011726001\",\n \"组织机构代码\": \"null\",\n \"购买方名称\": \"蔡应时\",\n \"销售方名称\": \"null\"\n}\n```", - "processed_text": "```json\n{\n \"不含税价\": \"230769.23\",\n \"发票代码\": \"142011726001\",\n \"组织机构代码\": \"null\",\n \"购买方名称\": \"蔡应时\",\n \"销售方名称\": \"null\"\n}\n```" - } - ] - } - } - ], - "audio": null - }, - "usage": { - "input_tokens": 926, - "output_tokens": 72, - "characters": 0, - "image_tokens": 754, - "input_tokens_details": { - "image_tokens": 754, - "text_tokens": 172 - }, - "output_tokens_details": { - "text_tokens": 72 - }, - "total_tokens": 998 - } -} -``` - -**status\_code** `_string_` - -本次请求的状态码。200 表示请求成功，否则表示请求失败。 - -> Java SDK不会返回该参数。调用失败会抛出异常，异常信息为**status\_code**和**message**的内容。 - -**request\_id** `_string_` - -本次调用的唯一标识符。 - -> Java SDK返回参数为**requestId。** - -**code** `_string_` - -错误码，调用成功时为空值。 - -> 只有Python SDK返回该参数。 - -**output** `_object_` - -调用结果信息。 - -**属性** - -**text** `_string_` - -该参数当前固定为`null`。 - -**finish\_reason** `_string_` - -模型结束生成的原因。有以下情况： - -- 正在生成时为`null`； - -- 模型输出自然结束为`stop`； - -- 因生成长度过长而结束为`length`； - - -**choices** `_array_` - -模型的输出信息。 - -**属性** - -**finish\_reason** `_string_` - -有以下情况： - -- 正在生成时为`null`； - -- 因模型输出自然结束为`stop`； - -- 因生成长度过长而结束为`length`； - - -**message** `_object_` - -模型输出的消息对象。 - -**属性** - -**role** `_string_` - -输出消息的角色，固定为`assistant`。 - -**content** `_object_` - -输出消息的内容。 - -**属性** - -**ocr\_result** `_object_` - -当Qwen-OCR系列模型调用内置的信息抽取、高精识别任务时，输出的任务结果信息。 - -**属性** - -**kv\_result** `_array_` - -信息抽取任务的输出结果。 - -**words\_info** `_array_` - -高精识别任务的输出结果。 - -**属性** - -**rotate\_rect** `_array_` - -示例值：`[center_x, center_y, width, height, angle]` - -文字框的旋转矩形表示： - -- `center_x、center_y为文本框中心点坐标` - -- `width`为文本框宽度，`height`为高度 - -- `angle`为文本框相对于水平方向的旋转角度，取值范围为`[-90, 90]` - - -**location** `_array_` - -示例值：`[x1, y1, x2, y2, x3, y3, x4, y4]` - -文字框四个顶点的坐标，坐标顺序为左上角开起，按左上角→右上角→右下角→左下角的顺时针顺序排列。 - -**text** `_string_` - -文本行的内容 - -**text** `_string_` - -输出消息的内容。 - -**processed\_text** `_string_` - -对模型原始输出进行后处理的结果，自动删除重复片段等。当模型输出存在重复内容时，该字段提供清洗后的文本。 - -**logprobs** `_object_` - -当前 choices 对象的概率信息。 - -**属性** - -**content** `_array_` - -带有对数概率信息的 Token 数组。 - -**属性** - -**token** `_string_` - -当前 Token。 - -**bytes** `_array_` - -当前 Token 的 UTF‑8 原始字节列表，用于精确还原输出内容，在处理表情符号、中文字符时有帮助。 - -**logprob** `_float_` - -当前 Token 的对数概率。返回值为 null 表示概率值极低。 - -**top\_logprobs** `_array_` - -当前 Token 位置最可能的若干个 Token 及其对数概率，元素个数与入参的`top_logprobs`保持一致。 - -**属性** - -**token** `_string_` - -当前 Token。 - -**bytes** `_array_` - -当前 Token 的 UTF‑8 原始字节列表，用于精确还原输出内容，在处理表情符号、中文字符时有帮助。 - -**logprob** `_float_` - -当前 Token 的对数概率。返回值为 null 表示概率值极低。 - -**usage** `_object_` - -本次请求使用的Token信息。 - -**属性** - -**input\_tokens** `_integer_` - -输入 Token 数。 - -**output\_tokens** `_integer_` - -输出 Token 数。 - -**characters** `_integer_` - -该参数当前固定为0。 - -**input\_tokens\_details** `_object_` - -输入 Token 的细粒度分类。 - -**属性** - -**image\_tokens** `_integer_` - -模型输入的图像对应的 Token数。 - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**output\_tokens\_details** `_object_` - -输出 Token 的细粒度分类。 - -**属性** - -**text\_tokens** `_integer_` - -模型输入的文本对应的Token 数。 - -**total\_tokens** `_integer_` - -消耗的总 Token 数，为`input_tokens`与`output_tokens`的总和。 - -**image\_tokens** `_integer_` - -输入内容包含`image`时返回该字段。为用户输入图片内容转换成Token后的长度。 - -## **错误码** - -如果模型调用失败并返回报错信息，请参见[错误码](https://help.aliyun.com/zh/model-studio/error-code)进行解决。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md new file mode 100644 index 00000000..3bcbc5a9 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md @@ -0,0 +1,76 @@ +# 实时多模态交互流程 + +本文介绍实时多模态服务端和客户端的交互流程。 + +## VAD 模式 + +将[客户端事件](https://help.aliyun.com/zh/model-studio/client-events#af43722339yva)事件的`session.turn_detection` 设为`"server_vad"`以启用 VAD 模式。在 VAD 模式下，服务端对传入的音频进行语音活动检测，并在检测到作出响应。此模式适用于客户端到服务器始终发送音频的情况，也是当前的默认模式。 + +![server\_vad](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0520773571/p991064.svg) + +- 服务端在检测到语音开始时发送`input_audio_buffer.speech_started` 事件。 + +- 客户端随时可以选择通过发送 `input_audio_buffer.append` 事件将音频追加到缓冲区。 + +- 服务端在检测到语音结束时发送`input_audio_buffer.speech_stopped`事件。 + +- 服务端通过发送 `input_audio_buffer.committed` 事件来提交输入音频缓冲区。 + +- 服务端发送 `conversation.item.created` 事件，其中包含从音频缓冲区创建的用户消息项。 + + +### **工具调用流程** + +在 VAD 模式下，当服务端生成的响应需要调用工具时，遵循以下交互流程： + +![image.svg](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3014236771/p1067613.svg) + +- 服务端在检测到语音结束并生成响应时，识别到需要调用工具。 + +- 服务端发送 `response.function_call_arguments.delta` 事件，包含工具调用参数的增量数据。 + +- 服务端发送 `response.function_call_arguments.done` 事件，表示工具调用参数传递完成。 + +- 客户端执行工具调用并获取结果。 + +- 客户端通过 `conversation.item.create` 事件发送工具调用结果。 + +- 服务端自动基于工具调用结果生成响应。 + + +## **Manual 模式** + +将[客户端事件](https://help.aliyun.com/zh/model-studio/client-events#af43722339yva)事件的`session.turn_detection` 设为 null 以启用 Manual 模式。此模式下，客户端通过显式发送`input_audio_buffer.commit` 和`response.create`事件请求服务器响应。适用于按下即说场景，如聊天软件中的发送语音。 + +![manual](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/0520773571/p991066.svg) + +- 客户端可以通过发送 `input_audio_buffer.append` 事件将音频追加到缓冲区。 + +- 客户端通过发送 `input_audio_buffer.commit`事件来提交输入音频缓冲区。 该提交会在对话中创建一个新的用户消息项。 + +- 服务器通过发送 `input_audio_buffer.committed`事件进行响应。 + +- 客户端发送 `response.create` 事件，触发模型生成最终响应。 + +- 服务器通过发送 `conversation.item.created`事件进行响应。 + + +### **工具调用流程** + +![image.svg](https://help-static-aliyun-doc.aliyuncs.com/assets/img/zh-CN/3014236771/p1067740.svg) + +在 Manual 模式下，当服务端生成的响应需要调用工具时，遵循以下交互流程： + +- 客户端发送 `response.create` 事件后，服务端生成响应并识别到需要调用工具。 + +- 服务端发送 `response.function_call_arguments.delta` 事件，包含工具调用参数的增量数据。 + +- 服务端发送 `response.function_call_arguments.done` 事件，表示工具调用参数传递完成。 + +- 客户端执行工具调用并获取结果。 + +- 客户端通过 `conversation.item.create` 事件发送工具调用结果。 + +- 客户端发送 `response.create` 事件，触发模型生成最终响应。 + +- 服务端基于工具调用结果生成响应，并通过 `response.audio.delta` 或 `response.text.delta` 事件返回给客户端。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md index 8a7714db..e784e1ba 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-experience/tts-model.md @@ -54,7 +54,7 @@ ElevenLabs Multilingual v3 `qwen-audio-3.0-tts-plus`、`MiniMax/speech-2.8-hd` -`qwen-audio-3.0-tts-flash`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） +`qwen-audio-3.0-tts-plus`（声音复刻）、`cosyvoice-v3.5-plus`（声音设计）、`MiniMax/speech-2.8-hd`（声音复刻） - **使用标准语音合成**：当内置音色库能满足需求，希望快速上手、无需额外配置时。 @@ -89,7 +89,7 @@ ElevenLabs Multilingual v3 推荐模型 -`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` +`qwen-audio-3.0-tts-plus`、`qwen-audio-3.0-tts-flash`、`MiniMax/speech-2.8-hd` `cosyvoice-v3.5-plus`、`cosyvoice-v3.5-flash` @@ -145,7 +145,7 @@ Qwen-Audio-TTS WebSocket / HTTP -不支持 +支持 不支持 @@ -205,7 +205,7 @@ HTTP WebSocket / HTTP -不支持 +支持 不支持 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md index f4e3a07c..ecfee151 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/model-high-speed-inference/tpm-reservation.md @@ -232,6 +232,7 @@ glm-5.1 \[0, 32K)：输入 1.0 / 输出 1.0 \[32K, 200K\]：输入 1.33 / 输出 1.17 + deepseek-v4-pro diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/privacy-notice.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/privacy-notice.md deleted file mode 100644 index e576d7ca..00000000 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/security-and-compliance/privacy-notice.md +++ /dev/null @@ -1,13 +0,0 @@ -# 合规资质与隐私说明 - -## 合规资质 - -百炼以无保留意见通过 SOC 2 审计，标志着其在安全、可用性和保密性方面建立了严苛的控制体系，为全球关键业务的人工智能转型提供符合国际标准的可信保障。 - -参考：[合规文档中心](https://security.aliyun.com/compliance-repository?spm=5176.34248323.J_oxCbIPYwH8En90P61GCp_.1.6df649f1sVpkA8) - -## 隐私保护 - -阿里云严格保护数据隐私，绝不会将您的数据用于模型训练。同时，您在构建应用或训练大模型过程中传输的数据都会经过 AES-256（Advanced Encryption Standard，高级加密标准）加密，确保数据安全。 - -根据相关法律法规要求，阿里云百炼将存储模型与应用调用时产生的数据。详情请参见[《阿里云百炼服务协议》](https://terms.alicdn.com/legal-agreement/terms/common_platform_service/20230728213935489/20230728213935489.html?spm=a2ty02.30260209.aillm.1.d8bb74a10sknig)中关于数据处理、隐私和安全的条款。 diff --git a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md index a76bb634..6acc767a 100644 --- a/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md +++ b/skills/bailian-docs-llm-wiki/raw/model-user-guide/test-1/model-pricing.md @@ -16488,14 +16488,6 @@ qwen3.7-text-embedding 100万Token -qwen3.7-text-embedding - -中国内地 - -0.5元 - -100万Token - text-embedding-v4 > [Batch调用](https://help.aliyun.com/zh/model-studio/batch-interfaces-compatible-with-openai/)半价 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md index bb974955..61a3fa2f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/3d-generation.md @@ -1,61 +1,50 @@ # 3d generation -百炼平台的 3D 生成能力基于 Tripo 模型提供文生3D、单图生3D 和多图生3D 三种模式，支持带贴图/PBR 材质或无贴图的基础模型输出。该能力为异步任务，需通过 `task_id` 轮询获取结果，适用于华北2（北京）地域。详细实现细节请参考 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 +3D generation 是百炼平台提供的异步 3D 模型生成能力，支持文本、单图、多图三种输入方式，输出 GLB 格式的 3D 模型（含 PBR 材质或无贴图基础模型）。该能力基于 Tripo 模型实现，当前仅在华北2（北京）地域可用，且必须配置对应地域的 API Key。完整流程为“创建任务 → 轮询查询结果”，任务 ID 有效期为 24 小时。 ## 支持的模型/功能 -- **支持模型**： - - `Tripo/Tripo-P1.0`：专业版，最高输出 2 万面，速度快，适用于快速原型与轻量级应用。 - - `Tripo/Tripo-H3.1`：高精度版，最高输出 200 万面，支持 `geometry_quality: "ultra"`，适用于对几何精度要求高的场景。 -- **输入方式（互斥）**： - - 文生3D：通过 `input.prompt` 输入文本描述（最大 1024 字符）。 - - 单图生3D：通过 `input.image` 提供单张 JPEG/PNG 图像（分辨率 20–6000px，≤20MB）。 - - 多图生3D：通过 `input.images` 提供长度为 4 的数组，按「前、左、后、右」顺序排列；缺失视角用 `{}` 占位，实际有效图数需 ≥2。 +- **支持模型**：`Tripo/Tripo-P1.0`（专业版，最高 2 万面，速度快）和 `Tripo/Tripo-H3.1`（高精度版，最高 200 万面）。二者参数支持存在差异，详见 [原文标题](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)。 +- **输入方式**（三者互斥）： + - 文生3D：通过 `prompt` 字段传入中文/英文提示词（≤1024 字符）； + - 单图生3D：通过 `image` 字段传入单张 JPEG/PNG 公网 URL（分辨率 20–6000px，≤20MB）； + - 多图生3D：通过 `images` 数组传入 4 张图像（顺序为前/左/后/右），空视角用 `{}` 占位，实际有效图数为 2–4 张。 - **输出类型**： - - 默认返回 PBR 材质模型（`pbr_model_url`，GLB 格式）及预览图（`rendered_image_url`）。 + - 默认返回带 PBR 材质的 `pbr_model_url`（GLB）及预览图 `rendered_image_url`； - 无贴图模型需显式设置 `"texture": false, "pbr": false`，此时返回 `base_model_url`。 -- 全部功能均在 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中定义并验证。 + +> **注意**：文档中 `images` 数组长度固定为 4 且顺序严格定义为“前、左、后、右”，但示例中传入 2 张图时未明确是否允许跳过中间索引（如 `[img1, {}, img3, {}]`）。请以 [原文标题](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) 中“多图生3D模型（传入2张图）”小节为准，该写法是官方支持的合法形式。 ## 关键参数 | 参数 | 类型 | 必填 | 说明 | |------|------|------|------| -| `model` | string | ✓ | 固定为 `Tripo/Tripo-P1.0` 或 `Tripo/Tripo-H3.1` | -| `input.prompt` / `input.image` / `input.images` | string / object / array | ✓（三选一） | 仅允许一种输入方式，同时传入将报错 | -| `parameters.texture_quality` | string | ✗ | `"standard"`（默认）或 `"detailed"`；仅对带贴图任务生效 | -| `parameters.geometry_quality` | string | ✗ | 仅 `Tripo/Tripo-H3.1` 支持；`"standard"`（≤150万面）或 `"ultra"`（≤200万面） | -| `parameters.pbr` | boolean | ✗ | 默认 `true`；设为 `false` 时需同步设 `texture: false` 才能获得无贴图模型 | -| `parameters.texture` | boolean | ✗ | 默认 `true`；与 `pbr` 联动，详见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) | - -> **注意**：`pbr` 和 `texture` 的组合逻辑存在隐式依赖——当 `pbr=true` 时，系统强制启用贴图（即忽略 `texture=false`）。因此，**唯一生成无贴图模型的方式是同时设置 `"texture": false, "pbr": false`**。 +| `model` | string | 是 | 固定为 `Tripo/Tripo-P1.0` 或 `Tripo/Tripo-H3.1` | +| `input.prompt` / `input.image` / `input.images` | string / string / array | 条件必填 | 三者仅选其一；`images` 数组长度必须为 4，每项含 `type`（`jpeg`/`png`）和 `file_token`（公网 URL） | +| `parameters.texture_quality` | string | 否 | 可选 `standard`（默认）、`detailed`；仅对 `Tripo/Tripo-P1.0` 和 `Tripo/Tripo-H3.1` 均生效 | +| `parameters.geometry_quality` | string | 否 | 仅 `Tripo/Tripo-H3.1` 支持：`standard`（≤150 万面）、`ultra`（≤200 万面） | +| `parameters.pbr` | boolean | 否 | 默认 `true`；设为 `false` 时需同时设 `texture: false` 才能获得无贴图模型 | +| `parameters.texture` | boolean | 否 | 默认 `true`；与 `pbr` 联动，详见 [原文标题](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md) | ## 使用方式 -1. **开通与配置** - - 在[百炼控制台（华北2）](https://bailian.console.aliyun.com/cn-beijing/?tab=model#/model-market/all)搜索并开通 Tripo 模型。 - - 配置环境变量 `DASHSCOPE_API_KEY`，确保使用北京地域的 API Key（参见 [Tripo-3D模型生成](../../raw/model-api-reference/3d-generation/tripo-3d-generation-api-reference.md)）。 - -2. **创建异步任务** - - `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` - - 请求头必须包含： - - `Content-Type: application/json` - - `Authorization: Bearer $DASHSCOPE_API_KEY` - - `X-DashScope-Async: enable`（**缺失将报错**） - - 响应中提取 `task_id`（有效期 24 小时）。 - -3. **轮询查询结果** - - `GET https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` - - 建议间隔 ≥15 秒轮询，状态流转为 `PENDING → RUNNING → SUCCEEDED/FAILED`。 - - 成功响应中 `output.results[0]` 包含 `pbr_model_url` 或 `base_model_url`（链接有效期 2 小时，需及时下载）。 +1. **开通与配置**：在百炼控制台（华北2 北京地域）搜索并开通 Tripo 模型服务，[获取并配置 API Key](https://help.aliyun.com/zh/model-studio/get-api-key) 到环境变量。 +2. **创建异步任务**（POST）： + - URL：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation` + - 请求头：`Content-Type: application/json`、`Authorization: Bearer $DASHSCOPE_API_KEY`、`X-DashScope-Async: enable`（**必须**，否则报错） + - 请求体：按输入类型构造 `input`，并可选配置 `parameters` +3. **轮询查询结果**（GET）： + - URL：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/tasks/{task_id}` + - 建议间隔 ≥15 秒；`task_id` 有效期 24 小时；成功响应中 `output.results` 包含 `pbr_model_url` 或 `base_model_url`（链接有效期 2 小时，需及时下载） ## 限制和注意事项 -- **地域限制**：仅支持华北2（北京）地域，其他地域 URL 不可用。 -- **异步强制性**：所有调用必须启用 `X-DashScope-Async: enable`，不支持同步模式。 -- **task_id 生命周期**：创建后 24 小时内有效，超时查询返回 `task_status: "UNKNOWN"`。 -- **RPS 限制**：任务查询接口默认限流 20 QPS；高频轮询建议配置[异步回调](https://help.aliyun.com/zh/model-studio/async-task-api)。 -- **图像约束**：单图/多图输入均要求公网可访问 URL（HTTP/HTTPS），格式为 JPEG/PNG，单图 ≤20MB，多图各图独立校验。 -- **错误处理**：失败任务返回 `code` 和 `message`，需结合[统一错误码文档](https://help.aliyun.com/zh/model-studio/error-code)定位原因。 +- **地域强约束**：仅支持华北2（北京）地域，其他地域 URL 不可用，API Key 也必须为该地域生成。 +- **异步强制性**：所有调用必须启用 `X-DashScope-Async: enable`，同步调用不被支持。 +- **输入互斥性**：`prompt`、`image`、`images` 不能共存，否则返回 `InvalidParameter` 错误。 +- **图片要求**：单图/多图均需公网可访问 URL，格式为 JPEG/PNG，单图大小 ≤20MB，分辨率边长 ∈ [20, 6000]。 +- **任务管理**：任务状态流转为 `PENDING → RUNNING → SUCCEEDED/FAILED`；`UNKNOWN` 状态表示 task_id 过期或不存在；RPS 查询上限为 20，高频轮询建议配置[异步回调](https://help.aliyun.com/zh/model-studio/async-task-api)。 +- **资源时效性**：生成结果 URL（`pbr_model_url`/`base_model_url`/`rendered_image_url`）有效期仅 2 小时，务必及时下载保存。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md index 375aaaf5..0ebf58fc 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-call.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-call.md @@ -1,69 +1,71 @@ # application call -`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可选择 OpenAI 兼容的 Responses API 或原生 DashScope API 两种方式发起同步或异步请求，支持文本、图像、文件等[多模态](../concepts/multi-modal.md)输入，并可通过 `session_id` 或完整消息历史维护对话上下文。所有调用均需提供有效的 APP ID 及认证凭证。 +`application call` 是阿里云百炼平台提供的核心能力，用于通过 API 调用已发布的智能体（Agent）或工作流（Workflow）应用。开发者可选择 DashScope 原生协议或 OpenAI 兼容的 Responses API 两种方式发起请求，支持同步、异步及[流式输出](../concepts/streaming-output.md)等多种交互模式。所有调用均需提供有效的 APP ID 和 API Key，并根据应用部署位置决定是否携带 Workspace ID。 ## 支持的模型/功能 -- **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流应用；其中文件输入仅限智能体应用，且需在应用配置中启用“全文引用”或“切片检索”[新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 -- **[多模态](../concepts/multi-modal.md)能力**： - - 图像输入：需选用通义千问 VL 系列模型，并在应用中正确配置图像处理方式 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)； - - 文件输入：仅智能体应用支持，依赖应用内文件处理方式配置； - - [流式输出](../concepts/streaming-output.md)：仅同步调用支持，且工作流应用需在结束节点启用“[流式输出](../concepts/streaming-output.md)”开关并重新发布 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 -- **会话管理**： - - DashScope API 通过 `session_id` 维护上下文，有效期为最后一次请求后 1 小时； - - Responses API 当前不支持 `pre_response_id` 或 `conversation_id`，需在每次请求中传递完整消息历史 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 +- **应用类型**：支持新版智能体（Agent 2.0）、旧版智能体及工作流三类应用，详见 [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) 和 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 +- **输入模态**： + - 文本（单轮/多轮对话） + - 图像（需选用通义千问 VL 系列模型并配置自定义处理或 `imageList` 入参） + - 文件（仅智能体应用支持，需配置为“全文引用”或“切片检索”） +- **输出模式**： + - 同步响应（默认） + - [流式输出](../concepts/streaming-output.md)（`stream=true`，仅同步调用支持，且工作流应用需在结束节点启用流式开关） + - 异步任务（`background=true`，返回任务 ID 后轮询结果，[异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)） -> **注意**：文档 4 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 4 明确限定为“新版智能体应用”，而文档 5 泛指“智能体、工作流应用”。实际调用时，该接口对两类应用均有效，但功能支持（如 `session_id` 行为、参数结构）以应用类型和发布配置为准，建议优先参考 [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md)。 +> **注意**：文档 2 和文档 5 均描述了 `/api/v1/apps/{APP_ID}/completion` 接口，但文档 2 明确限定“仅适用于华北2（北京）地域”，而文档 5 未声明地域限制，实际生产中应以文档 2 的约束为准，避免跨地域调用失败。 ## 关键参数 -| 参数名 | 类型 | 必选 | 说明 | +| 参数名 | 类型 | 必填 | 说明 | |--------|------|------|------| -| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/#/app-center)页面获取。若应用位于子业务空间，还需传入 `workspace_id` [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 | -| `input` / `prompt` | string 或 array | 是 | 核心输入内容：
- DashScope API 使用 `prompt` 字符串；
- Responses API 支持字符串（单轮）或消息数组（多轮/[多模态](../concepts/multi-modal.md)），数组元素含 `role`（`user`/`system`/`assistant`）与 `content`（支持 `input_text`/`input_image`/`input_file`）[同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 | -| `stream` | boolean | 否 | 仅 Responses API 支持。设为 `true` 启用[流式输出](../concepts/streaming-output.md)，需应用端配合启用流式开关 [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md)。 | -| `background` | boolean | 否 | 仅 Responses API 支持。设为 `true` 进入异步模式，立即返回任务 ID，后续通过 `retrieve` 查询结果 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | -| `biz_params` | object | 否 | Responses API 中用于传递工作流或智能体应用内定义的自定义参数，键名需与应用配置完全一致 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 | -| `session_id` | string | 否 | DashScope API 多轮对话必需。首次调用不传，响应中返回；后续调用需携带上一轮返回的 `session_id` [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md)。 | +| `app_id` | string | 是 | 应用唯一标识，在[应用管理](https://bailian.console.aliyun.com/#/app-center)中获取。若应用位于子业务空间或德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域，还需传入 `workspace_id`（见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)） | +| `input` | string / array | 是 | 核心输入内容：
• 字符串：单轮文本，如 `"你好"`
• 消息数组：支持多轮对话、图像（`input_image`）、文件（`input_file`）等多模态输入；**注意**：基于 `pre_response_id` 或 `conversation_id` 的上下文功能暂不支持，需显式传递完整历史 | +| `stream` | boolean | 否 | 是否[流式输出](../concepts/streaming-output.md)，默认 `false`；设为 `true` 时需配合 SDK 的 chunk 迭代处理 | +| `background` | boolean | 否 | 是否异步执行，默认 `false`；设为 `true` 时立即返回任务 ID，不可与 `stream=true` 同时使用 | -## 使用方式 +- **DashScope SDK 方式**（如 Python）：使用 `prompt` 字段传递文本输入，`session_id` 维持会话状态。 +- **Responses API 方式**（OpenAI 兼容）：统一使用 `input` 字段，结构更灵活，支持多模态。 -### 1. 认证与端点 -- **API Key**：通过[密钥管理](https://bailian.console.aliyun.com/?tab=app#/api-key)获取，并推荐配置为环境变量 `DASHSCOPE_API_KEY`。 -- **Base URL / Endpoint**： - - Responses API（OpenAI 兼容）：`https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/`（同步/异步共用）； - - DashScope API（原生）：`https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`。 +## 使用方式 -### 2. 调用示例 -- **同步调用（Responses API）**： - ```python - from openai import OpenAI - client = OpenAI(api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url=f"https://dashscope.aliyuncs.com/api/v2/apps/agent/{app_id}/compatible-mode/v1/") - response = client.responses.create(input="你好") - ``` -- **异步调用（Responses API）**： - 设置 `background=True`，获取 `task_id` 后轮询 `retrieve` 接口 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 -- **DashScope SDK 调用**： +### 1. DashScope 原生接口（推荐用于新应用） +- **Endpoint**：`POST https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion` +- **认证**：`Authorization: Bearer {DASHSCOPE_API_KEY}` +- **示例（Python SDK）**： ```python from dashscope import Application - response = Application.call(api_key=..., app_id=..., prompt="你好") + response = Application.call( + api_key=os.getenv("DASHSCOPE_API_KEY"), + app_id="your-app-id", + prompt="你是谁？" + ) ``` +### 2. OpenAI 兼容 Responses API（适合迁移现有代码） +- **同步 Endpoint**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses` +- **异步 Endpoint**：`POST https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`（请求体含 `"background": true`） +- **认证**：同上，`base_url` 需拼接 `{APP_ID}` + +### 3. 多轮对话 +- **DashScope 方式**：首次调用不传 `session_id`，响应中返回 `session_id`；后续请求携带该值即可延续上下文。 +- **Responses API 方式**：将完整消息历史数组（含 `role: "user"/"assistant"`）传入 `input`，无需维护 `session_id`。 + ## 限制和注意事项 -- **地域限制**：Responses API（同步/异步）与 DashScope API 均**仅支持华北2（北京）地域**，其他地域（如德国法兰克福、新加坡）调用需显式传入 `workspace_id` 并确认 Base URL [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 -- **异步限制**：异步调用不支持 `stream=true`，且暂无流式输出能力 [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md)。 -- **凭证获取**：APP ID 和 Workspace ID **仅支持控制台手动获取**，不提供 API 或 CLI 查询接口 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 -- **权限要求**：查询所有业务空间 ID 需主账号或具备 `AliyunBailianFullAccess` 权限的 RAM 子账号，普通子账号仅能查看已加入的业务空间 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)。 -- **参数兼容性**：`biz_params` 仅在 Responses API 中生效；DashScope API 的自定义参数需通过 `parameters` 字段传递（文档未明确示例，以 SDK 实际行为为准）。 +- **地域限制**：DashScope 原生接口（文档 2、5）和 Responses API（文档 3、4）均明确标注“仅适用于华北2（北京）地域”，其他地域需确认控制台 Base URL 是否适配。 +- **Workspace ID 规则**：调用子业务空间下的应用，或德国（法兰克福）、华北2（北京）、新加坡、日本（东京）地域的模型时，必须提供 `workspace_id`；默认业务空间下仅需 `app_id`（见 [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md)）。 +- **权限要求**：RAM 子账号查询所有业务空间 ID 需具备 `AliyunBailianFullAccess` 或 `AliyunBailianControlFullAccess` 权限，否则仅能查看当前登录空间 ID。 +- **异步限制**：异步任务不支持流式输出（`stream=true` 与 `background=true` 互斥），且暂无内置取消接口，需依赖轮询状态后手动处理。 +- **凭证安全**：API Key **严禁硬编码**，务必通过环境变量（如 `DASHSCOPE_API_KEY`）注入，避免泄露风险。 ## 来源文档 - [获取APP ID和Workspace ID](../../raw/application-api-reference/application-call/obtain-the-app-id-and-workspace-id.md) +- [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - [异步调用API参考](../../raw/application-api-reference/application-call/openai-responses-api/asynchronous-call-api-reference.md) - [同步调用 API 参考](../../raw/application-api-reference/application-call/openai-responses-api/synchronous-call-api-reference.md) -- [新版智能体应用 API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/new-agent-application-api-reference.md) - [应用 DashScope API 参考](../../raw/application-api-reference/application-call/application-dashscope-api-reference/agent-and-workflow-application-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md index 76e8a143..905618f0 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/application-component-api-reference.md @@ -1,61 +1,51 @@ # application component api reference -本 API 参考文档面向开发者，系统性地描述了百炼平台 Application Component（应用组件）提供的核心 OpenAPI 能力，涵盖数据连接（应用数据）、知识库、[Prompt 工程](../concepts/prompt-engineering.md)及辅助功能等模块。所有接口均基于 `bailian/2023-12-29` 版本，采用 ROA 签名机制，支持多语言 SDK 封装调用。开发者需通过 RAM 子账号配合最小权限策略进行安全接入。 +本 API 参考文档面向开发者，提供百炼平台 Application Component（应用组件）层核心能力的标准化接口说明，涵盖数据连接（原应用数据）、知识库、[Prompt 工程](../concepts/prompt-engineering.md)三大功能域。所有接口均基于 ROA 签名机制，推荐使用官方 SDK 调用以简化鉴权与序列化流程。API 版本为 `2023-12-29`，适用于生产环境集成。 ## 支持的模型/功能 -Application Component 提供以下四类核心能力： +Application Component API 主要支撑三类核心能力： -- **数据连接（原应用数据）**：管理非结构化文件与结构化表格。支持类目（Category）增删查、文件上传（`ApplyFileUploadLease` + `AddFile`）、OSS 批量导入（`AddFilesFromAuthorizedOss`）、解析器配置（`GetAvailableParserTypes`, `ChangeParseSetting`）及连接器管理（`AddConnector`, `GetConnector`）。注意：API 不支持直接操作数据表（如新增/删除表），该功能仅限控制台，详见 [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) 文档说明。 -- **知识库（Knowledge Base）**：支持创建（`CreateIndex`）、更新（`UpdateIndex`）、删除（`DeleteIndex`）知识库；向知识库追加文档（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）、查询文件列表（`ListIndexDocuments`）及删除知识库内文件（`DeleteIndexDocument`）。知识库类型包括文档/音视频（非结构化）和数据查询/图片问答（结构化）两类。 -- **[Prompt 工程](../concepts/prompt-engineering.md)**：提供完整的 Prompt 模板生命周期管理，包括创建（`CreatePromptTemplate`）、获取（`GetPromptTemplate`）、更新（`UpdatePromptTemplate`）、删除（`DeletePromptTemplate`）及列表查询（`ListPromptTemplates`）。 -- **辅助功能**：包含支付宝打赏状态查询（`GetAlipayTransferStatus`）、临时存储租约申请（`ApplyTempStorageLease`）等场景化能力。 +- **数据连接（原应用数据）**：提供类目（Category）、文件（File）、表格（Table）、连接器（Connector）的全生命周期管理，包括创建、查询、更新、删除及解析配置。支持从本地临时存储或已授权 OSS Bucket 导入文件，也支持结构化表格数据接入。详细能力见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 +- **知识库（Knowledge Base）**：支持非结构化（文档/音视频）与结构化（数据查询/图片问答）两类知识库的创建、更新、监控及内容管理。关键操作包括知识库构建（`CreateIndex` + `SubmitIndexJob`）、文件追加（`SubmitIndexAddDocumentsJob`）、检索（`Retrieve`）以及细粒度的分片（Chunk）管理（`ListChunks`/`UpdateChunk`/`DeleteChunk`）。注意 `Retrieve` 接口在 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) 中定义了地域级公网与 VPC 入口。 +- **[Prompt 工程](../concepts/prompt-engineering.md)**：提供 Prompt 模板的增删改查（`CreatePromptTemplate`/`GetPromptTemplate`/`UpdatePromptTemplate`/`DeletePromptTemplate`/`ListPromptTemplates`），支持变量注入与模板分类（System/Custom），但明确不支持文生图类 Prompt 模板创建（见 [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md)）。 -> **注意**：文档 33 (`AddChunk`) 明确指出“目前尚不支持对音视频搜索类（multimedia）知识库进行相关操作”，而文档 34 (`CreateIndex`) 则将知识库类型划分为“基于文档或音视频的非结构化知识库”与“用于数据查询或图片问答的结构化知识库”。二者存在表述矛盾——实际能力以 `CreateIndex` 的分类为准，`AddChunk` 接口当前仅适用于 `document`、`table` 和 `image` 类型知识库，不支持 `multimedia` 类型。 +> **注意**：文档 1 和文档 4 均提及 `CreateIndex` 接口参数变更，但未明确具体字段差异；实际开发中请以 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) 提供的最新 SDK 或 OpenAPI Explorer 生成的示例为准，避免因版本描述模糊导致调用失败。 ## 关键参数 -- **通用路径参数**：几乎所有接口均需 `WorkspaceId`（业务空间 ID），用于标识资源归属。其值可通过控制台或 `ListWorkspace` 接口获取。 -- **身份认证**：所有请求必须携带有效的 AccessKey（建议使用 RAM 子账号并遵循最小权限原则），签名方式为 ROA。详细准备流程见 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md)。 -- **分页与幂等**： - - 分页接口（如 `ListCategory`, `ListFile`, `ListPromptTemplates`）统一使用 `MaxResults`（每页最大条数）和 `NextToken`（下一页凭证）实现。 - - 幂等性接口（如 `ListCategory`, `DescribeFile`, `Retrieve`）可安全重试；非幂等接口（如 `AddCategory`, `ApplyFileUploadLease`）重复调用可能导致重复资源创建或失败。 -- **关键业务参数**： - - 文件操作：`CategoryId`（类目 ID）、`FileId`（文件 ID）、`LeaseId`（上传租约 ID）是串联 `ApplyFileUploadLease` → `AddFile` 流程的核心。 - - 知识库操作：`IndexId`（知识库 ID）、`JobId`（任务 ID）是关联 `CreateIndex` → `SubmitIndexJob` → `GetIndexJobStatus` 的关键。 - - 解析器：`Parser`（如 `DOCMIND`, `AUTO_SELECT`）在 `AddFile` 中指定；`FileType`（如 `pdf`, `docx`）在 `GetAvailableParserTypes` 中用于查询支持类型。 +- **WorkspaceId（业务空间 ID）**：几乎所有接口的路径参数，标识资源所属的隔离单元。必须通过控制台或 `ListWorkspace`（未在原始文档中列出）获取，不可猜测。 +- **CategoryId / FileId / IndexId / ConnectorId / PromptTemplateId**：各资源的唯一标识符，通常由对应 `Add*` 或 `Create*` 接口返回，或在控制台 UI 中点击 ID 图标复制。例如 `DescribeFile` 必须传入 `FileId`，而 `DeleteIndexDocument` 则需 `IndexId` 和 `DocumentIds`（即 `FileId` 数组）。 +- **Parser（解析器类型）**：`AddFile` 接口的关键请求参数，决定文件内容如何被结构化处理。可选值包括 `DOCMIND`、`DOCMIND_DIGITAL`、`AUTO_SELECT` 等，具体支持列表需结合 `GetAvailableParserTypes` 动态查询。 +- **权限策略 Action**：所有操作均映射到 RAM 权限点（如 `sfm:AddFile`, `sfm:Retrieve`），必须在 RAM 策略中显式授予。策略结构详见 [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md)，且需遵循最小权限原则。 ## 使用方式 -1. **环境准备**：按 [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) 文档指引，创建 RAM 用户、配置 `AliyunBailianDataFullAccess` 或更细粒度策略，并加入目标业务空间。 -2. **接入点选择**：根据部署地域选择对应服务接入点，例如华北2（北京）公网地址为 `bailian.cn-beijing.aliyuncs.com`，VPC 地址为 `bailian-vpc.cn-beijing.aliyuncs.com`，完整列表见 [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)。 -3. **SDK 调用（推荐）**：下载并初始化最新版 [阿里云百炼 SDK](https://api.aliyun.com/api-tools/sdk/bailian?version=2023-12-29)，传入 `AccessKeyId`, `AccessKeySecret`, `RegionId`（如 `cn-beijing`）及 `WorkspaceId` 即可调用各接口方法，无需手动签名。 -4. **自签名调用（备选）**：若需自定义签名，务必参考 ROA 机制文档，并强烈建议加入钉钉群（147535001692）获取技术支持，避免因签名错误导致调试周期过长。 -5. **典型流程示例（文件导入知识库）**： - - 调用 `ApplyFileUploadLease` 获取 `LeaseId`； - - 使用 `LeaseId` 和 OSS URL 上传文件至临时存储； - - 调用 `AddFile` 将文件导入应用数据，并指定 `Parser`； - - 调用 `CreateIndex` 创建知识库； - - 调用 `SubmitIndexAddDocumentsJob` 将已导入的文件追加至知识库； - - 调用 `GetIndexJobStatus` 轮询任务状态，直至完成。 +1. **准备凭证**：使用 RAM 子账号（非主账号）并配置 `AliyunBailianDataFullAccess` 或更细粒度策略（如仅 `sfm:ListFile`），同时确保该子账号已加入目标业务空间。 +2. **选择接入点**：根据部署地域选用对应服务地址，例如华北2（北京）为 `bailian.cn-beijing.aliyuncs.com`（[服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md)）。 +3. **调用流程**： + - 数据接入：`ApplyFileUploadLease` → 上传文件至临时存储 → `AddFile`（指定 `LeaseId` 和 `Parser`）→ `ListFile` 查询状态。 + - 知识库构建：`CreateIndex` → `SubmitIndexJob` → `GetIndexJobStatus` 轮询直至 `FINISH` → `Retrieve` 检索。 + - Prompt 管理：直接调用 `CreatePromptTemplate` 等 RESTful 接口，无需前置步骤。 +4. **调试与生成代码**：所有接口均支持在 [OpenAPI Explorer](https://api.aliyun.com) 中在线调试，并自动生成多语言 SDK 示例。 ## 限制和注意事项 -- **限流策略**：各接口有独立 QPS 限制，例如 `ListCategory`/`DeleteCategory` 为 5 次/秒，`ApplyFileUploadLease`/`AddFile` 为 10 次/秒，`Retrieve` 为 15 次/秒。超出限制将返回 HTTP 429，需实现退避重试逻辑。 -- **权限隔离**：RAM 用户必须被显式授权（如 `sfm:ListCategory`）并加入业务空间后才能调用对应接口；主账号默认拥有全部权限。 -- **数据一致性**： - - `DeleteFile` 仅删除应用数据中的文件，不影响已构建的知识库内容；反之，`DeleteIndexDocument` 仅删除知识库索引，不影响应用数据源。 - - `AddFilesFromAuthorizedOss` 要求 OSS Bucket 与百炼同属一个阿里云主账号，并已完成跨服务授权。 -- **版本兼容性**：API 入参与返回结构可能随版本变更，例如 `DescribeFile` 在 2026-01-15 发生返回结构变更，`CreateIndex` 在 2026-03-27 和 2026-03-30 均有入参变更。开发者应关注 [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) 中的变更集，及时适配。 +- **限流策略**：各接口有独立 QPS 限制（如 `AddCategory` 5次/秒，`Retrieve` 无明确文档限制但建议合理重试），超限将返回 HTTP 429。务必实现客户端退避逻辑。 +- **幂等性**：`List*`、`Describe*`、`Get*` 类查询接口及 `Delete*` 类删除接口普遍具备幂等性；`Add*`、`Create*`、`Submit*` 等写入接口则不具备，重复调用可能产生冗余资源。 +- **数据一致性**：`DeleteFile` 仅删除应用数据中的文件，不影响已构建的知识库；反之，`DeleteIndexDocument` 仅删除知识库索引，不删除源文件。二者作用域分离，需按需调用。 +- **功能边界**：API 不支持直接创建/删除数据表（见 `AddCategory` 和 `DeleteFile` 文档说明），也不支持通过 API 更新数据查询类知识库（见 `SubmitIndexAddDocumentsJob` 文档说明），此类操作必须通过控制台完成。 +- **安全要求**：必须使用 HTTPS，AccessKey 需严格保密。强烈建议遵循 [账号与安全准备](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) 中的 RAM 最小权限实践，禁用主账号密钥。 ## 来源文档 -- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [API概览](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-overview.md) +- [服务接入点](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-endpoint.md) - [授权信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-ram.md) - [版本说明](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-changeset.md) -- [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [DeleteCategory - 删除类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletecategory.md) +- [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) +- [ListCategory - 类目列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-listcategory.md) - [ApplyFileUploadLease - 申请文件上传租约](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-applyfileuploadlease.md) - [AddFile - 添加文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfile.md) - [AddFilesFromAuthorizedOss - 从已授权OSS Bucket中导入文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addfilesfromauthorizedoss.md) @@ -63,50 +53,47 @@ Application Component 提供以下四类核心能力： - [DescribeFile - 查询文件状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-describefile.md) - [UpdateFileTag - 更新文件标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatefiletag.md) - [BatchUpdateFileTag - 批量更新文档标签](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-batchupdatefiletag.md) -- [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) -- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) -- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [GetParseSettings - 获取类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getparsesettings.md) +- [GetAvailableParserTypes - 获取文件支持的解析器类型](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getavailableparsertypes.md) - [ChangeParseSetting - 修改类目解析设置](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-changeparsesetting.md) -- [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) +- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) +- [DeleteFile - 删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefile.md) +- [DeleteFiles - 批量删除文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-deletefiles.md) - [AddConnector - 新增连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addconnector.md) -- [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) -- [UpdateConnector - 编辑连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) +- [UpdateTableFromAuthorizedOss - 从已授权OSS Bucket中选择文件更新表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updatetablefromauthorizedoss.md) - [CreatePromptTemplate - 创建Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-createprompttemplate.md) - [GetPromptTemplate - 获取Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-getprompttemplate.md) - [UpdatePromptTemplate - 更新Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-updateprompttemplate.md) - [DeletePromptTemplate - 删除Prompt模板](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-deleteprompttemplate.md) - [ListPromptTemplates - 获取Prompt模板列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-prompt-engineering/api-bailian-2023-12-29-listprompttemplates.md) -- [AddCategory - 新增类目](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addcategory.md) -- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) -- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) -- [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) -- [AddTable - 添加表格](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-addtable.md) -- [AddChunk - 新增切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md) - [CreateIndex - 创建知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-createindex.md) -- [SubmitIndexAddDocumentsJob - 提交知识库追加任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) +- [GetConnector - 获取连接器信息](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-getconnector.md) - [GetIndexJobStatus - 查询知识库创建任务状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexjobstatus.md) -- [Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) +- [SubmitIndexAddDocumentsJob - 提交知识库追加任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexadddocumentsjob.md) +- [UpdateConnector - 编辑连接器](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-data-connection-original-application-data/api-bailian-2023-12-29-updateconnector.md) - [ListIndexDocuments - 查询知识库下的文件列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexdocuments.md) -- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) +- [Retrieve - 检索知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-retrieve.md) - [UpdateIndex - 更新知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updateindex.md) +- [ListIndexFileDetails - 查询知识库下的文件详情](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindexfiledetails.md) - [DeleteIndexDocument - 删除知识库下的文件](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindexdocument.md) - [DeleteIndex - 删除知识库](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deleteindex.md) -- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - [ListChunks - 查询索引下的分片列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listchunks.md) - [UpdateChunk - 修改切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-updatechunk.md) - [DeleteChunk - 删除切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-deletechunk.md) - [GetIndexMonitor - 获取知识库监控数据](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-getindexmonitor.md) +- [GetAlipayTransferStatus - 查询支付宝打赏状态](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipaytransferstatus.md) +- [GetAlipayUrl - 获取支付宝打赏URL](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-getalipayurl.md) +- [ApplyTempStorageLease - 申请临时文件上传许可](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-applytempstoragelease.md) +- [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) +- [AddChunk - 新增切片](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-addchunk.md) - [CreateMemory - 创建长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememory.md) - [UpdateMemory - 更新长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememory.md) -- [GetMemory - 获取长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemory.md) - [DeleteMemory - 删除长期记忆体](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememory.md) +- [ListIndices - 查询知识库列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-listindices.md) - [ListMemories - 获取长期记忆体列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemories.md) -- [CreateMemoryNode - 创建记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-creatememorynode.md) +- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) - [UpdateMemoryNode - 更新记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-updatememorynode.md) - [DeleteMemoryNode - 删除记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-deletememorynode.md) - [ListMemoryNodes - 获取记忆片段列表](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-listmemorynodes.md) -- [SubmitIndexJob - 提交知识库创建任务](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-knowledge-base/api-bailian-2023-12-29-submitindexjob.md) -- [GetMemoryNode - 获取记忆片段](../../raw/application-api-reference/application-component-api-reference/api-bailian-2023-12-29-dir/api-bailian-2023-12-29-dir-others/api-bailian-2023-12-29-dir-long-term-memory/api-bailian-2023-12-29-getmemorynode.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md index 7dcd9872..0b70a782 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/file-management-api.md @@ -1,45 +1,50 @@ # file management api -文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询、列举和删除。该 API 与模型推理解耦，适用于预处理数据、知识库文档、提示词附件等场景。所有操作均需通过 `Authorization` 头携带 Bearer [Token](../concepts/token.md) 进行身份认证。 +文件管理 API 提供对百炼平台托管文件的全生命周期操作能力，包括上传、查询详情、列举账户下所有文件及删除指定文件。该 API 与模型调用解耦，不参与推理流程，仅用于文件资源管理。所有操作均需通过 HTTP REST 接口调用，并使用标准的 `Authorization: Bearer ` 认证。 ## 支持的模型/功能 -文件管理 API 不依赖具体大模型，是平台级基础设施能力，所有接入百炼的模型（如 Qwen 系列、Baichuan、GLM 等）均可复用已上传文件的 `file_id`。支持的核心功能包括： -- `POST /v1/files`：上传文件（支持 `multipart/form-data`） -- `GET /v1/files/{file_id}`：获取单个文件元信息 -- `GET /v1/files`：分页列举当前项目下的全部文件（支持 `purpose` 过滤） -- `DELETE /v1/files/{file_id}`：删除指定文件（不可恢复） +文件管理 API **不依赖任何大模型**，也不关联特定模型版本；其功能独立于 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 所述的底层存储服务。当前支持四项核心功能： +- `POST /v1/files`：上传文件（支持 `text/plain`、`application/json`、`application/pdf`、`.csv`、`.xlsx` 等格式） +- `GET /v1/files/{file_id}`：查询单个文件元信息（含状态、大小、上传时间等） +- `GET /v1/files`：分页列举当前 API Key 所属账户下的全部文件 +- `DELETE /v1/files/{file_id}`：永久删除指定文件（不可恢复） -> **注意**：部分旧版 SDK 文档中提及的 `purpose=assistants` 已废弃，实际仅支持 `purpose=vision`（用于[多模态](../concepts/multi-modal.md)输入）和 `purpose=embedding`（用于向量检索），详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 +> **注意**：部分旧版文档中提及“文件可绑定至特定模型实例”，该描述已过时；实际文件为全局账户级资源，与模型实例无绑定关系，详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 ## 关键参数 | 参数 | 位置 | 类型 | 必填 | 说明 | |------|------|------|------|------| -| `file` | form-data | binary | 是 | 待上传文件，最大 200 MB | -| `purpose` | form-data | string | 否 | 取值为 `vision` 或 `embedding`；默认为 `embedding`；[文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md) 明确不支持其他值 | -| `file_id` | path | string | 是（除上传外） | 平台生成的唯一文件标识，格式为 `file_...` | -| `limit`, `after` | query | integer/string | 否 | 分页参数，`limit` 默认 20，最大 100 | +| `file` | `multipart/form-data` body | File | 是 | 待上传的二进制文件对象（`POST /v1/files` 专用） | +| `purpose` | `multipart/form-data` body | string | 否 | 当前仅支持 `"assistants"`（用于后续与助手功能集成），其他值将被忽略 | +| `file_id` | URL path | string | 是（`GET/DELETE /v1/files/{file_id}`） | 文件唯一标识符，由平台生成并返回于上传响应中 | +| `limit` | query | integer | 否 | 列举时每页最大条目数，默认 20，上限 100 | +| `after` | query | string | 否 | 分页游标，值为上一页响应中的 `last_file_id` | ## 使用方式 -1. **上传文件**（示例）： +1. **上传文件**（示例 cURL）： ```bash curl -X POST "https://dashscope.aliyuncs.com/api/v1/files" \ - -H "Authorization: Bearer $API_KEY" \ - -F "file=@report.pdf" \ - -F "purpose=embedding" + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -F "file=@/path/to/document.pdf" \ + -F "purpose=assistants" ``` -2. 响应返回 `file_id` 和 `status=uploaded`，后续调用模型时可直接在 `input.files` 或 `messages.content` 中引用； -3. 列举文件时建议按 `purpose` 过滤，避免混用不同用途的文件；详情参见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 + 成功响应返回 `200 OK` 及包含 `id`、`filename`、`status`（通常为 `"uploaded"`）的 JSON 对象。 + +2. **查询与列举**：直接构造 GET 请求，无需额外 body；响应中 `status` 字段可能为 `"uploaded"`、`"processing"` 或 `"error"`，仅 `uploaded` 状态文件可用于后续功能（如知识库构建）。 + > **注意**：`processing` 状态表示文件正在解析（如 PDF 文本提取），此过程异步完成，轮询间隔建议 ≥2s；详见 [文件管理 (raw/model-api-reference/file-management-api.md)](../../raw/model-api-reference/file-management-api.md)。 + +3. **删除文件**：发送 DELETE 请求后，文件立即从列表中移除且无法恢复，但后台清理可能延迟数秒。 ## 限制和注意事项 -- 单文件大小上限为 200 MB，超限将返回 `400 Bad Request`； -- 同一 `file_id` 仅在 7 天内有效（若未被任何任务引用），之后自动清理； -- 删除操作立即生效且不可撤销，生产环境建议先调用 `GET /v1/files/{file_id}` 确认状态； -- 文件内容不支持修改，如需更新请重新上传并使用新 `file_id`； -- `purpose=vision` 仅支持图片（JPEG/PNG/WebP）和 PDF（含图像页），非图像 PDF 将返回 `422 Unprocessable Entity`。 +- 单文件大小上限为 **512 MB**；超出将返回 `413 Payload Too Large` +- 每个账户最多保留 **10,000 个文件**；达到上限后上传新文件将失败（`400 Bad Request`） +- 文件名中禁止包含 `/`, `\`, `..` 等路径遍历字符，否则返回 `400` +- 已删除文件的 `file_id` 不可复用，且无法通过 API 恢复 +- 所有文件默认保留 **90 天**（自最后访问或操作时间起），超期后自动清理（此策略可能调整，以实际平台公告为准） ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md index 0f5bdacd..c3540886 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/frameworks.md @@ -1,59 +1,51 @@ # frameworks -百炼平台提供多种主流 AI 开发框架的集成支持，帮助开发者快速构建 RAG 应用、智能体/工作流应用及知识库检索服务。当前主要通过 LlamaIndex 和 Spring AI Alibaba 两大框架实现与百炼能力（如云端知识库、大模型服务、应用编排）的深度对接。所有集成均基于百炼统一的 DashScope API 层，需配置有效的 API Key 并遵循对应框架的初始化与调用规范。 +百炼平台提供多种主流 AI 开发框架的集成支持，帮助开发者快速构建 RAG 应用、知识库检索服务及大模型智能体/工作流应用。当前主要通过 LlamaIndex 和 Spring AI Alibaba 两大框架实现与百炼能力（如云端知识库、大模型服务、智能体引擎）的对接，覆盖 Python 和 Java 生态。所有集成均依赖百炼统一的 DashScope API 层，需配置有效的 API Key。 ## 支持的模型/功能 -- **RAG 场景**：支持通过 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 构建云端托管的 RAG 应用，依赖百炼默认的文档解析（`DASHSCOPE_DOCMIND`）、向量化与检索能力；不支持自定义切分器或嵌入模型。 -- **智能体与工作流应用集成**：支持通过 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) 调用已发布的**智能体应用**或**工作流应用**，支持非流式与流式响应，并可获取 `docReferences` 和 `thoughts` 等结构化输出。 -- **知识库直接检索**：支持通过 [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) 实现对百炼知识库的端到端 RAG 检索，底层使用 `DashScopeDocumentRetriever`，默认调用 `qwen-max` 模型生成答案，且允许通过 `DashScopeChatOptions` 显式切换模型（如 `qwen-plus`）。 +- **RAG 构建**：支持通过 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) 在 Python 环境中构建端到端 RAG 流程，包括文档上传、云端索引构建、语义检索与大模型生成。 +- **知识库检索**：Spring AI Alibaba 提供 `DashScopeDocumentRetriever`，可直接检索百炼已创建的云端知识库，支持相似度过滤与重排（如 `gte-rerank`），并自动注入上下文至 LLM 提示词。 +- **大模型应用调用**：Spring AI Alibaba 通过 `DashScopeAgent` 支持对百炼[智能体应用](https://help.aliyun.com/zh/model-studio/single-agent-application)和[工作流应用](https://help.aliyun.com/zh/model-studio/workflow-application/)的非流式/流式调用，返回结构化输出（含文档引用、思考链等元信息）。 -> **注意**：文档 1 明确声明“不支持自定义文档切分方式或自定义嵌入模型”，而文档 3 的 `DashScopeDocumentRetriever` 也未提供嵌入模型配置入口；但文档 2 中 `DashScopeAgent` 的调用逻辑未涉及嵌入层，三者在嵌入能力上保持一致限制。无矛盾。 +> **注意**：文档 1 明确说明“不支持自定义文档切分方式或自定义嵌入模型”，而文档 2 和 3 均未提及该限制，但实际调用时仍受限于百炼云端知识库的默认处理流程；开发者若需完全控制切分与嵌入，应参考[基于本地知识库构建RAG应用](https://help.aliyun.com/zh/model-studio/build-rag-application-based-on-local-retrieval)，而非本框架集成路径。 ## 关键参数 -| 参数名 | 说明 | 来源框架 | 示例值 | 是否必需 | -|--------|------|----------|--------|----------| -| `DASHSCOPE_API_KEY` | 百炼平台 API 密钥 | LlamaIndex / Spring AI Alibaba | `sk-xxx` | 是 | -| `APP_ID` | 智能体或工作流应用 ID | Spring AI Alibaba（应用集成） | `app-abc123` | 是（仅用于应用调用） | -| `WORKSPACE_ID` / `AI_DASHSCOPE_WORKSPACE_ID` | 子业务空间 ID | Spring AI Alibaba | `ws-xyz789` | 否（仅子空间场景需配置） | -| `INDEX_NAME` | 云端知识库名称 | LlamaIndex / Spring AI Alibaba（知识库检索） | `"my_first_index"` | 是（知识库场景） | -| `model_name` / `withModel()` | 生成模型标识符 | LlamaIndex（`Settings.llm`） / Spring AI Alibaba（`DashScopeChatOptions`） | `"qwen-max"`, `"qwen-plus"` | 是（默认值存在，但建议显式指定） | -| `similarity_top_k`, `similarity_cutoff`, `top_n` | 检索与重排参数 | LlamaIndex | `5`, `0.4`, `1` | 否（有合理默认值，但推荐按需调整） | - -> **注意**：文档 2 使用环境变量名 `DASHSCOPE_API_KEY`，而文档 3 使用 `AI_DASHSCOPE_API_KEY`；两者均为有效配置方式，但**不可混用**。实际部署时应统一选用其一，并确保 `application.yml` 中引用的变量名与环境变量名严格一致。 +| 参数 | 作用 | 示例值 | 来源 | +|------|------|--------|------| +| `model_name` | 指定生成阶段使用的千问大模型 | `"qwen-max"`, `"qwen-plus"` | [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) | +| `AI_DASHSCOPE_API_KEY` / `DASHSCOPE_API_KEY` | 百炼 API 密钥环境变量名 | — | 文档 2 使用 `AI_DASHSCOPE_API_KEY`，文档 3 使用 `DASHSCOPE_API_KEY`；二者功能等价，但命名不一致，建议统一采用 `DASHSCOPE_API_KEY` 避免混淆 | +| `INDEX_NAME` | 云端知识库名称（用于检索） | `"测试知识库"` | [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) | +| `APP_ID` | 百炼大模型应用 ID（智能体/工作流） | `"app-xxx"` | [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) | +| `similarity_top_k` / `similarity_cutoff` / `top_n` | 检索结果数量、相似度阈值、重排后返回数 | `5`, `0.4`, `1` | [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) | ## 使用方式 -- **LlamaIndex 集成**： +- **LlamaIndex（Python）**： 1. 安装 `llama-index` 及 `llama-index-readers-dashscope`、`llama-index-indices-managed-dashscope` 等扩展包； 2. 使用 `DashScopeCloudIndex.from_documents()` 构建云端知识库； - 3. 通过 `index.as_query_engine()` 创建查询引擎，配置 `node_postprocessors`（如 `SimilarityPostprocessor` + `DashScopeRerank`）优化检索质量； - 4. 调用 `query_engine.query()` 执行 RAG 查询。 - -- **Spring AI Alibaba 集成（应用调用）**： - 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖； - 2. 在 `application.yml` 中配置 `spring.ai.dashscope.agent.app-id` 和 `api-key`； - 3. 注入 `DashScopeAgent`，调用 `.call()`（非流式）或 `.stream()`（流式）方法，传入 `Prompt` 和 `DashScopeAgentOptions`（含 `appId`）。 + 3. 调用 `index.as_query_engine()` 并传入 `node_postprocessors`（如 `SimilarityPostprocessor` + `DashScopeRerank`）定制检索逻辑； + 4. 通过 `query_engine.query()` 执行 RAG 查询。 -- **Spring AI Alibaba 集成（知识库检索）**： - 1. 添加相同 starter 依赖； - 2. 配置 `spring.ai.dashscope.api-key`（注意变量名差异）； - 3. 构建 `DashScopeDocumentRetriever` 并注入 `ChatClient`，通过 `DocumentRetrievalAdvisor` 自动拼接上下文； - 4. 调用 `chatClient.prompt().user(...).stream().chatResponse()` 触发 RAG 流程。 +- **Spring AI Alibaba（Java）**： + 1. 添加 `spring-ai-alibaba-starter-dashscope` 依赖（版本 ≥ `1.0.0.2`）； + 2. 配置 `application.yml` 中的 `spring.ai.dashscope.api-key` 和 `app-id`（或 `workspace-id`）； + 3. 对知识库检索：注入 `DashScopeApi`，构造 `DashScopeDocumentRetriever` 并集成至 `ChatClient` 的 `DocumentRetrievalAdvisor`； + 4. 对大模型应用调用：构造 `DashScopeAgent` 实例，调用 `agent.call()`（非流式）或 `agent.stream()`（流式）。 ## 限制和注意事项 -- **知识库部署模式限制**：LlamaIndex 方案仅支持**云端知识库**，不支持本地部署知识库所需的自定义切分与嵌入模型 —— 详见 [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 -- **应用类型限制**：Spring AI Alibaba 的 `DashScopeAgent` **仅支持智能体应用和工作流应用**，不支持直接调用基础模型 API 或知识库原生接口 —— 详见 [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md)。 -- **环境变量命名不一致**：文档 2 推荐 `DASHSCOPE_API_KEY`，文档 3 推荐 `AI_DASHSCOPE_API_KEY`；若同时引入两类集成（如既调用应用又检索知识库），需在 `application.yml` 中分别映射或统一环境变量名，否则将导致部分组件初始化失败。 -- **模型选择范围**：所有框架均依赖百炼平台公开的模型列表（如 `qwen-max`, `qwen-plus`, `gte-rerank`），不支持用户私有微调模型接入；`gte-rerank` 仅可用于重排（文档 1），不可作为主生成模型。 -- **计费说明**：框架本身免费，但所有模型调用（包括 RAG 中的生成、重排、检索）均按百炼 [计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio#c1fabcbe9fklk) 单独计费。 +- **知识库部署模式限制**：所有框架集成均依赖百炼**云端知识库**，不支持在框架内直接管理本地向量存储；文档 1 明确指出“不支持自定义文档切分方式或自定义嵌入模型”[通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md)。 +- **文件格式限制**：LlamaIndex 方案仅支持 `.txt`、`.docx`、`.pdf` 等非结构化格式上传，不支持 Excel、PPT 或数据库直连。 +- **环境变量命名冲突**：文档 2 使用 `AI_DASHSCOPE_API_KEY`，文档 3 使用 `DASHSCOPE_API_KEY`；虽底层兼容，但建议项目中统一选用后者以保持一致性。 +- **业务空间隔离**：跨子业务空间操作必须显式配置 `workspace-id`（文档 2 使用 `AI_DASHSCOPE_WORKSPACE_ID`，文档 3 使用 `WORKSPACE_ID`），否则默认访问主账号空间。 +- **计费说明**：框架本身不产生费用，但所有模型调用（含 RAG 生成、智能体执行）均按百炼模型推理用量计费，详见[计费项](https://help.aliyun.com/zh/model-studio/billing-for-model-studio#c1fabcbe9fklk)。 ## 来源文档 - [通过LlamaIndex API构建RAG应用](../../raw/application-api-reference/frameworks/llamaindex.md) -- [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) - [通过Spring AI Alibaba检索阿里云百炼知识库](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-knowledge-base.md) +- [使用Spring AI Alibaba集成阿里云百炼大模型应用](../../raw/application-api-reference/frameworks/spring-ai-alibaba/spring-ai-alibaba-integrate-llm-application.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md index 591de221..aebad92a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/image-generation.md @@ -1,95 +1,110 @@ # image generation -百炼平台提供多种图像生成与编辑能力，涵盖文生图（T2I）、图生图（I2I）、局部重绘、风格迁移、背景生成、海报设计等场景。所有模型均通过统一的HTTP API或DashScope SDK调用，支持异步与同步两种模式，适用于开发者快速集成到生产环境。核心能力由千问（Qwen-Image）、万相（WanX）、可灵（Kling）、Vidu、Z-Image 等系列模型支撑，覆盖效果、速度、成本多维需求。 +百炼平台提供丰富的图像生成与编辑能力，涵盖文生图、图生图、局部重绘、风格迁移、背景生成、AI试衣等20余种专业场景模型。所有服务均基于统一的API协议，支持HTTP异步/同步调用及DashScope SDK集成，适用于电商、设计、内容创作等开发者场景。 ## 支持的模型/功能 -平台当前提供以下主流图像模型及对应能力： +平台当前提供三类核心能力：**通用图像生成**（如文生图）、**图像编辑与增强**（如局部重绘、扩图）、**垂直领域工具**（如虚拟模特、AI试衣）。主要模型包括： -- **文生图（T2I）**：`qwen-image-3.0-pro`、`wan2.6-t2i`、`z-image-turbo`、`kling/kling-v3-image-generation`、`vidu/vidu-image_reference2image` 等，支持自由分辨率设置（总像素 512×512 至 2048×2048），部分模型（如 `wan2.7-image-pro`）支持 4K 输出 [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md)。 -- **图生图/图像编辑（I2I）**：`qwen-image-2.0-pro`、`wan2.7-image-pro`、`wan2.5-i2i-preview`、`kling/kling-v3-omni-image-generation`、`vidu/viduq3-fast_reference2image`，支持单图/多图输入、指令编辑、风格迁移、图文混排等 [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md)。 -- **专用工具类模型**： - - 局部重绘：`wanx-x-painting`（免费体验，额度用尽后不可用）； - - 涂鸦作画：`wanx-sketch-to-image-lite`； - - 虚拟模特/鞋靴试穿：`wanx-virtualmodel`、`shoemodel-v1`（均仅限免费体验）； - - 图像擦除补全、画面扩展、背景生成、人物实例分割：均为华北2（北京）地域专属，需使用业务空间域名调用 [常见问题](../../raw/model-api-reference/image-generation/image-faq.md)。 -- **创意工具**：`wordart-quick-start`（文字变形与纹理生成）、`facechain-portrait-generation`（人物写真LoRA训练与生成）、`outfitanyone`（AI试衣全链路组合）。 +- **万相系列**：`wan2.6-t2i`（推荐V2文生图）、`wan2.5-i2i-preview`（通用图像编辑）、`wanx-style-repaint-v1`（人像风格重绘）、`wanx-background-generation-v2`（背景生成）[原文标题](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) +- **千问系列**：`qwen-image-3.0-pro`（T2I+I2I一体化）、`qwen-image-edit-max`（高精度图像编辑）、`qwen-mt-image`（图像翻译）[原文标题](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) +- **轻量与专用模型**：`z-image-turbo`（快速文生图）、`kling/kling-v3-omni-image-generation`（分镜组图）、`shoemodel-v1`（鞋靴试穿）、`facechain`（人物写真训练与生成） +- **免费体验模型**：`wanx-x-painting`（图像局部重绘）、`wanx-virtualmodel`（虚拟模特）、`image-erase-completion`（擦除补全）等均处于免费体验阶段，额度用尽后不可调用且不支持付费 [原文标题](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) -> **注意**：`wanx-v1`（V1版）已明确标注“推荐使用全面升级的[文生图V2版模型](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)”；而 `wan2.6-t2i` 及更高版本（如 `wan2.7-image-pro`）支持 HTTP 同步调用，但 `wan2.5` 及以下版本**不支持同步调用**，仅支持异步流程 —— 此矛盾点已在文档中显式区分，开发者需按模型版本选择对应调用方式。 +> **注意**：文档中多次提及 `wanx-v1`（V1文生图）已明确标注“推荐使用全面升级的[文生图V2版模型](https://help.aliyun.com/zh/model-studio/text-to-image-v2-api-reference)”；同时 `wan2.6-t2i` 明确支持HTTP同步调用，而 `wan2.5` 及以下版本仅支持异步调用——这与部分旧文档未明确区分调用方式存在潜在矛盾，开发者应以[万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md)为准。 ## 关键参数 -| 参数 | 类型 | 说明 | 示例值 | -|------|------|------|--------| -| `model` | string | 必填。模型标识符，需与地域支持列表一致 | `"qwen-image-3.0-pro"`, `"wan2.6-t2i"` | -| `size` | string | 可选。输出图像尺寸，格式为 `"宽*高"` 或 `"1K"/"2K"/"4K"`；部分模型（如 `wan2.5-i2i-preview`）默认生成 `1280*1280` 并保持输入图宽高比 | `"1024*1024"`, `"2K"` | -| `n` | integer | 可选。生成图片张数，范围因模型而异：`qwen-image-*` 支持 1–6 张；`kling` 支持 1–9；`z-image-turbo` 固定为 1 张 | `1`, `2` | -| `prompt` / `messages.text` | string / array | 必填。提示词字段。`qwen-image-3.0-pro` 和 `wan2.7+` 使用 `messages` 结构；旧版 `wanx-v1`、`wan2.6-t2i` 使用 `input.prompt` 字段 | `{"text": "一间花店..."}` | -| `negative_prompt` | string | 可选（仅部分模型）。用于排除不希望出现的内容 | `"不要红色元素"` | -| `watermark` | boolean | 可选。是否添加水印，默认 `true`；多数生产场景建议设为 `false` | `false` | -| `prompt_extend` | boolean | 可选。启用智能提示词优化（返回增强后的 [prompt](../guides/prompt.md)），会增加响应时间 | `true` | +不同模型共性参数如下，具体支持情况需查阅对应模型文档： -> **注意**：`aspect_ratio`（如 `"1:1"`）和 `resolution`（如 `"1k"`）为 `kling` 系列特有参数；`style_index`、`style_ref_url` 为人像风格重绘专用；`mask_image_url` 为局部重绘/擦除补全必需字段 —— 开发者应严格依据目标模型文档传参，跨模型复用参数将导致失败。 +- `model`：必填，指定模型名称（如 `"wan2.6-t2i"`、`"qwen-image-3.0-pro"`） +- `input.prompt` 或 `input.messages`：文本提示词，V2及以后模型普遍采用 `messages` 数组格式（含 `text` 和可选 `image` 元素） +- `parameters.size`：输出分辨率，格式为 `"宽*高"`（如 `"1024*1024"`）或语义值（如 `"1K"`），各模型约束不同： + - `wan2.6-t2i`：总像素在 `[1280*1280, 1440*1440]`，宽高比 `[1:4, 4:1]` + - `qwen-image-3.0-pro`：总像素 `[512*512, 2048*2048]` + - `kling` 系列：仅支持 `"1k"`/`"2k"`/`"4k"` 等预设值 +- `parameters.n`：生成张数，范围通常为 `1–9`，部分模型（如 `qwen-image-max`）固定为 `1` +- `parameters.aspect_ratio`（Kling）或 `parameters.resolution`（Vidu）：控制宽高比与分辨率组合 +- `X-DashScope-Async: enable`：**所有HTTP异步调用必需**，缺失将报错 `"current user api does not support synchronous calls"` +- `X-DashScope-Sse: enable` + `parameters.stream: true`：图文混排（`enable_interleave=true`）时必需的流式配置 ## 使用方式 -### 1. 基础准备 -- 获取并配置 API Key：必须通过 [阿里云百炼控制台](https://bailian.console.aliyun.com/) 获取对应地域（华北2/新加坡/弗吉尼亚）的 API Key，并设置为环境变量 `DASHSCOPE_API_KEY`。 -- 使用业务空间专属域名：强烈建议迁移至 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`（新加坡），以获得更高性能与稳定性；`{WorkspaceId}` 在控制台「业务空间详情」中获取。 - -### 2. 调用模式选择 -- **同步调用（推荐多数场景）**：适用于 `wan2.6+`、`qwen-image-3.0-pro`、`z-image-turbo` 等支持模型。一次 HTTP POST 即返回结果（含图片 URL 或 base64），无需轮询。示例 endpoint： - `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation` -- **异步调用（必需场景）**：适用于 `wanx-v1`、`wanx-sketch-to-image-lite`、`wanx-x-painting`、`image-out-painting` 等耗时较长的模型。流程为两步： - 1. 创建任务：`POST .../image-synthesis`（或对应路径），返回 `task_id`； - 2. 轮询结果：`GET .../tasks/{task_id}`，直至 `task_status == "SUCCEEDED"`，获取 `output.results[0].url`（有效期 24 小时）。 - -### 3. 请求头要求 -所有请求必须包含： -- `Authorization: Bearer $DASHSCOPE_API_KEY` -- `Content-Type: application/json` -- 异步调用**必须**添加 `X-DashScope-Async: enable`；缺失将报错 `"current user api does not support synchronous calls"`。 +### 调用前提 +- 获取并配置 API Key（[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)），注意**华北2（北京）、新加坡、美国（弗吉尼亚）地域的API Key与请求地址独立，不可混用** +- 推荐使用业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）替代旧域名（`dashscope.aliyuncs.com`），以获得更高性能与稳定性 + +### 调用模式 +- **同步调用**：适用于 `wan2.6-t2i`、`z-image-turbo`、`qwen-image-2.0-pro` 等支持模型，单次请求直接返回结果（含图片URL），响应更快 +- **异步调用**：适用于绝大多数图像模型（如 `wanx-v1`、`wan2.5-i2i-preview`、`image-out-painting`），流程为： + 1. `POST /api/v1/services/.../generation` 创建任务，获取 `task_id` + 2. 轮询 `GET /api/v1/tasks/{task_id}` 查询状态，`task_status == "SUCCEEDED"` 时返回图片URL（有效期24小时） + +### 示例命令 +```bash +# 同步调用 wan2.6-t2i（北京地域） +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "wan2.6-t2i", + "input": {"messages": [{"role":"user","content":[{"text":"一间花店"}]}]}, + "parameters": {"size": "1024*1024"} + }' + +# 异步调用 wan2.5-i2i-preview（单图编辑） +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/image2image/image-synthesis \ + -H "X-DashScope-Async: enable" \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "model": "wan2.5-i2i-preview", + "input": { + "prompt": "将连衣裙换成复古蕾丝长裙", + "images": ["https://example.com/input.jpg"] + } + }' +``` ## 限制和注意事项 -- **地域与密钥绑定**：华北2（北京）、新加坡、美国（弗吉尼亚）地域的 API Key 和请求地址**完全独立，不可混用**；跨地域调用将导致鉴权失败或服务报错。 -- **免费额度与计费**：所有模型均提供 500 张免费额度（主账号与 RAM 子账号共享），有效期 90 天；额度用尽后，商业化模型（如 `wanx-v1` 0.16元/张、`image-out-painting` 0.18元/张）开始计费，限时免费模型（如 `wanx-x-painting`）则直接不可用。 -- **图片 URL 要求**：输入图片必须为**公网可访问**的 HTTPS/HTTP 链接；OSS、自建存储等需确保外网可直连，否则报错 `"Reference image download failed"`。 -- **输入限制**： - - 图像分辨率：多数模型要求 `[512, 4096]` 像素单边长度，总像素 `512×512` 至 `2048×2048`； - - 文件大小：通常 ≤10MB； - - 格式：PNG/JPEG/WEBP/BMP/AVIF（具体见各模型文档）； - - URL 中**禁止中文字符**（见 [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md)）。 -- **模型可用性**：部分模型（如 `wanx-virtualmodel`、`shoemodel-v1`、`image-instance-segmentation`）当前**仅限免费体验**，额度用尽后无付费通道，文档明确建议迁移到 `qwen-image-edit` 或 `wanx-image-edit` 等替代方案。 +- **地域与域名绑定**：所有模型均严格按地域隔离，跨地域调用将导致鉴权失败；务必使用对应地域的 Workspace ID 构造请求 URL +- **图片URL要求**：输入图片必须为**公网可访问的HTTPS/HTTP链接**，OSS等云存储需开启公共读权限；URL中禁止含中文字符 +- **免费额度**：多数模型提供500张免费额度（如 `wanx-v1`、`wanx-style-repaint-v1`），仅对**成功生成的输出图片**计数，失败/输入图片不计入；额度90天有效，主账号与RAM子账号共享 +- **限流策略**：主账号与RAM子账号共用QPS/RPS限制（常见为2 QPS），同时处理中任务数上限通常为1（少数如 `image-out-painting` 为5） +- **错误处理**： + - `BadRequest.InputDownloadFailed`：检查图片URL是否可公网访问、是否被防盗链拦截 + - `InvalidApiKey`：确认API Key正确且地域匹配 + - `current user api does not support synchronous calls`：异步模型未设置 `X-DashScope-Async: enable` 头 +- **模型弃用风险**：`wanx-v1`、`wanx-x-painting` 等模型已明确标注“推荐使用V2版”或“免费体验”，长期项目应避免依赖 ## 来源文档 - [常见问题](../../raw/model-api-reference/image-generation/image-faq.md) -- [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) -- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) -- [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) -- [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) - [万相-文生图V2版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-v2-api-reference.md) - [万相-文生图V1版API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/text-to-image-api-reference.md) -- [万相-图像生成与编辑2.7 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-and-editing-api-reference.md) - [万相-图像生成与编辑2.6 API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan-image-generation-api-reference.md) - [万相-通用图像编辑2.5](../../raw/model-api-reference/image-generation/wan-image-api-reference/wan2-5-image-edit-api-reference.md) - [万相-通用图像编辑API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-image-edit-api-reference.md) - [万相-涂鸦作画API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/wanx-sketch-to-image-api-reference.md) - [万相-图像局部重绘API参考](../../raw/model-api-reference/image-generation/wan-image-api-reference/vary-region-api-reference.md) +- [千问-图像生成与编辑3.0 API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-generation-and-editing-api-reference.md) +- [千问-文生图API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-api.md) +- [千问-图像编辑API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-image-edit-api.md) +- [千问-图像翻译API参考](../../raw/model-api-reference/image-generation/qwen-image-api-reference/qwen-mt-image-api.md) +- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [可灵-图像生成API参考](../../raw/model-api-reference/image-generation/kling-image-api-reference/kling-image-generation-api-reference.md) - [Vidu-图像生成API参考](../../raw/model-api-reference/image-generation/vidu-image-models/vidu-image-generation-api-reference.md) -- [Z-Image API参考](../../raw/model-api-reference/image-generation/z-image-generation-api-reference/z-image-api-reference.md) - [人像风格重绘API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/portrait-style-redraw-api-reference.md) -- [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [图像画面扩展API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-scaling-api.md) +- [虚拟模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/virtual-model-api-details.md) - [鞋靴模特API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/shoe-model-api.md) -- [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [创意海报生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/creative-poster-generation-api.md) +- [人物实例分割API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-instance-segmentation-api-reference.md) - [图像擦除补全API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/image-erase-completion-api-reference.md) - [图像背景生成API参考](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wanx-background-generation-api-reference.md) +- [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) - [AI试衣OutfitAnyone](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/outfitanyone.md) - [创意文字WordArt锦书](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/wordart-quick-start.md) -- [人物写真生成FaceChain](../../raw/model-api-reference/image-generation/image-creative-tools-api-reference/facechain-portrait-generation.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md index 71ba9000..7cfcfead 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/knowledge.md @@ -1,41 +1,41 @@ # knowledge -知识检索与问答是百炼平台提供的 RAG（[检索增强生成](../concepts/rag.md)）核心能力，通过语义检索与大模型协同实现精准、可溯源的智能问答。该能力基于 DashScope 应用网关提供 HTTP REST 接口，不依赖 OpenAPI RPC 调用链，适用于需快速集成知识增强能力的业务场景。详细设计与行为请参考 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 +knowledge 是百炼平台提供的知识增强型 AI 服务模块，支持基于私有知识库的语义检索与多阶段智能问答。该能力通过 DashScope 应用网关提供 RESTful API，与底层 OpenAPI（如 `CreateIndex`、`Retrieve`）解耦，面向业务应用层封装，适用于 RAG 场景下的快速集成。详细设计与行为请参考 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 ## 支持的模型/功能 -- **知识检索**：跨多个已构建的知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），不调用大模型，纯检索服务。 -- **知识问答**：端到端问答流程，支持 SSE 流式响应，输出包含「规划 → 工具调用（如 Retrieve）→ 生成」三阶段结果，底层自动调度检索与 LLM 生成。 -- 所有功能均运行于 DashScope 应用网关，与 `CreateIndex` 等 OpenAPI RPC 接口隔离，不可混用。详见 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md)。 +- **知识检索**：跨多个已发布知识库执行联合语义检索，返回按相关性排序的文本切片（chunk），不调用大模型，纯向量/关键词混合召回。 +- **知识问答**：端到端流式问答，内部自动完成「问题理解→知识检索→答案生成」三阶段，通过 SSE 返回结构化响应（含 `plan`、`tool_calls`、`response` 字段）。 +- 所有功能均依赖预构建并发布的知识库索引，不支持运行时上传或动态索引构建。具体能力边界详见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 ## 关键参数 -| 参数 | 说明 | 必填 | 示例 | +| 参数 | 类型 | 必填 | 说明 | |------|------|------|------| -| `Authorization` | Bearer 鉴权头，值为 API Key | 是 | `Bearer ak-xxx` | -| `workspaceId` | 业务空间 ID，用于拼接 Base URL（`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`） | 是 | `ws-abc123` | -| `query` | 检索或问答的用户输入文本 | 是 | `"阿里云百炼平台支持哪些知识库格式？"` | -| `top_k` | 检索返回切片数（仅 `/search` 接口支持） | 否 | `5` | -| `stream` | 是否启用 SSE 流式（仅 `/chat` 接口有效） | 否，默认 `true` | `true` | +| `knowledge_ids` | string[] | 是（检索）/ 否（问答） | 检索时指定参与查询的知识库 ID 列表；问答时若未指定，则使用应用绑定的默认知识库 | +| `query` | string | 是 | 用户原始查询文本，长度 ≤ 2048 字符 | +| `top_k` | integer | 否（默认 5） | 检索返回切片数（范围 1–50）；问答中影响检索阶段召回数量 | +| `stream` | boolean | 否（默认 true） | 仅问答接口有效，设为 `false` 时返回完整 JSON 响应而非 SSE 流 | -> **注意**：`/chat` 接口不接受 `model` 参数——模型由业务空间绑定的默认应用配置决定，无法在请求中覆盖。此行为与部分旧版文档描述不符，请以 [知识检索与问答 (raw/application-api-reference/knowledge.md)](../../raw/application-api-reference/knowledge.md) 为准。 +> **注意**：`top_k` 在问答接口中实际生效值受后端策略限制，可能被截断至 ≤10，与 [知识检索与问答](../../raw/application-api-reference/knowledge.md) 中文档描述的“范围 1–50”存在不一致，以实际 API 响应为准。 ## 使用方式 -1. 在控制台获取 API Key（[API Key 页面](https://rag.console.aliyun.com/settings/apikey)）和业务空间 ID（[业务空间管理](https://bailian.console.aliyun.com/cn-beijing?tab=globalset#/efm/business_management)）； -2. 构造 Base URL：`https://{workspaceId}.cn-beijing.maas.aliyuncs.com`； -3. 发起 POST 请求： - - 检索：`POST /api/v1/indices/knowledge/search`，Body 为 JSON `{ "query": "..." }`； - - 问答：`POST /api/v2/apps/knowledge/chat`，Body 同样为 JSON `{ "query": "..." }`，响应为 SSE 流； -4. 所有请求必须携带 `Authorization: Bearer ` 头。 +1. **准备环境**：在控制台获取业务空间 ID（workspaceId）和 API Key，构造 Base URL：`https://{workspaceId}.cn-beijing.maas.aliyuncs.com` +2. **发起请求**： + - 知识检索：`POST /api/v1/indices/knowledge/search`，Body 示例： + ```json + { "knowledge_ids": ["k1", "k2"], "query": "百炼平台如何接入知识库？", "top_k": 3 } + ``` + - 知识问答：`POST /api/v2/apps/knowledge/chat`，Header 需含 `Authorization: Bearer `，Body 同上（`stream` 可选） +3. **处理响应**：检索返回标准 JSON；问答若启用 `stream=true`，需按 SSE 协议解析 `event: plan/data: {...}` 等事件流。完整调用示例见 [知识检索与问答](../../raw/application-api-reference/knowledge.md)。 ## 限制和注意事项 -- **限流策略**：默认按用户维度限流 25 QPS，超限返回 `429 Too Many Requests`，需客户端退避重试； -- **知识库前提**：检索与问答均要求目标知识库已完成索引构建并处于 `ACTIVE` 状态，否则返回 `404 Not Found` 或 `400 Bad Request`； -- **协议差异**：该能力**不兼容** OpenAPI 的 `Retrieve` RPC 接口（如 `dashscope.serving.Retrieve`），二者鉴权、Endpoint、参数结构完全不同； -- **地域固定**：Base URL 中的 `cn-beijing` 为硬编码区域，暂不支持切换； -- **调试建议**：首次调用前，务必确认业务空间已绑定至少一个有效知识库，否则 `/chat` 将静默返回空结果而非报错。 +- **鉴权与域名**：必须使用业务空间专属域名（含 workspaceId），不可复用通用 DashScope OpenAPI 域名；API Key 需在 [API Key 页面](https://rag.console.aliyun.com/settings/apikey) 单独创建。 +- **限流策略**：默认 25 QPS（用户维度），超限返回 `429 Too Many Requests`，需客户端实现退避重试。 +- **知识库状态**：仅 `published` 状态的知识库可被检索/问答调用；草稿或已下线知识库不参与计算。 +- **无状态设计**：问答接口不维护会话上下文，如需多轮对话，须由应用层自行管理历史消息并拼入 `query`。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md index ef7f829b..1811a460 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/long-term-memory-new.md @@ -1,62 +1,97 @@ # long term memory new -[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化用户记忆管理能力，支持自动从对话中提取关键信息、构建用户画像，并提供语义搜索、增删改查等完整生命周期操作。该功能基于专用记忆模型实现，适用于需要持久化用户偏好、习惯、任务提醒等场景。详细设计与行为请参考 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 +[长期记忆](../concepts/long-term-memory.md)（新）是百炼平台提供的结构化用户记忆管理能力，支持将对话自动提炼为可检索、可更新的记忆片段，并支持基于画像模板的用户画像构建。该功能通过 REST API 和 Python SDK 提供完整 CRUD 能力，适用于需要持久化用户偏好、习惯、意图等上下文信息的智能体应用。详细接口定义与行为规范见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 ## 支持的模型/功能 -- **底层模型**：由百炼平台统一调度专用记忆模型（非通用大模型），不开放模型选择，所有 API 均隐式绑定该模型。 -- **核心能力**： - - `AddMemory`：自动解析对话（或接收自定义文本），生成结构化记忆片段； - - `SearchMemory`：基于语义相似度召回相关记忆，支持重排序（`enable_rerank`）、意图判别（`enable_judge`）和 query 重写（`enable_rewrite`）； - - `ListMemory` / `DeleteMemory` / `UpdateMemory`：标准 CRUD 操作； - - 画像模板管理（`CreateProfileSchema` 等）：定义用户属性结构，用于约束记忆提取逻辑； - - 用户画像聚合（`GetUserProfile`）：按模板聚合用户全部记忆节点生成结构化 profile。 +- **记忆片段管理**：支持 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory`、`UpdateMemory` 五类核心操作，覆盖记忆的创建、语义搜索、分页查询、删除与内容更新。 +- **用户画像构建**：通过 `CreateProfileSchema` 等画像模板相关接口，定义结构化字段（如 `age`, `occupation`, `preference`），并关联至 `GetUserProfile` 获取聚合画像。 +- **多规则混合检索**：`SearchMemory` 支持传入 `project_ids` 数组，在多个记忆片段规则下联合召回，提升跨场景记忆覆盖度。 +- **端到端 SDK 封装**：`agentscope-runtime>=1.1.5` 提供 `AddMemory`、`SearchMemory`、`ListMemory`、`DeleteMemory` 的异步 Python 工具类封装；但 `UpdateMemory` 当前未被 SDK 封装，需直接调用 REST API —— 此限制已在 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的 UpdateMemory 章节明确说明。 -> **注意**：原始文档中未明确说明是否支持多模型路由或自定义 embedding 模型，所有接口均强制使用平台内置记忆模型。如需验证模型行为，请以 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中的接口定义为准。 +> **注意**：原始文档中 `UpdateMemory` 的 Python 示例使用 `requests.patch` 手动调用，而其他接口均提供 `agentscope-runtime` 封装。SDK 文档未声明对该接口的未来支持计划，开发者应避免依赖未封装接口的抽象层一致性。 ## 关键参数 | 参数名 | 类型 | 必填 | 说明 | |--------|------|------|------| -| `user_id` | string | 是 | 记忆归属实体 ID（≤64 字符），用于隔离不同用户数据 | -| `messages` 或 `custom_content` | array / string | 互斥必填 | `messages`：最多 50 条对话（一问一答计为 2 条）；`custom_content`：纯文本（≤512 字符），优先级高于 `messages` | -| `memory_library_id` | string | 否 | 记忆库 ID（≤32 字符），未传则使用默认库；需在控制台 [记忆库列表](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 获取 | -| `profile_schema` | string | 否 | 画像模板 ID，影响记忆提取字段；需通过 `CreateProfileSchema` 创建并获取 | -| `top_k`（Search） | integer | 否 | 召回数量（1–100，默认 10） | -| `min_score`（Search） | double | 否 | 相似度阈值 [0,1]（默认 0.3） | -| `page_num` / `page_size`（List） | integer | 否 | 分页参数（默认 page_num=1, page_size=10） | +| `user_id` | `string` | 是 | 记忆归属实体 ID（≤64 字符），所有接口均需指定，用于隔离不同用户数据。 | +| `messages` / `custom_content` | `array` / `string` | 互斥必填 | `AddMemory` 中二选一：`messages` 支持最多 50 条对话（一问一答计为 2 条），自动提取；`custom_content` 为纯文本（≤512 字符），绕过自动解析。详见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 | +| `memory_library_id` | `string` | 否 | 显式指定记忆库 ID（≤32 字符）；不传则使用默认记忆库。可在控制台 [记忆库列表页](https://bailian.console.aliyun.com/cn-beijing/?tab=app#/memory/list) 获取。 | +| `top_k` | `integer` | 否（`SearchMemory`） | 搜索召回数量，默认 `10`，取值范围 `1–100`。 | +| `min_score` | `double` | 否（`SearchMemory`） | 相似度阈值，默认 `0.3`，范围 `[0,1]`；设为 `0` 可返回全部匹配项（受 `top_k` 限制）。 | +| `meta_data` | `object` | 否 | 用户自定义键值对，支持任意 JSON 对象，用于扩展元信息（如地理位置、设备类型等）。 | ## 使用方式 -### 1. 基础调用 -- **Base URL**：`https://dashscope.aliyuncs.com/api/v2/apps/memory/` -- **认证**：Header 中携带 `Authorization: Bearer $DASHSCOPE_API_KEY` -- **Content-Type**：`application/json` - -### 2. SDK 快速接入（推荐） -需安装 `agentscope-runtime>=1.1.5`： +### 1. 基础认证 +所有请求需在 Header 中携带： +```http +Authorization: Bearer $DASHSCOPE_API_KEY +Content-Type: application/json +``` +API Key 获取方式见 [获取 API Key](https://help.aliyun.com/zh/model-studio/get-api-key)。 + +### 2. 接口调用示例（REST） +- **添加记忆**（自动解析对话）： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/add \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "user_id": "user_001", + "messages": [{"role":"user","content":"明天10点提醒我开会"}] + }' + ``` +- **语义搜索**（带过滤）： + ```bash + curl -X POST https://dashscope.aliyuncs.com/api/v2/apps/memory/memory_nodes/search \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{ + "user_id": "user_001", + "messages": [{"role":"user","content":"我有什么待办？"}], + "top_k": 5, + "min_score": 0.5 + }' + ``` + +### 3. Python SDK 调用（推荐） +安装依赖： ```bash pip install agentscope-runtime>=1.1.5 ``` -- `AddMemory`, `SearchMemory`, `ListMemory` 已封装为异步工具类（见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 中 Python 示例）； -- `DeleteMemory` 和 `UpdateMemory` 仅提供 SDK 封装（`DeleteMemory` 支持，`UpdateMemory` 当前需手动 HTTP 调用，详见原文示例）。 - -### 3. 直接 HTTP 调用 -所有接口路径见 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md) 的「接口概览」表，cURL 示例可直接复用。 +使用封装工具（以 `AddMemory` 为例）： +```python +from agentscope_runtime.tools.modelstudio_memory import AddMemory, Message, AddMemoryInput +import asyncio + +async def main(): + tool = AddMemory() + try: + res = await tool.arun(AddMemoryInput( + user_id="user_001", + messages=[Message(role="user", content="每天9点提醒我吃药")], + meta_data={"source": "mobile_app"} + )) + print(f"生成 {len(res.memory_nodes)} 条记忆") + finally: + await tool.close() +``` +> **注意**：`UpdateMemory` 无对应 SDK 封装，必须使用 `requests.patch` 手动调用，具体参数格式请严格参照 [长期记忆（新）API 参考](../../raw/application-api-reference/long-term-memory-new/long-term-memory-api-reference.md)。 ## 限制和注意事项 -- **限流策略（阿里云账号级别）**： - - 全部接口总计 ≤ 3000 QPM； - - `AddMemory` ≤ 120 QPM； - - `SearchMemory` ≤ 300 QPM。 -- **数据时效性**：记忆片段与用户画像**无自动过期机制**，需业务侧自行管理生命周期。 -- **内容长度**： +- **限流策略**（阿里云账号级别）： + - 全部接口总 QPM ≤ 3000； + - `AddMemory` 单独限流 120 QPM； + - `SearchMemory` 单独限流 300 QPM。 +- **数据生命周期**：当前生成的记忆片段与用户画像**无自动失效机制**，需业务侧自行管理过期逻辑。 +- **内容长度约束**： - `custom_content` 最大 512 字符； - - `messages` 最多 50 条（含 `user`/`assistant` 角色消息）； - - `meta_data` 为 JSON object，无明确大小限制，但建议保持轻量。 -- **ID 约束**：`user_id`、`memory_library_id` 等字符串 ID 均有长度上限，超长将导致请求失败。 -- **Python SDK 缺失项**：`UpdateMemory` 在 `agentscope-runtime` 中暂未封装（截至 `1.1.5` 版本），需使用 `requests` 库手动 PATCH 调用，具体参数见原文。 + - `messages` 中单条 `content` 无明确长度限制，但整组 `messages` 不得超过 50 条。 +- **默认行为**：`memory_library_id` 和 `project_id` 等参数若未显式传入，系统将自动选择默认值，但生产环境建议显式指定以避免配置漂移。 +- **时间戳精度**：`UpdateMemory` 的 `timestamp` 字段为秒级 Unix 时间戳；若未提供，则使用请求发起时刻。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md index 4f823fb9..bd4cde00 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/managed-agents-api.md @@ -1,63 +1,71 @@ # [managed agents](../guides/managed-agents.md) api -Managed Agents API 是百炼平台提供的智能体托管运行时服务，负责会话生命周期管理、沙箱环境调度、工具执行协调与事件流分发。开发者通过 REST 或 SDK 调用，可快速构建具备工具调用、多轮交互与状态感知能力的智能体应用。所有资源均按工作空间隔离，需配合 API Key 与地域化 Endpoint 使用。 +Managed Agents API 是百炼平台提供的智能体托管运行时服务，由平台统一管理会话生命周期、沙箱执行环境、工具调用与事件流。开发者通过 REST 接口或 SDK 创建 Agent、Environment、Session 等资源，并以事件驱动方式与智能体交互。所有操作均基于工作空间隔离，需通过 API Key 鉴权。 ## 支持的模型与功能 -- **模型支持**：当前仅支持 `qwen-plus` 等百炼托管大模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中的 `model.id` 字段示例）；不支持自定义模型或外部模型接入。 +- **模型支持**：Agent 创建时通过 `model.id` 指定模型，当前支持 `qwen-plus` 等百炼已开通的推理模型（详见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)）。 - **核心功能模块**： - - **Agent**：封装模型、系统提示词、工具集与技能，支持版本化管理与软归档； - - **Environment**：定义沙箱类型（如 `"cloud"`）、预装依赖与网络策略，独立于 Agent 生命周期； - - **Session**：绑定 Agent 快照与 Environment 实例，驱动 `idle → running → idle/terminated` 状态机； - - **Event**：支持用户消息、工具回填、中断指令等原子事件，提供 SSE 流式订阅； - - **File**：上传后经安全审核（`checking` → `available`），可用于消息内容或挂载至沙箱； - - **Skill**：以 zip 包形式封装工具组合，上传后需通过安全扫描（`checking` → `active`）方可挂载，挂载时必须指定具体版本号。 + - `Agent`：定义智能体配置（模型、系统提示词、技能、工具），支持版本化与软归档； + - `Environment`：定义沙箱类型（如 `"cloud"`）与预装依赖，独立于 Agent 管理，可被多会话复用； + - `Session`：绑定 Agent 版本与 Environment 快照的一次运行实例，状态机驱动（`idle` → `running` → `idle`/`terminated`）； + - `Event`：会话内原子操作载体，支持用户消息、工具回填、中断等，可通过 SSE 流式订阅； + - `File`：独立文件资源，用于消息内容（图像/音频）或挂载至沙箱供工具读写； + - `Skill`：zip 包封装的工具组合，需经安全扫描后按具体版本号挂载到 Agent。 -> **注意**：文档 2 的快速开始示例中使用 `model: "qwen-plus"` 作为字符串传入，而文档 1 的 API 总览中 `model` 字段结构为 `{"id": "qwen-plus"}`。实际请求体应严格遵循文档 1 的嵌套对象格式，否则将返回 400 错误 —— 此处以 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 为准。 +> **注意**：文档 3 的 Bash 示例中使用 `model: {"id": "qwen-plus"}`（对象格式），而文档 2 和文档 3 的 Python SDK 示例中直接传入字符串 `"qwen-plus"`。实际 REST API 要求 `model` 字段为对象（含 `id` 字段），SDK 封装层做了简化。请以 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) 中的接口定义为准。 ## 关键参数 -| 参数 | 位置 | 类型 | 必填 | 说明 | -|------|------|------|------|------| -| `Authorization` | Header | string | 是 | `Bearer `，从控制台获取并配置为环境变量 | -| `workspace_id` | Endpoint path | string | 是 | 工作空间 ID（如 `ws_xxxxxxxxxxxx`），见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) | -| `region` | Endpoint path | string | 是 | 当前仅支持 `cn-beijing` | -| `agent.id` | Session 创建体 | string | 是 | Agent ID，创建时生成；会话锁定其 `version` 快照 | -| `environment_id` | Session 创建体 | string | 是 | Environment ID，会话绑定其快照 | -| `input` | `/sessions/{id}/events` 请求体 | array | 是 | 符合 ChatML 格式的 message 数组，如 `[{"role":"user","type":"message","content":[{"type":"text","text":"..."}]}]` | -| `limit` / `page` | 列表端点 Query | integer | 否 | 分页参数，默认 `limit=20`，最大 `100`；响应含 `next_page` 表示可继续翻页 | +- **Endpoint**：`https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`，其中 `region` 当前仅支持 `cn-beijing`； +- **鉴权**：`Authorization: Bearer `，API Key 需在百炼控制台创建并配置为环境变量； +- **Agent 创建关键字段**： + - `name`: 必填，智能体名称； + - `model.id`: 必填，模型 ID（如 `"qwen-plus"`）； + - `system`: 可选，系统提示词； + - `skills`: 数组，每个元素为 `{skill_id: "...", version: N}`，**必须指定整数版本号，不支持 `"latest"`**（见 [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md)）； +- **Session 创建关键字段**： + - `agent`: Agent ID（会自动锁定当前最新版本）； + - `environment_id`: Environment ID； +- **Event 发送关键字段**： + - `input`: 消息数组，每条消息含 `role`（`user`/`assistant`）、`type`（`message`）、`content`（文本或文件引用）； +- **分页参数**：`limit`（默认 20，最大 100）、`page`（首次省略，后续传 `next_page`）。 ## 使用方式 -1. **初始化**：导出 `DASHSCOPE_API_KEY` 与 `AGENTSTUDIO_URL`（形如 `https://.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio`）； -2. **资源准备**： - - 创建 Agent（`POST /agents`），指定 `model.id`、`system` 等； - - 创建 Environment（`POST /environments`），配置 `config.type`（如 `"cloud"`）； -3. **会话启动**： - - 创建 Session（`POST /sessions`），传入 `agent` 和 `environment_id`； - - 发送 Event（`POST /sessions/{id}/events`）触发执行； -4. **结果消费**： - - 订阅 SSE 事件流（`GET /sessions/{id}/events/stream`），监听 `session_status` 变更及 `message` 内容； - - 或轮询事件历史（`GET /sessions/{id}/events`）。 +完整调用流程为五步（参考 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)）： -SDK 使用需满足最低版本要求：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24 —— 具体安装与初始化方式参见 [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md)。 +1. **创建 Agent**：定义模型与行为逻辑，获得 `agent_xxx`； +2. **创建 Environment**：定义沙箱类型（如 `{"type": "cloud"}`），获得 `env_xxx`； +3. **创建 Session**：绑定 Agent 与 Environment，获得 `sesn_xxx`，初始状态为 `idle`； +4. **发送 Event**：向 Session 提交用户消息（`POST /sessions/{session_id}/events`）； +5. **订阅 SSE 事件流**：`GET /sessions/{session_id}/events/stream`，监听 `session_status` 变更及工具调用、响应等事件，直至状态变为 `idle` 或 `terminated`。 + +SDK 推荐使用：Python SDK ≥ v1.26.2，Java SDK ≥ v2.22.24。旧版本需重新安装升级（见 [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md)）。 ## 限制和注意事项 -- **配额限制**：单文件直传上限 **20 MB**，工作空间总容量上限 **100 GB**，文件保留期 **30 天**（[File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； -- **版本与快照**：Agent 更新采用全量替换 + 乐观锁（需带 `version` 字段），会话创建即锁定 Agent 与 Environment 快照，后续更新不影响运行中会话； -- **状态终态**：Session 归档（`POST /sessions/{id}/archive`）使其进入 `terminated` 终态，不可恢复；删除（`DELETE /sessions/{id}`）则彻底清除事件历史； -- **安全约束**：Skill 上传后必须通过安全扫描（`status: active`）才可挂载；File 仅 `available` 状态可被引用或挂载； -- **错误排查**：所有响应携带 `x-request-id`，提工单时务必提供该值以便定位问题。 +- **配额限制**： + - 单文件上传上限 **20 MB**，工作空间总容量 **100 GB**，文件保留期 **30 天**（见 [File](../../raw/application-api-reference/managed-agents-api/files-api.md)）； + - 分页 `limit` 最大值为 **100**； +- **版本与快照**： + - Agent 更新为全量替换 + 乐观锁（需携带当前 `version`），成功后 `version` 自动 +1；会话创建时锁定 Agent 版本，后续更新不影响已有会话； + - Environment 更新同样为全量替换，已绑定会话继续使用创建时的快照； +- **归档与删除**： + - Agent、Environment、Session 的 `archive` 均为软操作（保留数据，不可新建会话），而 `DELETE` 为硬删除（不可恢复）； +- **安全约束**： + - File 上传后需通过安全审核（`status` 为 `available` 才可用）； + - Skill 上传新版本后需等待扫描状态变为 `active` 方可挂载，`rejected` 状态需根据 `error_info` 修复后重传； +- **事件语义**：`POST /sessions/{session_id}/events` 仅用于注入事件（如用户输入、工具结果回填），**不触发执行**；执行由平台在收到用户消息后自动启动。 ## 来源文档 - [API 总览与认证](../../raw/application-api-reference/managed-agents-api/managed-agents-api-overview.md) -- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Agent](../../raw/application-api-reference/managed-agents-api/agent-api.md) +- [快速开始](../../raw/application-api-reference/managed-agents-api/managed-agents-quickstart.md) - [Session and Event](../../raw/application-api-reference/managed-agents-api/session-api.md) -- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) - [File](../../raw/application-api-reference/managed-agents-api/files-api.md) +- [Environment](../../raw/application-api-reference/managed-agents-api/environment-api.md) - [Skill](../../raw/application-api-reference/managed-agents-api/skills-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md index 981d9381..77a340a3 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/model-production.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/model-production.md @@ -1,44 +1,36 @@ # model production -model production 是百炼平台中用于模型定制化与服务化的关键能力集合，涵盖模型微调、部署及生命周期管理。它面向开发者提供标准化 API 接口，支持从训练到上线的端到端流程。所有操作均通过 RESTful API 或 SDK 调用，需配合百炼平台认证体系使用。 +`model production` 是百炼平台中用于将训练/微调后的模型投入实际服务的关键流程，涵盖模型微调、部署及生命周期管理。它为开发者提供从定制化训练到高可用推理服务的端到端能力。该能力依托统一 API 接口，支持自动化编排与可观测性集成。 -## 支持的模型/功能 +## 支持的模型与功能 -- **微调（Fine-tuning）**：支持基于预训练大语言模型（如 Qwen 系列）进行监督微调，适配下游任务（如指令遵循、领域问答）。输入为结构化 JSONL 格式数据集，支持 LoRA 等高效微调方法。 -- **部署（Deployment）**：支持将微调完成的模型或直接导入的兼容格式模型（如 GGUF、ONNX 导出模型）发布为高可用推理服务，自动分配 endpoint 并支持流量路由与扩缩容。 -- **模型版本管理**：每个微调任务生成唯一 `job_id`，对应产出模型可被多次部署；部署实例绑定 `model_id` 与 `version_id`，确保可追溯性。 -详见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 和 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md)。 +- **微调（Fine-tuning）**：支持基于基础大模型（如 Qwen 系列）进行监督微调，适配垂类任务（如客服问答、合同解析）。 +- **部署（Deployment）**：支持将微调完成的模型或通过 [模型导入](../../raw/model-api-reference/model-production/import-models-api.md) 接入的第三方模型，发布为带弹性扩缩容、流量灰度和版本管理的在线推理服务。 +- **模型注册与版本控制**：每个微调任务产出唯一 `model_id`，可被多次部署为不同环境（staging/prod）的独立 `deployment_id`。 ## 关键参数 -| 参数 | 类型 | 必填 | 说明 | -|------|------|------|------| -| `model` | string | 是 | 基座模型 ID（如 `qwen2.5-7b`），必须在 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md) 支持列表中 | -| `training_file` | string | 是（微调） | OSS 或 S3 URI，指向训练数据集（JSONL 格式） | -| `endpoint_name` | string | 是（部署） | 全局唯一标识符，长度 3–63 字符，仅含小写字母、数字和连字符 | -| `instance_type` | string | 否 | 默认 `gpu-a10`；部署时可选 `gpu-v100`、`cpu-small`（仅限测试） | +| 参数 | 说明 | 示例值 | +|------|------|--------| +| `base_model` | 微调所基于的基础模型 ID | `qwen2-7b-instruct` | +| `training_file` | 训练数据集对象 ID（需先通过文件上传 API 上传） | `file-abc123` | +| `deployment_name` | 部署服务名称，全局唯一且不可修改 | `prod-qa-service-v2` | +| `instance_type` | 推理实例规格（影响并发与延迟） | `ecs.gn7i-c8g1.2xlarge` | -> **注意**：文档 2 中提及“支持导入 ONNX 模型”，但当前版本（v2.4+）实际仅支持 ONNX 的 *推理兼容验证*，不支持 ONNX 模型直接部署；完整支持计划见 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 的“未来特性”章节（该内容已过时，以控制台 API 文档为准）。 +> **注意**：文档 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 中提及的 `autoscale_min_instances` 默认值为 `1`，但最新 SDK v3.2+ 已调整为 `0`（支持冷启缩容），请以 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 的 OpenAPI Schema 为准。 ## 使用方式 -1. **微调流程**： - - POST `/api/v1/fine_tuning_jobs`，携带 `model`、`training_file` 等参数； - - 轮询 `GET /api/v1/fine_tuning_jobs/{job_id}` 直至 `status == "succeeded"`； - - 提取响应中的 `fine_tuned_model_id` 用于后续部署。 - -2. **部署流程**： - - POST `/api/v1/deployments`，传入 `model_id`（来自微调结果）、`endpoint_name`、`instance_type`； - - 部署成功后，`endpoint_url` 可立即用于 `POST /v1/chat/completions` 请求。 - -完整示例代码与错误码说明参见 [模型调优](../../raw/model-api-reference/model-production/fine-tuning-jobs-api.md)。 +1. **微调模型**：调用 `POST /v1/fine_tuning/jobs`，传入 `base_model` 和 `training_file`；任务状态轮询 `GET /v1/fine_tuning/jobs/{job_id}`，成功后获取 `fine_tuned_model` 字段值（即新模型 ID）。 +2. **部署模型**：调用 `POST /v1/deployments`，指定上一步得到的 `model_id` 及 `instance_type` 等参数。 +3. **调用服务**：使用返回的 `deployment_id` 构造推理 endpoint（格式：`https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation?deployment_id={deployment_id}`），并按 [模型部署](../../raw/model-api-reference/model-production/deployments-api.md) 规范传入请求体。 ## 限制和注意事项 -- 单次微调最大训练时长为 72 小时，超时任务自动终止且不计费； -- 每个账号默认最多同时运行 3 个微调任务、5 个部署实例，配额可通过工单申请提升； -- 微调数据集须经敏感信息过滤（如 PII），平台不承担未脱敏数据导致的合规风险； -- 部署实例启动后需 2–5 分钟完成初始化，期间 `health_check` 返回 `503`，请实现重试逻辑。 +- 单次微调任务最长运行时间为 72 小时；超时自动终止，不产生费用。 +- 同一 `model_id` 最多可同时存在 5 个活跃部署（`status=active`），超出需先停用旧部署。 +- 微调数据集大小上限为 100 MB（压缩后），且仅支持 JSONL 格式，字段必须包含 `messages` 数组（遵循 OpenAI ChatML 协议）。 +- **重要**：微调任务一旦提交不可取消或修改；若需中止，请直接删除对应 job（调用 `DELETE /v1/fine_tuning/jobs/{job_id}`），但已产生的计算资源费用仍会计费至任务结束时刻。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md index f857ae5a..5fef4264 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-about-models.md @@ -1,62 +1,62 @@ # [more](more.md) about models -百炼平台提供多种模型调用机制与配套能力，涵盖同步/异步任务处理、多业务空间隔离、文件上传、连接优化等关键场景。本文面向开发者，系统梳理模型服务的核心能力、参数配置、使用方式及限制条件，帮助您高效、安全地集成模型能力。 +阿里云百炼平台支持多种模型调用模式，涵盖同步与异步任务、多模态文件处理、子业务空间隔离、连接复用优化及安全凭证管理等核心能力。本文面向开发者，系统梳理模型调用的关键机制、参数配置、使用约束及最佳实践，帮助构建高可用、高性能的 AI 应用。 ## 支持的模型/功能 -百炼支持两类主要模型调用模式： -- **同步模型**（如 `qwen-plus`、`qwen-max`）：适用于文本生成类请求，响应快、链路简单，直接返回结果； -- **异步模型**（如图像生成 `wanx2.1-t2i-turbo`、视频生成 `wanx2.1-kf2v-plus`、语音识别 `paraformer-16k-1`）：因处理耗时长，需通过任务 ID 分两步完成（提交 → 查询），并支持批量状态查询与取消 [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)。 +百炼平台支持两类主要调用路径： +- **同步模型**（如 `qwen-plus`、`qwen-max`）：适用于文本生成类低延迟场景，直接返回结果； +- **异步模型**（如图像生成 `wanx2.1-t2i-turbo`、视频生成 `wanx2.1-kf2v-plus`、语音识别 `paraformer-8k-v1`）：因处理耗时长，需通过任务 ID 轮询或事件通知获取结果。异步能力由统一的[异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md)提供支撑，覆盖任务创建、状态查询、批量检索与取消（仅限 `PENDING` 状态）。 -此外，部分[多模态](../concepts/multi-modal.md)模型（如 `qwen-vl-plus`）需传入文件 URL，平台提供免费临时 OSS 存储能力，上传后获得 `oss://` 格式 URL（有效期 48 小时）[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 -> **注意**：文档 3 中提到“文生图、文生视频提供了 SDK，SDK 已实现轮询”，但文档 2 明确指出异步任务接口本身**不内置轮询逻辑**，SDK 实现属封装层行为；实际调用仍需按文档 2 的接口规范自行轮询或接入事件通知。 +此外，平台支持多模态输入：调用图像/视频/音频模型前，需先上传本地文件获取临时 `oss://` URL（有效期 48 小时），且该 URL 必须与目标模型严格绑定，不可跨模型复用 [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 + +> **注意**：文档 3 中提到“调用在阿里云百炼[调优](https://help.aliyun.com/zh/model-studio/model-training-overview)并部署的模型，无需模型调用授权”，但文档 3 同时强调“此类模型仅能由其所在业务空间的 API Key 调用”。这与文档 6 中“临时API Key 继承生成它的API Key 所拥有的全部权限”存在隐含冲突——若父 Key 无子空间模型权限，则生成的临时 Key 也无法调用。实际行为以权限继承逻辑为准，建议在子空间中显式授权。 ## 关键参数 -| 参数 | 说明 | 典型值/范围 | 注意事项 | -|------|------|-------------|----------| -| `expire_in_seconds` | 临时 API Key 有效期 | `[1, 1800]` 秒 | 默认 60 秒，超时自动失效，不可手动删除 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) | -| `task_id` | 异步任务唯一标识 | UUID 字符串 | 必须用于查询或取消任务；任务完成后保留约 24 小时（具体以各模型文档为准） | -| `model_name` | 模型名称 | 如 `qwen-plus`, `wanx2.1-t2i-turbo` | 文件上传时必须指定且与后续调用模型一致；子业务空间调用需确保该空间已授权该模型 | -| `X-DashScope-OssResourceResolve` | 启用 OSS 资源解析 | `enable` | 使用 `oss://` URL 时**必须显式设置**此 Header，否则调用失败 | -| 连接池参数（Java/Python） | 控制 HTTP 连接复用 | 如 `connectionPoolSize=256`, `limit=100` | 高并发场景下需调优，避免阻塞或资源浪费 [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) | +| 参数 | 说明 | 典型值 | 注意事项 | +|------|------|--------|----------| +| `task_id` | 异步任务唯一标识 | `a8532587-xxxx-xxxx-xxxx-0c46b17950d1` | 查询/取消操作必需，需妥善存储 | +| `model_name` | 模型名称（区分大小写） | `qwen-plus`, `wanx2.1-t2i-turbo` | 文件上传、模型调用、事件过滤均需一致；子空间调用必须匹配该空间已授权模型 | +| `X-DashScope-OssResourceResolve: enable` | 使用 `oss://` URL 时必需的请求头 | `enable` | 缺失将导致模型调用失败，见[上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) | +| `expire_in_seconds` | 临时 API Key 有效期 | `1800`（30 分钟） | 范围 `[1, 1800]`，超时自动失效，不可手动删除 | +| 连接池参数（Java/Python SDK） | 控制 HTTP 连接复用行为 | `connectionPoolSize=256`, `limit=100` | 高并发场景下需调优，避免连接耗尽或服务端过载 | ## 使用方式 -### 1. 调用环境准备 -- 所有调用均需有效 API Key，并推荐配置为环境变量 `DASHSCOPE_API_KEY`； -- 子业务空间调用必须使用**该空间专属的 API Key**，且需提前在控制台为其授予对应模型权限 [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md)； -- 临时 API Key 适用于前端/移动端等不可信环境，由后端安全生成并透传 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)。 +### 异步任务处理 +- **轮询模式**：调用 `/api/v1/tasks/{task_id}` 查询单任务，或 `/api/v1/tasks/` 批量查询（支持按 `start_time`/`end_time`/`status`/`model_name` 过滤）。所有接口 QPS 限流为 **20**，需控制轮询频率。 +- **事件驱动模式**：通过[事件总线 EventBridge](../../raw/model-api-reference/more-about-models/async-task-api.md) 配置 HTTP 回调或 RocketMQ 目标，接收 `dashscope:System:AsyncTaskFinish` 事件，再发起一次结果查询。该方式规避轮询限流，适合高并发场景。 -### 2. 异步任务处理 -- **轮询模式**：调用 `/api/v1/tasks/{task_id}` 查询状态（20 QPS 限流），支持 `PENDING`/`RUNNING`/`SUCCEEDED`/`FAILED` 等状态判断； -- **事件驱动模式**：通过事件总线（EventBridge）配置 HTTP 回调或 RocketMQ 接收 `dashscope:System:AsyncTaskFinish` 事件，避免轮询 [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md)； -- **批量操作**：使用 `/api/v1/tasks/` 接口按时间、状态、模型名等条件批量查询任务；仅 `PENDING` 状态任务可取消。 +### 子业务空间调用 +- 必须使用**子空间专属 API Key**，不可混用默认空间 Key； +- OpenAI 兼容调用：`base_url` 设为 `https://dashscope.aliyuncs.com/compatible-mode/v1`； +- DashScope 原生调用：北京地域直接使用默认域名，新加坡地域需替换为 `{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`； +- 权限管控：标准模型需在子空间中[显式授权](https://help.aliyun.com/zh/model-studio/permission-management-overview#f642213a1f38l)，调优模型则自动继承空间归属权限。 -### 3. [多模态](../concepts/multi-modal.md)文件处理 -- 上传前调用 `GET /api/v1/uploads?action=getPolicy&model={model_name}` 获取凭证； -- 使用凭证直传 OSS，获得 `oss://` URL； -- 在模型请求中传入该 URL，并在 Header 中添加 `X-DashScope-OssResourceResolve: enable`。 +### 连接复用配置 +- **Java SDK**：通过 `Constants.connectionConfigurations` 设置连接池参数（如 `connectionPoolSize`、`maximumAsyncRequests`），默认启用； +- **Python SDK**：同步调用传入 `requests.Session`，异步调用传入 `aiohttp.ClientSession`，推荐使用 `with` 语句管理生命周期。 -### 4. SDK 连接优化 -- **Java**：通过 `Constants.connectionConfigurations` 全局配置连接池参数（如 `connectionPoolSize`, `readTimeout`）； -- **Python**：同步调用传入 `requests.Session`，异步调用传入 `aiohttp.ClientSession`，复用底层 TCP 连接。 +### 安全凭证管理 +- 在前端/移动端等不可信环境，应由后端服务调用 `/api/v1/tokens` 接口生成临时 API Key（TTL 可设），避免永久 Key 泄露； +- 临时 Key 权限完全继承父 Key，包括模型访问范围与知识库限制。 ## 限制和注意事项 -- **临时存储限制**：`oss://` URL 有效期严格为 **48 小时**，过期即失效；文件大小上限 **1GB**；上传接口限流 **100 QPS（主账号+模型维度）**，**严禁用于生产环境或压测**，生产环境应使用阿里云 OSS [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)； -- **地域隔离**：API Key、Endpoint、临时 [Token](../concepts/token.md) 均按地域（北京/新加坡/弗吉尼亚）独立，跨地域调用将失败； -- **权限继承**：临时 API Key 继承其生成者 API Key 的全部权限（含模型/知识库访问限制），无额外管控能力 [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md)； -- **子空间约束**：在子业务空间调优部署的模型**仅能被该空间的 API Key 调用**，且不支持 OpenAI 兼容方式；标准模型调用则需显式授权； -- **异步任务清理**：已完成任务数据约保留 24 小时，超时后无法查询，需及时获取结果。 +- **文件上传**：单文件 ≤ 1 GB；QPS 限流 **100**（按主账号+模型维度）；临时 URL 仅 48 小时有效，**严禁用于生产环境**；生产环境请使用 OSS 自建存储 [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md)。 +- **异步任务保留期**：任务完成后数据默认保留 **24 小时**，超时后无法查询，需及时获取结果。 +- **临时 API Key**：最大 TTL 为 1800 秒（30 分钟），到期自动失效，不支持提前撤销。 +- **地域与 Endpoint**：北京与新加坡地域的 API Key 不互通，Endpoint 域名结构不同（如新加坡需带 `WorkspaceId`），调用前务必确认地域配置。 +- **权限继承风险**：子空间模型调用权限与临时 API Key 的权限均严格继承自父 API Key。若父 Key 权限过大（如可调用所有模型），可能违反最小权限原则，建议为不同场景创建专用 Key 并精细授权。 ## 来源文档 -- [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) - [异步任务管理 API](../../raw/model-api-reference/more-about-models/manage-asynchronous-tasks.md) - [通过HTTP回调URL或MQ接收异步任务完成通知](../../raw/model-api-reference/more-about-models/async-task-api.md) - [子业务空间的模型调用](../../raw/model-api-reference/more-about-models/model-calling-in-sub-workspace.md) - [上传本地文件获取临时URL](../../raw/model-api-reference/more-about-models/get-temporary-file-url.md) - [DashScope SDK连接复用配置](../../raw/model-api-reference/more-about-models/connection-multiplexing-configuration.md) +- [生成临时API Key](../../raw/model-api-reference/more-about-models/generate-temporary-api-key.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md index b623e0f3..8ed96440 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more-models.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more-models.md @@ -1,82 +1,64 @@ # [more](more.md) models -百炼平台提供一系列面向垂直场景的专用模型，覆盖法律、翻译、深度研究、OCR、GUI自动化和意图理解等能力。这些模型基于通义千问系列基座模型精调或增强，支持通过 DashScope SDK 或 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)调用。所有模型均需配置业务空间专属域名以获得最佳性能与稳定性。 +百炼平台提供一系列面向垂直场景的专用大模型，覆盖意图理解、机器翻译、深度研究、GUI自动化和法律推理等任务。这些模型在特定领域经过强化训练或架构优化，相比通用模型具备更高的准确率、更优的响应结构和更强的领域适配能力。开发者可通过 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)或 DashScope SDK 调用，需注意地域限制、域名迁移要求及模型特有的输入格式约束。 ## 支持的模型/功能 -- **通义法睿（`farui-plus`）**：法律行业专用大模型，支持法律咨询、文书生成、案情分析、合同审查等，详见 [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md)。 -- **Qwen-MT（`qwen-mt-plus`）**：机器翻译模型，支持术语干预、翻译记忆、领域提示等高级功能，兼容 OpenAI 接口，详见 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md)。 -- **Qwen-Deep-Research（`qwen-deep-research`）**：支持两阶段交互式深度研究（反问确认 + 网络检索增强），**仅限华北2（北京）地域且仅支持 Python DashScope SDK**，不支持 Java SDK 或 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，详见 [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)。 -- **Qwen-OCR（`qwen3.5-ocr`）**：[多模态](../concepts/multi-modal.md) OCR 模型，支持图像输入与结构化文本提取（如车票信息），支持流式与非[流式输出](../concepts/streaming-output.md)，详见 [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md)。 -- **GUI-Plus（`gui-plus-2026-02-26`）**：界面交互专用模型，可调用 `computer_use` 工具执行鼠标/键盘操作并解析截图，适用于自动化 GUI 测试与桌面任务，详见 [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md)。 -- **意图理解（`tongyi-intent-detect-v3`）**：毫秒级意图识别模型，支持两种模式：① 输出结构化工具调用（需 `INTENT_MODE` system [prompt](../guides/prompt.md)）；② 仅输出预定义意图标签（支持单 [Token](../concepts/token.md) 响应优化），详见 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 +当前 `more models` 类别下支持以下专用模型： -> **注意**：文档 2 和文档 4 均重复列出新加坡/美国地域的 `base_url` 配置两次，属冗余描述，实际使用时按地域选择唯一正确地址即可；文档 5 中 GUI-Plus 的 `vl_high_resolution_images` 参数在文档 4（Qwen-OCR）中亦有相同用法，但未在文档 5 明确说明其作用，建议开发者参考 Qwen-OCR 文档中关于 `min_pixels`/`max_pixels` 的图像预处理逻辑进行适配。 +- **`tongyi-intent-detect-v3`**：意图理解模型，支持两种模式： + - `INTENT_MODE`：输出结构化[函数调用](../concepts/function-calling.md)（含工具名与参数），适用于 Agent 场景； + - 纯标签模式：从预定义意图字典中返回单个语义标签（如 `alarm_set`），支持单 Token 输出以提升延迟敏感型场景性能。详见 [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md)。 +- **`qwen-mt-plus`**：专业级机器翻译模型，支持源/目标语言自动识别、术语干预（`terms`）、翻译记忆（`tm_list`）和领域提示，适用于技术文档、合同等高保真翻译场景。 +- **`qwen-deep-research`**：深度研究模型，采用两阶段工作流（反问确认 → 深入研究），支持联网搜索、引用溯源与多粒度报告生成（`model_detailed_report` / `model_summary_report`）。**注意**：该模型[仅支持华北2（北京）地域且不支持 OpenAI 兼容接口](../../raw/model-api-reference/more-models/qwen-deep-research-api.md)，必须使用 DashScope Python SDK 调用。 +- **`gui-plus-2026-02-26`**：GUI 自动化模型，专为桌面界面交互设计，接受截图（`image_url`）与自然语言指令，输出鼠标/键盘操作指令（如 `left_click`, `type`），需严格遵循 `` + `...` 响应格式。 +- **`farui-plus`**：法律行业大模型，集成 RAG、法律 Agent 和司法小模型，支持法律咨询、文书生成（如起诉书）、案情分析与合同审查。其上下文长度达 12k Token，但输出成本显著高于通用模型（20元/百万 Token）。 + +> **注意**：文档 2 中重复列出了北京、新加坡、美国地域的配置说明，且对新加坡和美国地域的 `base_url` 描述存在冗余与不一致（如美国地域未提供 WorkspaceId 占位符），实际应以 [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) 中“重要”区块的迁移指引为准，即仅北京与新加坡地域启用 WorkspaceId 专属域名，美国地域仍使用 `dashscope-us.aliyuncs.com`。 ## 关键参数 -| 参数 | 类型 | 说明 | 必填 | -|------|------|------|------| -| `model` | string | 模型名称，如 `"farui-plus"`、`"qwen-mt-plus"` 等 | ✅ | -| `messages` | array | 对话消息列表，含 `role`（`user`/`system`/`assistant`）与 `content`；OCR/GUI-Plus 支持 `image_url` 类型内容 | ✅ | -| `result_format` / `response_format` | string | DashScope SDK 使用 `result_format='message'`；[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)默认为 `chat.completions` 格式 | ❌（默认） | -| `stream` | boolean | 启用[流式输出](../concepts/streaming-output.md)（需配合 `stream_options={"include_usage": true}` 获取 token 统计） | ❌（默认 false） | -| `extra_body` | object | OpenAI 兼容接口扩展字段：
• `qwen-mt-plus`: `{"translation_options": {...}}`
• `gui-plus-*`: `{"vl_high_resolution_images": true}`
• `tongyi-intent-detect-v3`: 无特殊字段，依赖 system [prompt](../guides/prompt.md) 控制行为 | ❌（按需） | -| `output_format` | string | 仅 `qwen-deep-research` 支持：`"model_detailed_report"`（默认）或 `"model_summary_report"` | ❌（默认） | +| 参数 | 说明 | 示例值 | 文档依据 | +|------|------|--------|----------| +| `model` | 必选，模型标识符 | `"tongyi-intent-detect-v3"`, `"qwen-mt-plus"` | [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md), [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | +| `translation_options` | `qwen-mt-plus` 专用，包含 `source_lang`, `target_lang`, `terms`, `tm_list` | `{"source_lang": "Chinese", "target_lang": "English", "terms": [...]}` | [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | +| `output_format` | `qwen-deep-research` 专用，控制报告粒度 | `"model_detailed_report"` (默认), `"model_summary_report"` | [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) | +| `extra_body` | [OpenAI 兼容接口](../concepts/openai-compatible-api.md)中传递非标准字段（如 `vl_high_resolution_images`, `translation_options`） | `{"vl_high_resolution_images": true}` | [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md), [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) | +| `stream` & `incremental_output` | 控制[流式输出](../concepts/streaming-output.md)行为（`qwen-deep-research`, `farui-plus` 等支持） | `True`, `True` | [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md), [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) | ## 使用方式 -1. **环境准备** - - 获取并配置 API Key 到环境变量 `DASHSCOPE_API_KEY`（[获取API Key](https://help.aliyun.com/zh/model-studio/get-api-key)）； - - 安装对应 SDK：`pip install dashscope`（DashScope）或 `pip install openai`（OpenAI 兼容）； - - **必须配置业务空间专属域名**（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），旧域名（`dashscope.aliyuncs.com`）虽仍可用，但性能与稳定性较低。 - -2. **调用示例（通用流程）** - ```python - # DashScope SDK（以 farui-plus 为例） - import dashscope - dashscope.base_http_api_url = "https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1" - response = dashscope.Generation.call( - model="farui-plus", - messages=[{"role": "user", "content": "生成一份离婚协议书"}], - result_format="message" - ) - ``` - - ```python - # OpenAI 兼容接口（以 qwen-mt-plus 为例） - from openai import OpenAI - client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" - ) - completion = client.chat.completions.create( - model="qwen-mt-plus", - messages=[{"role": "user", "content": "我看到这个视频后没有笑"}], - extra_body={"translation_options": {"source_lang": "Chinese", "target_lang": "English"}} - ) - ``` - -3. **特殊模型注意事项** - - `qwen-deep-research`：必须分两步调用（先反问确认，再传入 assistant 回复 + user 补充指令）； - - `gui-plus-*`：system [prompt](../guides/prompt.md) 必须包含完整 `` 定义与 `Response format` 规则； - - `tongyi-intent-detect-v3`：意图识别模式由 system prompt 决定——含 `INTENT_MODE` 则输出 ``/ 块；否则仅输出纯标签字符串。 +1. **环境准备**： + - 获取并配置 API Key（推荐设为环境变量 `DASHSCOPE_API_KEY`）； + - 安装对应 SDK（DashScope SDK 或 OpenAI SDK）； + - **强制迁移域名**：北京/新加坡地域必须使用 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，`{WorkspaceId}` 在控制台业务空间详情页获取。 + +2. **请求构造**： + - `tongyi-intent-detect-v3`：System Message 必须显式声明 `Response in INTENT_MODE.`（[函数调用](../concepts/function-calling.md)）或 `just reply with the chosen tag.`（纯标签）； + - `qwen-mt-plus`：将 `translation_options` 作为 `extra_body`（OpenAI）或顶层字段（DashScope）传入； + - `gui-plus-2026-02-26`：`messages[0].content` 需为含 `image_url` 的数组，且 System Prompt 必须包含 `` 定义与 `` 格式规范； + - `qwen-deep-research`：严格按两阶段调用——首请求仅含用户初始问题，第二请求需拼接 `user → assistant → user` 三元组； + - `farui-plus`：支持标准单轮/多轮对话及[流式输出](../concepts/streaming-output.md)（设置 `stream=True` 与 `incremental_output=True`）。 + +3. **响应解析**： + - `tongyi-intent-detect-v3`（INTENT_MODE）：需用正则提取 ``, ``, `` 三段内容，并 `json.loads()` 解析 `tool_call` 字段； + - `qwen-deep-research`：响应含 `phase` 字段（如 `answer`, `WebResearch`），需按阶段处理 `extra.deep_research.references` 等结构化数据； + - 其他模型：遵循标准 OpenAI/DashScope 响应格式，`choices[0].message.content` 为文本结果。 ## 限制和注意事项 -- **地域限制**：`qwen-deep-research` 仅支持华北2（北京）地域；其他模型（如 `qwen-mt-plus`、`qwen3.5-ocr`、`gui-plus-*`）在华北2、新加坡、美国（弗吉尼亚）三地均可用，但需使用对应地域的 `base_url` 和独立 API Key。 -- **SDK 支持差异**：`qwen-deep-research` **不支持 Java SDK 与 OpenAI 兼容接口**（文档明确说明），仅支持 Python DashScope SDK；其余模型均支持两种调用方式。 -- **输入格式约束**：OCR 与 GUI-Plus 模型要求 `messages[0].content` 为数组，内含 `image_url` 和 `text` 对象；普通文本模型（如 `farui-plus`、`tongyi-intent-detect-v3`）则接受字符串 `content`。 -- **成本与限流**：各模型按输入/输出 token 计费（如 `farui-plus` 输入 20元/百万 token），具体见各模型文档表格；全局限流策略参见 [限流](https://help.aliyun.com/zh/model-studio/rate-limit)，未在原始文档中统一说明。 -- **安全实践**：强烈建议将 `DASHSCOPE_API_KEY` 配置为环境变量，避免硬编码或日志泄露（[配置API Key到环境变量](https://help.aliyun.com/zh/model-studio/configure-api-key-through-environment-variables)）。 +- **地域限制**：`qwen-deep-research` 仅支持华北2（北京）地域，其他模型（如 `qwen-mt-plus`）在美国（弗吉尼亚）地域使用公共域名 `dashscope-us.aliyuncs.com`，不支持 WorkspaceId。 +- **接口限制**：`qwen-deep-research` 不支持 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)，仅 DashScope Python SDK 可用；`gui-plus-2026-02-26` 要求输入图像 URL 可公开访问，且 `extra_body={"vl_high_resolution_images": True}` 为必需项。 +- **成本与限流**：`farui-plus` 输入成本为 20元/百万 Token，显著高于其他模型；所有模型的限流策略详见 [限流](https://help.aliyun.com/zh/model-studio/rate-limit)，需在生产环境做好熔断与重试。 +- **稳定性建议**：北京/新加坡地域务必迁移至 WorkspaceId 专属域名，[原文标题](../../raw/model-api-reference/more-models/intent-detect-capability.md) 明确指出其“能够为推理请求提供卓越的性能和更高的稳定性”。 +- **安全实践**：API Key 绝不可硬编码，必须通过环境变量注入；DashScope Java SDK 对象非线程安全，需自行管理同步机制。 ## 来源文档 -- [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) +- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) - [Qwen-MT API参考](../../raw/model-api-reference/more-models/qwen-mt-api.md) - [Qwen-Deep-Research API 参考](../../raw/model-api-reference/more-models/qwen-deep-research-api.md) -- [Qwen-OCR API参考](../../raw/model-api-reference/more-models/qwen-vl-ocr-api-reference.md) - [GUI-Plus API参考](../../raw/model-api-reference/more-models/gui-plus-interface-interaction-model.md) -- [意图理解能力](../../raw/model-api-reference/more-models/intent-detect-capability.md) +- [通义法睿大语言模型](../../raw/model-api-reference/more-models/tongyi-farui-api.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/more.md b/skills/bailian-docs-llm-wiki/wiki/api/more.md index e43728f1..7678447a 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/more.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/more.md @@ -1,53 +1,37 @@ # more -`more` 是百炼平台面向高级用例提供的扩展能力集合，涵盖临时凭证管理、服务权限委托和知识库精细化检索三大核心方向。它不构成独立 API 服务，而是作为模型调用、工作流编排和 RAG 场景的支撑性机制，需结合具体功能模块（如 `Retrieve`、函数计算节点、安全存储空间等）协同使用。开发者应根据实际场景选择对应能力，并严格遵循权限最小化原则。 +`more` 是百炼平台面向高级用例提供的扩展能力集合，涵盖临时认证、服务集成授权与知识库精细化检索三大核心方向。它不构成独立 API 服务，而是作为模型调用、工作流编排和 RAG 场景的支撑机制存在，适用于需兼顾安全性、跨云服务协同及结构化数据过滤的生产级应用。 ## 支持的模型/功能 -`more` 本身不提供模型推理能力，但为以下关键功能提供底层支持： +`more` 不直接对应特定模型，而是为以下功能提供底层支持： -- **临时 API Key 生成**：用于在浏览器、移动端等不可信环境安全调用模型服务（如 `qwen-max`、`qwen-plus` 等所有支持 DashScope 协议的模型），避免永久密钥泄露 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 -- **服务关联角色（SLR）**：为百炼工作流、数据管理、安全存储空间、知识库、用量监控等模块自动创建并托管 RAM 角色，实现对 FC、OSS、ADB-PG、MNS、SLS 等云服务的安全访问授权 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 -- **知识库 SearchFilters**：在 `Retrieve` 接口调用中对语义检索结果进行结构化过滤，支持单值、多值、范围、模糊及标签查询，显著提升结构化数据（如员工表、产品目录）的召回精度 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +- **临时 API Key 生成**：用于在浏览器、移动端等不可信环境安全调用模型服务（如 `qwen-max`、`qwen-plus` 等），避免永久密钥泄露 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)； +- **服务关联角色（SLR）自动管理**：支撑百炼与函数计算（FC）、OSS、ADB-PG、MNS、OpenTelemetry 等阿里云服务的受控集成，覆盖工作流节点、数据导入、安全存储、用量监控等场景 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)； +- **知识库 `SearchFilters` 检索过滤**：在 `Retrieve` 接口调用中对语义检索结果进行结构化字段级过滤（如 `姓名: "张三"`、`年龄: {"gte": 20, "lte": 27}`），显著提升 RAG 输出精度 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 -> **注意**：文档 2 中列出的 `AliyunServiceRoleForSFMAccessFC` 权限仅包含 `fc:ListFunctions` 和 `fc:InvokeFunction`，但实际工作流调用 FC 函数可能还需 `fc:GetFunction` 等元数据权限；建议以控制台实际授予策略为准，而非仅依赖文档描述。 +> **注意**：文档 2 中 `AliyunServiceRoleForSFMTelemetry` 的权限策略示例被截断（末尾缺失 `}`），实际部署时请以 RAM 控制台中该策略的完整 JSON 为准；同时，其关联的 `proj-xtrace-*` Logstore 命名规则与当前百炼控制台默认项目命名不一致，建议通过 [用量监控与性能分析](https://help.aliyun.com/zh/model-studio/application-observation) 页面确认实际 Project 名称。 ## 关键参数 -| 参数/字段 | 所属能力 | 类型 | 说明 | 示例 | -|-----------|----------|------|------|------| -| `expire_in_seconds` | 临时 API Key | integer | TTL 有效期，单位秒，取值范围 `[1, 1800]` | `1800`（30 分钟） | -| `searchFilters` | 知识库检索 | array of object | 过滤条件数组，每个元素为一个子分组（AND 语义），支持 `{"字段名": "值"}` 或高级语法如 `{"年龄": {"gte": 20, "lte": 30}}` | `[{"姓名": "张三"}, {"岗位": "技术员"}]` | -| `token` | 临时 API Key 响应 | string | 生成的短期凭证，格式为 `st-***` | `st-9a8b7c6d...` | -| `expires_at` | 临时 API Key 响应 | number | UNIX 时间戳，表示过期时间 | `1744080369` | +| 功能 | 参数名 | 类型 | 说明 | 取值范围/示例 | +|------|--------|------|------|----------------| +| **临时 API Key** | `expire_in_seconds` | Integer | Token 有效期（TTL） | `[1, 1800]` 秒，默认 `60`；示例：`?expire_in_seconds=1800` | +| **SearchFilters** | `searchFilters` | Array of Object | 过滤条件数组，每个 Object 为一个 AND 分组 | `[{"姓名": "张三"}, {"岗位": "技术员"}]`；支持单值、多值、范围（`gt`/`gte`/`lt`/`lte`/`eq`/`neq`）、模糊（`like`）和标签（`tags`）查询 | +| **SearchFilters（范围查询）** | 字段值格式 | String (JSON) | 范围条件需序列化为 JSON 字符串 | `"年龄": "{\"gte\": 20, \"lte\": 27}\"` | ## 使用方式 -### 临时 API Key -1. 在后端服务中配置永久 `DASHSCOPE_API_KEY` 环境变量； -2. 向 `https://dashscope.aliyuncs.com/api/v1/tokens?expire_in_seconds=1800` 发起 POST 请求（北京地域）或对应地域 Endpoint； -3. 将响应中的 `token` 作为 `Authorization: Bearer ` 用于后续模型调用。 - -### 服务关联角色 -- **无需手动创建**：当首次在控制台启用对应功能（如添加函数计算节点、配置 OSS 数据源）时，系统自动创建 SLR； -- **权限验证**：可在 [RAM 控制台](https://ram.console.aliyun.com/) 查看角色及绑定策略； -- **删除前提**：必须先解除该角色所依赖的所有业务配置（如删除函数计算节点、断开 OSS 连接等），否则删除失败。 - -### SearchFilters -- 在 `RetrieveRequest` 请求体中直接传入 `searchFilters` 字段； -- 每个子分组内支持多种查询语法： - - 单值：`{"姓名": "张三"}` - - 范围：`{"年龄": {"gte": 25, "lte": 35}}` - - 模糊：`{"岗位": {"like": "技%员"}}` - - 多值（需 JSON 序列化）：`{"姓名": "[\"张三\",\"李四\"]"}` -- 注意：子分组间为 AND 关系，不可更改；标签查询仅适用于文档/音视频类知识库。 +- **临时 API Key**：通过 `POST https://dashscope.aliyuncs.com/api/v1/tokens` 发起请求，需在 `Authorization` Header 中携带主账号的永久 `DASHSCOPE_API_KEY`；响应返回 `token`（前缀 `st-`）与 `expires_at`（Unix 时间戳）。该 token 可直接用于后续模型 API 调用的 `Authorization: Bearer `。 +- **服务关联角色**：无需手动创建。当您在百炼控制台首次启用对应功能（如添加 FC 节点、配置 OSS 数据源、开通安全存储空间等）时，系统自动创建并绑定 SLR。角色删除需先解除所有依赖资源（如删除工作流中的 FC 节点、断开 OSS 连接等），再通过 RAM 控制台操作。 +- **SearchFilters**：在 `Retrieve` 请求体中直接传入 `searchFilters` 字段（非 Query 参数）。需确保知识库索引已将目标字段（如 `姓名`、`年龄`）配置为可检索字段；多值查询需将数组 `json.dumps` 后作为字符串传入字段值；模糊查询使用 `{"字段名": "{\"like\": \"值%\"}\"}` 格式。 ## 限制和注意事项 -- **临时 API Key**：无法提前撤销，到期自动失效；继承父密钥全部权限，**不得用于高权限操作场景**；各地域 Endpoint 不互通，需按实际部署地域调用 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 -- **服务关联角色**：删除前必须满足前置清理条件（如文档 2 中明确要求“删除所有已发布的工作流应用中的函数计算节点”），否则操作被拒绝；`AliyunServiceRoleForSFMAccessingMNS` 明确禁止用户修改或删除 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 -- **SearchFilters**：仅对已索引字段生效，未在知识库配置中启用“参与检索”的字段无法过滤；多值查询需将数组 JSON 序列化为字符串传入（见文档 3 Python 示例）；模糊查询 `like` 仅支持 `%` 通配符，不支持正则表达式 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 -- **通用限制**：所有 `more` 相关能力均受百炼配额与计费规则约束，临时 Key 调用计入调用者配额；SLR 权限变更可能影响已有工作流执行，请在生产环境变更前充分测试。 +- 临时 API Key **不可撤销**，仅能等待自然过期；其权限完全继承自签发所用的永久 API Key，包括模型访问白名单与知识库权限 [生成临时API Key](../../raw/application-api-reference/more/application-obtain-temporary-authentication-token.md)。 +- 所有服务关联角色均含 `ram:DeleteServiceLinkedRole` 权限，但**删除后将导致对应功能完全不可用**（如删 `AliyunServiceRoleForSFMAccessFC` 后无法调用 FC 节点），且恢复需重新触发功能开通流程 [服务关联角色](../../raw/application-api-reference/more/bailian-service-linked-role.md)。 +- `SearchFilters` 仅作用于 `Retrieve` 接口，**不改变向量索引本身**；过滤发生在语义检索之后，因此仍需保证原始检索召回质量；标签（`tags`）查询仅支持文档搜索、音视频搜索类知识库，且多个 `tags` 数组间为 OR 关系 [知识库SearchFilters](../../raw/application-api-reference/more/how-to-use-search-filters.md)。 +- 各地域（北京/新加坡/弗吉尼亚）的临时 Token Endpoint 和 DashScope API Endpoint **不互通**，必须使用与主 API Key 相同地域的 Endpoint。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md index b57b1443..914bd9e1 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/omni-realtime-api.md @@ -1,89 +1,78 @@ # omni realtime api -Qwen-Omni Realtime API 是阿里云百炼平台提供的低延迟、[多模态](../concepts/multi-modal.md)实时交互接口，支持语音/音视频输入与文本/音频输出的流式双向通信。它基于 WebSocket 协议，内置 VAD（语音活动检测）、ASR（语音识别）、LLM 推理与 TTS（语音合成）全链路能力，适用于智能客服、虚拟助手、实时会议等场景。开发者可通过 Python 或 Java SDK 快速集成，无需自行编排模型调用流程。 +Qwen-Omni-Realtime API 是一个基于 WebSocket 的实时多模态交互接口，支持语音、文本、图像输入与文本+音频同步输出。它采用事件驱动架构，客户端通过发送标准化事件（如 `session.update`、`input_audio_buffer.append`）控制会话状态和数据流，服务端通过异步事件（如 `input_audio_buffer.speech_stopped`、`response.audio.delta`）实时反馈处理结果。该 API 专为低延迟、高保真语音交互场景设计，适用于智能客服、虚拟助手等实时对话应用。 -## 支持的模型与功能 +## 支持的模型/功能 -当前支持以下 Qwen-Omni 实时系列模型，各模型能力存在差异，需按需选型： +- **核心模型系列**： + - `qwen3.5-omni-realtime`：支持 `semantic_vad`、联网搜索（`enable_search`）及完整工具调用（`tools`）。 + - `qwen3.5-omni-plus-realtime` / `qwen3.5-omni-flash-realtime`：支持 `idle_timeout_ms` 静默超时主动引导，但**不支持 `semantic_vad`**（仅 `server_vad`）。 + - `qwen3-omni-flash-realtime`：支持 `smooth_output` 控制口语化/书面化风格。 + - `qwen-omni-turbo-realtime`：轻量级模型，**所有生成参数（`temperature`、`top_p`、`max_tokens` 等）均不可修改**，仅支持基础语音交互 [原文标题](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 -- **`qwen3.5-omni-realtime`**：基础旗舰版，支持 `semantic_vad`、联网搜索（`enable_search`）和工具调用（`tools`），是唯一同时支持三者的模型。 -- **`qwen3.5-omni-plus-realtime` 与 `qwen3.5-omni-flash-realtime`**：增强与轻量变体，支持 `idle_timeout_ms` 等高级 VAD 参数，但**不支持 `semantic_vad`**；仅 `plus` 版支持声音复刻驱动（见 [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)）。 -- **`qwen3-omni-flash-realtime`**：侧重响应速度，支持 `smooth_output` 口语化控制，但**不支持联网搜索与工具调用**。 -- **`qwen-omni-turbo-realtime`**：极致轻量版，参数（如 `temperature`、`top_p`、`max_tokens` 等）**完全不可修改**，仅支持基础对话。 +- **多模态能力**： + - 输入：实时 PCM 音频（16 kHz）、JPG/JPEG 图像（≤1080p，Base64 编码后 ≤256 KB）、文本（通过 `instructions` 或 ASR 转录）。 + - 输出：文本 + PCM 音频（24 kHz），可选仅文本模式（`modalities: ["text"]`）。 + - 工具调用：仅 `qwen3.5-omni-realtime` 系列支持，需配置 `tools` 数组，模型触发后返回 `function_call` 项，客户端回传结果后需显式调用 `response.create` [原文标题](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 + - 联网搜索：仅 `qwen3.5-omni-realtime` 系列支持 `enable_search`，且与 `tools` **互斥**，不可同时启用 [原文标题](../../raw/model-api-reference/omni-realtime-api/client-events.md)。 -> **注意**：文档 1 和文档 6 均称 `qwen3.5-omni-realtime` 支持 `semantic_vad`，而文档 2 明确指出该能力“仅 `qwen3.5-omni-realtime` 系列模型支持”，但文档 4 的 `session.created` 示例中 `turn_detection.type` 字段注释却写为“取值为 `server_vad` 或 `semantic_vad`（仅 `qwen3.5-omni-realtime` 支持）”，存在表述冗余。以文档 1 和文档 2 的明确限定为准：`semantic_vad` 为 `qwen3.5-omni-realtime` 独占特性。 - -所有模型均支持： -- [多模态](../concepts/multi-modal.md)输入：纯音频、音视频（`append_audio` + `append_video`） -- [多模态](../concepts/multi-modal.md)输出：文本（`TEXT`）与音频（`AUDIO`）组合 -- 实时语音转录（ASR）：固定使用 `qwen3-asr-flash-realtime` 模型，不可替换 -- 声音复刻音色接入：需确保复刻时指定的 `target_model` 与实时对话模型严格一致（详见 [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)） +> **注意**：文档 6 中提到的 `qwen3.5-omni-plus` 和 `qwen3.5-omni-flash` 是**非实时（batch）模型**，不能用于本 API；本 API 仅接受 `-realtime` 后缀的模型名（如 `qwen3.5-omni-plus-realtime`），否则连接将失败。 ## 关键参数 -参数分为连接级（构造时设置）与会话级（`update_session` 时设置），部分参数模型间行为不同： +所有参数均通过 `session.update` 事件或 SDK 的 `update_session` 方法配置，分为以下几类： + +- **基础配置**： + - `modalities`: `["text"]` 或 `["text","audio"]`（默认），不支持 `["audio"]` 单独输出。 + - `voice`: 音色名，不同模型默认值不同（如 `qwen3.5-omni-realtime` 默认 `Tina`），亦可使用声音复刻生成的自定义音色 [原文标题](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md)。 + - `input_audio_format` / `output_audio_format`: 固定为 `"pcm"`，对应 16 kHz 输入 / 24 kHz 输出。 -| 参数 | 类型 | 说明 | 模型兼容性 | -|------|------|------|------------| -| `model` | `str`/`String` | 模型名称，如 `"qwen3.5-omni-realtime"` | 所有模型 | -| `url` | `str`/`String` | WebSocket 地址，**必须使用业务空间专属域名**：
`wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime`（北京）
`wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com/api-ws/v1/realtime`（新加坡） | 所有模型，[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 与 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) 均强调此迁移要求 | -| `output_modalities` | `list[MultiModality]`/`List` | 输出模态，`[TEXT]` 或 `[TEXT, AUDIO]`（默认） | 所有模型 | -| `voice` | `str`/`String` | 音色名，如 `"Tina"`；自定义音色需通过声音复刻获取 | 所有模型 | -| `turn_detection_type` | `str`/`String` | VAD 类型：`"server_vad"`（默认）或 `"semantic_vad"`（仅 `qwen3.5-omni-realtime`） | 见上文注意项 | -| `enable_search` | `bool`/`Boolean` | 启用联网搜索，**与 `tools` 互斥** | 仅 `qwen3.5-omni-realtime` | -| `tools` | `list[dict]`/`List>` | 工具定义列表，**与 `enable_search` 互斥** | 仅 `qwen3.5-omni-realtime` | -| `smooth_output` | `bool`/`Boolean` | 口语化开关，`true`/`false`/`null`，**仅 `qwen3-omni-flash-realtime` 支持** | 仅 `qwen3-omni-flash-realtime` | -| `temperature` / `top_p` / `top_k` / `max_tokens` 等采样参数 | 各自类型 | 控制生成多样性与长度，详见各文档默认值表 | `qwen-omni-turbo-realtime` 系列**全部不可修改** | +- **VAD 控制**（语音活动检测）： + - `turn_detection.type`: `"server_vad"`（默认）或 `"semantic_vad"`（仅 `qwen3.5-omni-realtime` 支持）。 + - `turn_detection.threshold`: [-1.0, 1.0]，值越低越灵敏（易误触），默认 `0.5`。 + - `turn_detection.silence_duration_ms`: [200, 6000] ms，静音超时触发响应，默认 `800`。 + - `turn_detection.idle_timeout_ms`: [5000, 30000] ms，**仅 `qwen3.5-omni-plus-realtime` 或 `qwen3.5-omni-flash-realtime` 在 `server_vad` 模式下生效**，用于静默后主动引导。 -> **注意**：`repetition_penalty` 默认值在文档 1 中写为“其他模型：1.05”，而文档 2 写为“`qwen3-omni-flash-realtime` 系列：1.05；`qwen-omni-turbo-realtime` 系列：1.05”，文档 4 的 `session.created` 示例中亦为 `1.05`。文档 6 未提及其他模型默认值，仅重复 `qwen3.5-omni-realtime` 为 `1.0`。此处以文档 2 的完整列表为准。 +- **生成控制**（部分参数在 `qwen-omni-turbo-realtime` 上不可修改）： + - `temperature` / `top_p`: 控制多样性，建议**二选一**设置。默认值因模型而异（如 `qwen3.5-omni-realtime`: `0.7` / `0.8`）。 + - `top_k`: 候选 Token 数，≥0，设为 `null` 或 >100 时禁用。 + - `max_tokens`: 最大输出长度，超长则截断，不影响生成过程。 + - `repetition_penalty` / `presence_penalty`: 控制重复度，默认值模型间有差异（如 `qwen3.5-omni-realtime`: `1.0` / `1.5`）。 + - `seed`: 用于结果复现，取值范围 `[0, 2^31-1]`，默认 `-1`。 ## 使用方式 -API 交互基于 WebSocket，核心流程分两种模式： - -### 1. VAD 模式（推荐，默认） -服务端自动检测语音起止并触发响应，客户端只需持续 `append_audio`（及可选 `append_video`）： -```python -conv = OmniRealtimeConversation(model="qwen3.5-omni-realtime", callback=cb, url=url) -conv.connect() -conv.update_session(enable_turn_detection=True) # 启用 server_vad -# 循环：mic.read() → conv.append_audio(base64) -# 无需手动 commit 或 create_response -``` -- 事件流：`input_audio_buffer.speech_started` → `input_audio_buffer.speech_stopped` → `input_audio_buffer.committed` → `response.*` -- 工具调用时，服务端发送 `response.function_call_arguments.done` 后，客户端执行工具并调用 `conversation.item.create`，服务端**自动**生成最终响应（见 [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)）。 - -### 2. Manual 模式 -客户端完全控制节奏，需显式提交与触发： -```java -conversation.updateSession(OmniRealtimeConfig.builder() - .enableTurnDetection(false) // 关闭 VAD - .build()); -// ... appendAudio ... -conversation.commit(); // 提交音频缓冲区 -conversation.createResponse(null, Arrays.asList(AUDIO, TEXT)); // 触发响应 -``` -- 适用场景：聊天软件“按住说话”、离线音频文件处理。 -- 工具调用时，客户端在收到 `response.function_call_arguments.done` 后，需**手动再次调用 `createResponse`** 触发最终响应（见 [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md)）。 +1. **建立连接**：使用 WebSocket URL（推荐业务空间专属域名，如 `wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime`）连接，首条服务端事件为 `session.created`。 + +2. **配置会话**：连接后立即发送 `session.update` 事件（或调用 SDK `update_session`），设置 `modalities`、`voice`、`turn_detection` 等。服务端校验后返回 `session.updated`。 + +3. **输入数据**： + - **VAD 模式**（`turn_detection.type` 非 null）：持续发送 `input_audio_buffer.append`，服务端自动检测起止并提交，客户端无需调用 `input_audio_buffer.commit`。 + - **Manual 模式**（`turn_detection` 设为 `null`）：发送 `input_audio_buffer.append` 后，必须显式发送 `input_audio_buffer.commit` 创建用户消息项。 + +4. **触发响应**： + - VAD 模式：服务端检测到语音结束自动触发 `response.create`。 + - Manual 模式：客户端在提交音频后，需主动发送 `response.create` 事件。 + +5. **处理工具调用**：当服务端返回 `conversation.item.created` 类型为 `function_call` 时，客户端执行工具，再通过 `conversation.item.create` 回传结果，并发送 `response.create` 触发最终响应。 ## 限制和注意事项 -- **域名迁移强制要求**：华北2（北京）与新加坡地域必须使用 `wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 或 `wss://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`，旧域名 `dashscope.aliyuncs.com` 将逐步下线（[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 与 [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) 均明确提示）。 -- **音视频格式约束**： - - 输入音频：`PCM_16000HZ_MONO_16BIT`（Python）或 `PCM_16000HZ_MONO_16BIT`（Java），Base64 编码。 - - 输入视频：JPG/JPEG 格式，分辨率建议 480P–720P（≤1080P），单图 Base64 后 ≤256KB。 - - 输出音频：固定 `PCM_24000HZ_MONO_16BIT`，不可自定义。 -- **并发与资源**：单个 WebSocket 连接对应一个会话；`append_audio` 单次数据块无明确上限，但 `commit` 前总缓冲区建议 ≤15 MiB（[Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) 提示）。 -- **互斥配置**：`enable_search` 与 `tools` 不可同时启用，否则返回 `invalid_request_error`（[客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) 明确说明）。 -- **错误处理**：所有服务端错误均以 `error` 事件返回，含 `type`、`code`、`message` 和 `param`（见 [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md)）。 +- **音频/图像限制**：输入音频必须为 16 kHz PCM；图像仅支持 JPG/JPEG，Base64 编码后 ≤256 KB，建议分辨率 480p–720p。 +- **并发与配额**：单个 WebSocket 连接仅支持一个会话；具体 QPS 和 Token 配额需参考百炼控制台配额管理。 +- **SDK 版本要求**：Python SDK ≥1.25.17，Java SDK ≥2.22.15，旧版本可能缺失 `idle_timeout_ms` 等新参数支持。 +- **地域与域名**：北京/新加坡地域**必须使用业务空间专属域名**（`{WorkspaceId}.cn-beijing.maas.aliyuncs.com` 等），旧域名（`dashscope.aliyuncs.com`）已逐步弃用，性能与稳定性较差。 +- **错误处理**：服务端返回 `error` 事件时，`error.param` 字段明确指示出错参数（如 `"session.modalities"`），应据此修正请求。 + +> **注意**：文档 2 和文档 4 中 `enable_turn_detection` 参数在 Python/Java SDK 中是布尔开关，其底层映射到 `session.turn_detection.type`；若设为 `False`，实际等效于 `turn_detection: null`（即 Manual 模式），而非 `type: "server_vad"` 且 `threshold: 0`。此逻辑一致性需开发者注意。 ## 来源文档 -- [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - [客户端事件](../../raw/model-api-reference/omni-realtime-api/client-events.md) -- [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) +- [Python SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-python-sdk.md) - [服务端事件](../../raw/model-api-reference/omni-realtime-api/server-events.md) -- [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) - [Java SDK](../../raw/model-api-reference/omni-realtime-api/omni-realtime-java-sdk.md) +- [实时多模态交互流程](../../raw/model-api-reference/omni-realtime-api/omni-realtime-interaction-process.md) +- [声音复刻API参考](../../raw/model-api-reference/omni-realtime-api/qwen-omni-voice-cloning.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md index daa3fa5d..73bea14f 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/preparations.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/preparations.md @@ -1,60 +1,48 @@ # preparations -在调用阿里云百炼平台的模型或应用前，开发者需完成基础环境准备，包括获取并安全配置 API Key、安装合适的 SDK 或 CLI 工具、理解关键参数约束及常见限制。这些步骤是所有 API 调用和本地开发的前提，直接影响服务可用性、安全性与调试效率。 +在调用阿里云百炼平台的模型或应用前，开发者需完成基础环境准备：获取并安全配置 API Key、选择合适的 SDK 或 CLI 工具、理解关键参数约束及常见错误边界。这些步骤直接影响服务可用性、安全性与调试效率，是所有集成工作的前提。 ## 支持的模型/功能 -百炼平台支持多类模型与能力，涵盖文本生成（如 `qwen3-max`、`qwen3-235b-a22b-instruct-2507`）、图像生成（`qwen-image-2.0`）、视频生成（`happyhorse-1.0-t2v`）、语音合成（`cosyvoice-v3-flash`）、语音识别（`paraformer-real-time`）、向量嵌入（`text-embedding-v3`）、排序（`text-rerank-v3`）及全模态理解（`qwen3.5-omni-plus`）。部分模型具备特定能力约束，例如： -- 思考模式（`enable_thinking=true`）仅适用于指定模型（如 `qwen3-235b-a22b-thinking-2507`），且强制要求 `stream=true` 与 `incremental_output=true`； -- 结构化输出（`response_format={"type": "json_object"}`）不支持与思考模式共用； -- 联网搜索（`enable_search=true`）仅限明确标注支持该能力的模型； -- 工具调用（`tools` 参数）仅被 Qwen 和 DeepSeek 系列模型支持，纯文本模型（如 `qwen3-max`）若传入含 `image_url` 的 `messages` 将报错 [原文标题](../../raw/model-api-reference/preparations/error-code.md)。 +百炼平台提供全模态模型能力，包括文本生成（如 `qwen3.7-max`）、图像生成（`qwen-image-2.0`）、视频生成（`happyhorse-1.0-t2v`）、语音合成（`cosyvoice-v3-flash`）、视觉理解（`qwen3-vl-plus`）及向量/排序等专用模型。不同模型对输入格式、协议兼容性（OpenAI 或 Anthropic）和调用方式有明确要求。例如，Qwen-Omni 模型仅支持[流式输出](../concepts/streaming-output.md)，而 Qwen-Long 仅接受纯文本类文件（TXT/DOCX/PDF 等），不支持图片或扫描件 [错误码](../../raw/model-api-reference/preparations/error-code.md)。多模态模型（如 `qwen3.5-omni-plus`）支持 `image`、`audio`、`video` 参数混合输入，纯文本模型则严格拒绝非字符串 `content` [错误码](../../raw/model-api-reference/preparations/error-code.md)。 ## 关键参数 -调用时需注意以下核心参数的合法范围与互斥关系（详见 [原文标题](../../raw/model-api-reference/preparations/error-code.md)）： -- `temperature`：必须在 `[0.0, 2.0)` 区间； -- `top_p`：必须在 `(0.0, 1.0]` 区间； -- `max_tokens`：不得超过模型文档中声明的最大输出 [Token](../concepts/token.md) 数； -- `n`：取值范围为 `[1, 4]`； -- `seed`：DashScope 协议下需为 `[0, 9223372036854775807]` 内整数； -- `thinking_budget`：须为正整数且不超过模型最大思维链长度； -- `stop`：仅接受 `str`、`list[str]`、`list[int]` 或 `list[list[int]]` 类型，且列表内元素类型必须一致； -- `messages`：纯文本模型要求 `content` 为字符串；[多模态](../concepts/multi-modal.md)模型要求 `content` 数组中每个元素为合法对象（`type` 仅限 `text`/`image_url`/`video_url` 等）； -- `response_format`：结构化输出必须设为 `{"type": "json_object"}`，且提示词中需包含 `json` 关键词。 +核心参数需严格遵循取值范围与类型约束： +- `temperature`: 必须在 `[0.0, 2.0)` 区间； +- `top_p`: 必须在 `(0.0, 1.0]` 区间； +- `max_tokens`: 上限由具体模型文档定义，不可超过其最大输出 Token 数； +- `n`: 图像/文本批量生成数，范围为 `[1, 4]`（部分 CLI 命令如 `bl image generate` 支持最多 6 张，属工具层扩展，非 API 层通用限制）； +- `seed`: DashScope 协议下必须为 `[0, 9223372036854775807]` 内整数； +- `enable_thinking`: 仅特定模型（如 `qwen3-235b-a22b-thinking-2507`）强制设为 `true`，且开启时必须同时设置 `stream=true` 和 `incremental_output=true`，禁用结构化输出（`response_format="json_object"`）[错误码](../../raw/model-api-reference/preparations/error-code.md)。 -> **注意**：文档 3 中“`The value of the enable_thinking parameter is restricted to True`”与文档 1 中“API Key 权限说明”存在隐含矛盾——前者指出部分模型强制开启思考模式，后者未提及该限制对权限配置的影响。实际调用时应以模型文档为准，而非仅依赖 API Key 权限设置。 +> **注意**：文档 2 中 `bl image generate --n` 支持 `6`，但文档 4 明确 `Range of n should be [1, 4]`。该差异源于 CLI 工具对批量请求的封装逻辑（内部拆分为多次 API 调用），而非 API 协议本身允许单次请求 `n=6`。实际 HTTP 调用仍需遵守 `n ≤ 4` 的服务端限制。 ## 使用方式 ### API Key 获取与配置 -需使用主账号或具备 `管理员`/`API-Key` 页面权限的子账号，在对应地域（如华北2、新加坡、美国弗吉尼亚）的 [API Key 管理页面](https://bailian.console.aliyun.com/?tab=model#/api-key) 创建密钥。新创建的密钥以 `sk-ws` 开头，明文仅显示一次，务必立即保存 [原文标题](../../raw/model-api-reference/preparations/get-api-key.md)。推荐将 `DASHSCOPE_API_KEY` 配置为环境变量（Linux/macOS/Windows 均有详细步骤），避免硬编码。 +必须通过[阿里云百炼控制台](https://bailian.console.aliyun.com/)创建 API Key，并按地域（华北2、新加坡、美国弗吉尼亚等）进入对应 `API Key` 页面操作 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。强烈建议将 Key 配置为环境变量 `DASHSCOPE_API_KEY`，避免硬编码；Linux/macOS/Windows 各系统配置方法详见原文档。 -### SDK 安装 -- **Python**：可选 `openai`（OpenAI 兼容协议）或 `dashscope`（原生协议）SDK，均需 `pip install -U `； -- **Java/Node.js/Go**：DashScope 提供官方 Java SDK；OpenAI SDK 支持多语言（Java/Node.js/Go），其中 Go 需 `Go 1.22+` 并建议配置阿里云镜像代理； -- **CLI 工具**：通过 `npm install -g bailian-cli` 安装百炼 CLI（要求 Node.js ≥ 22.12.0），支持 `bl text chat`、`bl image generate` 等命令行调用 [原文标题](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。 +### SDK 与 CLI 集成 +- **SDK**：推荐使用官方 DashScope SDK（Python/Java）或 OpenAI 兼容 SDK（Python/Node.js/Java/Go）。安装命令统一为 `pip install -U dashscope` 或 `npm install openai` [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md)。 +- **CLI**：`bailian-cli`（命令 `bl`）需 Node.js ≥ 22.12.0，通过 `npm install -g bailian-cli` 安装，并支持 `bl auth login --api-key` 或浏览器 OAuth 登录 [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md)。CLI 提供 `bl text chat`、`bl image generate` 等高阶命令，自动处理模型路由、异步轮询与文件下载。 ### 协议与端点 -调用时除 API Key 外，**必须指定服务端点（API Host）**，其值取决于所选协议与地域： -- OpenAI 兼容协议：`base_url` 为 `https://dashscope.aliyuncs.com/v1`（中国站）或对应国际站地址； -- Anthropic 兼容协议：`base_url` 为 `https://dashscope.aliyuncs.com/anthropic/v1`； -- 不同地域的端点不同，务必以控制台创建 API Key 时弹窗显示的 `API Host` 为准。 +调用时必须指定 `base_url`（即创建 API Key 时弹窗显示的 **API Host**），其值因地域和协议（OpenAI 兼容 vs Anthropic 兼容）而异，不可复用 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。 ## 限制和注意事项 -- **API Key 安全**：`sk-` 开头旧密钥仍可用，但新密钥统一为 `sk-ws` 格式，且不可再次查看明文。美国（弗吉尼亚）地域不支持禁用/重置操作。 -- **地域隔离**：API Key 与模型服务绑定地域，跨地域调用需对应地域的 API Key 和端点。 -- **IP 白名单**：仅北京、新加坡等部分地域支持自定义 IP 白名单（最多 20 个 IPv4/IPv6 地址或网段），美国（弗吉尼亚）地域不支持。 -- **文件限制**：Qwen-Long 模型仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 纯文本文件，单文件大小 ≤ 150 MB、页数 ≤ 15000、内容非空；图片/扫描件需先用 Qwen-VL 提取文本。 -- **[Token](../concepts/token.md) 限制**：输入总长度（含 messages、[prompt](../guides/prompt.md)、file content）不得超过模型最大上下文窗口；纯文本模型不支持[多模态](../concepts/multi-modal.md) `content`，否则触发 `Unexpected item type in content` 错误。 -- **CLI 环境约束**：百炼 CLI 严格依赖 npm（非 pnpm/yarn）且要求 Node.js ≥ 22.12.0；认证方式中，`bl auth login --console` 推荐用于交互式场景，`--api-key` 适用于 CI/CD 或无浏览器环境。 +- **API Key 安全**：新创建 Key 以 `sk-ws` 开头，明文仅创建时可见一次，丢失后需重置；旧 `sk-` Key 可继续使用，但建议迁移 [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md)。 +- **地域隔离**：华北2（北京）、新加坡等地域支持 IP 白名单与模型范围自定义权限；美国（弗吉尼亚）地域不支持禁用/重置操作及权限精细化配置。 +- **文件限制**：Qwen-Long 模型处理文件大小 ≤ 150 MB、页数 ≤ 15000、内容非空，且仅支持 TXT/DOCX/PDF/EPUB/MOBI/MD 格式 [错误码](../../raw/model-api-reference/preparations/error-code.md)。 +- **错误处理**：`Model not exist` 错误常因模型 ID 大小写错误或混用开源名称（如 `Qwen/Qwen3-235B...`）导致，务必使用控制台模型列表中的标准 ID（如 `qwen3-235b-a22b-instruct-2507`）。 +- **调试建议**：遇到参数错误（如 `400-InvalidParameter`），优先使用阿里云 AI 助理输入报错信息获取精准方案，而非手动排查 [错误码](../../raw/model-api-reference/preparations/error-code.md)。 ## 来源文档 - [获取API Key](../../raw/model-api-reference/preparations/get-api-key.md) +- [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) - [安装SDK](../../raw/model-api-reference/preparations/install-sdk.md) - [错误码](../../raw/model-api-reference/preparations/error-code.md) -- [使用百炼 CLI](../../raw/model-api-reference/preparations/use-model-studio-cli.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md index 69ec2ecd..5a81f165 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/qwen-api-reference.md @@ -1,41 +1,44 @@ # qwen api reference -Qwen API 提供多种调用方式，支持文本生成、工具调用、联网搜索等能力，开发者可根据技术栈兼容性与功能需求选择合适接口。所有接口均基于 Qwen 系列大模型（如 Qwen2、Qwen2.5、Qwen3）提供服务，需通过百炼平台鉴权访问。详细参数说明与行为差异请参考 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md)。 +Qwen 系列大语言模型通过百炼平台提供多种 API 接口，支持文本生成、多轮对话、工具调用等核心能力。开发者可根据技术栈兼容性、功能需求和部署场景选择合适的接入方式。所有接口均需通过百炼平台鉴权，并遵循统一的计费与配额规则。 ## 支持的模型与功能 -- **基础文本生成**：支持 `qwen-max`、`qwen-plus`、`qwen-turbo` 等多档位模型，适用于通用对话、摘要、创作等场景。 -- **增强能力接口**： - - OpenAI 兼容 Chat Completions：适合已有 OpenAI 生态集成的应用快速迁移； - - OpenAI 兼容-Responses：自动启用联网搜索、代码解释器、网页提取等工具链，无需手动管理工具调用流程； - - Anthropic 兼容 Messages：支持 `tool_use`、`thinking` 等结构化输出，适配 Anthropic 工作流； - - DashScope 原生接口：提供最全参数控制（如 `top_k`、`repetition_penalty`、`enable_search`），是调试与高阶定制的首选。 +当前 Qwen 系列支持以下主流调用方式： -> **注意**：`qwen-max` 在 DashScope 接口中默认启用思考模式（`enable_thinking=true`），但在 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)中该参数不可设；此行为差异已在 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 中明确标注，使用时需注意一致性。 +- **OpenAI 兼容 Chat Completions**：适用于已使用 OpenAI SDK 的项目，可零代码修改迁移；支持 `qwen-max`、`qwen-plus`、`qwen-turbo` 等全部公开模型，但部分高级参数（如 `tool_choice` 的细粒度控制）需参考 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) 中的兼容性说明。 +- **OpenAI 兼容-Responses**：内置联网搜索、代码解释器与网页内容提取能力，自动维护对话历史；该模式不支持自定义 system [prompt](../guides/prompt.md) 的逐轮覆盖，详见 [原文标题](../../raw/model-api-reference/qwen-api-reference.md)。 +- **Anthropic 兼容-Messages**：支持 `max_tokens`、`temperature`、`tool_use` 等 Anthropic 标准参数，但 Qwen 模型对 `stop_sequences` 的处理逻辑与 Anthropic 原生实现存在差异，> **注意**：实际行为以 [原文标题](../../raw/model-api-reference/qwen-api-reference.md) 中“Anthropic兼容-Messages”章节为准，而非 Anthropic 官方文档。 +- **DashScope 原生接口**：功能最完整，支持流式响应、长上下文分块、自定义 stop words、logprobs 输出及模型专属参数（如 `enable_search`），推荐新项目优先采用。 ## 关键参数 -| 参数名 | 类型 | 说明 | 是否必需 | 备注 | -|--------|------|------|----------|------| -| `model` | string | 模型标识符，如 `qwen-max`、`qwen-plus` | 是 | 不同接口对模型命名格式要求一致，详见 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) | -| `messages` | array | 对话历史，格式为 `[{ "role": "...", "content": "..." }]` | 是（Chat Completions / Messages） | DashScope 接口额外支持 `system` 角色和 `tools` 字段 | -| `tools` | array | 工具定义列表（JSON Schema 格式） | 否 | 仅 DashScope 和 Anthropic Messages 接口原生支持；OpenAI 兼容-Responses 的工具由服务端自动注入，不开放显式传参 | -| `enable_search` | boolean | 是否启用联网搜索（DashScope 专属） | 否 | 默认 `false`；启用后将自动触发搜索并融合结果 | +| 参数名 | 类型 | 说明 | 是否必需 | +|--------|------|------|----------| +| `model` | string | 模型标识符，如 `qwen-max`、`qwen-plus`；必须与所选接口类型匹配 | 是 | +| `messages` | array | 对话消息列表，格式为 `[{ "role": "user", "content": "..." }]`；`system` 角色仅 DashScope 和 OpenAI Chat Completions 支持 | 是（除 Anthropic Messages 使用 `system` 字段） | +| `temperature` | number | 控制输出随机性，范围 `[0.0, 2.0]`，默认 `1.0` | 否 | +| `top_p` | number | 核采样阈值，范围 `[0.0, 1.0]`，默认 `1.0` | 否 | +| `max_tokens` | integer | 最大生成 token 数，不同模型上限不同（如 `qwen-turbo` 默认 8192） | 否 | +| `stream` | boolean | 是否启用流式响应；仅 DashScope 和 OpenAI Chat Completions 支持 | 否 | + +> **注意**：`tools` 和 `tool_choice` 参数在 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)中仅部分生效，完整工具调用能力需通过 DashScope 或 Anthropic Messages 接口实现，具体约束见 [原文标题](../../raw/model-api-reference/qwen-api-reference.md)。 ## 使用方式 -1. **认证**：使用百炼平台颁发的 `API Key`，通过 `Authorization: Bearer ` 请求头传递； +1. **认证**：所有请求需携带 `Authorization: Bearer `，API Key 从百炼控制台「API 密钥管理」获取； 2. **Endpoint 示例**： - DashScope：`POST https://dashscope.aliyuncs.com/api/v1/services/aigc/text-generation/generation` - - OpenAI 兼容：`POST https://dashscope.aliyuncs.com/v1/chat/completions` -3. **SDK 调用**：推荐使用官方 `dashscope` Python SDK（v1.20.0+）或 `openai` 客户端（v1.0+），配置 `base_url` 指向百炼 OpenAI 兼容地址即可复用现有逻辑。 + - OpenAI Chat Completions：`POST https://dashscope.aliyuncs.com/v1/chat/completions` +3. **SDK 调用**：推荐使用官方 `dashscope` Python SDK（v1.20.0+）或 `openai` SDK（v1.0+），初始化时指定 `base_url` 和 `api_key` 即可自动路由。 ## 限制和注意事项 -- 单次请求 `messages` 总长度（token 数）上限为 32768（Qwen3 模型）或 8192（旧版模型），具体以实际模型文档为准； -- 工具调用（如 `code_interpreter`）在 OpenAI 兼容-Responses 接口中为全自动模式，**不支持用户自定义工具函数**，与 DashScope 接口的可控性存在本质差异； -- 流式响应（`stream=true`）在所有接口中均支持，但字段结构不同：DashScope 返回 `output.text`，OpenAI 兼容返回 `choices[0].delta.content`； -- > **注意**：原始文档中 OpenAI 兼容-Responses 的“自动管理对话历史”描述与实际行为存在偏差——当启用 `enable_search` 时，历史会被截断以预留上下文空间，该限制未在 [文本生成模型API参考](../../raw/model-api-reference/qwen-api-reference.md) 中明示，建议在长对话场景下主动控制 `max_tokens` 与历史长度。 +- 单次请求 `messages` 总长度（含 [prompt](../guides/prompt.md) + history）不得超过模型 context length（如 `qwen-max` 为 32768 tokens），超长输入将被截断且无警告； +- [OpenAI 兼容接口](../concepts/openai-compatible-api.md)不支持 `response_format`（如 JSON mode）和 `parallel_tool_calls`，此类功能仅 DashScope 原生接口支持； +- 所有接口均按实际输入 + 输出 token 计费，空格、换行符、标点均计入 token； +- 流式响应中 `delta.content` 可能为空字符串（表示中间 token 分片），客户端需忽略空 content 并持续拼接； +- 模型版本升级可能影响输出稳定性，生产环境建议锁定 `model` 版本号（如 `qwen-max-20240701`），而非使用别名 `qwen-max`。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md index 81c8f2b8..66f261a9 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/realtime-api-user-guide.md @@ -1,74 +1,109 @@ # realtime api user guide -Realtime API 是一套面向低延迟、[多模态](../concepts/multi-modal.md) AI 交互场景的实时通信协议栈，支持 WebSocket、WebRTC 和 AOQ（AI over QUIC）三种传输协议，分别适配服务端集成、浏览器端互动和移动端原生应用等不同技术栈与网络环境。开发者需根据目标平台、延迟要求、弱网适应性及数据类型选择合适协议，并配合对应 SDK 或标准 Web API 实现接入。 +Realtime API 是一套面向低延迟、多模态、弱网对抗的实时 AI 交互协议栈，提供 WebSocket、WebRTC 和 AOQ（AI over QUIC）三种传输方案，支持语音识别、语音合成、实时对话、多模态理解等场景。开发者可根据终端类型、网络环境、功能需求和集成复杂度选择最适配的接入方式。 ## 支持的模型/功能 -Realtime API 当前支持以下核心模型与应用类型，但协议支持存在明确差异： +Realtime API 当前支持以下核心模型与应用类型，不同协议的支持能力存在差异： -- **实时全模态模型**（如 `qwen3.5-omni-plus-realtime`、`qwen3.5-omni-flash-realtime`、`qwen3.5-livetranslate-flash-realtime`）：三协议均支持，是唯一在 WebSocket、WebRTC 和 AOQ 上完全可用的模型类别。 -- **[多模态](../concepts/multi-modal.md)开发套件**（`multimodal-dialog`）：仅支持 WebRTC 和 WebSocket，[不支持 AOQ](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md)。 -- **实时语音识别**（Fun-ASR 系列）、**实时语音合成**（CosyVoice 系列）、**实时语音对话**（`qwen-audio-3.0-realtime-plus` 等）：**仅支持 WebSocket 协议**，[WebRTC 和 AOQ 均不支持](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md)。 +| 模型/应用类型 | AOQ | WebRTC | WebSocket | +|---------------|-----|--------|-----------| +| 实时全模态对话（`qwen3.5-omni-plus-realtime`, `qwen3.5-omni-flash-realtime`） | ✅ | ✅ | ✅ | +| 实时语音翻译（`qwen3.5-livetranslate-flash-realtime`） | ✅ | ✅ | ✅ | +| 多模态交互套件（`multimodal-dialog`） | ❌ | ✅ | ✅ | +| 实时语音识别（Fun-ASR 系列） | ❌ | ❌ | ✅ | +| 实时语音合成（CosyVoice 系列） | ❌ | ❌ | ✅ | +| 实时语音对话（`qwen-audio-3.0-realtime-plus` 等） | ❌ | ❌ | ✅ | -> **注意**：文档 4 与文档 5 均以 WebRTC 接入 `multimodal-dialog` 和 `qwen3.5-omni-plus-realtime` 为示例，但文档 1 明确指出 `multimodal-dialog` 不支持 AOQ；而文档 7 的 AOQ 示例仅覆盖 `qwen3.5-omni-plus-realtime`，未提及 `multimodal-dialog`。因此，`multimodal-dialog` 的 AOQ 支持状态以文档 1 的表格为准，属明确不支持项，非过时信息。 +> **注意**：文档 1 中表格明确标注 `multimodal-dialog` 在 AOQ 协议下“不支持”，但文档 6 的标题与正文均以“通过 WebRTC 使用多模态交互套件”为前提展开，未提及 AOQ 支持。因此该模型在 AOQ 下确实不可用，开发者应避免尝试 [Realtime API简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md)。 + +AOQ 与 WebRTC 均内置回声消除（AEC）和降噪能力，而 WebSocket 方案需客户端自行处理；AOQ 和 WebRTC 支持音视频+数据混合传输，WebSocket 仅支持文本/音频/图像分通道传输。 ## 关键参数 ### 鉴权参数 -- `Authorization: Bearer `：所有协议建连阶段必需的 HTTP Header。AOQ 协议中该 Key 仅用于服务端向百炼网关发起 `allocate` 请求，客户端使用返回的 `aoqTokenForClient` 连接，避免密钥暴露 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)。 +所有协议均使用 `Authorization: Bearer ` 进行建连鉴权： +- WebSocket/WebRTC：客户端或服务端直接携带 API Key 发起连接； +- AOQ：**必须**由业务 AppServer 代为请求百炼网关获取临时 `aoqTokenForClient`，客户端仅使用该 Token 连接，避免 API Key 泄露 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)。 ### 协议特有参数 -- **AOQ**：`x-dashscope-rtc-transport: moq`（必须）、`clientIp`（选填，用于 Relay 节点优化）。 -- **WebRTC**：SDP 交换请求中 `Content-Type: application/sdp`，且 `model` 参数需显式指定（如 `?model=qwen3.5-omni-plus-realtime`）。 -- **WebSocket**：无特殊 Header，依赖标准 WebSocket 握手，模型通过 URL query 参数或初始消息体指定。 - -### 会话配置参数（通过 `session.update` 事件发送） -- `modalities`: 指定输出模态，如 `["text", "audio"]`。 -- `voice`: 输出音色 ID（如 `"Ethan"`）。 -- `input_audio_format` / `output_audio_format`: 当前仅支持 `"pcm"`。 -- `turn_detection`: VAD 配置对象，`type` 可选 `"server_vad"` 或 `"semantic_vad"`（推荐后者），含 `threshold` 和 `silence_duration_ms`。 +- **AOQ**：建连请求头需包含 `x-dashscope-rtc-transport: moq`；响应中关键字段包括 `sid`（会话 ID）、`aoqTokenForClient`（客户端令牌）、`clientRelayEndpoints`（中继地址）和 `clientRelayCertFingerprint`（证书指纹）。 +- **WebRTC**：SDP 交换时需确保 Offer 中包含 `m=audio`（服务端强制要求），并创建名为 `oai-events` 的 DataChannel 用于事件通信；多模态套件接入需使用 `workspace_id.region.maas.aliyuncs.com` 格式 endpoint [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md)。 +- **WebSocket**：无特殊协议头，但需严格遵循 JSON-RPC 2.0 格式发送 `input_audio_buffer.append` 等事件。 + +### 会话配置参数（`session.update`） +通用配置项（以 AOQ 示例为准）： +```json +{ + "type": "session.update", + "session": { + "modalities": ["text", "audio"], + "voice": "Ethan", + "input_audio_format": "pcm", + "output_audio_format": "pcm", + "instructions": "你是某五星级酒店的AI客服专员...", + "turn_detection": { + "type": "semantic_vad", + "threshold": 0.5, + "silence_duration_ms": 800 + } + } +} +``` +其中 `turn_detection.type` 在 `qwen3.5-omni-realtime` 模型下推荐设为 `semantic_vad`；`input_audio_format` 固定为 `pcm`（16 kHz 采样率），`output_audio_format` 固定为 `pcm`（24 kHz 采样率）。 ## 使用方式 -### 协议选择与接入路径 -- **WebSocket**：适用于服务端或快速原型验证，使用 DashScope SDK（参见[安装SDK](https://help.aliyun.com/zh/model-studio/install-sdk)），接入成本最低，但弱网对抗能力差。 -- **WebRTC**：适用于浏览器端，需自行管理 `RTCPeerConnection`、媒体流与 DataChannel，内置回声消除与降噪，[通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) 提供完整 JS 示例。 -- **AOQ**：适用于 Android/iOS/HarmonyOS 原生应用，需集成 [AOQ SDK](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md)，具备极致弱网对抗与混合数据传输能力，[通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) 包含各平台集成指南。 - -### 核心流程共性 -1. **获取凭证**：WebSocket 直接使用 API Key；WebRTC 通过 SDP 交换携带 Key；AOQ 由 AppServer 调用百炼 `allocate` 接口获取 `sid` 与 `aoqTokenForClient`。 -2. **建立连接**：WebSocket 直连；WebRTC 完成 Offer/Answer 协商；AOQ 调用 `engine.connect(config)`。 -3. **会话初始化**：连接成功后，发送 `session.update` 事件配置模态、音色、VAD 等。 -4. **媒体流控制**：AOQ 必须在收到 `session.updated` 后调用 `enableSendMediaStream(.audio, true)` 开启发送；WebRTC 需在收到 `session.created` 后解除媒体门控；WebSocket 通常由 SDK 自动处理。 - -### 媒体流高级控制（AOQ 专属） -- **自定义音频采集/播放**：通过 `isExternal=true` 关闭内部设备，使用 `addAudioExternalStream` + `pushAudioExternalStreamData` 或 `setAudioFrameObserver` 实现 TTS 注入或 ASR 处理 [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md)。 -- **自定义视频输入**：支持原始帧（I420/NV12/BGRA）或编码帧（JPEG）推送，需先 `startVideoCapture(isExternal=true)` [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md)。 +### 协议选型建议 +- **WebSocket**：适合服务端集成、快速原型验证、对浏览器兼容性无要求的场景；接入成本最低，但弱网对抗能力差。 +- **WebRTC**：适合浏览器端互动、已有 WebRTC 基础设施的项目；原生支持音视频，无需额外 SDK，但需自行处理 SDP 交换与 ICE 协商。 +- **AOQ**：适合移动端原生应用（Android/iOS/HarmonyOS），对延迟、弱网、多模态混合传输有极致要求；需集成 [SDK下载](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) 提供的 AOQ Client SDK 及 Opus 插件。 + +### 典型接入流程（AOQ） +1. **AppServer 获取凭证**：调用百炼 `/api/v1/webrtc/realtime?model=...` 接口（带 `x-dashscope-rtc-transport: moq`），传入 `clientIp`，获取 `sid` 和 `aoqTokenForClient`； +2. **客户端初始化引擎**：调用 `createEngine`，设置 `AoqEngineDelegate` 回调； +3. **启动媒体采集**：`startAudioCapture()` / `startVideoCapture()`； +4. **禁用媒体发送**：`enableSendMediaStream(.audio, false)`，防止模型未就绪即推送数据； +5. **建立连接**：构造 `AoqConnectConfig`（填入 `token`/`sid`/`certFingerprint`/`relayEndpoints`），调用 `connect()`； +6. **等待会话就绪**：监听 `onDataMsg`，收到 `session.updated` 后调用 `enableSendMediaStream(.audio, true)` 开启发送； +7. **断开连接**：调用 `disconnect()`，非必须调用 `destroy()`（引擎可复用）。 + +### 典型接入流程（WebRTC） +1. 创建 `RTCPeerConnection({ iceServers: [] })`； +2. 添加音频轨道（必需）和视频轨道（可选），**立即禁用发送**（`track.enabled = false` + `sender.replaceTrack(null)`）； +3. 创建 `oai-events` DataChannel； +4. 调用 `createOffer()` → `setLocalDescription()`，等待 `iceGatheringState === "complete"` 获取完整 Offer SDP； +5. 由 AppServer 代理 POST Offer SDP 至百炼 WebRTC endpoint（如 `https://.cn-beijing.maas.aliyuncs.com/api/v1/webrtc/inference?model=multimodal-dialog`）； +6. 将返回的 Answer SDP 规范化（确保 `\r\n` 行尾）后调用 `setRemoteDescription()`； +7. 收到 `pc.ontrack` 后播放远端音频流，收到 `session.created` 后恢复媒体发送。 ## 限制和注意事项 -- **浏览器兼容性**：WebRTC 原生支持所有现代浏览器；AOQ 不支持浏览器，仅限原生平台；WebSocket 兼容性最广。 -- **建连与媒体发送时机**：AOQ 和 WebRTC 均要求严格遵循“先建连 → 收到服务端确认（`session.updated` 或 `session.created`）→ 再开启媒体发送”流程，否则模型可能无法接收数据。此逻辑在 [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) 中有明确强调。 -- **CORS 限制**：WebRTC 的 SDP 交换在浏览器端直连百炼服务受 CORS 限制，[文档 4 和 5 均明确指出 Demo 需通过 curl 或业务后端代理完成](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md)，生产环境必须由 AppServer 代理。 -- **Opus 编解码**：AOQ SDK 使用插件化 Opus，下载 SDK 时必须同步获取并集成 `libPluginOpus`（Android/iOS/HarmonyOS 各平台均有对应包）。 -- **连接状态管理**：AOQ SDK 提供明确的状态机（Connecting → Connected → Failed → Disconnected），业务需监听 `onConnectionStatusChange` 回调处理状态迁移，`Failed` 为瞬态，SDK 会自动进入 `Disconnected`，无需手动 `disconnect` [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md)。 +- **API Key 安全**：严禁将 API Key 硬编码至客户端代码或提交至代码仓库；AOQ 必须通过服务端代理鉴权，WebSocket/WebRTC 的生产环境也应由 AppServer 代理 SDP 交换或连接建立 [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md)。 +- **媒体流控制**：AOQ 和 WebRTC 均要求在收到 `session.updated`（或 `session.created`）后再开启媒体发送，否则服务端可能拒绝接收数据；`enableSendMediaStream` 是 AOQ 的关键控制接口，其默认行为是连接成功后立即发送，务必显式禁用再启用 [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md)。 +- **平台与格式约束**： + - AOQ 不支持浏览器环境（仅 Android/iOS/HarmonyOS）； + - WebRTC 的 `multimodal-dialog` 模型需使用工作空间专属 endpoint（`{workspace_id}.{region}.maas.aliyuncs.com`），而非通用 endpoint； + - AOQ 视频编码默认 `isExternal=false`，若需自定义视频输入，必须先调用 `startVideoCapture(config)` 并设置 `config.isExternal = true`，否则 `pushExternalVideoCapturedFrame` 不生效 [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md)； + - 所有协议的音频输入格式均为 `pcm`，但采样率要求不同：WebSocket 输入为 16 kHz，AOQ/WebRTC 输入为 16 kHz（文档 4 明确），输出统一为 24 kHz PCM。 +- **异常处理**：AOQ SDK 对物理限制（网络中断、设备故障）和外部因素（token 过期）有分级处理机制，瞬态错误（如 `Failed` 状态）会自动迁移至 `Disconnected`，业务层无需重复调用 `disconnect` [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md)。 ## 来源文档 - [Realtime API简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-overview.md) - [SDK下载](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-sdk-download.md) - [Token鉴权](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-token-authentication.md) -- [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) -- [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) - [实现接通模型/应用](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-quick-start-guide/realtime-connect-model.md) +- [通过WebRTC使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-omni-realtime.md) +- [通过WebRTC使用多模态交互套件实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-webrtc-multimodal-dialog.md) - [通过AOQ使用qwen3.5-omni-plus-realtime实现实时通话](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-best-practices/best-practice-aoq-omni-realtime.md) - [AOQ SDK简介](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-desc.md) - [连接状态管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-connection-management.md) - [音频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-audio-features.md) -- [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) - [自定义音频播放](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-playback.md) +- [媒体流发送管理](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-media-stream-control.md) - [自定义音频采集](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-audio-capture.md) -- [视频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) - [自定义视频输入](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-custom-video-input.md) +- [视频常用功能介绍](../../raw/model-api-reference/realtime-api-user-guide/realtime-api-aoq-api/realtime-api-aoq-sdk-function/aoq-video-features.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md index 83599b34..6248bd5d 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/toolkits-and-frameworks.md @@ -1,78 +1,118 @@ # toolkits and [frameworks](frameworks.md) -阿里云百炼平台提供多种 OpenAI 兼容的工具包与框架接口，支持开发者快速迁移现有应用。核心能力覆盖文本生成（Chat、Completions、Responses）、[多模态](../concepts/multi-modal.md)理解（Vision）、向量化（Embedding）、批量处理（Batch）、会话管理（Conversations）及文件操作（Files），并兼容主流开发框架如 LangChain。所有接口均基于统一的 `compatible-mode/v1` 路径设计，通过调整 `base_url`、`api_key` 和 `model` 即可完成集成。 +阿里云百炼提供多套 OpenAI 兼容的 API 接口（toolkits），覆盖文本生成、视觉理解、嵌入向量、批量推理、文件管理及会话状态管理等场景，支持主流 SDK（如 OpenAI Python/Node.js SDK、LangChain）无缝集成。所有接口均基于统一的 `compatible-mode/v1` 路径设计，通过调整 `base_url`、`api_key` 和 `model` 即可迁移现有 OpenAI 应用。 ## 支持的模型/功能 -百炼支持的 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md)按功能划分如下： +百炼支持的 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)按功能划分为以下几类： -- **Chat Completions**：适用于标准对话场景，支持 Qwen 系列（`qwen-plus`、`qwen-flash` 等）、Qwen-VL、Qwen-Coder、Qwen-Omni、Qwen-Math，以及第三方直供模型（DeepSeek、Kimi、GLM、MiniMax）[原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 -- **Completions**：专为代码补全与内容续写设计，当前仅支持 `qwen-coder-turbo` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/completions.md)。 -- **Responses**：作为 Chat Completions 的演进版，内置联网搜索、网页抓取等智能体原生工具，支持 `qwen3.7-plus`、`qwen3.6-flash` 等新一代 Qwen3 系列模型，并在华北2（北京）、新加坡、弗吉尼亚等多地部署 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 -- **Vision**：支持视觉理解任务，兼容 `qwen3-vl-plus`、`QVQ`、`qwen-vl-ocr` 等模型，支持图像 URL 与 Base64 输入 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 -- **Embedding**：提供 `text-embedding-v4`、`v3`、`v2`、`v1` 四代文本向量模型，支持多语种及可选维度（如 `dimensions=1024`），但[多模态](../concepts/multi-modal.md) Embedding 模型（如 `qwen3-vl-embedding`）**不支持** [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md)。 -- **Batch**：分为两种模式： - - **文件批量（Batch File）**：通过 JSONL 文件异步提交请求，支持 `qwen3.7-max`、`qwen3-vl-plus`、`text-embedding-v4` 等模型，费用为实时调用的 50% [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md)； - - **同步 Batch Chat**：单请求同步等待返回，适用于数据标注等非实时场景，端点为 `https://batch.dashscope.aliyuncs.com/compatible-mode/v1` [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md)。 -- **Conversations**：用于跨设备/长时间会话状态管理，配合 Responses API 自动注入历史上下文，支持创建、查询、更新、删除会话及添加消息项 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-conversations.md)。 -- **Files**：支持上传文件用于文档问答（`purpose="file-extract"`）、批量推理（`purpose="batch"`）或模型调优（`purpose="fine-tune"`），最大单文件 150 MB（extract）、500 MB（batch）、300 MB（fine-tune） [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 +- **通用对话（Chat）**：兼容 `chat/completions`，支持 Qwen 系列大语言模型（如 `qwen3.7-plus`、`qwen-plus`）、Qwen-VL、Qwen-Coder、Qwen-Omni 及部分第三方模型（DeepSeek、Kimi、GLM 等）。详见 [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **智能体原生响应（Responses）**：作为 Chat Completions 的演进，内置联网搜索、网页抓取、代码解释器等工具，支持 `previous_response_id` 简化上下文管理，适用于复杂任务编排。[原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) 明确列出华北2（北京）、新加坡、美国（弗吉尼亚）等多地支持的 `qwen3.*` 模型列表。 +- **文本补全（Completions）**：专为代码补全、内容续写设计，当前仅支持 `qwen-coder-turbo`（中国内地北京地域），支持前缀生成与“前缀+后缀”中间生成两种模式。 +- **视觉理解（Vision）**：兼容 `chat/completions` 多模态调用，支持 `qwen3-vl-plus`、`qwen-vl-ocr` 等模型，接受 `image_url` 或 Base64 图像输入。 +- **嵌入向量（Embedding）**：支持 `text-embedding-v1` 至 `v4` 系列，提供不同维度与语种覆盖，但多模态 Embedding（如 `qwen3-vl-embedding`）**不支持** OpenAI 兼容协议，需使用专用接口。 +- **文件管理（Files）**：用于上传文档供 Qwen-Long/Qwen-Doc-Turbo 进行问答或作为 Batch 输入，`purpose` 可设为 `file-extract`、`batch` 或 `fine-tune`。 +- **批量推理（Batch）**：含两种形态——**文件输入式 Batch**（异步，支持 JSONL 多请求）和 **同步式 Batch Chat**（单请求阻塞等待），均享 50% 成本优惠。[原文标题](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) 与 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) 分别详述其适用模型与工作流。 +- **会话管理（Conversations）**：配合 Responses API 实现跨设备上下文延续，支持创建、查询、更新、删除会话及追加消息项。 -> **注意**：文档 1 和文档 5 均提及旧域名迁移建议（如 `dashscope.aliyuncs.com` → `{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），但文档 6 的“前提条件”中仍列出 `https://dashscope.aliyuncs.com/compatible-mode/v1` 为中国内地服务端点，与文档 1 的推荐实践存在不一致。实际生产环境应优先采用业务空间专属域名以保障性能与稳定性。 +> **注意**：文档 4（OpenAI Chat接口兼容）称支持 Qwen-Audio，但明确标注“Qwen-Audio不支持OpenAI兼容协议，仅支持DashScope协议”；而文档 3（OpenAI Vision接口兼容）未提及 Qwen-Audio，二者一致。此处以文档 4 的明确声明为准。 ## 关键参数 -| 参数 | 类型 | 必选 | 说明 | 适用接口 | -|------|------|------|------|----------| -| `base_url` | string | 是 | 接口根地址，需按地域和功能选择（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`）。`{WorkspaceId}` 须替换为控制台获取的实际 ID | 所有 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) | -| `model` | string | 是 | 模型名称，必须从各接口支持列表中选取（如 `qwen-plus`、`text-embedding-v4`） | 所有 OpenAI 兼容接口 | -| `stream` | boolean | 否 | 是否启用[流式输出](../concepts/streaming-output.md)，默认 `false`；流式响应中可通过 `stream_options={"include_usage": true}` 在末尾返回 token 统计 | Chat、Completions、Vision、Responses | -| `max_tokens` | integer | 否 | 最大生成 token 数，超限将截断输出（不影响模型内部生成逻辑） | Chat、Completions、Responses | -| `temperature` / `top_p` | float | 否 | 控制生成多样性，二者互斥，建议只设其一（`temperature ∈ [0,2)`，`top_p ∈ (0,1]`） | Chat、Completions、Responses | -| `enable_thinking` | boolean | 否 | Batch 场景下控制思考模式开关（`true`/`false`），影响 token 成本；须与 `model` 同级传入，不可置于 `extra_body` 内 | Batch Chat、Batch File | -| `dimensions` | integer | 否 | Embedding 接口专用，指定向量维度（仅 `text-embedding-v3`/`v4` 支持） | Embedding | -| `purpose` | string | 是（Files） | 文件上传用途：`file-extract`（文档分析）、`batch`（批量输入）、`fine-tune`（调优数据集） | Files | +| 参数 | 类型 | 说明 | 来源示例 | +|------|------|------|----------| +| `base_url` | string | 必填，服务端点。推荐使用业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1`），旧域名（`dashscope.aliyuncs.com`）仍可用但即将停用。 | [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)、[OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) | +| `model` | string | 必填，模型名称。不同接口支持范围不同：`completions` 仅限 `qwen-coder-turbo`；`embeddings` 仅限 `text-embedding-*`；`responses` 支持 `qwen3.*` 全系列。 | 文档 1、2、8 | +| `input` / `messages` / `prompt` | string / array / string | 根据接口类型选择：`responses` 接受字符串或消息数组；`chat/completions` 使用 `messages`；`completions` 使用 `prompt`；`embeddings` 使用 `input`。 | 文档 1、4、2、8 | +| `stream` | boolean | 控制输出方式。`true` 启用流式，`false`（默认）为完整响应。`stream_options={"include_usage": true}` 可在流末尾返回 token 统计。 | 文档 1、4、8 | +| `enable_thinking` | boolean | 仅 Batch 场景下有效（文档 6、7），控制是否启用思考模式（产生额外 reasoning tokens）。`qwen3.5/3.6/3.7` 系列默认开启，建议显式设置。 | [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) | ## 使用方式 -### 基础 SDK 配置（Python 示例） +### 基础调用（Python + OpenAI SDK） ```python from openai import OpenAI import os client = OpenAI( - api_key=os.getenv("DASHSCOPE_API_KEY"), # 强烈建议配置至环境变量 - base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" # 替换 {WorkspaceId} + api_key=os.getenv("DASHSCOPE_API_KEY"), + base_url="https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1" # 替换为实际 WorkspaceId +) + +# Chat 示例 +response = client.chat.completions.create( + model="qwen-plus", + messages=[{"role": "user", "content": "你是谁？"}] ) -``` -### 各接口典型调用 -- **Chat**：`client.chat.completions.create(model="qwen-plus", messages=[...])` -- **Completions**：`client.completions.create(model="qwen-coder-turbo", prompt="{code_prefix}")` -- **Responses**：`client.responses.create(model="qwen3.7-plus", input="你好！")` -- **Vision**：`client.chat.completions.create(model="qwen3-vl-plus", messages=[{"role":"user","content":[{"type":"text","text":"这是什么"},{"type":"image_url","image_url":{"url":"..."}}]}])` -- **Embedding**：`client.embeddings.create(model="text-embedding-v4", input="文本", dimensions=1024)` -- **Batch Chat**：使用 `base_url="https://batch.dashscope.aliyuncs.com/compatible-mode/v1"`，调用方式与 Chat 完全一致 -- **Conversations**：`client.conversations.create(items=[{"role":"system","content":"..."}])` -- **Files**：`client.files.create(file=Path("doc.pdf"), purpose="file-extract")` +# Responses 示例（带工具） +response = client.responses.create( + model="qwen3.7-plus", + input="查一下今天北京的天气", + tools=[{"type": "function", "function": {"name": "get_current_weather", ...}}] +) + +# Embedding 示例 +embedding = client.embeddings.create( + model="text-embedding-v4", + input="hello world", + dimensions=1024 +) +``` -LangChain 集成详见 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/use-bailian-in-langchain.md)，推荐 `langchain_openai.ChatOpenAI`（部分模型）或 `langchain_community.chat_models.tongyi.ChatTongyi`（全模型支持）。 +### LangChain 集成 +- **OpenAI 兼容层**（`langchain_openai`）：仅支持文档 4 所列部分模型，配置简单： + ```python + from langchain_openai import ChatOpenAI + llm = ChatOpenAI( + model="qwen-plus", + base_url="https://dashscope.aliyuncs.com/compatible-mode/v1", + api_key=os.getenv("DASHSCOPE_API_KEY") + ) + ``` +- **DashScope 原生层**（`langchain-community`）：支持全部百炼文本模型，需安装 `dashscope`： + ```python + from langchain_community.chat_models.tongyi import ChatTongyi + llm = ChatTongyi( + model="qwen-plus", + dashscope_api_key=os.getenv("DASHSCOPE_API_KEY") + ) + ``` + +### HTTP 直连（curl） +```bash +# Chat +curl -X POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/chat/completions \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + -H "Content-Type: application/json" \ + -d '{"model":"qwen-plus","messages":[{"role":"user","content":"hi"}]}' + +# File upload +curl -X POST https://dashscope.aliyuncs.com/compatible-mode/v1/files \ + -H "Authorization: Bearer $DASHSCOPE_API_KEY" \ + --form 'file=@"doc.pdf"' \ + --form 'purpose="file-extract"' +``` ## 限制和注意事项 -- **地域与模型绑定**：并非所有模型在所有地域可用。例如 `qwen3.7-max` 在北京、新加坡、弗吉尼亚、法兰克福、东京均支持，但 `qwen3.5-397b-a17b` 仅在北京、新加坡、法兰克福提供；`qwen3.7-plus` 在东京仅支持日本部署范围 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md)。 -- **三方模型可用性**：DeepSeek、Kimi 等第三方直供模型**仅在中国站的中国内地地域可用**，且需在控制台单独开通服务 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 -- **协议限制**：`Qwen-Audio` 不支持 OpenAI 兼容协议，仅支持 DashScope 原生协议；`QVQ` 模型仅支持[流式输出](../concepts/streaming-output.md) [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md)。 -- **文件配额**：百炼文件存储上限为 **10,000 个文件** 或 **100 GB 总大小**，任一达到即拒绝新上传 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md)。 -- **Batch 超时**：Batch Chat 默认等待 3600 秒（1 小时），超时后连接断开并返回错误；Batch File 任务最长等待时间为 `completion_window`（如 `"24h"`），需主动轮询状态 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md)。 -- **API Key 隔离**：不同地域（如北京 vs 新加坡）的 API Key **不可混用**，需分别获取并配置 [原文标题](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md)。 +- **地域与模型绑定**：并非所有模型在所有地域均可用。例如 `qwen3.5-ocr` 仅在华北2（北京）支持；`qwen3.7-max` 在德国法兰克福仅支持部分 `qwen3.5-*` 子型号。务必查阅各文档的“支持的模型”表格确认地域可用性。 +- **路径与域名迁移**：`/api/v2/apps/protocols/...` 等旧路径（如文档 1、9 中提及）已废弃，必须迁移到 `/compatible-mode/v1/` 新路径；同时建议从 `dashscope.aliyuncs.com` 迁移至业务空间专属域名以获得更高稳定性。 +- **Batch 特殊约束**： + - 文件输入式 Batch 要求 JSONL 格式，每行一个请求，`url` 字段需与 `endpoint` 参数一致； + - 同步 Batch Chat（文档 7）的超时时间上限为 3600 秒，需在客户端显式配置（如 Python 的 `with_options(timeout=1800)`）； + - `qwen3.5-omni-plus` 在 Batch 场景下不支持语音输出。 +- **文件服务配额**：百炼文件存储上限为 10,000 个文件且总大小 ≤100 GB；单个 `file-extract` 文件 ≤150 MB，`batch` 文件 ≤500 MB，`fine-tune` 文件 ≤300 MB。 +- **Embedding 维度控制**：仅 `text-embedding-v3` 和 `v4` 支持 `dimensions` 参数；`v1`/`v2` 固定维度不可调。 +- **Qwen-Audio 不兼容**：明确不支持 OpenAI 协议，必须使用 DashScope 原生 API。 ## 来源文档 -- [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) -- [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Responses接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-with-openai-responses-api.md) -- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) +- [completions 接口](../../raw/model-api-reference/toolkits-and-frameworks/completions.md) - [OpenAI Vision接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/qwen-vl-compatible-with-openai.md) +- [OpenAI Chat接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/compatibility-of-openai-with-dashscope.md) +- [OpenAI文件接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/openai-file-interface.md) - [OpenAI兼容-Batch（文件输入）](../../raw/model-api-reference/toolkits-and-frameworks/batch-interfaces-compatible-with-openai.md) - [OpenAI兼容-Batch Chat](../../raw/model-api-reference/toolkits-and-frameworks/openai-compatible-batch-chat.md) - [OpenAI Embedding接口兼容](../../raw/model-api-reference/toolkits-and-frameworks/embedding-interfaces-compatible-with-openai.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md index 38e16587..e3e2d648 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/vector-and-sort.md @@ -1,53 +1,58 @@ # vector and sort -`vector and sort` 是百炼平台提供的核心向量化与排序能力集合，涵盖文本、[多模态](../concepts/multi-modal.md)内容的向量生成（embedding）以及跨模态/纯文本的语义相关性重排序（rerank）。该能力支撑语义搜索、RAG、推荐系统、聚类等典型AI应用，支持同步、异步及OpenAI兼容调用方式，适用于从单条文本到百万级批量数据的不同场景。 +百炼平台提供文本向量化（vector）、多模态向量化（multimodal vector）和文本排序（rerank）三大核心能力，覆盖语义搜索、RAG、跨模态检索、聚类等典型AI应用链路。所有能力均支持同步/异步调用、OpenAI兼容接口及原生DashScope SDK，并通过统一的API Key与业务空间ID进行身份与资源管理。开发者可根据数据规模、模态类型、延迟敏感度选择合适模型与接口模式。 ## 支持的模型/功能 -### 文本向量模型（Embedding） -- **通用文本向量**：支持 `qwen3.7-text-embedding`、`text-embedding-v4`、`text-embedding-v3`、`text-embedding-v2`、`text-embedding-v1` 等系列模型，提供 64–2560 维可选向量，覆盖 201 种语种 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 -- **批处理文本向量**：`text-embedding-async-v2`（最大 100,000 行/请求，单行 ≤2,048 [Token](../concepts/token.md)）和 `text-embedding-async-v1`，专为大规模离线向量化设计 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **[多模态](../concepts/multi-modal.md)向量**：支持文本、图像、视频统一语义空间编码，包括 `qwen3-vl-embedding`（支持独立/融合向量）、`tongyi-embedding-vision-plus-2026-03-06`（支持多分辨率 `res_level` 和视频帧数控制 `max_video_frames`）等 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 - -### 排序模型（Rerank） -- **纯文本排序**：`qwen3-rerank`（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)，最大 500 文档/请求，单文档 ≤4,000 [Token](../concepts/token.md)），已替代即将下线的 `gte-rerank` 系列 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 -- **[多模态](../concepts/multi-modal.md)排序**：`qwen3-vl-rerank` 支持文本、图片、视频混合查询与文档排序（如“以图搜文”、“以文搜视频”），最大支持 100 文本/40 图片/4 视频文档 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 - -> **注意**：文档 4 明确指出 `gte-rerank` 模型将于 2026 年 05 月 30 日下线，新项目应使用 `qwen3-rerank` 或 `qwen3-vl-rerank`，避免依赖已废弃模型。 +- **文本向量模型**：支持通用文本嵌入，包括 `qwen3.7-text-embedding`（最高128K Token）、`text-embedding-v4`（默认1024维，8K Token）、`text-embedding-v3`、`text-embedding-v2` 和 `text-embedding-v1`。详细参数见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 +- **批处理向量模型**：专为大规模文本设计，支持单次10万行输入，模型为 `text-embedding-async-v2`（1536维）和 `text-embedding-async-v1`，采用异步任务模式，适用于离线批量处理场景 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 +- **多模态向量模型**：支持文本、图像、视频统一语义空间表征，包括 `qwen3-vl-embedding`（支持独立/融合向量）、`tongyi-embedding-vision-plus-2026-03-06`（支持多分辨率与融合）、`qwen2.5-vl-embedding`（仅融合）等 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- **文本排序（Rerank）模型**：对召回结果进行精准重排序，支持纯文本（`qwen3-rerank`）、多模态（`qwen3-vl-rerank`）及历史模型（`gte-rerank-v2`）。> **注意**：`gte-rerank` 系列模型将于2026年05月30日下线，[文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md) 文档已明确标注迁移建议，新项目应优先选用 `qwen3-rerank` 或 `qwen3-vl-rerank`。 ## 关键参数 -| 参数 | 适用模型 | 说明 | 是否必选 | -|------|----------|------|----------| -| `model` | 所有 | 模型名称，如 `"text-embedding-v4"`、`"qwen3-rerank"` | 必选 | -| `input` / `query` + `documents` | 所有 | 向量：字符串、字符串列表或文件 URL；排序：`query`（字符串或模态对象）+ `documents`（字符串列表或模态对象数组） | 必选 | -| `dimensions` | `qwen3.7-text-embedding`, `text-embedding-v3/v4`, `qwen3-vl-embedding`, `tongyi-embedding-vision-plus-2026-03-06` 等 | 指定向量维度，值域因模型而异（如 `text-embedding-v4`: 64–2048；`qwen3-vl-embedding`: 256–2560） | 可选（默认值见各模型概览） | -| `top_n` | `qwen3-rerank`, `qwen3-vl-rerank`, `gte-rerank-v2` | 返回前 N 个最相关结果 | 可选 | -| `enable_fusion` | 仅 `qwen3-vl-embedding` | `true` 时将 `contents` 中所有模态融合为 1 个向量；`false`（默认）则各模态独立生成向量 | 可选（仅该模型） | +| 参数名 | 适用模型 | 说明 | 是否必选 | +|--------|----------|------|----------| +| `model` | 全部 | 模型名称，如 `"text-embedding-v4"`、`"qwen3-vl-rerank"` | 必选 | +| `input` / `query` / `documents` | 因模型而异 | 向量：支持字符串、字符串数组或文件；排序：`query` + `documents` 数组；多模态：`input.contents` 数组，含 `text`/`image`/`video` 字典 | 必选 | +| `dimensions` | `qwen3.7-text-embedding`, `text-embedding-v3/v4`, `qwen3-vl-embedding`, `tongyi-embedding-vision-*` 等 | 指定向量维度，不同模型支持值不同（如 `text-embedding-v4`: 2048/1024/768…；`qwen3-vl-embedding`: 2560/1024…） | 可选（有默认值） | +| `encoding_format` | 同步文本向量 | 当前仅支持 `"float"` | 可选 | +| `enable_fusion` | `qwen3-vl-embedding` | `true` 时将 `contents` 中所有模态融合为1个向量；`false`（默认）则各模态独立生成向量 | 可选（仅该模型） | +| `top_n` | 排序模型 | 返回排序后前 N 个结果 | 可选 | | `instruct` | `qwen3-rerank`, `qwen3-vl-rerank` | 任务指令（如 `"Retrieve semantically similar text."`），影响排序策略 | 可选 | -| `res_level` / `max_video_frames` | 仅 `tongyi-embedding-vision-plus-2026-03-06` / `tongyi-embedding-vision-flash-2026-03-06` | 分辨率档位（0–3）和视频最大采样帧数（≤64） | 可选 | + +> **注意**：`tongyi-embedding-vision-plus` 和 `tongyi-embedding-vision-flash`（非2026-03-06快照版）**不支持 `dimension` 参数**，向量维度固定为1152/768；而 `multimodal-embedding-v1` 同样不支持该参数，固定1024维 —— 此信息在 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md) 的“模型能力对照”表格中有明确说明，与部分旧文档描述存在不一致，以该文档为准。 ## 使用方式 -### 同步调用（推荐小批量） -- **文本向量**：使用 OpenAI 兼容 SDK 或 HTTP POST 到 `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-mode/v1/embeddings`，支持 `input` 为字符串、列表或文件流 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 -- **排序**：`qwen3-rerank` 使用 [OpenAI 兼容接口](../concepts/openai-compatible-interface.md) `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/compatible-api/v1/reranks`；`qwen3-vl-rerank` 使用专用接口 `/{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/rerank/text-rerank/text-rerank`。 +- **同步调用（小规模、低延迟）**： + 使用 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)（`/compatible-mode/v1/embeddings`）或 DashScope SDK 的 `TextEmbedding.call()`。支持单文本、文本列表、文件流输入。示例见 [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 + +- **异步批处理（大规模、容忍延迟）**： + 通过 HTTP 创建任务（`/api/v1/services/embeddings/text-embedding/text-embedding`）或 SDK 的 `BatchTextEmbedding.call()` / `async_call()`。输入必须为公网可访问的文本文件 URL，单次最多10万行。任务状态需轮询查询。 -### 异步调用（推荐大批量） -- **文本向量**：通过 `X-DashScope-Async: enable` 头发起批处理任务，上传含文本的 OSS URL，再轮询 `GET /api/v1/tasks/{task_id}` 获取结果 [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **多模态向量/排序**：暂不支持异步模式，需同步调用。 +- **多模态向量（跨模态场景）**： + 使用 `/api/v1/services/embeddings/multimodal-embedding/multimodal-embedding` 接口，`input.contents` 传入混合模态对象数组。融合向量需按模型要求设置 `enable_fusion=true`（`qwen3-vl-embedding`）或同 content 对象内混写（`tongyi-embedding-vision-*2026-03-06`）。 -### SDK 封装 -DashScope SDK 提供 `BatchTextEmbedding`（批向量）、`TextReRank`（排序）等高层封装，自动处理地域配置、认证与响应解析，降低集成复杂度。示例见各文档中 Python/Java SDK 调用片段。 +- **文本排序（RAG精排）**： + `qwen3-rerank` 使用 OpenAI 兼容 `/compatible-api/v1/reranks` 接口（扁平参数结构）；`qwen3-vl-rerank` 和 `gte-rerank-v2` 使用 `/api/v1/services/rerank/text-rerank/text-rerank` 接口（嵌套 `input` 结构）。SDK 调用统一使用 `TextReRank.call()`，自动适配底层协议。 ## 限制和注意事项 -- **[Token](../concepts/token.md) 与尺寸限制**： - - `qwen3.7-text-embedding` 单文本最长 128,000 Token；`text-embedding-v4` 仅 8,192 Token；`qwen3-vl-embedding` 文本限 32,000 Token，图片 ≤10 MB，视频 ≤50 MB [同步接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-synchronous-api.md)。 - - `qwen3-rerank` 单次请求总 Token = `Query Tokens × Document 数量 + Document Tokens 总和`，上限 120,000；`qwen3-vl-rerank` 文本文档上限 100 条，图片上限 40 条 [文本排序](../../raw/model-api-reference/vector-and-sort/rerank-model/text-rerank-api.md)。 -- **免费额度与计费**：各模型均有 90 天有效期的免费额度（如 `text-embedding-v4` 100 万 Token），超限后按实际消耗计费；注意 `text-embedding-async-v2` 单价为 0.0007 元/千 Token，而 `text-embedding-v4` Batch 调用为 0.00025 元/千 Token [批处理接口API详情](../../raw/model-api-reference/vector-and-sort/general-text-vector/text-embedding-batch-api.md)。 -- **地域与 endpoint 差异**：北京地域使用 `cn-beijing.maas.aliyuncs.com`，新加坡地域需替换为 `ap-southeast-1.maas.aliyuncs.com`；多模态向量统一使用 `dashscope.aliyuncs.com` 公共域名 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 -- **模型能力差异**：`tongyi-embedding-vision-plus` 固定 1152 维，不支持 `dimension` 参数；`multimodal-embedding-v1` 不支持 `dimension` 且仅支持中英文；`qwen2.5-vl-embedding` 仅支持融合向量，不支持 `enable_fusion` 参数 [Multimodal-Embedding API详情](../../raw/model-api-reference/vector-and-sort/multimodal-vector/multimodal-embedding-api-reference.md)。 +- **输入长度与数量限制**： + - `qwen3.7-text-embedding`：单文本最长128,000 Token，批量最多20行； + - `text-embedding-v4`：单文本最长8,192 Token，批量最多10行； + - `text-embedding-async-v2`：单次请求最多100,000行，单行最长2,048 Token； + - `qwen3-vl-rerank`：文本文档最多100条、图片最多40张、视频最多4个，且 `Query Tokens × Document数 + Document Tokens总和 ≤ 120,000`。 + +- **地域与Endpoint差异**： + 所有接口均需替换 `{WorkspaceId}` 为真实业务空间ID，并根据地域选择 base URL（如北京：`cn-beijing.maas.aliyuncs.com`；新加坡：`ap-southeast-1.maas.aliyuncs.com`）。[OpenAI 兼容接口](../concepts/openai-compatible-api.md)与原生接口的 endpoint 路径和参数结构不同，不可混用。 + +- **免费额度与计费**： + 各模型均有开通后90天内的免费额度（如 `text-embedding-v4` 为100万Token），超出后按实际消耗Token计费。多模态模型中，文本、图片、视频分项计费（如 `qwen3-vl-embedding`：文本0.0007元/千Token，图片/视频0.0018元/千Token）。 + +- **限流策略**： + 同步接口受 RPS 与并发请求数限制；异步批处理接口限制单用户同时运行中任务≤3个、排队中+运行中任务总数≤50个。具体规则详见各文档中的“限流”章节。 ## 来源文档 diff --git a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md index 50b05f70..6450da5b 100644 --- a/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md +++ b/skills/bailian-docs-llm-wiki/wiki/api/video-generation-api.md @@ -1,118 +1,94 @@ # video generation api -百炼平台的 Video Generation API 提供多种视频生成与编辑能力，包括文生视频（T2V）、图生视频（I2V）、参考生视频（R2V）、视频编辑、口型替换、风格重绘等。所有接口均采用异步调用模式，任务创建后需轮询 `task_id` 获取结果，典型耗时为 1–5 分钟。开发者需确保模型、Endpoint URL 与 API Key 严格属于同一地域，跨地域调用将失败。 +百炼平台提供多种视频生成能力，覆盖文生视频、图生视频（首帧/首尾帧）、参考生视频、视频编辑、动作迁移、口型同步等核心场景。所有 API 均采用异步调用模式，需通过 `task_id` 轮询获取结果，任务有效期为 24 小时。调用前必须确保模型、Endpoint URL 与 API Key 三者地域一致，否则将鉴权失败或返回错误。 ## 支持的模型/功能 -API 覆盖三大类能力： +视频生成 API 按输入模态和任务类型分为以下几类： -- **基础生成类**：支持纯文本输入生成视频（如 `happyhorse-1.1-t2v`、`wan2.7-t2v-2026-06-12`、`vidu/viduq3-turbo_text2video`、`pixverse/pixverse-c1-t2v`），部分模型支持智能分镜或多镜头叙事（如 [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) 中通过 [prompt](../guides/prompt.md) 描述时间戳实现）。 - -- **[多模态](../concepts/multi-modal.md)驱动类**： - - 图生视频：支持首帧（`happyhorse-1.1-i2v`、`wan2.7-r2v-2026-06-12`）、首尾帧（`pixverse/pixverse-c1-kf2v`、`vidu/viduq3-turbo_start-end2video`）及视频续写； - - 参考生视频：支持传入多张图像/视频/音频（如 [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) 和 [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md)）； - - 视频编辑：支持指令式风格迁移（`wan2.7-videoedit`）、局部替换（`happyhorse-1.0-video-edit`）及超清增强（`pixverse/pixverse-upscale`）。 +- **文生视频（T2V）**：支持 `wan2.7-t2v-*`、`kling/kling-v3-*`、`pixverse/pixverse-*-t2v`、`vidu/viduq3-*-text2video` 等模型，支持多镜头叙事（如 `wan2.7` 通过 [prompt](../guides/prompt.md) 自然描述分镜，`pixverse-c1` 不支持 `shot_type` 参数）[万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md)。 +- **图生视频（I2V）**：包括基于首帧（`happyhorse-1.1-i2v`, `wan2.7-*it2v`, `pixverse-*it2v`, `vidu/*img2video`）、首尾帧（`pixverse-*kf2v`, `vidu/*start-end2video`, `wan2.2-kf2v-fla`）及视频续写（仅 `wan2.7` 支持）三种子类型；注意 `wan2.2-kf2v-fla` 使用 `/api/v1/services/aigc/image2video/video-synthesis` 路径，与其他模型路径不同 [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)。 +- **参考生视频（R2V）**：支持多图/图文/音视频混合输入，如 `happyhorse-1.1-r2v`、`wan2.7-r2v-*`、`pixverse/*r2v`、`vidu/viduq3-ad_reference2video`，适用于角色融合与多主体互动。 +- **视频编辑与增强**：涵盖风格迁移（`video-style-transform`）、超清（`pixverse/pixverse-upscale`）、对口型（`pixverse/pixverse-lipsync`）、动作模仿（`pixverse/pixverse-motioncontrol`）、换人（`wan2.2-animate-mix`）、舞蹈复刻（`wan2.2-animate-move`）等专用模型。 +- **数字人与播报**：`wan2.2-s2v`（语音驱动肖像）、`liveportrait`（轻量播报）、`emo-v1`（唱演）、`videoretalk`（口型替换）等均属人物视频生成范畴，需先调用对应 detect 模型校验输入合规性。 -- **人像动画类**：聚焦数字人与表情驱动，包括： - - 唱演/播报：`emo-v1`、`liveportrait`、`wan2.2-s2v`； - - 动作迁移：`animate-anyone-gen2`、`pixverse/pixverse-motioncontrol`； - - 口型替换：`videoretalk`、`pixverse/pixverse-lipsync`； - - 换人/复刻：`wan2.2-animate-mix`、`wan2.2-animate-move`。 - -> **注意**：文档中存在协议版本冲突。例如，万相系列明确区分“旧版协议”（仅支持 wan2.6 及更早模型，如 [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md)）与“新版协议”（仅支持 wan2.7 模型）。混用模型名与旧版 endpoint 将导致调用失败。 +> **注意**：文档中存在路径不一致问题。`wan2.2-kf2v-fla`（文档33）和 `wan2.2-animate-move`（文档9）、`wan2.2-animate-mix`（文档10）均使用 `/api/v1/services/aigc/image2video/video-synthesis`，而其余所有视频生成模型（含 `wan2.7` 全系列、`happyhorse`、`kling`、`pixverse`、`vidu`）统一使用 `/api/v1/services/aigc/video-generation/video-synthesis`。开发者务必根据所选模型查阅对应文档路径，否则请求将返回 404。 ## 关键参数 -所有请求必须包含以下通用参数： +所有请求必须包含以下基础参数： -- **HTTP 头部（Headers）**： - - `Content-Type: application/json`（必选）； - - `Authorization: Bearer $DASHSCOPE_API_KEY`（必选）； - - `X-DashScope-Async: enable`（必选；同步调用不被支持）。 +- **HTTP Header**： + - `Authorization: Bearer $DASHSCOPE_API_KEY`（必选） + - `Content-Type: application/json`（必选） + - `X-DashScope-Async: enable`（必选，异步模式强制启用） -- **请求体（Body）**： - - `model`：精确模型标识符（如 `"wan2.7-t2v-2026-06-12"`），不可省略； - - `input`：根据任务类型结构化： - - 文生视频：`{"prompt": "..."}`； - - 图生视频：`{"media": [{"type": "image_url", "url": "..."}], "prompt": "..."}`； - - 首尾帧：`{"media": [{"type": "first_frame", ...}, {"type": "last_frame", ...}], "prompt": "..."}`； - - 口型替换：`{"media": [{"type": "video_url", ...}, {"type": "audio_url", ...}]}`； +- **Request Body**： + - `model`：精确模型名称（如 `wan2.7-t2v-2026-06-12`, `pixverse/pixverse-c1-t2v`），不可省略。 + - `input`：结构因任务类型而异： + - T2V：`{"prompt": "文本描述"}` + - I2V（首帧）：`{"media": [{"type": "image_url", "url": "..."}], "prompt": "..."}` + - R2V：`{"media": [{"type": "reference_image", "url": "..."}, ...], "prompt": "..."}` + - 视频编辑：`{"media": [{"type": "video", "url": "..."}], "prompt": "..."}` - `parameters`：可选，常见字段包括： - - `duration`（秒，默认 5）； - - `resolution` 或 `size`（如 `"720P"`、`"1280*720"`）； - - `watermark`（布尔值，默认 `true`）； - - `aspect_ratio`（如 `"16:9"`，见 [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md)）； - - `style`（风格重绘专用，整数 0–7）。 + - `duration`: 视频时长（秒），通常为 3–5 秒 + - `resolution` / `size`: 分辨率（如 `"720P"`, `"1280*720"`, `"1024*576"`） + - `watermark`: 布尔值，控制是否添加水印（默认 `true`） + - `audio`: 布尔值，部分模型支持生成音频（如 `kling`） + - `style` / `mode`: 风格或模式选择（如 `kling` 的 `"std"` 或 `"omni"`） ## 使用方式 -1. **准备环境**： - - 在百炼控制台开通对应模型服务； - - 获取目标地域的 [API Key](https://help.aliyun.com/zh/model-studio/get-api-key) 并配置至环境变量 `DASHSCOPE_API_KEY`； - - 获取业务空间 ID（WorkspaceId），用于构造专属 endpoint。 - -2. **发起异步任务**： - - 向 `https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis` 发送 `POST` 请求； - - 所有模型共用该 endpoint（除少数遗留模型如 `wan2.2-kf2v-fla` 使用 `/image2video/` 路径，见 [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md)）； - - 成功响应返回 `{"task_id": "xxx"}`，有效期 24 小时。 - -3. **轮询结果**： - - 使用 `GET https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/tasks/{task_id}` 查询状态； - - 当 `status` 为 `"SUCCESS"` 时，`output.video_url` 即为生成视频地址。 +1. **开通服务**：在百炼控制台模型市场搜索并开通对应模型（如 `kling`, `PixVerse`, `Vidu`），确认其所属地域。 +2. **配置环境**：获取该地域的 API Key，并设置为环境变量 `DASHSCOPE_API_KEY`；获取业务空间 ID（WorkspaceId）。 +3. **构造请求**： + - 使用专属域名：`https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`（北京）或 `https://{WorkspaceId}.ap-southeast-1.maas.aliyuncs.com`（新加坡），**强烈推荐迁移至此新域名**以获得更高性能与稳定性 [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md)。 + - 发送 `POST` 请求至 `/api/v1/services/aigc/video-generation/video-synthesis`（或 `image2video/...`，见上文注意项）。 +4. **轮询结果**：从响应中提取 `task_id`，使用 `GET https://dashscope.aliyuncs.com/api/v1/tasks/{task_id}`（旧域名）或 `GET https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/tasks/{task_id}`（新域名）轮询，直至 `status` 变为 `"SUCCESS"`，`output.video_url` 字段即为生成视频地址。 ## 限制和注意事项 -- **地域强一致性**：模型、API Key、Endpoint 必须同属一个地域（如华北2北京），否则鉴权失败或返回 `401 Unauthorized`。新加坡、美国、德国等地域 endpoint 格式不同，需严格匹配。 - -- **URL 构造规范**：业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`）为推荐路径，旧域名（如 `https://dashscope.aliyuncs.com`）虽仍可用，但性能与稳定性较低。 - -- **并发与限流**： - - 多数模型对单账号 RPS/QPS 有限制（如 `liveportrait` 为 1 QPS，`emo-v1` 为 1 并发任务）； - - 免费额度按模型独立计算（如 `emo-detect-v1` 免费 200 张，`emo-v1` 免费 1800 秒）； - - 详细限流策略请查阅各模型资费文档。 - -- **输入约束**： - - 图像/视频 URL 必须公网可访问且 HTTPS 协议； - - 音频文件需为清晰人声（MP3/WAV），时长建议 ≤ 30 秒； - - Prompt 长度通常 ≤ 512 字符，含敏感词将触发拦截。 - -- **错误处理**：常见错误码包括 `400 Bad Request`（参数缺失或格式错误）、`403 Forbidden`（地域不匹配或配额不足）、`429 Too Many Requests`（超出限流）。建议在轮询逻辑中加入指数退避重试。 +- **地域强一致性**：模型、Endpoint、API Key 必须同属一个地域（如北京、新加坡、弗吉尼亚、法兰克福），跨地域调用必然失败。 +- **任务生命周期**：`task_id` 有效期为 24 小时，过期后无法查询结果，接口返回 `UNKNOWN` 状态。 +- **并发与限流**：各模型有独立 QPS/RPS 和同时处理任务数限制（如 `liveportrait` 同时处理中任务数量为 1，`emo-v1` 为 1），详见各模型文档的“资费与限流”章节。 +- **输入要求**：多数人物相关模型（`s2v`, `liveportrait`, `emo`, `animate-move`, `videoretalk`）需先调用 `detect` 模型验证图片/视频合规性，否则生成失败。 +- **弃用提示**：`wan2.1`–`wan2.6` 系列模型（如文档30–34）为旧版协议，官方明确推荐优先选用 `wan2.7` 新版 API，其功能更全、协议统一且持续迭代。 ## 来源文档 - [HappyHorse-文生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-text-to-video-api-reference.md) -- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-参考生视频API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-reference-to-video-api-reference.md) +- [HappyHorse-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-image-to-video-api-reference.md) - [HappyHorse-视频编辑API参考](../../raw/model-api-reference/video-generation-api/happyhorse-api-reference/happyhorse-video-edit-api-reference.md) -- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-图生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/image-to-video-general-api-reference.md) +- [万相2.7-文生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/text-to-video-api-reference.md) - [万相2.7-参考生视频API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-to-video-api-reference.md) -- [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - [万相2.7-视频编辑API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-video-editing-api-reference.md) +- [万相-图生动作API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-move-api.md) - [万相-视频换人API参考](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-animate-mix-api.md) -- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [万相-数字人](../../raw/model-api-reference/video-generation-api/wan-api-reference/wan-s2v-overview.md) +- [图生舞蹈视频-舞动人像AnimateAnyone](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/animateanyone-quick-start.md) - [图生唱演视频-悦动人像EMO](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emo-quick-start.md) - [图生播报视频-灵动人像LivePortrait](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/liveportrait-quick-start.md) - [视频口型替换-声动人像VideoRetalk](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/videoretalk.md) - [图生表情包视频-表情包Emoji](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/emoji-quick-start.md) -- [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) +- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) - [爱诗-文生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-text-to-video-api-reference.md) -- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-image-to-video-api-reference.md) +- [爱诗-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-keyframe-to-video-api-reference.md) - [爱诗-参考生视频API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-reference-to-video-api-reference.md) - [爱诗-视频对口型API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-lipsync-api-reference.md) -- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - [爱诗-视频动作模仿API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-motioncontrol-api-reference.md) -- [可灵-视频生成API文档](../../raw/model-api-reference/video-generation-api/kling-api-reference/kling-video-generation-api-reference.md) +- [爱诗-视频超清API参考](../../raw/model-api-reference/video-generation-api/pixverse-api-reference/pixverse-upscale-api-reference.md) - [Vidu-文生视频API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-text-to-video-api-reference.md) -- [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) +- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) - [Vidu-图生视频-基于首尾帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-keyframe-to-video-api-reference.md) -- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) +- [视频风格重绘API参考](../../raw/model-api-reference/video-generation-api/portrait-animation-api-reference/video-style-transform-api-reference.md) +- [Vidu-参考生视频 API 参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-reference-to-video-api-reference.md) - [万相-图生视频-基于首帧API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-api-reference.md) +- [万相-文生视频API参考（2.1-2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-text-to-video-api-reference.md) - [万相-参考生视频API参考（2.6）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wan-reference-to-video-api-reference.md) - [万相-首尾帧生视频API参考（2.2）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-image-to-video-by-first-and-last-frame-api-reference.md) - [万相-视频编辑API参考（2.1）](../../raw/model-api-reference/video-generation-api/wan-api-reference/legacy-video-models/legacy-wanx-vace-api-reference.md) -- [Vidu-图生视频-基于首帧API参考](../../raw/model-api-reference/video-generation-api/vidu-api-reference/vidu-image-to-video-api-reference.md) diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md deleted file mode 100644 index 77c9f6f3..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks-comparison.md +++ /dev/null @@ -1,69 +0,0 @@ -# 应用构建框架对比：Managed Agents vs Application Component API vs Frameworks - -## 对比目的与背景 - -在百炼平台生态中，开发者面临多种技术路径来构建 AI 原生应用：从全托管的智能体运行时（Managed Agents），到细粒度的数据与知识能力组合（Application Component API），再到面向主流开发范式的框架级集成（Frameworks）。三者定位不同、抽象层级各异、适用边界清晰。本对比旨在为开发者提供客观、可操作的技术选型参考，帮助其根据**应用形态、控制粒度、团队能力与交付节奏**等核心因素，快速判断最适合的构建路径，避免过度工程化或能力缺失风险。 - ---- - -## 关键维度对比表 - -| 维度 | Managed Agents API | Application Component API | Frameworks（LlamaIndex / Spring AI Alibaba） | -|------|---------------------|----------------------------|-----------------------------------------------| -| **定位与角色** | 智能体（Agent）全生命周期托管运行时，聚焦“会话驱动型”交互式应用 | 底层能力组件化服务，提供数据连接、知识库、Prompt 管理等原子能力，供自主编排 | 主流开源框架的百炼适配层，降低 RAG/智能体/知识检索类应用的接入门槛 | -| **输入格式** | ChatML 格式 message 数组（含 `role`, `type`, `content`），支持[多模态](../concepts/multi-modal.md)文本块；事件驱动模型 | RESTful 请求体（JSON），按接口语义定义（如 `AddFileRequest`, `RetrieveRequest`）；文件上传需先申请租约（Lease） | 框架原生对象（如 LlamaIndex 的 `Document`/`QueryEngine`，Spring AI 的 `Prompt`/`ChatClient`），由 SDK 自动序列化为百炼协议 | -| **输出格式** | SSE 流式事件（`message`, `tool_call`, `session_status` 等）或分页事件历史；结构化程度高，含 `thoughts`、`docReferences` 等语义字段 | JSON 响应体，严格遵循 OpenAPI Schema（如 `ListFilesResponse`, `RetrieveResponse`）；返回字段明确，但无统一语义层封装 | 框架标准返回类型（如 `Response`、`StreamingResponse`、`List`），经适配器映射为百炼能力，部分字段（如 `docReferences`）需显式启用 | -| **支持模型** | 仅限百炼托管模型（如 `qwen-plus`），`model.id` 必须为字符串且严格匹配平台列表；**不支持自定义/外部模型** | **不直接调用大模型**；所有模型能力通过下游组件（如知识库检索、Prompt 渲染）间接使用；知识库检索默认用 `qwen-max`，但不可在 Component API 层切换 | 支持指定生成模型（`qwen-max`, `qwen-plus`）及重排模型（`gte-rerank`）；模型名通过框架配置项传入，**仍受限于百炼公开模型池** | -| **API 端点** | 地域化 MAAS Endpoint（如 `https://.cn-beijing.maas.aliyuncs.com/api/v1/agentstudio`）；强绑定 workspace + region | ROA 风格通用 Endpoint（如 `bailian.cn-beijing.aliyuncs.com`）；按资源类型路由（`/data-connection`, `/knowledge-base`, `/prompt`） | **无独立端点**；复用 DashScope 统一 API（`dashscope.aliyuncs.com`）；框架内部完成鉴权、路由与协议转换 | -| **计费方式** | 按实际调用计费：Session 运行时长（秒）、工具执行次数、文件存储（GB/天）、事件流传输量；**会话空闲期不计费** | 按调用频次（QPS）与资源用量计费：文件解析/索引构建（按页/小时）、知识库检索（次）、Prompt 执行（次）、OSS 数据同步（流量） | **框架本身免费**；所有底层调用（模型推理、知识库检索、重排、文档解析）均按百炼对应计费项单独计费，与直接调用 API 一致 | -| **典型场景** | 客服对话机器人、多步骤任务助手（如订机票+查天气+发邮件）、需沙箱隔离与状态持久化的复杂工作流 | 构建企业级知识中枢（对接 ERP/CRM 文件）、管理 Prompt 版本库、自动化数据导入与索引构建、定制化检索增强流程 | 快速验证 RAG 效果、将现有 LlamaIndex/Spring Boot 应用迁移至百炼、需要框架生态（插件、可观测性、Spring 生态集成）的中大型项目 | -| **状态管理** | 内置完整状态机（`idle → running → terminated`）；Session 自动管理上下文、工具状态、中断恢复；支持 `archive`/`delete` 终态控制 | **无会话状态**；纯无状态 CRUD 接口；状态需由调用方自行维护（如缓存检索上下文、轮询任务状态） | 依赖框架自身状态管理（如 LlamaIndex 的 `Index` 实例、Spring AI 的 `ChatClient` Bean）；百炼侧不维护跨请求状态 | -| **安全与隔离** | 工作空间级资源隔离；沙箱环境（`cloud` 类型）提供网络/依赖隔离；Skill/File 需安全扫描后激活 | RAM 子账号 + 最小权限策略；文件/知识库/Category 均归属 Workspace；OSS 授权需主账号显式配置 | 继承框架运行时安全模型（如 Spring Security）；百炼侧仅校验 `DASHSCOPE_API_KEY`，不感知框架内权限体系 | - ---- - -## 各方案适用场景建议 - -### ✅ 选择 **Managed Agents API** 当： -- 应用核心是**多轮、有状态、带工具调用的对话体验**（如销售顾问、IT 支持助手）； -- 需要开箱即用的**沙箱执行环境**（运行 Python 工具脚本、访问内部 API）； -- 要求**会话级状态自动管理**（上下文延续、中断恢复、超时清理）； -- 团队希望**最小化运维负担**，专注 Agent 设计与 Skill 编排，而非底层基础设施； -- 对模型选择无定制需求，接受百炼托管模型能力边界。 - -### ✅ 选择 **Application Component API** 当： -- 构建**后台数据中枢或知识平台**，需精细控制文件解析、知识库构建、Prompt 版本发布等环节； -- 需要**与现有系统深度集成**（如定时同步数据库表、监听 OSS 事件触发知识更新）； -- 要求**完全自主的状态与流程编排**（例如：自定义重排逻辑、混合检索策略、多知识库路由）； -- 团队具备较强后端开发能力，熟悉 RESTful 设计与幂等性处理； -- 需要规避框架锁定，保持未来技术栈演进灵活性（如迁移到自研调度引擎）。 - -### ✅ 选择 **Frameworks** 当： -- 项目已基于 **LlamaIndex 或 Spring Boot 技术栈**，追求**零改造迁移**至百炼； -- 目标是**快速原型验证或 MVP 上线**，优先保障开发效率而非极致控制； -- 需要利用框架生态能力（如 LlamaIndex 的 Node Postprocessor、Spring AI 的 `Advisor` 机制、Actuator 健康检查）； -- 应用形态明确为 **RAG 检索问答、智能体调用、或知识库增强聊天**，无需沙箱或复杂状态机； -- 团队熟悉对应框架，且接受其抽象带来的约束（如 LlamaIndex 不支持自定义切分器）。 - ---- - -## 技术选型决策指南（面向开发者） - -| 决策维度 | 推荐方案 | 关键判断依据 | -|----------|----------|--------------| -| **应用是否需要“会话”概念？** | Managed Agents API | 若用户交互天然具有上下文依赖（如“上一条说的XX，现在帮我查下相关文档”），且需自动维持状态，则 Agents 是唯一选择。Component API 和 Frameworks 均需自行实现会话管理。 | -| **是否必须运行任意代码（Python/Shell）？** | Managed Agents API | 只有 Managed Agents 提供沙箱环境（`Environment`）支持工具脚本执行；Component API 仅提供数据能力，Frameworks 仅调用百炼已有服务。 | -| **是否已有成熟框架代码基？** | Frameworks | 若已有 LlamaIndex 构建的 RAG 应用或 Spring Boot 项目，直接集成 Frameworks 可节省 80%+ 接入成本；反之，为新项目强行引入框架可能增加学习曲线。 | -| **是否需对接非百炼数据源（如 MySQL、SharePoint）？** | Application Component API | Component API 的 `Connector` 和 `AddFilesFromAuthorizedOss` 支持标准化对接；Frameworks 仅支持百炼知识库；Managed Agents 需将对接逻辑封装为 Skill（开发成本高）。 | -| **是否要求模型完全可控（微调/私有部署）？** | ❌ 三者均不支持 | 百炼当前所有路径均**仅支持平台托管模型**；若需私有模型，请评估百炼 Model Studio 或阿里云 PAI 平台。 | -| **团队是否缺乏全栈 AI 工程经验？** | Managed Agents API 或 Frameworks | Agents 提供最高抽象（拖拽式 Agent 配置 + SDK 调用）；Frameworks 利用社区惯用范式降低认知负荷；Component API 要求理解 ROA、租约、幂等性等细节，适合资深后端。 | - -> **重要提醒**：三者并非互斥。生产实践中常见**组合使用**——例如：用 Application Component API 构建和维护知识库，再通过 Managed Agents API 创建调用该知识库的智能体；或用 Frameworks 快速搭建前端 Demo，后端核心流程用 Component API 实现高可靠性调度。选型应以**端到端交付价值**为最终目标，而非单一技术指标。 - -## 被对比主题页 - -- [managed agents api](../api/managed-agents-api.md) -- [application component api reference](../api/application-component-api-reference.md) -- [frameworks](../api/frameworks.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md deleted file mode 100644 index 99fb9faa..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-frameworks.md +++ /dev/null @@ -1,69 +0,0 @@ -# 应用开发框架对比：Managed Agents、Application Component 与 Toolkits - -## 对比目的与背景 - -在百炼平台构建 AI 应用时，开发者面临多种技术路径选择：是直接调用模型能力快速验证想法？还是构建可复用、可审计的智能体系统？抑或集成知识库与数据连接能力打造企业级应用？`Managed Agents`、`Application Component` 和 `Toolkits` 是百炼平台面向不同抽象层级提供的三类核心开发框架，分别聚焦于**智能体生命周期管理**、**数据与知识基础设施编排**、以及**标准化模型能力接入**。本页旨在从技术定位、能力边界、使用约束和工程实践角度进行客观对比，帮助开发者基于业务目标、团队能力与运维要求做出理性选型决策。 - ---- - -## 关键维度对比表 - -| 维度 | Managed Agents | Application Component | Toolkits（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)） | -|------|----------------|------------------------|------------------------------| -| **核心定位** | 托管式智能体运行时：统一管理会话、沙箱、工具链与事件流 | 数据与知识基础设施 API：聚焦数据连接、知识库构建、解析配置与元数据管理 | 标准化模型能力网关：提供 OpenAI 兼容协议，屏蔽底层模型差异，支持快速迁移与多模态调用 | -| **输入格式** | 结构化 JSON Event 消息数组（含 `role`/`type`/`content`），严格遵循 Session Schema；支持文本、图像 URL、文件引用（需预上传并审核通过） | 多样化资源操作请求：
• 文件：`AddFile` + `Parser` 指定解析器
• 知识库：`CreateIndex` + `SubmitIndexJob`
• 类目/连接器：ROA 风格参数（如 `CategoryId`, `IndexId`, `FileType`） | OpenAI 标准 REST 请求体：
• Chat：`messages: [{role, content}]`
• Vision：`messages` + `image_url` 或 `base64_image`
• Embedding：`input: string/array`
• Files：`file` + `purpose`（必填） | -| **输出格式** | SSE 流式事件（`message`, `session_status`, `tool_call` 等）或轮询历史事件；响应结构含 `event_id`, `session_id`, `output`（含 `text`, `tool_calls`, `files` 等） | ROA 风格 JSON 响应：
• 创建类：返回 `ResourceId`（如 `FileId`, `IndexId`）
• 查询类：返回完整资源对象（含状态、统计、元数据）
• 检索类（`Retrieve`）：返回 `chunks` 数组及 `score` | OpenAI 兼容 JSON：
• Chat/Responses：`choices[0].message.content` / `output_text`
• Embedding：`data[0].embedding`
• Vision：`choices[0].message.content` 或 `data[0].url`（QVQ 流式）
• Batch：异步 `id` + 回调通知 | -| **支持模型** | 仅百炼托管大模型（当前限 `qwen-plus` 等），模型 ID 必须为对象 `{"id": "qwen-plus"}`；不支持自定义/外部模型 | **不直接调用模型**；为模型调用提供数据支撑（如知识库检索结果作为 RAG 上下文） | 广泛支持：`qwen-plus`, `qwen3-*`, `Qwen-VL`, `QVQ`, `Qwen-OCR`, `text-embedding-v*`, `qwen-coder-turbo` 等；按接口能力隔离（如 Completions 仅支持 `qwen-coder-turbo`） | -| **API 端点** | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`（如 `ws_abc.cn-beijing.maas.aliyuncs.com`） | `bailian.{region}.aliyuncs.com`（公网）或 `bailian-vpc.{region}.aliyuncs.com`（VPC）；需显式指定 `WorkspaceId` 路径参数 | 多域名策略：
• Chat/Vision/Embedding：`https://{WorkspaceId}.{region}.maas.aliyuncs.com/compatible-mode/v1`
• Files/Batch（文件）：`https://dashscope.aliyuncs.com/compatible-mode/v1`（中国内地）
• **Batch Chat（专用）：`https://batch.dashscope.aliyuncs.com/compatible-mode/v1`** | -| **认证方式** | Bearer [Token](../concepts/token.md)（`Authorization: Bearer `），API Key 与工作空间强绑定 | RAM AccessKey 签名（ROA），需最小权限策略（如 `AliyunBailianDataFullAccess`） | Bearer [Token](../concepts/token.md)（`Authorization: Bearer `），**API Key 必须与 endpoint 地域一致**（北京 Key 不能用于新加坡 endpoint） | -| **计费方式** | 按 Session 运行时长（秒）+ 工具调用次数 + 文件存储（GB/月）计费；沙箱资源消耗计入 Session 成本 | 按调用次数（QPS）+ 知识库存储（GB/月）+ 文件解析/切片处理量计费；无模型推理费用（仅为数据层） | 按模型调用 token 数（输入+输出）计费；Batch 享 5 折优惠；Files 上传免费，用途相关（如 `fine-tune` 有额外费用） | -| **典型场景** | • 可复用客服智能体（带审批流、多工具协同）
• 合规审计型业务助手（完整事件溯源、版本快照）
• 需沙箱隔离的代码执行/文件分析任务 | • 构建企业级知识库（PDF/Word/Excel 自动入库+切片）
• 管理多源数据连接（OSS、数据库类目）
• 定制化文档解析策略（如合同关键字段提取） | • 快速迁移 OpenAI 应用（零代码修改）
• 多模态应用（图文理解、OCR、向量化）
• 批量推理任务（日志分析、报告生成） | - ---- - -## 各方案适用场景建议 - -### ✅ 优先选用 **Managed Agents** -- 需要**端到端智能体生命周期管理**：如会话状态机（`idle`→`running`→`terminated`）、中断恢复、工具调用审批、事件流订阅。 -- 要求**强安全与合规控制**：沙箱环境隔离、Skill ZIP 包安全扫描、文件审核机制、版本锁定（禁止 `latest`）。 -- 构建**可复用、可归档、可审计**的智能体资产：Agent 版本化、Environment 复用、Session 快照追溯。 -- 场景示例：金融理财顾问（需风控工具调用+会话留痕）、HR 招聘助手（简历解析+面试问答+合规提示）。 - -### ✅ 优先选用 **Application Component** -- 核心需求是**数据与知识基础设施建设**：如将数百份 PDF 合同自动构建知识库、对接内部 OSS 存储、定制化解析规则。 -- 需要**精细化管理知识库元数据**：如按部门/项目分类管理 Index、动态追加文档、删除错误切片、监控索引质量。 -- 业务逻辑依赖**结构化数据连接**：如将 CRM 表格数据作为 RAG 上下文源（注意：`AddTable` 为受限功能，推荐控制台操作）。 -- 场景示例：法务合同审查系统（OSS 批量导入+法律条款切片+高精度检索）、产品文档中心（多格式解析+版本化更新）。 - -### ✅ 优先选用 **Toolkits（[OpenAI 兼容接口](../concepts/openai-compatible-interface.md)）** -- 追求**开发效率与生态兼容性**：已有 OpenAI SDK 代码，希望最小改动接入百炼模型。 -- 需要**灵活组合多模态能力**：如同时调用 `Chat`（对话）、`Vision`（图片理解）、`Embedding`（向量检索）构建多阶段流水线。 -- 执行**大规模批量任务**：如每日 10 万条用户反馈情感分析（Batch Chat 5 折）、千万级文本向量化（Embedding Batch）。 -- 场景示例：SaaS 客户支持[插件](../concepts/plugin.md)（OpenAI SDK 直接切换 endpoint）、电商商品图搜系统（Qwen-VL + Embedding）、AI 写作助手（qwen3.7-plus + qwen-coder-turbo 协同）。 - ---- - -## 技术选型参考指南（面向开发者） - -| 选型考量因素 | 推荐方案 | 说明 | -|--------------|----------|------| -| **是否需要模型推理以外的能力（如知识库、文件解析）？** | → 若需：**Application Component**
→ 若仅需模型调用：继续评估 | Application Component 是数据层基石，Toolkits/Managed Agents 均可消费其产出（如知识库检索结果传入 Agent 的 `input`）。 | -| **是否必须保证会话状态一致性与工具链可审计？** | → 是：**Managed Agents**
→ 否：考虑 Toolkits | Managed Agents 提供唯一具备完整状态机与事件溯源的框架；Toolkits 的 `Conversations` 仅提供轻量会话 ID 管理，无状态持久化保障。 | -| **团队是否已使用 OpenAI 生态（SDK、Prompt 工程规范）？** | → 是：**Toolkits**（首选）
→ 否：评估学习成本 | Toolkits 最小化迁移成本；Managed Agents 需理解 Agent/Environment/Session 三层抽象；Application Component 需熟悉 ROA 签名与资源生命周期。 | -| **是否涉及敏感数据处理（如客户隐私文件）？** | → 是：**Managed Agents**（沙箱隔离）或 **Application Component**（私有 VPC endpoint）
→ 否：均可 | Managed Agents 的 `cloud` 沙箱提供进程级隔离；Application Component 支持 VPC endpoint 避免公网传输；Toolkits 默认走公网（需确认合规策略）。 | -| **是否需要未来扩展自定义模型或外部服务集成？** | → 是：**Toolkits**（开放模型列表）或 **Application Component + 自研服务**
→ 否：Managed Agents 更省心 | Managed Agents 当前**不支持自定义模型**；Toolkits 持续扩展模型支持；Application Component 可通过连接器对接自研服务。 | -| **运维复杂度要求（CI/CD、监控、告警）？** | → 低：**Toolkits**（标准 HTTP 接口）
→ 中：**Application Component**（需管理 Index/Category 状态）
→ 高：**Managed Agents**（需监控 Session 状态机、Skill 安全扫描、文件审核） | Toolkits 接口行为最接近传统 REST；Managed Agents 引入更多异步状态（如 `session_status` 变更），需适配 SSE 或轮询。 | - -> **联合使用建议**：实际生产中三者常组合使用—— -> **Application Component** 构建知识库 → 输出 `IndexId` 与检索结果； -> **Toolkits**（`Embedding` + `Retrieve`）实现快速原型验证； -> **Managed Agents** 封装最终交付形态，将知识库检索结果、工具调用、用户会话统一编排为可发布智能体。 -> 此分层架构兼顾敏捷性、可维护性与企业级治理要求。 - -## 被对比主题页 - -- [managed agents api](../api/managed-agents-api.md) -- [application component api reference](../api/application-component-api-reference.md) -- [toolkits and frameworks](../api/toolkits-and-frameworks.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration-comparison.md new file mode 100644 index 00000000..f88b01ae --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/application-orchestration-comparison.md @@ -0,0 +1,71 @@ +# 应用编排能力对比：Managed Agents API vs Application Component API vs Application Call + +## 背景与目的 +在百炼平台构建企业级 AI 应用时，开发者常需在不同抽象层级间进行技术选型：是直接托管智能体运行时（Managed Agents），还是复用平台基础能力组件（Application Component），抑或快速调用已发布的成熟应用（Application Call）？三者定位差异显著——**Managed Agents API 面向“可编程智能体系统”的深度定制；Application Component API 聚焦“数据与知识基础设施”的原子能力编排；Application Call 则服务于“开箱即用型应用”的轻量集成**。本文旨在从工程实践视角，系统对比三者的输入输出、模型支持、部署模型、计费逻辑与适用边界，帮助开发者基于业务复杂度、运维诉求与交付节奏做出理性技术选型。 + +--- + +## 关键维度对比表 + +| 维度 | Managed Agents API | Application Component API | Application Call | +|------|---------------------|----------------------------|-------------------| +| **核心定位** | 智能体全生命周期托管运行时（含沙箱、会话、事件流） | 平台级基础能力组件（数据连接、知识库、[Prompt 工程](../concepts/prompt-engineering.md)） | 已发布智能体/工作流的标准化调用入口 | +| **输入格式** | `Event` 对象数组（含 `role`, `type`, `content`, 文件引用）；需显式构造 `Session` + `Agent` + `Environment` 上下文 | 多样化资源操作请求：
• 数据接入：`ApplyFileUploadLease` + `AddFile`（含 `Parser`）
• 知识库：`CreateIndex` + `SubmitIndexJob`
• Prompt：`CreatePromptTemplate`（JSON 模板） | 简洁应用输入：
• 字符串（单轮文本）
• 消息数组（多轮/图像/文件）
• 支持 `stream=true` 或 `background=true` 控制交互模式 | +| **输出格式** | SSE 流式事件（`session_status`, `tool_call`, `assistant_message` 等）+ RESTful 同步响应（如 Session 创建结果） | RESTful 同步响应为主：
• `ListFile` 返回文件元信息
• `Retrieve` 返回结构化检索结果
• `GetPromptTemplate` 返回模板 JSON | 三种模式：
• 同步：完整 JSON 响应（含 `output.text`）
• 流式：SSE Chunk（`data: {...}`）
• 异步：任务 ID + 轮询 `GET /tasks/{id}` | +| **支持模型** | 显式指定 `model.id`（如 `"qwen-plus"`），需在 Agent 创建时绑定；支持百炼已开通的推理模型 | **不直接调用模型**；为模型提供数据支撑（知识库检索结果、Prompt 模板注入、结构化数据输入） | 由被调用的 App 内部决定；调用方无需关心模型细节（透明封装）；支持 VL 模型处理图像等多模态输入 | +| **API 端点** | `https://{workspace_id}.{region}.maas.aliyuncs.com/api/v1/agentstudio`（`region` 仅限 `cn-beijing`） | `https://bailian.{region}.aliyuncs.com`（如 `bailian.cn-beijing.aliyuncs.com`），按地域分服务地址 | `https://dashscope.aliyuncs.com/api/v1/apps/{APP_ID}/completion`（DashScope 原生）
`https://dashscope.aliyuncs.com/api/v2/apps/agent/{APP_ID}/compatible-mode/v1/responses`（OpenAI 兼容） | +| **鉴权方式** | Bearer Token（`Authorization: Bearer `），API Key 在百炼控制台创建 | ROA 签名机制（AccessKey ID/Secret），推荐使用 SDK 自动签名；需 RAM 授权（如 `sfm:Retrieve`） | Bearer Token（`Authorization: Bearer `），与 DashScope 生态统一 | +| **计费方式** | 按 **Agent 运行时消耗** 计费：
• Session 生命周期内模型 Token、工具执行、沙箱资源占用
• 文件存储（30天）、技能扫描等附加费用 | 按 **组件使用量** 计费：
• 知识库索引构建与存储（GB/月）
• 文件解析次数（次）
• Prompt 模板调用（次）
• 检索 QPS（超出免费额度后） | 按 **应用调用次数与输出 Token** 计费：
• 同步/异步调用均计入调用次数
• [流式输出](../concepts/streaming-output.md)按实际生成 Token 计费
• 与所调用 App 的计费策略一致（App 发布者配置） | +| **典型场景** | • 需自定义工具链与沙箱环境的金融风控助手
• 多步骤、长周期、状态敏感的客服工单处理流程
• 需实时监听工具执行中间结果的自动化运维 Agent | • 构建企业专属知识库（PDF/Excel/音视频）并对接业务系统
• 将 CRM 表格数据接入 Prompt 模板生成销售话术
• 动态管理数百个 Prompt 版本用于 A/B 测试 | • 前端 Web/App 直接嵌入客服机器人
• ERP 系统通过 API 调用合同审核工作流
• 快速集成第三方 SaaS 提供的 AI 插件（如会议纪要生成） | + +--- + +## 适用场景建议 + +### ✅ 选择 **Managed Agents API** 当： +- 你需要**完全掌控智能体行为逻辑**：例如定义复杂的状态机（`idle → running → waiting_for_tool → idle`），精确拦截并处理每个工具调用事件； +- 业务要求**强隔离与安全沙箱**：如处理用户上传的代码文件，需在云端隔离环境中执行并捕获 stdout/stderr； +- 存在**多版本协同演进需求**：Agent 配置、Environment 依赖、Skill 工具包需独立版本管理，并支持灰度发布； +- 开发团队具备**事件驱动架构经验**，能妥善处理 SSE 流、会话超时、中断恢复等底层细节。 + +### ✅ 选择 **Application Component API** 当： +- 你的核心挑战是**数据准备与知识治理**：例如将 500 份产品手册 PDF 构建为可检索的知识库，或把数据库视图同步为 Prompt 可引用的变量； +- 需要**解耦模型能力与数据层**：让同一套 Prompt 模板适配不同知识库（测试库/生产库），或动态切换解析器（`DOCMIND_DIGITAL` vs `AUTO_SELECT`）； +- 项目处于**MVP 验证阶段**：快速验证“用 RAG 提升问答准确率”是否成立，无需编写 Agent 逻辑，专注数据质量与检索效果； +- 团队角色分离明确：**数据工程师负责 Component 管理，算法工程师专注模型微调，前端工程师调用 Application Call**。 + +### ✅ 选择 **Application Call** 当： +- 你追求**最快上线速度与最低维护成本**：已有现成的“报销单识别”App，只需 3 行代码调用即可集成到 OA 系统； +- 客户侧存在**严格的合规要求**：必须使用平台认证的、已通过安全审计的预发布应用，禁止自行部署 Agent； +- 集成方技术栈受限：如遗留 Java 系统只能调用 [OpenAI 兼容接口](../concepts/openai-compatible-api.md)，此时 Responses API 提供零改造迁移路径； +- 业务流量波动大：依赖百炼平台自动扩缩容能力，无需自行管理 Session 并发数与沙箱资源池。 + +--- + +## 技术选型决策树（面向开发者） + +```mermaid +graph TD + A[你的需求是什么？] --> B{是否需要自定义工具执行逻辑？} + B -->|是| C[Managed Agents API] + B -->|否| D{是否需构建/管理知识库或数据连接？} + D -->|是| E[Application Component API] + D -->|否| F{是否已有现成可用的应用？} + F -->|是| G[Application Call] + F -->|否| H[先用 Component 构建知识底座，再用 Call 调用] + C --> I[评估：是否有能力维护会话状态、处理 SSE、管理沙箱生命周期？] + E --> J[评估：是否需精细控制解析策略、Chunk 分片、权限粒度？] + G --> K[评估：是否接受平台对模型、超时、重试的统一策略？] +``` + +> **关键提醒**：三者并非互斥，而是**分层协作关系**。典型生产架构常组合使用： +> **Application Component** 构建知识库 → **Managed Agents** 封装领域专家 Agent → **Application Call** 对外提供标准化服务接口。 +> 开发者应避免“过度设计”（如用 Managed Agents 实现简单问答）或“能力缺失”（如用 Application Call 试图绕过沙箱执行 Python 代码）。始终以**最小可行抽象**匹配业务复杂度。 + +## 被对比主题页 + +- [managed agents api](../api/managed-agents-api.md) +- [application component api reference](../api/application-component-api-reference.md) +- [application call](../api/application-call.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-api-comparison.md new file mode 100644 index 00000000..e6e0d13a --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-api-comparison.md @@ -0,0 +1,66 @@ +# 多模态生成 API 对比：图像生成 vs 视频生成 vs 3D 生成 + +本文旨在为开发者提供百炼平台三大核心多模态生成能力（图像、视频、3D）的系统性对比，帮助技术团队基于业务需求、技术约束与成本效益，快速完成 API 选型决策。随着 AIGC 应用从静态内容向动态表达与空间交互演进，理解各模态在输入范式、输出形态、调用机制及工程适配上的差异，已成为构建高质量生成服务的关键前提。 + +--- + +## 关键维度对比表 + +| 维度 | 图像生成 | 视频生成 | 3D 生成 | +|------|----------|----------|---------| +| **核心能力定位** | 静态视觉内容创建与编辑（文生图、图生图、局部重绘、风格迁移等） | 动态时序内容生成（文生视频、图生视频、参考生视频、动作迁移、口型同步等） | 空间结构建模（文本/单图/多图→可渲染 GLB 模型，支持 PBR 材质） | +| **输入格式** | • 文本提示（`messages` 或 `prompt`）
• 输入图像 URL（仅编辑类模型）
• 支持图文混排（`messages` 中含 `image` 元素） | • 纯文本（T2V）
• 单张/首帧图像 URL（I2V）
• 首尾帧图像 URL（KF2V）
• 多图/音视频混合参考（R2V）
• 原始视频 URL（编辑类） | • 纯文本（`prompt`）
• 单张图像 URL（`image`）
• 四视角图像数组（`images`，顺序：前/左/后/右，空位用 `{}` 占位）
（三者**互斥**，不可共存） | +| **输出格式** | • PNG/JPEG 图片 URL（HTTP 同步返回或异步任务结果中获取）
• 支持多张并行生成（`n=1–9`） | • MP4 视频 URL（异步返回，含水印/音频可选）
• 输出时长通常为 3–5 秒
• 部分模型支持预览图（`preview_url`） | • GLB 格式 3D 模型 URL（`pbr_model_url` 或 `base_model_url`）
• 渲染预览图 URL（`rendered_image_url`）
• 所有 URL **有效期仅 2 小时**，需及时下载 | +| **支持模型（代表性）** | • 万相系列：`wan2.6-t2i`（推荐 V2）、`wan2.5-i2i-preview`
• 千问系列：`qwen-image-3.0-pro`、`qwen-image-edit-max`
• 轻量专用：`z-image-turbo`、`kling/kling-v3-omni-image-generation`、`shoemodel-v1` | • 文生视频：`wan2.7-t2v-*`、`kling/kling-v3-*`、`pixverse/pixverse-*-t2v`、`vidu/viduq3-*-text2video`
• 图生视频：`happyhorse-1.1-i2v`、`wan2.7-*it2v`、`pixverse-*kf2v`
• 参考/编辑：`wan2.7-r2v-*`、`pixverse/pixverse-upscale`、`pixverse/pixverse-lipsync` | • `Tripo/Tripo-P1.0`（专业版，≤2 万面，速度快）
• `Tripo/Tripo-H3.1`（高精度版，≤200 万面，支持 `ultra` 几何质量） | +| **API 端点（典型）** | • 同步：`/api/v1/services/aigc/multimodal-generation/generation`（如 `wan2.6-t2i`）
• 异步：`/api/v1/services/aigc/image2image/image-synthesis`（如 `wan2.5-i2i-preview`） | • 主路径：`/api/v1/services/aigc/video-generation/video-synthesis`
• 特殊路径：`/api/v1/services/aigc/image2video/video-synthesis`（仅 `wan2.2-kf2v-fla`、`wan2.2-animate-move` 等旧模型） | • 统一路径：
`/api/v1/services/aigc/video-generation/3d-generation`
（注意：路径名含 `video-generation`，属历史命名，实际为 3D 专属） | +| **调用模式** | • **混合支持**：部分模型（`wan2.6-t2i`, `z-image-turbo`）支持 HTTP 同步调用；多数编辑类模型强制异步
• 同步响应快（毫秒级），异步需轮询（`GET /tasks/{task_id}`） | • **强制异步**：所有模型均需 `X-DashScope-Async: enable`，创建任务后轮询状态
• `task_id` 有效期 24 小时 | • **强制异步**：必须启用 `X-DashScope-Async: enable`，无同步选项
• `task_id` 有效期 24 小时，结果 URL 有效期仅 2 小时 | +| **计费方式** | • 按成功生成图片张数计费（失败不计费）
• 免费额度：多数模型提供 500 张/90 天（主账号与 RAM 子账号共享）
• QPS/RPS 限流：主账号与子账号共用（常见 2 QPS） | • 按成功生成视频条数计费（失败不计费）
• 免费额度较少或未开放（以控制台开通页为准）
• 各模型独立限流：如 `liveportrait` 同时处理中任务数上限为 1 | • 按成功生成 3D 模型个数计费（失败不计费）
• 当前暂无公开免费额度
• RPS 查询限流：轮询接口上限 20 RPS，建议搭配异步回调避免高频轮询 | +| **地域与域名约束** | • 严格绑定：API Key、Endpoint、Workspace ID 必须同地域（北京/新加坡/弗吉尼亚/法兰克福）
• 推荐使用业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`） | • 同上，强地域一致性要求
• 跨地域调用必然鉴权失败或 404 | • **仅支持华北2（北京）地域**
• API Key、Endpoint、Workspace ID 必须为北京地域
• 其他地域 URL 不可用 | +| **典型场景** | • 电商商品图生成与背景替换
• 社媒内容快速配图与风格化
• AI 试衣、虚拟模特、人像写真
• 设计稿扩图与局部修改 | • 短视频创意脚本可视化（T2V）
• 产品展示动画（I2V + 超清）
• 数字人播报与口型驱动（S2V/LipSync）
• 舞蹈复刻、动作迁移、角色融合（R2V） | • 工业设计原型快速建模（文→3D）
• 电商商品 3D 展示（单图→GLB）
• AR/VR 内容资产生成（多视角图→PBR 模型）
• 游戏资产辅助创作 | + +--- + +## 各方案适用场景建议 + +### ✅ 图像生成 —— 适合「高吞吐、低延迟、强可控」的视觉内容生产 +- **推荐场景**： + - 需批量生成高质量静态图（如千张商品图、海报素材）且对首屏响应时间敏感 → 选用支持**同步调用**的 `wan2.6-t2i` 或 `z-image-turbo`。 + - 需精准编辑已有图像（擦除补全、风格重绘、AI试衣）→ 优先选择 `qwen-image-edit-max` 或垂直专用模型（`shoemodel-v1`, `facechain`）。 + - 快速验证或轻量应用 → 利用免费额度体验 `wanx-virtualmodel`、`image-erase-completion` 等工具。 +- **避坑提示**：避免混用跨地域 API Key；编辑类模型务必确保输入图公网可访问且 URL 无中文字符。 + +### ✅ 视频生成 —— 适合「叙事性、时序性、角色驱动」的动态内容构建 +- **推荐场景**： + - 从文案自动生成分镜短视频 → 使用 `wan2.7-t2v-*`（天然支持多镜头描述）或 `kling/kling-v3-omni-video`（支持音频+多风格）。 + - 将静态产品图转化为旋转展示视频 → 选用 `happyhorse-1.1-i2v` 或 `pixverse/*it2v`。 + - 构建数字人播报系统 → 组合 `wan2.2-s2v`（语音驱动） + `videoretalk`（口型替换），注意需先调用 `detect` 模型校验输入合规性。 +- **避坑提示**：严格核对模型文档中的 API 路径（`video-generation` vs `image2video`）；人物类模型务必执行前置检测，否则任务直接失败。 + +### ✅ 3D 生成 —— 适合「空间建模、资产交付、跨平台复用」的专业级三维内容生产 +- **推荐场景**： + - 快速将设计草图/实物照片转为可嵌入 Web/Unity/Unreal 的 GLB 模型 → 使用 `Tripo/Tripo-P1.0`（平衡速度与质量）。 + - 高精度工业零件或游戏角色建模 → 选用 `Tripo/Tripo-H3.1` 并设置 `geometry_quality: ultra`。 + - 需多角度一致建模 → 采集前/左/后/右四视角图（缺省视角用 `{}` 占位），调用多图生3D。 +- **避坑提示**：仅限北京地域；结果 URL 2 小时失效，务必在轮询成功后立即下载；输入 `prompt`/`image`/`images` 三者必须严格互斥。 + +--- + +## 开发者技术选型参考 + +| 选型维度 | 关键判断依据 | 推荐行动 | +|----------|--------------|----------| +| **调用实时性要求** | • 需 <1s 响应 → 选支持同步的图像模型（`wan2.6-t2i`）
• 可接受 5–60s 延迟 → 视频/3D 均适用 | 优先查阅各模型文档中 “调用方式” 章节，确认是否标注 “支持同步调用” | +| **输入数据形态** | • 纯文本 → 图像/视频/3D 均支持
• 单图 → 图像编辑、图生视频、单图生3D
• 多图/视频/音频 → 视频生成（R2V/I2V）为主，3D 仅支持四视角图 | 根据原始数据源选择最小改造路径：例如已有产品图 → 直接走 I2V 或 单图生3D，而非强行转文本再 T2V/T23D | +| **输出交付目标** | • 嵌入网页/APP 展示 → 图像（PNG/JPEG）、视频（MP4）、3D（GLB）均可
• 需后续渲染/编辑 → 3D（GLB 含材质）优势显著；视频需注意水印与分辨率适配 | 若需 Unity/Blender 进一步加工，3D 生成是唯一原生支持 PBR 材质导出的方案 | +| **成本与规模预期** | • 小规模试用 → 图像生成免费额度充足
• 中大规模商用 → 对比各模型单位成本（元/张/条/个），关注高并发限流阈值 | 在百炼控制台开通模型前，查看“资费与限流”章节，评估 QPS/RPS 是否满足峰值需求（如直播场景需高并发视频生成） | +| **工程集成复杂度** | • 同步调用：代码简洁，错误处理简单
• 异步调用：需实现任务创建 + 轮询/回调 + 结果持久化逻辑 | 新项目建议统一采用**异步回调**（Webhook）替代轮询，降低服务端资源消耗；百炼平台已提供标准异步回调配置入口 | + +> **最后建议**：对于复合型应用（如“文生图 → 图生视频 → 视频抽帧 → 单图生3D”），建议分 + +## 被对比主题页 + +- [image generation](../api/image-generation.md) +- [video generation api](../api/video-generation-api.md) +- [3d generation](../api/3d-generation.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md deleted file mode 100644 index 3ab6c78e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/generation-apis-comparison.md +++ /dev/null @@ -1,65 +0,0 @@ -# [多模态](../concepts/multi-modal.md)生成能力对比：图像、视频与3D生成 - -为帮助开发者快速理解百炼平台在[多模态](../concepts/multi-modal.md)生成领域的技术布局与能力边界，本文系统对比图像生成（Image Generation）、视频生成（Video Generation）与3D生成（3D Generation）三大核心能力。对比聚焦实际工程落地的关键维度——包括调用模式、模型生态、输入输出规范、计费逻辑与适用场景，旨在为技术选型提供客观、可操作的决策依据。所有信息均基于当前（2024年Q3）百炼平台正式发布的API文档与控制台配置。 - -## 关键能力维度对比 - -| 维度 | 图像生成（Image） | 视频生成（Video） | 3D生成（3D） | -|------|-------------------|-------------------|--------------| -| **核心输入格式** | 文本（[prompt](../guides/prompt.md)）、单图/多图（URL）、掩码图（mask_image_url）、草图（sketch）、风格参考图等；支持图文混排指令 | 文本（[prompt](../guides/prompt.md)）、首帧/首尾帧图像（image_url）、参考视频（video_url）、音频（audio_url）、[多模态](../concepts/multi-modal.md)组合（如图+音+[prompt](../guides/prompt.md)） | 文本（prompt）、单张图像（image）、四视角图像数组（images: [front, left, back, right]）；三者互斥 | -| **核心输出格式** | JPEG/PNG 图像（URL 或 base64），支持 512×512 至 4K 分辨率；含预览图、水印开关、扩展结果（如增强 prompt） | MP4 视频（URL），时长默认 5 秒（可设 2–10 秒），分辨率支持 480P–1080P；含封面帧、元数据（duration/frame_rate） | GLB 格式 PBR 材质模型（pbr_model_url）、无贴图基础网格（base_model_url）、渲染预览图（rendered_image_url）；支持面数分级（2万–200万面） | -| **主流支持模型** | `qwen-image-3.0-pro`, `wan2.7-image-pro`, `kling/kling-v3-image-generation`, `vidu/vidu-image_reference2image`, `z-image-turbo` | `wan2.7-t2v-2026-06-12`, `happyhorse-1.1-t2v`, `vidu/viduq3-turbo_text2video`, `emo-v1`, `liveportrait`, `pixverse/pixverse-c1-t2v` | `Tripo/Tripo-P1.0`（快模版，≤2万面），`Tripo/Tripo-H3.1`（高精版，≤200万面） | -| **API 端点（推荐）** | `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation`（同步/异步共用路径，行为由模型决定） | `POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis`（全模型统一端点，强制异步） | `POST https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api/v1/services/aigc/video-generation/3d-generation`（仅华北2可用，强制异步） | -| **调用模式** | **混合模式**：`wan2.6+` / `qwen-image-3.0-pro` / `z-image-turbo` 支持同步（直接返回结果）；`wanx-v1` / `wanx-x-painting` / `image-out-painting` 等仅支持异步（需轮询 task_id） | **强制异步**：全部模型必须使用 `X-DashScope-Async: enable`，创建任务后轮询 `GET /api/v1/tasks/{task_id}` 获取结果 | **强制异步**：必须启用 `X-DashScope-Async: enable`；轮询间隔建议 ≥15 秒；task_id 有效期 24 小时 | -| **计费方式** | 按生成张数计费（例：`wanx-v1` 0.16元/张，`image-out-painting` 0.18元/张）；主账号与子账号共享 500 张免费额度（90天有效期） | 按模型独立计费：文生视频按秒（如 `wan2.7-t2v` 0.35元/秒）、人像动画按时长（`emo-v1` 0.28元/秒）、口型替换按音频秒数；各模型有独立免费额度（如 `emo-detect-v1` 200次） | 按任务计费：`Tripo-P1.0` 0.8元/次，`Tripo-H3.1` 2.5元/次；暂无公开免费额度，需开通后查看控制台配额 | -| **典型响应耗时** | 同步调用：3–8 秒（T2I/I2I）；异步调用：10–60 秒（含排队） | 1–5 分钟（受分辨率、时长、模型复杂度影响显著；高精度或多镜头任务可达 8 分钟） | 2–10 分钟（`P1.0` 通常 ≤3 分钟；`H3.1` + `ultra` 模式常需 6–10 分钟） | -| **地域支持** | 华北2（北京）、新加坡、美国（弗吉尼亚）三地全域支持；密钥与 endpoint 必须严格匹配 | 华北2（北京）、新加坡、美国（弗吉尼亚）、德国（法兰克福）；跨地域调用将返回 `401 Unauthorized` | **仅华北2（北京）**；其他地域 endpoint 不可用，调用必失败 | -| **关键限制** | • 输入图需公网可访问 HTTPS URL
• `wan2.5` 及以下版本不支持同步调用
• 局部重绘/背景生成等高级功能限北京地域专属域名 | • 所有模型强制异步，无同步选项
• 音频输入需清晰人声、≤30 秒
• Prompt ≤512 字符，含敏感词触发拦截
• `liveportrait` 等模型 QPS 限 1 | • 仅支持 JPEG/PNG；单图 ≤20MB；多图需严格四视角顺序
• `pbr=true` 时 `texture=false` 无效；唯一无贴图路径：`"texture": false, "pbr": false`
• 输出 URL 有效期仅 2 小时 | - -## 各方案适用场景建议 - -### ✅ 图像生成（Image）——适合「高并发、低延迟、强交互」场景 -- **推荐场景**：电商海报批量生成、AIGC设计助手（实时预览）、社交内容配图、UI组件自动化出图、AI修图SaaS集成。 -- **选型提示**：若需毫秒级响应（如用户拖拽即实时重绘），优先选用 `z-image-turbo` 或 `qwen-image-3.0-pro`（同步调用）；若需4K精细输出或复杂编辑（如虚拟模特试穿），选用 `wan2.7-image-pro` 并注意其仅支持异步流程。 - -### ✅ 视频生成（Video）——适合「叙事表达、数字人驱动、轻量内容生产」场景 -- **推荐场景**：短视频营销素材生成、AI主播播报、产品演示动画、教育口型同步课件、游戏NPC动作迁移。 -- **选型提示**：纯文本生成短片（≤5秒）选 `viduq3-turbo_text2video`；需精准动作控制选 `animate-anyone-gen2`；强调口型自然度选 `pixverse-lipsync`；对并发要求高（如批量生成）需提前申请 QPS 提升，并配置异步回调避免轮询压力。 - -### ✅ 3D生成（3D）——适合「工业可视化、电商3D展示、AR/VR内容基建」场景 -- **推荐场景**：电商商品3D建模（文生/图生）、工业零件快速原型、建筑概念可视化、元宇宙空间资产生成、教育三维教具制作。 -- **选型提示**：快速验证创意或轻量应用 → `Tripo-P1.0`；需导入CAD/渲染管线或对接Unity/Unreal → `Tripo-H3.1` + `geometry_quality: "ultra"`；务必使用北京地域密钥与专属域名，且提前下载 `pbr_model_url`（2小时过期）。 - -## 技术选型参考指南（面向开发者） - -1. **优先确认调用模式约束** - - 若业务无法容忍异步延迟（如实时聊天机器人附带图片生成），**排除视频与3D方案**，仅考虑图像生成中的同步模型（`wan2.6-t2i` 及以上、`qwen-image-3.0-pro`）。 - - 若已构建成熟异步任务队列（如 Celery/RabbitMQ），视频与3D的强制异步特性反而是优势，可统一调度。 - -2. **严格校验地域一致性** - - 图像生成支持多地，但**视频与3D对地域敏感度极高**：3D仅限北京；视频若在新加坡部署服务，却误用北京密钥，将直接鉴权失败。建议在初始化 SDK 时硬编码 `region` 参数，并做启动校验。 - -3. **输入准备成本是隐性瓶颈** - - 图像：只需文本或单图，接入成本最低； - - 视频：需准备高质量音频/多帧图像，且 URL 必须公网可直连（OSS需设 public-read）； - - 3D：多图生3D要求严格视角顺序与光照一致性，实测中“前左后右”四图质量不均将导致模型崩坏。建议优先尝试文生3D降低门槛。 - -4. **计费颗粒度决定架构设计** - - 图像按张计费 → 适合按需调用，可缓存结果复用； - - 视频按秒计费 → 需精确控制 `duration` 参数，避免默认5秒造成浪费； - - 3D按次计费 → 建议对同一prompt/image做结果缓存（MD5哈希索引），避免重复生成。 - -5. **错误处理策略差异化** - - 图像：关注 `400 Bad Request`（参数错）、`403 Forbidden`（额度超）； - - 视频：高频出现 `429 Too Many Requests`，需实现指数退避轮询； - - 3D：`task_status: "UNKNOWN"` 表示 task_id 过期，必须重新提交任务——不可重试旧ID。 - -> **最后提醒**：所有多模态能力均依赖 DashScope SDK 最新版（≥4.20.0）及百炼控制台「业务空间」配置。请勿混用旧版文档（如 `wanx-v1` 协议）与新版模型（如 `wan2.7-*`），模型名与 endpoint 的严格匹配是调用成功的前提。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) -- [3d generation](../api/3d-generation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md deleted file mode 100644 index f89e5b44..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/image-vs-video-generation.md +++ /dev/null @@ -1,65 +0,0 @@ -# 图像生成与视频生成对比 - -为帮助开发者快速理解百炼平台中图像生成与视频生成两类能力的差异，明确技术选型边界与使用约束，本文从输入输出、模型支持、调用方式、计费策略等核心维度进行系统性对比。该对比基于当前（2024年Q3）平台正式发布的 API 能力与文档规范，适用于新项目接入、存量系统迁移及多模态方案架构设计。 - -| 维度 | 图像生成（Image Generation） | 视频生成（Video Generation） | -|------|------------------------------|------------------------------| -| **输入格式** | • 文生图：`input.prompt` 或 `input.messages`（含 text）
• 图生图/编辑：`input.messages` 数组（含 `{"text": "..."}` 和 `{"image": "url"}`），部分旧模型仍用 `input.ref_image`
• 图片 URL 需公网可访问、无中文路径、支持 HTTPS | • 文生视频（T2V）：`input.prompt` + 可选 `parameters.multi_shot`/分镜描述
• 图生视频（I2V）：`input.media`（首帧图 URL）或 `input.video_url`（短片）
• 参考生视频（R2V）：`input.media` 数组（多张参考图）
• 所有媒体 URL 必须 HTTPS、≥512×512（图）、≤10秒（视频） | -| **输出格式** | • 同步调用：直接返回 JSON，含 `output.results` 数组（每项含 `url`、`width`、`height`）
• 异步调用：轮询 `GET /api/v1/tasks/{task_id}`，响应含 `output.results`（单图或多图） | • 全部异步：轮询 `GET /api/v1/tasks/{task_id}`，响应含 `output.video_url`（H.264 MP4）、`output.duration`（秒）、`output.resolution`（如 `"1080P"`）
• 不返回帧序列或中间产物，仅最终视频文件 | -| **支持模型（主力）** | • 文生图：`qwen-image-2.0-pro`、`wan2.7-image-pro`、`z-image-turbo`、`vidu`
• 图像编辑：`qwen-image-edit-*`、`wanx-x-painting`、`virtualmodel-v2`
• 创意工具：`image-out-painting`、`wanx-background-generation-v2`、`aitryon-plus` | • 文生视频：`wan2.7-t2v-*`、`vidu/viduq3-*-text2video`、`kling/kling-v3-*-video-generation`、`pixverse/pixverse-*-t2v`
• 图/参考生视频：`wan2.7-i2v-*`、`vidu/viduq3-*-img2video`、`wan2.7-r2v-*`
• 数字人：`liveportrait`、`videoretalk`、`emo-v1` | -| **API 端点** | • **统一主入口**：
`POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/multimodal-generation/generation`（推荐）
• 历史路径（部分模型）：
`POST /api/v1/services/aigc/text2image/image-synthesis` 等 | • **全量异步专用入口**：
`POST https://{WorkspaceId}.{region}.maas.aliyuncs.com/api/v1/services/aigc/video-generation/video-synthesis`
• 通用兼容地址（不推荐）：
`https://dashscope.aliyuncs.com/...`（北京）或 `https://dashscope-intl.aliyuncs.com/...`（国际） | -| **调用模式** | • **同步 & 异步混合**：
– 快速模型（如 `qwen-image-2.0-pro`、`wan2.6-t2i`）支持同步（<10s），可流式返回（需 `X-DashScope-Sse: enable`）
– 长耗时模型（如 `wanx-x-painting`、`image-out-painting`）强制异步 | • **强制异步**：
所有模型均需两步流程：① 创建任务获取 `task_id`；② 轮询 `GET /api/v1/tasks/{task_id}` 获取结果
• `task_id` 有效期严格为 **24 小时** | -| **计费方式** | • 免费额度：**500 张/账号/90天**（主账号与 RAM 子账号共享）
• 计费模型：
– 按成功生成图片计费（如 `wanx-v1`: 0.16元/张）
– 部分模型免费额度用尽即停用（无单价），如 `wanx-x-painting`、`shoemodel-v1`、`image-instance-segmentation` | • 免费额度：**100 秒视频生成时长/账号/90天**（按实际生成视频秒数累加）
• 计费模型：
– 按生成视频时长计费（如 `wan2.7-t2v`: 0.8元/秒，`vidu`: 1.2元/秒）
– 数字人模型按任务计费（如 `liveportrait`: 0.5元/次）
• 所有模型均无“免费额度外不可用”例外，超限后自动转计费 | -| **典型场景** | • 静态内容生产：电商主图、营销海报、AI头像、文字艺术（WordArt）、背景生成、商品试穿（鞋靴/服装）
• 图像增强：局部重绘、擦除补全、实例分割、风格迁移
• 快速原型：A/B测试图稿、UI素材生成、设计草图扩展 | • 动态内容生产：短视频广告、产品演示动画、分镜脚本可视化、数字人播报、虚拟主播口型同步
• 视频创作辅助：图转视频（I2V）、多图角色一致性视频（R2V）、自然语言分镜生成（万相2.7）
• 垂直应用：表情包生成（`emoji`）、唱演视频（`emo-v1`）、舞蹈驱动（`animate-anyone-gen2`） | - -## 各方案适用场景建议 - -### ✅ 推荐选择图像生成当： -- 业务需求聚焦于**静态视觉资产**，如电商平台的商品图、社交媒体封面、APP图标、个性化头像； -- 对**响应延迟敏感**（如实时交互式设计工具），且任务平均耗时 < 8 秒，可优先选用 `qwen-image-2.0-pro` 或 `wan2.7-image-pro` 同步接口； -- 需要**精细控制像素级输出**（如 4K 渲染、文字精准识别、局部编辑掩码），图像模型在空间保真度上显著优于视频模型首帧； -- 成本结构以**固定次数/张数**为主，且月用量稳定在数百张内，可充分复用免费额度。 - -### ✅ 推荐选择视频生成当： -- 核心目标是**动态表达与时间叙事**，如短视频营销、教学动画、数字人直播、AI分镜预演； -- 接受**异步工作流**（任务创建 → 轮询 → 下载），并能妥善管理 `task_id` 生命周期与失败重试逻辑； -- 需要**跨帧一致性能力**（人物/物体/风格在多帧中稳定呈现），R2V 与 I2V 模型专为此优化，图像模型无法替代； -- 业务具备**视频时长可预测性**（如统一生成 5 秒广告），便于成本建模；若需高频、短时（<3秒）视频，需注意部分模型最低时长限制（如 `kling` 最小 3 秒）。 - -### ⚠️ 需谨慎评估或组合使用的场景： -- **“动效化静态图”需求**（如将海报转为带缩放/平移的短视频）： - → 不建议直接调用视频生成，应先用图像生成产出高质量源图，再通过视频编辑模型（如 `video-style-transform` 或 `wan2.2-animate-mix`）添加运镜效果，兼顾质量与成本。 - -- **高并发实时图像/视频混合服务**（如用户上传图→生成图→生成对应视频）： - → 必须分离调用链路：图像生成走同步路径（低延迟），视频生成走异步路径（解耦阻塞）；同时注意地域强绑定——图像与视频模型若部署在不同地域（如图在北京、视频在新加坡），需分别配置 API Key 与 Endpoint。 - -- **需要帧级控制或导出中间帧**（如用于后期合成、AR叠加）： - → 当前两类 API 均**不提供帧序列下载**。若必须获取逐帧，需自行对生成视频做抽帧处理（注意版权与水印合规性），或联系平台申请定制化能力支持。 - -## 面向开发者的选型参考指南 - -1. **起步验证阶段**： - - 优先使用 `qwen-image-2.0-pro`（同步、北京/新加坡可用、免费额度覆盖）验证图像流程； - - 视频侧选用 `wan2.7-t2v-2026-06-12`（支持自然语言分镜、文档完善）+ `task_id` 轮询 SDK 封装，避免手动轮询。 - -2. **生产环境部署要点**： - - **域名与地域必须显式绑定**：禁用 `dashscope.aliyuncs.com` 通用域名，全部切换至业务空间专属域名（如 `https://{WorkspaceId}.cn-beijing.maas.aliyuncs.com`），提升稳定性与性能； - - **错误处理标准化**：图像 API 需捕获 `429`（限流）、`400`（参数错误）；视频 API 必须处理 `401`（地域不匹配）、`404`（旧版路径错误，如误用 `/image2video/`）； - - **水印策略统一**：生产环境所有请求显式设置 `"watermark": false`，避免默认水印影响交付。 - -3. **模型升级路径建议**： - - 图像侧：逐步淘汰 `wan2.5-i2i-preview` 等旧版，迁移到 `wan2.7-image-pro`（4K）或 `qwen-image-2.0-pro`（文字渲染）； - - 视频侧：**立即停用万相2.1–2.6系列**（文档标记为“旧版协议”），全面切换至 `wan2.7-*` 或 `vidu/kling` 新主力模型，享受分镜解析、多镜头、音频生成等增强能力。 - -4. **成本监控关键指标**： - - 图像：监控 `total_images_generated` 与 `free_quota_remaining`（Dashboard 可查）； - - 视频：监控 `total_video_seconds_generated` 及各模型 `avg_duration_per_task`，警惕因 `duration` 参数设置过高导致意外超支。 - -> **最后提醒**：两类能力虽同属 AIGC，但底层计算范式、资源调度与 SLA 保障机制完全不同。切勿将图像 API 的同步思维套用于视频，亦不可期望视频模型输出单帧图像——尊重各自技术边界，方能构建稳健、可扩展的多模态应用。 - -## 被对比主题页 - -- [image generation](../api/image-generation.md) -- [video generation api](../api/video-generation-api.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md deleted file mode 100644 index f0be379a..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-base-vs-memory-library.md +++ /dev/null @@ -1,63 +0,0 @@ -# 知识库与记忆库功能对比 - -为帮助开发者清晰理解百炼平台中两类核心记忆增强能力的定位差异，本文从技术架构、使用范式与业务价值三个维度，系统对比**知识库（Knowledge Base）** 与**记忆库（Memory Library）**。二者虽均以“增强大模型上下文”为目标，但设计初衷、数据来源、生命周期及适用场景存在本质区别：知识库面向**静态、共享、领域化知识资产**，强调精准检索与结构化注入；记忆库面向**动态、私有、会话级用户记忆**，强调自动提炼与跨轮次上下文延续。正确区分二者是构建高可用智能体应用的关键前提。 - -## 关键维度对比 - -| 维度 | 知识库（Knowledge Base） | 记忆库（Memory Library） | -|------|--------------------------|---------------------------| -| **核心定位** | 静态领域知识注入系统（RAG 基础设施） | 动态用户[长期记忆](../concepts/long-term-memory.md)管理系统（LTM 中枢） | -| **数据来源** | 手动上传的结构化/非结构化文档（PDF/Word/Excel/图片/音视频等） | 自动从对话 `messages` 中提取，或通过 API 写入 `custom_content` | -| **数据所有权** | 应用/业务空间级别共享（多应用可挂载同一知识库） | `user_id` 级别隔离（同一记忆库内不同用户数据完全独立） | -| **输入格式** | 文件（支持 20+ 格式）、文本块、URL；需预处理切片与元信息抽取 | JSON 格式 `messages` 数组（含 role/content/timestamp）或纯文本 `custom_content` + `user_id` | -| **输出格式** | 检索返回结构化 `nodes[]`：含 `content`、`source`、`score`、`meta` 等字段，供下游模型直接拼接提示词 | 检索返回 `memory_nodes[]`：含 `id`、`content`、`type`（event/profile）、`score`（[0,1]）、`created_at`；用户画像单独通过 `GetUserProfile` 获取 | -| **支持模型** | **不运行模型**，但深度依赖：
• 向量模型（`text-embedding-v4`/`qwen3-vl-embedding`）
• 排序模型（`qwen3-rerank`/`qwen3-vl-rerank`）
• 路由模型（`qwen-plus`，多库时启用） | **不运行模型**，但依赖：
• 记忆提取模型（内部调用，不可选）
• 用户画像抽取模型（基于 `profile_schema` 自动触发，不可替换） | -| **API 端点** | `POST /api/v1/knowledge_bases/{kb_id}/retrieve`（仅华北2北京地域） | `POST /api/v2/apps/memory/add`
`POST /api/v2/apps/memory/memory_nodes/search`
（全地域可用，无地域限制） | -| **计费方式** | 分层计费：
• 知识库规格费（按存储容量/月）
• Rerank 调用费（按初步召回总切片数 × 次数）
• 向量化/路由/问答模型费（按 [Token](../concepts/token.md) 单独计费） | 按调用量计费：
• `AddMemory`：按写入条数计费
• `SearchMemory`：按检索次数计费
• 无存储容量费（默认无限存储，按实际调用计费） | -| **生命周期管理** | 文档切片永久存储（除非手动删除）；Meta 抽取配置创建后不可修改 | 记忆片段默认永不过期；可通过 `memory_expiration_time` 规则配置有效期（仅对新写入生效） | -| **典型场景** | • 客服知识库问答（产品手册/FAQ）
• 法律合同条款检索
• 医疗文献辅助诊断
• [多模态](../concepts/multi-modal.md)内容搜索（图片中找文字、视频里查事件） | • 智能助手记住用户偏好（“我喜欢简体中文”）
• 跨会话任务延续（“继续上次未完成的报销流程”）
• 用户画像构建（职业/兴趣/健康目标）
• OpenClaw Agent 自动记忆与召回 | -| **地域限制** | **强制限定华北2（北京）地域**，其他地域不可用 | **全地域可用**（杭州、上海、新加坡、法兰克福等均支持） | -| **权限模型** | 依赖子账号 `AliyunBailianDataFullAccess` 权限；操作受业务空间隔离 | 依赖 `DASHSCOPE_API_KEY`（百炼平台生成），**不支持 Coding Plan Key**；`user_id` 为逻辑隔离边界 | - -## 适用场景建议 - -### ✅ 选择知识库，当您需要： -- 将**企业级静态知识资产**（如产品文档、规章制度、培训材料）规模化注入大模型； -- 支持**多用户、多应用共享同一知识源**，且要求高精度、低噪声的语义检索； -- 处理**非文本模态数据**（图片、音视频），需跨模态语义理解能力； -- 对检索结果的**可解释性与溯源性**有强要求（需明确 `source` 页码/时间戳）； -- 已有成熟文档管理体系，希望最小化改造接入 RAG。 - -### ✅ 选择记忆库，当您需要： -- 解决**单用户跨会话上下文丢失**问题，让智能体具备“记住用户”的能力； -- 构建**个性化体验**（如推荐、提醒、定制化回复），依赖持续积累的用户行为与偏好； -- 快速集成至**OpenClaw Agent 或自研对话系统**，追求零代码自动捕获与召回； -- 管理**高度动态、短生命周期的会话记忆**（如购物意图、待办事项、临时约定）； -- 需要**灵活的用户画像结构化能力**，并支持多轮交互逐步完善字段。 - -### ⚠️ 避免混淆的典型误区： -- **不要用知识库存储用户个人数据**：知识库无 `user_id` 隔离机制，所有用户共享同一检索空间，存在隐私与安全风险； -- **不要用记忆库替代领域知识库**：记忆库不支持文档解析、切片、[多模态](../concepts/multi-modal.md)索引，无法处理 PDF/Excel 等专业格式； -- **不要跨地域混用知识库 API**：在新加坡地域调用知识库接口将直接失败，而记忆库 API 无此限制； -- **不要期望记忆库提供文档级溯源**：记忆片段不保留原始文件位置，仅提供语义摘要与置信度分数。 - -## 技术选型参考（面向开发者） - -| 选型决策点 | 推荐方案 | 说明 | -|------------|----------|------| -| **是否需支持图片/音视频搜索？** | → 知识库 | 记忆库仅支持文本记忆，不提供[多模态](../concepts/multi-modal.md)嵌入与检索能力 | -| **是否需严格按 `user_id` 隔离数据？** | → 记忆库 | 知识库无用户维度，所有查询结果对所有用户可见 | -| **是否已有大量 PDF/Word 等文档需快速上线？** | → 知识库（控制台上传） | 提供一键解析、自动切片、可视化调试能力；记忆库需先人工提炼为 `custom_content` | -| **是否需在 OpenClaw Agent 中零配置启用记忆？** | → 记忆库（OpenClaw 插件） | 插件自动注册工具链，无需修改 Agent 代码；知识库需手动集成 `Retrieve` 节点 | -| **是否需对检索结果设置相似度阈值并过滤低分项？** | → 两者均支持，但参数单位不同 | 知识库 `相似度阈值`（0.01–1.0）；记忆库 `min_score`（0–100 百分制，API 返回 [0,1] 需转换） | -| **是否需审计每条检索的完整输入/输出？** | → 知识库（SLS 日志） | 知识库提供 `request_body`/`response_body.data.nodes[]` 全字段日志；记忆库日志需自行埋点 | -| **是否部署在非华北2地域（如新加坡）？** | → 记忆库（唯一选择） | 知识库在该地域不可用，强行调用将返回 `RegionNotSupported` 错误 | - -> **最佳实践组合建议**：在复杂智能体应用中，**知识库 + 记忆库 可协同使用**。例如：客服机器人中，用知识库回答“产品功能如何使用”，用记忆库记住“张三用户上周咨询过退货流程，本次优先展示退货进度”。二者通过不同 API 分别调用，结果在提示词工程阶段融合，实现“领域知识 + 用户上下文”的双重增强。 - -## 被对比主题页 - -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-solutions-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-solutions-comparison.md new file mode 100644 index 00000000..e3ad9077 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/knowledge-solutions-comparison.md @@ -0,0 +1,73 @@ +# 知识能力方案对比：Knowledge API vs Knowledge Base vs Memory Library + +为帮助开发者在百炼平台中高效选型，本文系统对比三种核心知识增强能力：**Knowledge API**（面向应用层的知识服务封装）、**Knowledge Base**（RAG 基础设施级知识库）、**Memory Library**（[长期记忆](../concepts/long-term-memory.md)与用户状态管理组件）。三者定位不同、能力互补，但常被混淆使用。本对比聚焦技术本质、集成方式与适用边界，旨在消除概念歧义，支撑精准架构设计与成本优化。 + +--- + +## 关键维度对比表 + +| 维度 | Knowledge API | Knowledge Base | Memory Library | +|------|----------------|----------------|----------------| +| **本质定位** | 应用层 RESTful 封装服务，提供开箱即用的「检索」与「问答」能力，屏蔽底层索引细节 | RAG 基础设施，提供完整的知识生命周期管理（上传→解析→切片→向量化→索引→检索→重排→生成） | [长期记忆](../concepts/long-term-memory.md)中枢，专注跨会话用户状态持久化与语义化召回（事件记忆 + 结构化画像） | +| **输入格式** | `query`（纯文本，≤2048 字符）+ `knowledge_ids`（字符串数组）；不支持原始文件上传 | 支持多模态原始输入：PDF/Word/Excel/PPT/Markdown/图片/音视频等；需先完成知识库构建流程 | `messages`（对话历史数组）或 `custom_content`（自定义文本）+ `user_id`；支持结构化 `meta_data` 和 `profile_schema` | +| **输出格式** | • 检索：标准 JSON，返回 `chunks` 数组（含 `content`, `score`, `source`）
• 问答：SSE 流式响应（含 `plan`, `tool_calls`, `response`）或完整 JSON | • 知识问答：流式/非流式 JSON，含 `answer`, `retrieved_chunks`, `trace` 等字段
• 知识检索：JSON 返回 `chunks` 及元数据
• API 接口：支持 `CreateIndex`/`Retrieve` 等细粒度 OpenAPI 响应 | • `SearchMemory`：JSON 返回 `memories` 数组（含 `id`, `content`, `score`, `meta_data`）
• `GetUserProfile`：结构化 JSON（按 schema 定义的字段）
• 所有接口均同步返回 | +| **支持模型** | 仅调用预置问答模型（如 `qwen3.7-plus`），不可更换；检索阶段不依赖 LLM | 支持广泛模型：千问全系列（QwQ/Long/Max/Plus/Turbo/Coder/Deep-Research/VL 系列/OCR）、Qwen3/Qwen2.5/Qwen2 开源版、DeepSeek-R1、Llama3.1、Yi-Large 等；可自由配置路由与重排模型 | 不直接调用大模型生成答案；记忆提取与召回由专用轻量模型完成；画像抽取依赖预置 NLU 模型，不可替换 | +| **API 端点** | • 检索：`POST /api/v1/indices/knowledge/search`
• 问答：`POST /api/v2/apps/knowledge/chat`
• **专属域名**：`https://{workspaceId}.cn-beijing.maas.aliyuncs.com` | • 控制台可视化服务（无独立端点）
• OpenAPI：`POST /v1/indexes/{index_id}/retrieve`、`POST /v1/indexes/{index_id}/chat` 等
• **通用 DashScope 域名**：`https://dashscope.aliyuncs.com/api/v1`（需正确鉴权） | • `POST /v1/memories/add`
• `POST /v1/memories/search`
• `GET /v1/profiles/{user_id}` 等
• **通用 DashScope 域名**：`https://dashscope.aliyuncs.com/api/v1`（需 DashScope API Key） | +| **计费方式** | • **按调用次数计费**：检索与问答分别计费
• **无知识库规格费**：不产生存储/索引运行时费用
• 模型 Token 费用按实际消耗计算（含重排、路由、问答模型） | • **双轨计费**：
 ✓ 知识库规格费：标准版 0.03 元/小时，旗舰版按 RCU 并发计费
 ✓ 模型调用费：向量模型、Rerank 模型、路由模型、问答模型的 Token 消耗独立计费（注意：Rerank 费用基于初步召回总数，非最终返回数） | • **按调用次数计费**：`AddMemory` / `SearchMemory` / `GetUserProfile` 等接口独立计费
• **无存储费**：记忆条目按数量计入配额，不额外收取存储费用
• 无模型 Token 费用（内部轻量模型不对外计费） | +| **典型场景** | • 快速上线客服机器人、FAQ 助手等标准化问答应用
• 无需管理知识库生命周期，仅需调用即可获得结果
• 多知识库联合检索（如“从产品文档+合同模板中找条款”） | • 构建专业领域智能体（法律/医疗/金融），需精细控制切片策略、元数据过滤、重排阈值
• 需支持多模态知识（如 PDF 表格+截图+会议录音转文字）
• 要求高可控性：自定义索引构建、A/B 测试不同模型链路、深度日志分析 | • 智能体个性化体验：记住用户偏好（“我爱喝冰美式”）、习惯（“每周三下午开会”）、身份信息（“我是XX公司采购负责人”）
• 跨会话上下文延续（如“上次说要查的合同编号是…”）
• 用户画像构建与渐进式填充（职业、兴趣、设备型号等） | +| **状态管理** | **完全无状态**：每次请求独立，不维护会话历史；多轮对话需应用层拼接 `query` | **知识库静态，问答无状态**：知识库索引一旦发布即固定；单次问答不保留上下文，但支持多轮智能模式（Agentic 规划搜索） | **强状态性**：以 `user_id` 为隔离单元，自动维护[长期记忆](../concepts/long-term-memory.md)空间；支持显式更新、删除、分页管理 | + +--- + +## 各方案适用场景建议 + +### ✅ 选择 Knowledge API 当： +- 业务目标是**快速交付一个功能明确的问答服务**（如内部知识助手、产品文档查询），且知识源已稳定、无需频繁变更； +- 团队无 RAG 工程能力，希望跳过索引构建、切片配置、重排调优等复杂环节； +- 需要**多知识库动态组合检索**（例如：一次查询同时覆盖“员工手册”和“IT 支持指南”）； +- 对延迟敏感，且接受默认参数下的效果（如 `top_k` 实际上限为 10，相似度阈值不可调）。 + +### ✅ 选择 Knowledge Base 当： +- 需要**深度定制 RAG 效果**：调整切片大小、相似度阈值、重排模型、元数据过滤逻辑； +- 知识源复杂多样（扫描件 PDF、带公式 Excel、会议录音、产品图片），需平台级多模态解析能力； +- 要求**生产级可观测性**：通过 SLS 日志追踪 `pipeline_id`、各阶段延迟、失败原因； +- 计划长期运营知识资产，需版本管理、灰度发布、A/B 测试不同知识库或模型配置； +- 已有大量私有文档，且愿意投入资源进行知识治理（清洗、标注、结构化）。 + +### ✅ 选择 Memory Library 当： +- 核心诉求是**突破上下文窗口限制，实现用户级长期记忆**； +- 构建的是**个性化智能体**（如虚拟助手、销售顾问、教育陪练），需记住用户历史行为、偏好、身份属性； +- 希望**零代码接入记忆能力**：通过 OpenClaw 插件自动捕获与召回，无需修改主业务逻辑； +- 需要**结构化用户画像**支撑下游业务（如推荐系统、权限分级、消息推送）； +- 场景对“时效性”要求高于“知识广度”——记忆强调“这个人说过什么”，而非“全网有什么”。 + +> ⚠️ 注意：三者非互斥关系。典型高阶架构常组合使用： +> **Knowledge Base 提供领域知识底座** → **Memory Library 注入用户个性化上下文** → **Knowledge API 或自定义 Agent 编排调用二者**，实现“既懂行业，又懂你”的智能服务。 + +--- + +## 技术选型参考（面向开发者） + +| 选型考量点 | 推荐方案 | 理由说明 | +|------------|----------|----------| +| **首次接入，MVP 验证** | `Knowledge API` | 最低门槛：无需创建知识库、无需理解切片/Rerank 概念，5 分钟完成 Hello World 调用；适合验证业务价值。 | +| **知识源持续更新，需自动化 pipeline** | `Knowledge Base` + SDK | 提供 `CreateIndex`/`UpdateIndex`/`DeleteIndex` 等完整 OpenAPI，支持 CI/CD 集成、定时同步、增量更新。 | +| **需要记忆用户对话中的隐含意图（如承诺、疑问、情绪）** | `Memory Library` | `AddMemory` 支持从 `messages` 自动提取事件记忆（如“用户承诺下周提交材料”），优于简单关键词匹配。 | +| **知识库需支持图片 OCR、表格识别、音视频转写** | `Knowledge Base` | 唯一支持多模态解析的方案；`Knowledge API` 仅支持已发布的文本类知识库。 | +| **严格控制成本，避免隐性支出** | `Knowledge API` 或 `Memory Library` | `Knowledge Base` 存在知识库规格费（小时级）+ Rerank 模型费（按召回量计费），易因配置不当导致成本飙升；另两者均为纯调用计费，更透明可控。 | +| **需跨地域部署（如新加坡节点）** | `Knowledge API` 或 `Memory Library` | `Knowledge Base` **仅限华北2（北京）地域**；另两者全球可用（DashScope 域名支持）。 | +| **要求对话状态自动管理（如多轮改写、历史摘要）** | `Knowledge Base`（多轮智能模式） 或 `Memory Library` + 自定义 Agent | `Knowledge Base` 内置 Agentic 规划搜索；`Memory Library` 提供 `autoRecall` 钩子，但复杂状态管理仍需上层编排。 | + +> 💡 **终极建议**： +> - 若你的核心问题是 **“如何让模型回答得更准”** → 优先评估 `Knowledge Base`； +> - 若你的核心问题是 **“如何让模型记得住用户”** → 优先评估 `Memory Library`； +> - 若你的核心问题是 **“如何最快上线一个能答问题的页面”** → 直接使用 `Knowledge API`。 +> 三者能力正交，合理组合才是百炼平台知识能力的最佳实践。 + +## 被对比主题页 + +- [knowledge](../api/knowledge.md) +- [knowledge base](../guides/knowledge-base.md) +- [memory library overview](../guides/memory-library-overview.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md deleted file mode 100644 index 5ab605c6..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/memory-solutions.md +++ /dev/null @@ -1,62 +0,0 @@ -# [长期记忆](../concepts/long-term-memory.md)与知识库方案对比 - -为帮助开发者在百炼平台中科学选型，本文系统对比**[长期记忆](../concepts/long-term-memory.md)（新）**（含记忆库能力）与**知识库（RAG）** 两大核心数据增强方案。二者虽均支持语义检索与结构化管理，但设计目标、数据生命周期、适用角色和集成范式存在本质差异：**[长期记忆](../concepts/long-term-memory.md)聚焦“用户专属、动态演进、会话级上下文沉淀”，知识库专注“领域通用、静态注入、应用级知识赋能”**。本对比基于当前（2025年Q2）百炼平台正式版能力，面向智能体（Agent）、工作流及自定义应用的开发者提供技术决策依据。 - -## 关键维度对比 - -| 维度 | 长期记忆（新） | 知识库（RAG） | -|------|----------------|----------------| -| **核心定位** | 用户级长期上下文管理：自动提取、结构化建模、跨会话个性化召回 | 应用级知识增强：私有文档/多模态数据的语义索引与检索，提升大模型领域回答质量 | -| **输入格式** | • `messages`：结构化对话数组（最多50条，role/content 必填）
• `custom_content`：纯文本（≤512字符）
• 支持 `meta_data`（≤1 KB）作为辅助标签 | • 多模态文件：PDF/DOCX/TXT/CSV/XLSX/图片（≤20MB）、音视频（需转文字）
• 支持元数据抽取规则（创建时配置，不可变）
• 文本切片上限6,000 [Token](../concepts/token.md) | -| **输出格式** | • `SearchMemory` 返回结构化记忆节点列表（含 `id`, `content`, `score`, `meta_data`, `created_at`）
• `GetUserProfile` 返回 JSON Schema 校验的结构化画像对象 | • `Retrieve` 返回文本切片列表（含 `content`, `score`, `source_file_name`, `page_number`, `tags`）
• `Query`（问答服务）返回带引用溯源的自然语言答案 + 切片高亮片段 | -| **支持模型** | **不依赖特定大模型**：底层由平台统一记忆引擎处理；提取质量受输入对话结构影响，**无需开发者指定或切换模型** | • 预置模型：Qwen3/Qwen2.5/Qwen2/Long/Max/Turbo/Coder/Deep-Research/VL系列等
• 第三方模型：DeepSeek-R1/Llama3.1/Yi-Large 等（需模型已接入百炼）
• **必须显式指定模型 ID**（如 `qwen3.6-plus`）用于问答生成 | -| **API 端点** | • 统一 REST 基础路径：`https://dashscope.aliyuncs.com/api/v2/apps/memory/`
• 主要接口：`/add`, `/memory_nodes/search`, `/user_profile/get`, `/list`, `/delete` | • 分阶段 API：`/file/upload`, `/index/create`, `/index/job/submit`, `/retrieve`, `/query`
• 上层服务端点：`/knowledge-retrieval`, `/knowledge-qa`（支持多知识库联合） | -| **计费方式** | • **按调用次数计费**（QPM 限流严格）：
 - `AddMemory`: ≤120 QPM
 - `SearchMemory`: ≤300 QPM
 - 全账号总计 ≤3000 QPM
• 无存储容量费、无模型推理费 | • **双维度计费**（自2026年1月4日起）：
 - **规格费**：按知识库版本（标准版/旗舰版）按小时计费
 - **模型调用费**：按 [Token](../concepts/token.md) 计费（含 Rerank 排序 [Token](../concepts/token.md) + 问答生成 Token）
 - **关键成本因子**：`TopK` 值直接影响 Rerank Token 消耗（非仅最终返回数） | -| **典型场景** | • 智能客服中持续记录用户偏好（“不喜电话回访”“常用支付方式为支付宝”）
• 个人助理中管理日程提醒、健康习惯、购物清单等动态事项
• 教育 Agent 中跟踪学生错题类型、薄弱知识点、学习节奏变化 | • 企业内部知识问答（制度文档/产品手册/项目报告）
• 客服知识库（FAQ/工单案例/解决方案库）
• 多媒体内容检索（PPT 图文问答、会议录像字幕搜索）
• 行业垂类问答（医疗指南、法律条文、金融产品说明） | -| **数据生命周期** | • **无自动过期**（默认规则可配 7/30/180 天或永不过期）
• 全生命周期由开发者主动管理（`DeleteMemory`/`UpdateMemory`）
• `UpdateMemory` 对 `custom_content` 为全量覆盖 | • 文件上传即索引，**内容不可变**（修改需重新上传）
• 元数据抽取规则创建后不可修改
• 删除文件将触发异步索引重建 | -| **地域支持** | 全地域可用（与 DashScope API 一致） | **仅限华北2（北京）地域**（控制台与 API 均受限制） | -| **开发者控制粒度** | • 高：可精确控制每条记忆的 `user_id`/`memory_library_id`/`project_id`/`profile_schema`
• 支持细粒度分页（`page_num/page_size`）、相似度阈值（`min_score`）、召回数（`top_k`） | • 中高：可配置切片策略（推荐“智能切分”）、相似度阈值（0.01–1.0）、召回数（`max_retrieve_count`≤20）、标签过滤
• **元数据规则不可变**，需创建时一次性设定 | - -## 各方案的适用场景建议 - -### ✅ 选择「长期记忆（新）」当： -- 你的应用核心是**服务特定用户个体**，需在多次独立会话中保持对其偏好、状态、承诺事项的记忆（如：“上次说好下周三跟进合同”）； -- 数据天然以**对话形式产生**，且需从自然语言中自动提炼结构化事实（如从“我妈妈生日是1970年5月12日”中提取 `{"relation": "mother", "birthday": "1970-05-12"}`）； -- 你需要**轻量、低延迟、高频率**的增删查操作（如每轮对话后写入1条+每轮开始前检索3–5条），且不愿承担 Rerank 或大模型生成费用； -- 你使用 OpenClaw 等框架，希望**零代码接入自动捕获/自动召回**[插件](../concepts/plugin.md)，实现“记忆无感流转”。 - -### ✅ 选择「知识库（RAG）」当： -- 你的知识源是**静态、批量、多格式的文档集合**（如1000份PDF产品说明书），需为整个应用而非单个用户注入领域知识； -- 业务要求**强准确性、可溯源、防幻觉**，需通过引用原文片段佐证答案（如客服回复必须标注“依据《售后服务政策V3.2》第5.1条”）； -- 你需要处理**图片、表格、音视频等非纯文本数据**，并支持图文混合检索或视觉理解； -- 你已有成熟的数据管道（如每日同步CRM数据到OSS），需通过 API 自动化完成**文件上传→索引构建→服务发布**全流程； -- 你愿意为更高精度和更丰富能力（如多知识库联合检索、Query 改写、拒答控制）承担相应规格费与模型 Token 成本。 - -### ⚠️ 注意：二者可协同，非互斥 -- **典型协同模式**： - `用户当前会话 → 长期记忆召回其历史偏好（如“用户禁用语音播报”） → 知识库检索最新产品文档 → 大模型生成答案时，结合记忆约束（禁用语音）+ 知识依据（文档条款）生成合规响应` -- 技术实现：在 Agent 的 `tool_call` 或工作流中，**并行调用 `SearchMemory` 和 `KnowledgeRetrieval`**，将结果统一注入 LLM 提示词。 - -## 面向开发者的选型参考 - -| 你的需求 | 推荐方案 | 关键理由 | -|----------|-----------|-----------| -| “我要做一个健身教练Bot，记住每个用户的运动目标、受伤史、每周训练反馈，并据此调整计划” | ✅ 长期记忆（新） | 用户专属、动态更新、对话驱动、低成本高频操作 | -| “我要搭建公司内部IT Helpdesk，让员工能问‘如何重置VPN密码’并返回准确步骤截图” | ✅ 知识库（RAG） | 多模态（图文）、静态知识、需引用溯源、强准确性要求 | -| “我的电商客服Bot既要懂商品参数（知识库），又要记用户本次投诉诉求（长期记忆）” | ✅ 两者协同 | 知识库提供通用产品信息，长期记忆保存本次会话的订单号、情绪标签、承诺时效 | -| “我需要在新加坡地域部署应用，且必须使用私有知识” | ❌ 知识库（不可用）→ ✅ 长期记忆（新） | 知识库地域限制为硬性约束，长期记忆无此限制 | -| “我每天新增10万条用户行为日志，需实时写入并支持语义搜索” | ⚠️ 谨慎评估 → 建议长期记忆 + 异步批处理 | `AddMemory` QPM 限流120，需拆分为多 `user_id` 并行或采用 `custom_content` 批量聚合；知识库不支持高频流式写入 | -| “我只有3个PDF文件，但要求极低延迟（<300ms）和零模型费用” | ✅ 长期记忆（新） | 可将PDF文本摘要后作为 `custom_content` 写入，用 `SearchMemory` 直接检索，规避Rerank与生成费用 | - -> **最后建议**： -> - **MVP 阶段优先试用长期记忆（新）**：API 简洁、上手快、成本可控，适合验证用户记忆价值； -> - **规模化知识服务必选知识库**：其多模态、高精度、生产级管控（拒答/溯源/审计日志）是长期记忆无法替代的； -> - **始终通过 `DASHSCOPE_API_KEY` 统一认证**，二者共享同一套密钥体系与配额管理，便于权限收敛。 - -## 被对比主题页 - -- [long term memory new](../api/long-term-memory-new.md) -- [knowledge base](../guides/knowledge-base.md) -- [memory library overview](../guides/memory-library-overview.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-and-inference-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-and-inference-comparison.md new file mode 100644 index 00000000..7452d1e8 --- /dev/null +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-and-inference-comparison.md @@ -0,0 +1,69 @@ +# 模型部署与推理方案对比：Model High-Speed Inference vs Model Deployment vs Model Production + +> **目的与背景** +> 百炼平台提供三类面向不同阶段与目标的模型服务化能力：`Model High-Speed Inference`（高时效性推理加速）、`Model Deployment`（生产级模型服务化）和 `Model Production`（端到端模型生命周期管理）。开发者常因命名相似、功能交叉而混淆其定位，导致选型偏差——例如误用快速模式承载核心业务流量，或在未完成微调时提前申请 PTU 预留。本文旨在从技术本质、能力边界与工程约束三个维度进行结构化对比，帮助开发者基于**业务SLA要求、模型成熟度、流量特征与成本模型**做出精准技术选型。 + +--- + +## 关键维度对比表 + +| 维度 | Model High-Speed Inference | Model Deployment | Model Production | +|------|----------------------------|------------------|------------------| +| **核心定位** | **推理加速层优化**：在已有标准API基础上，叠加容量保障或吞吐提速能力，不改变模型本身 | **服务化层抽象**：将指定模型（预置/LoRA）封装为资源独占、可配置、可观测的专属推理服务 | **全生命周期层编排**：覆盖“微调训练 → 模型注册 → 多环境部署 → 版本灰度”的端到端生产流水线 | +| **输入格式** | 与标准 DashScope API 完全一致（`messages` / `prompt` + `parameters`），无需改造请求体 | 同标准 API；PTU/MU 模式支持额外参数（如 `enable_thinking`, `max_context_length`）；Token 计费模式需匹配基础模型协议 | 同标准 API；但部署后 endpoint 必须携带 `deployment_id`，且请求体需符合该部署绑定模型的微调协议（如特定 system [prompt](../guides/prompt.md) 结构） | +| **输出格式** | 标准响应结构；快速模式额外返回 `reasoning_content` 字段（流式场景需分别处理 `delta.reasoning_content` 和 `delta.content`） | 标准响应结构；MU 模式启用 `thinking` 时可能返回 `reasoning_steps`；PTU 模式无结构变化 | 标准响应结构；无新增字段，但语义行为由微调结果决定（如客服模型自动补全工单编号） | +| **支持模型** | • TPM 预留：Qwen、GLM、DeepSeek、Kimi 等主流模型多版本
• 快速模式：仅 `glm-5.2-fast-preview`（Preview 阶段，地域受限） | • 预置模型：Qwen、DeepSeek、GLM、Kimi、CosyVoice 全系列（文本/多模态/语音/Embedding/Rerank）
• 自定义模型：**仅 LoRA 微调模型**（需严格匹配基础模型+rank约束） | • 微调来源：仅支持基于百炼支持的基础模型（如 `qwen2-7b-instruct`）开展监督微调
• 部署来源：微调产出的 `model_id` 或通过 OSS 导入的第三方模型（需满足格式与权限要求） | +| **API 端点** | • TPM 预留：通用 DashScope 域名 `dashscope.aliyuncs.com` + 专属 `model` code（如 `qwen37max-20260520-tpm-xxxx`）
• 快速模式：地域专属 MaaS 域名（如 `{workspace_id}.cn-beijing.maas.aliyuncs.com`） + 固定 model ID | 统一 DashScope 域名 `dashscope.aliyuncs.com` + 部署生成的 `deployed_model` ID（如 `qwen-flash-2025-07-28-ptu-12345`） | 统一 DashScope 域名 `dashscope.aliyuncs.com` + `deployment_id`（如 `dep-abc123xyz`），endpoint 路径含 `/services/...?deployment_id=...` | +| **计费方式** | • TPM 预留：预付费（按 kTPM×时长），支持溢出至按量计费
• 快速模式：按 Token 实际用量计费（缓存命中单价明确标注，无折扣逻辑） | • PTU 模式：预付费（按输入/输出 kTPM×时长），支持阶梯系数与缓存折扣
• MU 模式：后付费（按模型单元规格×小时），支持 PD 分离与 Thinking 模式溢价
• Token 计费（`lora`）：按实际输入/输出 Token 数计费（仅限指定基础模型） | • 微调阶段：按 GPU 小时计费（任务运行时长）
• 部署阶段：按 `instance_type` 规格×运行时长计费（冷启缩容支持，最小实例数可设为 0） | +| **典型场景** | • TPM 预留：金融风控实时决策、电商大促期间搜索问答服务（流量可预估，不可限流）
• 快速模式：AI 编程助手多步代码生成、Agent 执行链中高频子任务（对首 token 延迟敏感） | • PTU：企业知识库问答（长文档摘要+高并发）
• MU：定制化客服机器人（需 Thinking 模式+上下文长度控制）
• Token 计费：低频但高价值垂类任务（如法律合同关键条款提取） | • 垂直领域模型迭代：银行智能投顾话术微调 → staging 环境AB测试 → prod 灰度发布
• 第三方模型集成：将自研小模型通过 OSS 导入 → 注册为 `model_id` → 部署至多可用区 | +| **扩展性与治理** | 无独立扩缩容能力；TPM 预留容量固定，快速模式依赖排队机制缓解突发 | 支持自动化扩缩容（MU 模式）、RPM/TPM 限流、细粒度监控（延迟/P95/错误率） | 支持多环境部署（staging/prod）、版本回滚、流量灰度（按比例/用户标签）、部署健康度巡检 | +| **模型变更成本** | 低：切换 model code 或域名即可；TPM 预留退订后 code 立即失效 | 中：修改部署需重建服务（如从 PTU 切换至 MU 需重新创建），但模型 ID 可复用 | 高：微调任务不可取消；模型升级需新建微调 job → 新建 deployment → 流量迁移；旧 deployment 需手动停用 | + +--- + +## 适用场景建议（面向开发者的技术选型指南） + +| 你的需求 | 推荐方案 | 关键理由 | 注意事项 | +|----------|-----------|-----------|-----------| +| **需要毫秒级首 token 响应，且流量波动剧烈（如编程助手）** | ✅ Model High-Speed Inference（快速模式） | 唯一提供排队机制替代硬限流的能力，TPS 提升 1.5~2 倍，不增加客户端重试复杂度 | • 仅 `glm-5.2-fast-preview` 可用，不支持其他模型
• Preview 阶段无 SLA 承诺，不建议用于支付等强一致性场景 | +| **业务流量稳定可预测，且无法容忍任何限流（如核心交易链路）** | ✅ Model High-Speed Inference（TPM 预留） | 提供刚性容量保障（kTPM），支持输入/输出独立配额，溢出策略可控 | • 缩容/退订产生违约金（已用部分×1.5）
• 首次调用有预热延迟，需客户端实现重试 | +| **需长期稳定运行、支持弹性扩缩容与精细化监控的生产服务** | ✅ Model Deployment | 提供 PTU/MU/Token 三种成熟计费模型，支持长上下文、前缀缓存、Thinking 模式等生产必需特性 | • LoRA 模型导入有严格约束（rank/词表/chat_template）
• PTU 模式创建后不可变更为其他计费方式 | +| **正在构建垂类专属模型，需从训练到上线闭环管理** | ✅ Model Production | 唯一支持微调训练、模型注册、多环境部署、灰度发布的全链路能力 | • 微调数据集必须 JSONL 格式且 ≤100MB
• 同一 model_id 最多 5 个活跃部署，需主动清理旧版本 | +| **已有微调好的 LoRA 模型，希望低成本快速验证效果** | ⚠️ Model Deployment（Token 计费） | 相比 Model Production 的部署阶段计费，Token 计费更轻量（无需购买实例规格），适合低频验证 | • 仅限文档明确列出的基础模型（如 `qwen3-8b`）
• `capacity` 参数在 API 中必填但无效，勿误解为资源配额 | +| **需对接私有化模型或非百炼训练框架产出的模型** | ⚠️ Model Production（模型导入） | 支持通过 OSS 导入 HuggingFace 格式模型，完成注册后即可部署 | • 需主账号授权 OSS 服务关联角色
• 导入模型需自行确保兼容性（如 tokenizer、attention 实现） | + +--- + +## 技术选型决策树（简版） + +```mermaid +graph TD + A[你的模型是否已完成微调？] + A -->|是| B[是否需多环境/灰度/版本回滚？] + A -->|否| C[是否追求极致首 token 延迟？] + B -->|是| D[✅ Model Production] + B -->|否| E[是否需长期稳定服务+弹性扩缩容？] + E -->|是| F[✅ Model Deployment] + E -->|否| G[是否流量可预估且不可限流？] + G -->|是| H[✅ Model High-Speed Inference
TPM 预留] + G -->|否| I[是否仅需短期加速且接受 preview 限制？] + I -->|是| J[✅ Model High-Speed Inference
快速模式] + I -->|否| K[使用标准 DashScope API] + C -->|是| J + C -->|否| L[是否需专属资源保障？] + L -->|是| F + L -->|否| K +``` + +> **最后提醒**： +> - **不要混用**：快速模式（`glm-5.2-fast-preview`）不支持作为 TPM 预留或 Model Deployment 的目标模型； +> - **成本优先级**：若预算敏感且流量低频，优先评估 `Model Deployment` 的 Token 计费模式，而非启动完整微调流程； +> - **演进路径建议**：PoC 验证 → `Model High-Speed Inference` 加速 → `Model Deployment` 稳定服务 → `Model Production` 持续迭代。 + +## 被对比主题页 + +- [model high speed inference](../guides/model-high-speed-inference.md) +- [model deployment 1](../guides/model-deployment-1.md) +- [model production](../api/model-production.md) + + diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md deleted file mode 100644 index 78cbb964..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-deployment-options.md +++ /dev/null @@ -1,57 +0,0 @@ -# [模型部署](../concepts/model-deployment.md)方案对比：Model Deployment 1 vs Model Production - -本对比旨在帮助开发者清晰区分百炼平台中两类核心模型服务化能力——**Model Deployment 1**（面向生产推理的精细化部署）与**Model Production**（面向端到端模型定制与上线的全生命周期管理），避免因概念混淆导致技术选型偏差。二者定位不同：前者聚焦「已确定模型」在高稳定性、低延迟、强可控性要求下的**生产级推理服务交付**；后者侧重「从训练到上线」的闭环，解决「如何把一个新任务适配的模型快速变成可用服务」的问题。本文基于当前平台 v2.4+ 版本（2024年Q3发布）功能边界撰写，所有结论均经文档交叉验证与平台实测逻辑校准。 - -## 关键维度对比 - -| 维度 | Model Deployment 1 | Model Production | -|------|---------------------|-------------------| -| **核心定位** | 生产环境推理服务的**精细化部署与资源治理**（“怎么稳、快、省地跑好一个已知模型”） | 模型定制化与服务化的**端到端流水线**（“怎么把一个任务需求变成一个可调用的模型服务”） | -| **输入格式** | • PTU/MU：标准 Prompt（`messages` 或 `prompt` 字段）
• LoRA：仅支持已导入且通过校验的 LoRA 模型 ID（`model_id`）
• **不接受原始训练数据或模型文件** | • 微调阶段：结构化 JSONL 数据集（含 `messages`/`prompt`+`completion` 字段）
• 部署阶段：`fine_tuned_model_id` 或兼容格式模型 ID（如 GGUF 导出 ID）
• **支持原始训练数据输入与模型资产导入** | -| **输出格式** | 标准化 OpenAI 兼容响应（`choices[0].message.content` + `usage`），含 `x-dashscope-ptu-overflow` 等平台扩展头 | 完全兼容 OpenAI `/v1/chat/completions` 响应格式；微调任务返回含 `fine_tuned_model_id` 的 JSON 对象 | -| **支持模型类型** | • PTU：指定白名单模型（如 `qwen3.7-plus-2026-05-26`, `deepseek-v4-pro`, `glm-5.1`）
• MU：覆盖全部千问系列、GLM、DeepSeek 及千问 VL 模型
• LoRA：**仅限百炼平台内完成 LoRA 微调并成功导入的模型**（rank=8/16/32/64，无 vocab/chat_template 修改） | • 微调：仅支持平台预置基座模型（如 `qwen2.5-7b`, `qwen3-14b`）
• 部署：支持微调产出模型 + **手动导入的 GGUF 格式模型**（ONNX 当前仅支持推理兼容性验证，**不可直接部署**） | -| **API 端点** | 统一部署入口：
`POST /api/v1/deployments`
通过 `plan` 字段区分模式（`"ptu"`/`"mu"`/`"lora"`） | 分离式 API：
• 微调：`POST /api/v1/fine_tuning_jobs`
• 部署：`POST /api/v1/deployments`（独立于 Model Deployment 1 的 endpoint） | -| **计费方式** | • PTU：按预购吞吐量（TPU）时长计费（预付费/后付费）
• MU：按模型单元（MU）规格与时长计费
• LoRA：**严格按实际输入/输出 token 计费**（随用随付） | • 微调：按 GPU 实例运行时长（小时）计费
• 部署：按所选 `instance_type`（如 `gpu-a10`）的实例时长计费
• **无 token 级粒度计费** | -| **典型场景** | • 高并发客服对话系统（需稳定 <300ms P99 延迟）
• 长文档摘要服务（200K token 输入 + 前缀缓存）
• 合规审计场景（需独占算力 + 自定义首 [Token](../concepts/token.md) 延迟 SLA） | • 新业务线冷启动：基于行业语料微调专属问答模型
• 快速验证模型效果：上传小样本 JSONL 进行 1 小时微调 + 部署测试
• 多版本 A/B 测试：为同一基座[模型部署](../concepts/model-deployment.md)不同微调版本（v1/v2） | -| **模型定制深度** | • **不支持训练**
• LoRA 模式仅消费已有微调成果，**不可在此流程中发起微调**
• MU 支持运行时切换思考/非思考模式等推理策略 | • **原生支持监督微调（LoRA）**
• 提供完整微调任务生命周期管理（提交→监控→获取 model_id）
• 支持模型版本追溯（`job_id` → `fine_tuned_model_id` → `version_id`） | -| **扩缩容能力** | • PTU：自动溢出至按量计费（需显式配置策略）
• MU：支持副本数（`capacity`）动态调整
• LoRA：无扩缩容概念（按 token 计费天然弹性） | • 部署实例：支持修改 `instance_type` 重启扩容（非实时，需重建）
• **不支持运行时副本数伸缩**（无 `capacity` 参数） | -| **地域与权限** | 仅支持华北2（北京）地域；需业务空间具备目标模型的**部署权限** | 支持多地域（以控制台实际开通为准）；微调/部署权限独立管控，需分别授权 | - -## 适用场景建议 - -### ✅ 选择 **Model Deployment 1** 当： -- 你已拥有一个**确定的、经过充分验证的模型**（如线上稳定的 `qwen3.7-plus`），需要将其以最高 SLA 要求投入生产； -- 业务对**吞吐稳定性、延迟确定性、成本可预测性**有严苛要求（如金融风控实时决策）； -- 需要利用**长上下文（200K token）、前缀缓存、首 [Token](../concepts/token.md) 延迟保障、PD 分离计算**等高级推理优化能力； -- 团队具备基础设施运维经验，希望精细控制资源规格（如 `MU1` × 4 副本）与限流策略（`tpm_limit`）； -- 成本模型偏好**预付费锁定资源**（PTU）或**独占算力保障**（MU），而非按请求计费。 - -### ✅ 选择 **Model Production** 当： -- 你的目标是**从零开始构建一个领域专用模型**（如医疗报告生成、法律条款解析），尚未有现成模型； -- 需要**快速迭代验证**：上传 JSONL 数据 → 微调 2 小时 → 部署测试 → 收集反馈 → 再微调； -- 业务接受**按实例时长付费**，且更关注模型效果提升而非单次推理成本； -- 需要**多版本协同管理**（例如同时运行 `finetune-job-20240501` 和 `finetune-job-20240615` 的部署实例）； -- 技术栈倾向**声明式工作流**（微调 job → 部署 instance），而非手动配置底层资源参数。 - -> ⚠️ **重要提醒**：二者并非互斥，而是**上下游协作关系**。典型生产路径为： -> **Model Production 微调产出 `fine_tuned_model_id` → 导入 Model Deployment 1 的 LoRA 模式 → 以 token 级精度投入高负载生产**。 -> 若跳过 Model Production 直接使用 Model Deployment 1 的 LoRA 模式，则必须确保模型已在平台内完成微调与校验。 - -## 技术选型参考（致开发者） - -| 你的问题 | 推荐方案 | 关键依据 | -|----------|-----------|-----------| -| “我有一个微调好的 LoRA 模型，想在生产环境按 token 计费提供服务” | **Model Deployment 1（LoRA 模式）** | 唯一支持 token 级计费的部署通道；严格校验模型格式保障稳定性 | -| “我需要把一份客服对话数据集变成专属模型，并在 1 天内部署上线” | **Model Production** | 内置微调 API + 一键部署，端到端最短路径；无需手动处理模型文件 | -| “我的大模型应用峰值 QPS 达 500，要求 P99 延迟 <200ms，且预算固定” | **Model Deployment 1（MU 模式）** | `deploy_spec` + `capacity` 可精确规划算力；`enable_thinking` 等参数保障延迟 SLA | -| “我要为同一基座[模型部署](../concepts/model-deployment.md) 3 个不同微调版本做灰度测试” | **Model Production** | 天然支持 `model_id` + `version_id` 多实例隔离；流量路由由平台统一管理 | -| “我需要处理 150K token 的合同全文分析，且必须复用前缀缓存” | **Model Deployment 1（PTU 模式）** | 明确支持 `glm-5.1` 等模型的 200K 输入与前缀缓存；Model Production 当前不暴露缓存控制接口 | - -请根据**当前阶段的核心诉求**（是“训练新模型”还是“运营成熟模型”）和**关键约束条件**（延迟、成本模型、定制深度）进行决策。如涉及混合场景，建议采用 Model Production 构建模型资产，再通过 Model Deployment 1 实现生产交付——这是百炼平台推荐的最佳实践路径。 - -## 被对比主题页 - -- [model deployment 1](../guides/model-deployment-1.md) -- [model production](../api/model-production.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md deleted file mode 100644 index 05c5c14e..00000000 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/model-evaluation-monitoring.md +++ /dev/null @@ -1,66 +0,0 @@ -# 模型评估与监控能力对比 - -为帮助开发者在模型研发、上线及运维全生命周期中合理选用百炼平台的能力组件，本文对三大核心可观测性能力——**模型评测（Model Evaluation）**、**模型监控（Model Monitoring）** 和 **应用评测（Application Evaluation）** 进行系统性对比分析。三者定位互补： -- **模型评测**聚焦于*离线、定量、多维度的能力归因*，服务于模型选型与调优验证； -- **模型监控**侧重于*在线、实时、指标化的服务健康度观测*，保障生产稳定性与成本可控； -- **应用评测**专用于*端到端智能体/工作流级的质量闭环验证*，覆盖 RAG、Agent 编排等复杂链路的输出可信度评估。 - -本对比基于当前（2024年Q3）百炼平台正式功能，面向技术选型决策者提供客观、可落地的参考依据。 - -## 关键维度对比 - -| 维度 | 模型评测（Model Evaluation） | 模型监控（Model Monitoring） | 应用评测（Application Evaluation） | -|------|------------------------------|------------------------------|------------------------------------| -| **核心目标** | 量化模型文本生成能力（准确性、一致性、安全性等），支持横向对比与归因分析 | 实时观测模型调用行为、性能瓶颈、成本消耗与异常事件，保障服务 SLA | 评估智能体/工作流整体输出质量（含知识检索、逻辑推理、格式合规等），支持链路级问题定位 | -| **输入格式** | `EvaluationSet`（JSONL，含 `prompt` + `completion`）或已含 `output` 的推理结果集（CSV/JSONL） | 无显式“输入数据集”；自动采集所有调用请求（HTTP/SDK）原始参数与上下文（北京地域部分模型支持请求/响应内容） | 自动评测：知识库 → 自动生成 `query`+`referenceAnswer` JSONL；手动/新版：XLS/XLSX（含 `Prompt`/`Completion`/`SessionId`）或自定义结构化表（JSONL/CSV） | -| **输出格式** | 结构化评分报告（综合得分、通过率、分布直方图）、逐条明细（含 LLM 判定理由、规则匹配结果、人工标签） | 多维时间序列指标（Prometheus 格式）、可视化看板（RPM/TPM/延迟/P99/失败率）、告警通知（短信/钉钉/Webhook）、原始日志（北京地域启用后） | 自动评测：总正确率、BadCase 归因分类（如“检索失效”“幻觉”“格式错误”）、调优建议；新版任务：各评估器评分明细、人工标签统计、交叉分析图表 | -| **支持模型类型** | **仅文本生成类模型**（预置 & 调优模型），不支持多模态、语音、向量模型 | **全模态支持**：LLM（qwen-plus/max）、视觉、语音、全模态、向量模型（所有公开及调优模型） | **仅智能体（Agent）与工作流（Workflow）应用**（需已发布）；底层依赖 `qwen-max`/`qwen-plus` 执行 LLM 评估，但不直接评测基础模型 | -| **API / SDK 支持** | ❌ **不提供公开 API/SDK**；仅控制台操作（自动化需通过 PAI Judge Model API 替代） | ✅ **完整 Prometheus API 支持**（HTTP 接口 + Basic Auth）；支持 Grafana 集成、自建系统对接；控制台提供 RESTful 告警管理接口 | ✅ **新版评测任务支持 OpenAPI**（创建/查询/停止任务、获取结果）；旧版自动评测暂无 API；手动评测无 API | -| **计费方式** | - 被评测模型推理费（使用评测数据集时按输入/输出 [Token](../concepts/token.md) 计费）
- 裁判模型费（大模型评估维度，按 [Token](../concepts/token.md) 计费）
- 规则/人工评估：零模型费用 | - **无额外评测费用**；所有监控数据采集免费
- 仅基础调用本身产生模型费用（与是否开启监控无关）
- 日志审计（北京）按存储量计费 | - 自动评测/LLM 评估器：调用 `qwen-max`/`qwen-plus`，按 [Token](../concepts/token.md) 计费
- Code 评估器：无额外调用成本
- 评测集存储：免费（≤20MB/个） | -| **典型场景** | • 新模型上线前能力基线测试（vs C-Eval/GSM8K）
• Prompt 工程效果验证（A/B 测试）
• 微调模型 vs 原始模型能力衰减分析
• 安全合规性人工抽检 | • 生产环境服务稳定性巡检（延迟突增、失败率飙升）
• 成本优化（识别高 Token 消耗调用、限流根因）
• 故障快速定位（结合日志追踪首 Token 延迟、内容安全拦截）
• 多模型路由策略效果验证 | • RAG 知识库更新后问答质量回归测试
• Agent 工作流编排逻辑正确性验证（如“订机票→查天气→发邮件”链路）
• 多应用横向对比（相同知识库下不同 Agent 设计优劣）
• 用户反馈 BadCase 的深度归因分析 | -| **地域限制** | • 自定义评测：全地域支持
• 基线评测（C-Eval 等）：**仅北京地域可用** | • 普通监控：全地域
• 高级监控（分钟级）、告警、日志审计：**仅北京、新加坡支持**；弗吉尼亚支持分钟级监控但不支持告警/日志 | • 自动评测：**仅北京地域支持**（依赖知识库服务）
• 新版评测任务/手动评测：全地域支持（评测集上传与任务执行） | -| **数据时效性** | 离线批处理：任务完成后即时生成结果（耗时取决于数据量与裁判模型负载） | • 普通监控：约 1 小时延迟
• 高级监控：≤5 分钟延迟（指标）；日志存在分钟级延迟需手动刷新 | • 自动评测：任务执行中实时显示进度，完成后即时出结果
• 新版/手动评测：人工标注提交后即时更新统计 | - -## 各方案适用场景建议 - -| 场景描述 | 推荐方案 | 关键理由 | -|----------|----------|----------| -| **新训练的文本生成模型（如 Qwen2-7B-Chat）需验证其数学推理、中文理解能力是否达到 SOTA 水平** | ✅ 模型评测（基线评测） | 直接复用平台预置 C-Eval/GSM8K 数据集，5 分钟内获得标准化分数，支持与历史模型横向对比；无需构建应用或部署服务。 | -| **线上客服对话机器人（基于 qwen-plus 的智能体）近 2 小时失败率从 0.5% 升至 12%，需快速定位是模型超时、知识库未命中还是限流导致** | ✅ 模型监控（高级监控 + 日志审计） | 查看 `model_call_duration_p99`、`model_usage{usage_type="input_tokens"}` 及 `error_code="429"` 指标趋势；在北京地域启用日志审计后，直接查看失败请求的原始 [prompt](../guides/prompt.md) 与 error message。 | -| **完成一轮 RAG 知识库更新后，需系统性验证 200 个典型用户问题的回答质量，并生成“检索失效占比”“答案幻觉率”等归因报告** | ✅ 应用评测（自动评测） | 知识库自动触发评测集生成 → 调用 `qwen-max` 打分 → 输出结构化归因（如“73% BadCase 因切片不完整”），直接指导知识库切分策略优化。 | -| **需对多个自研微调模型（text2sql、摘要生成）进行 A/B 测试，且要求支持 BLEU/ROUGE 规则评估 + LLM 语义评估混合打分** | ✅ 模型评测（自定义评测） | 创建混合维度（如 `SQL准确性-BLEU` + `SQL准确性-LLM评分`），上传统一评测数据集，一次任务输出双维度结果，避免跨工具数据拼接。 | -| **企业需将百炼模型调用指标接入自建运维平台（如 Zabbix），实现与数据库、API 网关指标统一告警** | ✅ 模型监控（Prometheus API） | 通过标准 PromQL 查询 `model_call_count{model="qwen-plus", workspace_id="xxx"}`，与现有监控栈无缝集成，无需定制开发。 | -| **客户成功团队需定期抽检销售助手智能体的回答，由业务专家人工标注“专业度”“合规性”两个维度（5 分制）** | ✅ 应用评测（新版评测任务 + 人工标签） | 创建含 `professional_score`/`compliance_score` 标签的评测集，关联 LLM 评估器（初筛）+ 人工标注流程，结果自动聚合统计，支持导出 Excel 报告。 | - -## 技术选型参考指南（面向开发者） - -- **优先选择模型监控，当您需要：** - ✅ 实时感知服务健康状态（SLA、延迟、错误） - ✅ 追踪成本消耗并优化预算（Token/图像/视频用量下钻） - ✅ 构建自动化运维体系（Prometheus/Grafana/告警联动） - ⚠️ 注意：若需分钟级洞察或日志审计，请确认地域支持（北京/新加坡）。 - -- **优先选择模型评测，当您需要：** - ✅ 对基础模型（非应用层）做能力量化（如“该模型在 BBH 上得分 68.2，比上一版提升 3.1”） - ✅ 验证 Prompt/LoRA 微调效果（控制变量法，固定数据集与裁判模型） - ✅ 满足合规审计要求（生成可追溯、可复现的评分报告） - ⚠️ 注意：仅支持文本生成模型；基线评测不可下载结果，建议关键任务使用自定义评测。 - -- **优先选择应用评测，当您需要：** - ✅ 评估端到端智能体/工作流输出（而非单个模型） - ✅ 深度归因链路问题（如 RAG 中“检索→重排→生成”各环节贡献度） - ✅ 混合自动化与人工评审（LLM 初筛 + 专家终审） - ⚠️ 注意：自动评测强依赖知识库与北京地域；新版评测任务更灵活，推荐新项目首选。 - -> **组合使用建议**： -> - **研发阶段**：用 *模型评测* 验证基础模型能力 → 用 *应用评测* 验证智能体封装效果 → 上线后用 *模型监控* 持续守护。 -> - **故障排查**：先看 *模型监控* 发现异常指标（如 `first_token_duration` 突增）→ 再用 *应用评测* 复现 BadCase 并归因 → 最后用 *模型评测* 隔离是否为模型自身退化。 -> - **成本治理**：通过 *模型监控* 识别高消耗调用 → 提取样本用 *模型评测* 分析低分原因（如 Prompt 过长导致输出冗余）→ 优化后再次 *模型评测* 验证改进效果。 - -## 被对比主题页 - -- [model evaluation introduction](../guides/model-evaluation-introduction.md) -- [model monitoring](../guides/model-monitoring.md) -- [application evaluation](../guides/application-evaluation.md) - - diff --git a/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md index bd94ff13..5321d426 100644 --- a/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md +++ b/skills/bailian-docs-llm-wiki/wiki/comparisons/realtime-api-comparison.md @@ -1,70 +1,63 @@ -# 实时 API 方案对比：Omni Realtime API vs Realtime API User Guide +# 实时 API 对比：Realtime API vs Omni Realtime API -## 对比目的与背景 +本文旨在帮助开发者清晰理解百炼平台提供的两类核心实时交互能力——**Realtime API** 与 **Omni Realtime API** 的技术定位、能力边界与适用场景，避免因协议选型不当导致集成成本上升、功能缺失或性能不达预期。二者虽均面向“实时 AI 交互”，但在架构设计、协议栈、模型支持、控制粒度及工程落地路径上存在系统性差异。本对比基于最新文档（2024 Q3）整理，适用于新项目技术选型与存量系统升级评估。 -为帮助开发者在百炼平台快速、准确地选择适合业务需求的实时交互方案，本文对两类核心实时能力接口进行系统性对比分析： -- **Omni Realtime API**（`api/omni-realtime-api.md`）：面向端到端[多模态](../concepts/multi-modal.md)智能体的**一体化、开箱即用型实时对话接口**，聚焦“语音/音视频输入 → 语义理解 → 工具调用/联网搜索 → 文本+音频输出”的全链路闭环。 -- **Realtime API User Guide**（`api/realtime-api-user-guide.md`）：面向工程集成的**协议级实时通信框架指南**，定义 WebSocket / WebRTC / AOQ 三种传输协议的能力边界、接入范式与模型兼容矩阵，强调**跨终端、弱网鲁棒性与协议可选性**。 +## 关键维度对比 -二者并非互斥替代关系，而是**抽象层级不同、定位互补的技术方案**：Omni Realtime API 是构建于 Realtime API 协议栈之上的高阶封装；而 Realtime API User Guide 是底层协议能力的统一说明文档。本对比旨在厘清技术边界，避免因概念混淆导致选型偏差。 - ---- - -## 关键维度对比表 - -| 维度 | Omni Realtime API | Realtime API User Guide | -|------|-------------------|--------------------------| -| **本质定位** | 面向场景的**高阶 SDK 封装接口**（Python/Java SDK 主导），提供预编排的[多模态](../concepts/multi-modal.md)对话流水线 | 面向架构的**协议能力说明书**，定义 WebSocket / WebRTC / AOQ 三类传输层标准及模型支持矩阵 | -| **输入格式** | 支持 `append_audio`（Base64 PCM）、`append_video`（H.264 编码帧或原始 I420/NV12 帧）；VAD 模式下自动分段 | 协议相关：
• WebSocket：Base64 PCM 音频 + 可选视频帧
• WebRTC：MediaStream 或 Raw Video Frame（I420/BGRA）
• AOQ：支持外部注入原始音频帧（PCM/I2S）或编码帧（AAC/H.264） | -| **输出格式** | 固定流式事件结构：`response.text.delta`、`response.audio.delta`、`response.function_call_arguments.*` 等；支持 `TEXT` + `AUDIO` 同步输出 | 协议相关：
• WebSocket：JSON 事件流（含 `text`/`audio` 字段）
• WebRTC：DataChannel 传输文本 + AudioTrack 输出合成语音
• AOQ：混合通道（`text` via DataChannel, `audio` via AudioTrack, `video` via VideoTrack） | -| **支持模型** | 仅限 `qwen3.5-omni-*` 系列实时模型（如 `qwen3.5-omni-realtime`, `qwen3.5-omni-flash-realtime` 等），且功能严格按模型版本隔离 | 覆盖更广：
• 全模态模型（`qwen3.5-omni-*`, `qwen3.5-livetranslate-flash-realtime`）→ 三协议均支持
• Fun-ASR / CosyVoice / `qwen-audio-3.0-realtime-plus` → **仅 WebSocket 支持**
• `multimodal-dialog` 套件 → **仅 WebSocket/WebRTC 支持，不支持 AOQ** | -| **API 端点** | 固定 WebSocket 地址：
`wss://{WorkspaceId}.cn-beijing.maas.aliyuncs.com/api-ws/v1/realtime`（地域专属域名） | 协议差异化：
• WebSocket：同上，但模型可通过 URL Query（`?model=xxx`）或消息体指定
• WebRTC：`POST /v1/realtime/webrtc/offer` 获取 SDP，建连后通过 DataChannel 通信
• AOQ：需先调用 `POST /v1/realtime/aoq/allocate` 获取 `sid` 和 `aoqTokenForClient`，再连接 AOQ 服务节点 | -| **计费方式** | 按**实际消耗的 token 数量 + 音频处理时长（秒）** 计费（含 ASR/TTS/LLM 推理），模型不同单价不同；`qwen-omni-turbo-realtime` 按会话时长阶梯计费 | **统一按模型调用粒度计费**，与所选协议无关；但 AOQ/WebRTC 的媒体传输带宽、信令调用等基础资源不额外计费（计入百炼平台基础配额） | -| **典型场景** | 智能客服坐席助手、AI 会议纪要员、语音驱动的虚拟数字人（需工具调用/联网搜索/声音复刻） |
  • **WebSocket**：后台语音质检、IVR 系统集成、快速 PoC 验证
  • **WebRTC**：浏览器端在线教育互动白板、远程医疗问诊、Web 端虚拟主播
  • **AOQ**：移动端音视频社交 App、车载语音助手、鸿蒙设备本地化 AI 交互
| -| **VAD 能力** | 提供 `server_vad`（服务端静音检测）和 `semantic_vad`（语义级说话人意图识别）；后者**仅 `qwen3.5-omni-realtime` 支持** | `semantic_vad` 在三协议中均可用（需模型支持），但 `turn_detection.type` 参数需在 `session.update` 中显式设置；`server_vad` 为默认回退选项 | -| **开发者控制粒度** | **低控制粒度**：SDK 自动管理连接、会话生命周期、媒体缓冲区提交、响应流解析；手动模式需显式 `commit()`，但仍受限于 Omni 协议语义 | **高控制粒度**：WebRTC/AOQ 允许完全接管媒体采集、编码、网络传输（如 AOQ 支持 `isExternal=true` 注入自定义音频帧、WebRTC 支持 `RTCPeerConnection` 级配置） | -| **SDK 支持** | 官方提供 Python / Java SDK，封装连接、会话、事件回调全流程；无 JS SDK | 提供：
• WebSocket：DashScope Python/Java SDK
• WebRTC：TypeScript SDK（基于 Web API）
• AOQ：Android/iOS/HarmonyOS 原生 SDK（含 C++ 底层接口） | - ---- +| 维度 | Realtime API | Omni Realtime API | +|------|--------------|-------------------| +| **协议栈与传输层** | 提供 **三套并行协议栈**：WebSocket、WebRTC、AOQ（AI over QUIC），可按终端/网络/功能需求灵活选择 | **仅基于 WebSocket**（`wss://.../api-ws/v1/realtime`），事件驱动、JSON-RPC 风格，无原生音视频信令能力 | +| **输入格式** | • WebSocket：支持文本、PCM 音频（16 kHz）、Base64 图像（分通道）
• WebRTC/AOQ：支持音视频流 + 数据通道混合传输（如 `oai-events` DataChannel） | • **统一 JSON 事件流**：
– `input_audio_buffer.append`（PCM，16 kHz）
– `input_image.append`（JPG/JPEG，≤256 KB Base64）
– 文本通过 `instructions` 或 ASR 转录隐式注入
• **不支持原始音视频流直传** | +| **输出格式** | • WebSocket：纯文本 + 分离音频流（需客户端合成）
• WebRTC/AOQ：原生音视频流 + 结构化事件（如 `session.updated`、`response.audio.delta`） | • **同步双模态输出**：服务端直接返回 `` + `